Minimizing particles in water is a key goal for improving drinking water quality and safety.The media filtration process,as the last step of the solid–liquid separation process,is largely influenced by the characteri...Minimizing particles in water is a key goal for improving drinking water quality and safety.The media filtration process,as the last step of the solid–liquid separation process,is largely influenced by the characteristics of flocs,which are formed and controlled within the coagulation process.In a laboratory-based study,the impacts of the physical characteristics of flocs formed using aluminum sulfate on the filtration treatment of two comparative water samples were investigated using a photometric dispersion analyzer and a filterability apparatus.In general,the optimum dosage for maximizing filterability was higher than that for minimizing turbidity under neutral p H conditions.For a monomeric aluminum-based coagulant,the charge neutralization mechanism produced better floc characteristics,including floc growth speed and size,than the sweep flocculation mechanism.In addition,the charge neutralization mechanism showed better performance compared to sweep flocculation in terms of DOC removal and floc filterability improvement for both waters,and showed superiority in turbidity removal only when the raw water had high turbidity.For the different mechanisms,the ways that floc characteristics impacted on floc filterability also differed.The low variation in floc size distribution obtained under the charge neutralization mechanism resulted in the flocs being amenable to removal by filtration processes.For the sweep flocculation mechanism,increasing the floc size improved the settling ability of flocs,resulting in higher filter efficiency.展开更多
The effectiveness of deep-bed filtration with respect to suspension formed during the preceding processes is evaluated by the test of filterability. The properties and concentration of the suspension being filtered ar...The effectiveness of deep-bed filtration with respect to suspension formed during the preceding processes is evaluated by the test of filterability. The properties and concentration of the suspension being filtered are affected by the efficiency of the preceding aggregation and separation processes. The basic principles of the test of filterability are based on the mechanistic model of filtration. The equations in the mathematical model of the mechanistic conception of filtration are derived from the theory of filtration. The arrangement of the pilot filtration plant for the determination of filterability of flocculent suspension is presented in this paper. The test of filterability is carried out with a thin-layer filter element. The design of a filter element arrangement and its installation are also disclosed in this paper. The inter-dependence of the coefficient of filtration efficiency on the specific volume of intercepted suspensions, filter media grain sizes and different filtration velocities are graphically presented. In addition, the effect of the filter bed clogging resulting from the properties of different suspensions on the head loss generated, the length of filtration cycle and quality of filtrate produced are also shown in this paper.展开更多
The Savitzky-Golay(SG)filter,which employs polynomial least-squares approximations to smooth data and estimate derivatives,is widely used for processing noisy data.However,noise suppression by the SG filter is recogni...The Savitzky-Golay(SG)filter,which employs polynomial least-squares approximations to smooth data and estimate derivatives,is widely used for processing noisy data.However,noise suppression by the SG filter is recognized to be limited at data boundaries and high frequencies,which can significantly reduce the signal-to-noise ratio(SNR).To solve this problem,a novel method synergistically integrating Principal Component Analysis(PCA)with SG filtering is proposed in this paper.This approach avoids the is-sue of excessive smoothing associated with larger window sizes.The proposed PCA-SG filtering algorithm was applied to a CO gas sensing system based on Cavity Ring-Down Spectroscopy(CRDS).The perform-ance of the PCA-SG filtering algorithm is demonstrated through comparison with Moving Average Filtering(MAF),Wavelet Transformation(WT),Kalman Filtering(KF),and the SG filter.The results demonstrate that the proposed algorithm exhibits superior noise reduction capabilities compared to the other algorithms evaluated.The SNR of the ring-down signal was improved from 11.8612 dB to 29.0913 dB,and the stand-ard deviation of the extracted ring-down time constant was reduced from 0.037μs to 0.018μs.These results confirm that the proposed PCA-SG filtering algorithm effectively improves the smoothness of the ring-down curve data,demonstrating its feasibility.展开更多
1 Introduction Recently,the increasing demand for advanced telecommunication systems has spurred extensive research into bandpass filters(BPFs),with particular emphasis on miniaturization,reduction of insertion loss(I...1 Introduction Recently,the increasing demand for advanced telecommunication systems has spurred extensive research into bandpass filters(BPFs),with particular emphasis on miniaturization,reduction of insertion loss(IL),and enhancement of upper stopband rejection(Huang et al.,2021;Snyder et al.,2021;Lin et al.,2023;Zeng et al.,2023).展开更多
Accurate estimation of the State of Charge(SOC),State of Health(SOH),and Terminal Resistance(TR)is crucial for the effective operation of Battery Management Systems(BMS)in lithium-ion batteries.This study conducts a c...Accurate estimation of the State of Charge(SOC),State of Health(SOH),and Terminal Resistance(TR)is crucial for the effective operation of Battery Management Systems(BMS)in lithium-ion batteries.This study conducts a comprehensive comparative analysis of four Kalman filter variants Extended Kalman Filter(EKF),Extended Kalman-Bucy Filter(EKBF),Unscented Kalman Filter(UKF),and Unscented Kalman-Bucy Filter(UKBF)under varying battery parameter conditions.These include temperature fluctuation,self-discharge,current direction,cell capacity,process noise,and measurement noise.Our findings reveal significant variations in the performance of SOC and SOH predictions across filters,emphasizing that UKF demonstrates superior robustness to noise,while EKF performs better under accurate system dynamics.The study underscores the need for adaptive filtering strategies that can dynamically adjust to evolving battery parameters,thereby enhancing BMS reliability and extending battery lifespan.展开更多
In this study,a fifth-degree cubature particle filter(5CPF)is proposed to address the limited estimation accuracy in traditional particle filter algorithms for bearings-only tracking(BOT).This algorithm calculates the...In this study,a fifth-degree cubature particle filter(5CPF)is proposed to address the limited estimation accuracy in traditional particle filter algorithms for bearings-only tracking(BOT).This algorithm calculates the recommended density function by introducing a fifth-degree cubature Kalman filter algorithm to guide particle sampling,which effectively alleviates the problem of particle degradation and significantly improves the estimation accuracy of the filter.However,the 5CPF algorithm exhibits high computational complexity,particularly in scenarios with a large number of particles.Therefore,we propose the extended Kalman filter(EKF)-5CPF algorithm,which employs an EKF to replace the time update step for each particle in the 5CPF.This enhances the algorithm’s real-time capability while maintaining the high precision advantage of the 5CPF algorithm.In addition,we construct bearing-only dual-station and single-motion station target tracking systems,and the filtering performances of 5CPF and EKF-5CPF algorithms under different conditions are analyzed.The results show that both the 5CPF algorithm and EKF-5CPF have strong robustness and can adapt to different noise environments.Furthermore,both algorithms significantly outperform traditional nonlinear filtering algorithms in terms of convergence speed,tracking accuracy,and overall stability.展开更多
Color filters are essential components for optical modulation.However,conventional filters are restricted to operating exclusively in either reflective or transmissive mode.Furthermore,they suffer from limited UV and ...Color filters are essential components for optical modulation.However,conventional filters are restricted to operating exclusively in either reflective or transmissive mode.Furthermore,they suffer from limited UV and thermal stability,low color purity,and exhibit identical coloration on both surfaces.Herein,we propose a novel design strategy for trans-reflective color filters by integrating the absorptive properties of dye-doped polysulfone(PSU)with the diffractive capabilities of photonic crystals.This composite filter achieved broad-spectrum transmission with deep color outputs—yellow(0.410,0.510),magenta(0.446,0.231),and cyan(0.201,0.425)—closely aligned with standard color space coordinates.By tuning the refractive index of CeO_(2)@SiO_(2)nanoparticles to match dye-based PSU matrix,the transmittance of filters exceeded 70%.Moreover,dye-mediated absorption reduces the scattering light,thereby enhancing reflection color purity(full width at half maxima(FWHM)=25 nm)and producing vibrant blue,green,and red hues.The incorporation of UV-absorbing CeO_(2)@SiO_(2)nanoparticles effectively mitigated dye photodegradation,yielding exceptional UV stability(ΔT<2%under prolonged UV exposure).The filters also exhibited outstanding thermal stability(ΔT<1%after 30 min heat treatment at 230°C).This work establishes a robust materials design framework for multifunctional optical filters,advancing the development of highfidelity dual-mode color systems for next-generation display technologies.展开更多
Understanding fish movement trajectories in aquaculture is essential for practical applications,such as disease warning,feeding optimization,and breeding management.These trajectories reveal key information about the ...Understanding fish movement trajectories in aquaculture is essential for practical applications,such as disease warning,feeding optimization,and breeding management.These trajectories reveal key information about the fish’s behavior,health,and environmental adaptability.However,when multi-object tracking(MOT)algorithms are applied to the high-density aquaculture environment,occlusion and overlapping among fish may result in missed detections,false detections,and identity switching problems,which limit the tracking accuracy.To address these issues,this paper proposes FishTracker,a MOT algorithm,by utilizing a Tracking-by-Detection framework.First,the neck part of the YOLOv8 model is enhanced by introducing a Multi-Scale Dilated Attention(MSDA)module to improve object localization and classification confidence.Second,an Adaptive Kalman Filter(AKF)is employed in the tracking phase to dynamically adjust motion prediction parameters,thereby overcoming target adhesion and nonlinear motion in complex scenarios.Experimental results show that FishTracker achieves a multi-object tracking accuracy(MOTA)of 93.22% and 87.24% in bright and dark illumination conditions,respectively.Further validation in a real aquaculture scenario reveal that FishTracker achieves aMOTA of 76.70%,which is 5.34% higher than the baselinemodel.The higher order tracking accuracy(HOTA)reaches 50.5%,which is 3.4% higher than the benchmark.In conclusion,FishTracker can provide reliable technical support for accurate tracking and behavioral analysis of high-density fish populations.展开更多
Recommendation systems are an integral and indispensable part of every digital platform,as they can suggest content or items to users based on their respective needs.Collaborative filtering is a technique often used i...Recommendation systems are an integral and indispensable part of every digital platform,as they can suggest content or items to users based on their respective needs.Collaborative filtering is a technique often used in various studies,which produces recommendations by analyzing similarities between users and items based on their behavior.Although often used,traditional collaborative filtering techniques still face the main challenge of sparsity.Sparsity problems occur when the data in the system is sparse,meaning that only a portion of users provide feedback on some items,resulting in inaccurate recommendations generated by the system.To overcome this problem,we developed aHybrid Collaborative Filtering model based onMatrix Factorization andGradient Boosting(HCF-MFGB),a new hybrid approach.Our proposed model integrates SVD++,the XGBoost ensemble learning algorithm,and utilizes user demographic data and meta items.We utilize information,both explicitly and implicitly,to learn user preference patterns using SVD++.The XGBoost algorithm is used to create hundreds of decision trees incrementally,thereby improving model accuracy.Meanwhile,user demographic and meta-item data are clustered using the K-Means Clustering algorithm to capture similarities in user and item characteristics.This combination is designed to improve rating prediction accuracy by reducing reliance on minimal explicit rating data,while addressing sparsity issues in movie recommendation systems.The results of experiments on the MovieLens 100K,MovieLens 1M,and CiaoDVD datasets show significant improvements,outperforming various other baselinemodels in terms of RMSE and MAE.On theMovieLens 100K dataset,the HCF-MFGB model obtained an RMSE value of 0.853 and an MAE value of 0.674.On theMovieLens 1M dataset,the HCF-MFGB model obtained an RMSE value of 0.763 and an MAE value of 0.61.On the CiaoDCD dataset,the HCF-MFGB model achieved an RMSE value of 0.718 and an MAE value of 0.495.These results confirm a significant improvement in movie recommendation accuracy with the proposed approach.展开更多
A trace analytical method based on solid-phase extraction gas chromatography-tandem mass spectrometry(SPE–GC–MS/MS)was developed for the rapid detection of 256 semi-volatile organic compounds(SVOCs),including 25 pol...A trace analytical method based on solid-phase extraction gas chromatography-tandem mass spectrometry(SPE–GC–MS/MS)was developed for the rapid detection of 256 semi-volatile organic compounds(SVOCs),including 25 polycyclic aromatic hydrocarbons(PAHs),70 polychlorinated biphenyls(PCBs),123 pesticides,20 phthalate esters(PAEs),4 organophosphate esters(OPEs),9 synthetic musks(SMs),and 5 UV filters(UVs)in water.No-tably,this method provided a decent linearity of calibration standards(R^(2)>0.999),excellent method limits of quantification(MLOQs)(0.12–11.41 ng/L),satisfactory matrix spiking recovery rates(60.4%–126%),and high precision(intra-day relative standard deviations(RSDs):1.0%–10.0%,inter-day RSDs:3.0%–15.0%,and inter-week RSDs:3.4%–15.7%),making it suitable for trace-level studies.Statistical analysis revealed that SVOCs with higher volatility exhibited enhanced recovery rates.Validation of the methodology involved analyzing SVOCs in real spring water and river water samples.Twenty-seven SVOCs were detected in spring water and 58 in river water,with an average concentration of 631.73 and 16,095 ng/L,respectively.Among the detected SVOCs,PAEs constituted the predominant proportion.This study underscored the presence of SVOCs contamination specifi-cally within the spring water,although SVOCs concentrations in river water were significantly greater than those found in spring water.In summary,this sensitive method based on SPE–GC–MS/MS was successfully developed and validated for the rapid analysis of a diverse array of 256 SVOCs at trace levels in water,including not only the traditional highly valued PAHs,PCBs,pesticides,and PAEs,but also the emerging OPEs,UVs,and SMs.展开更多
Pull-ups are a very common fitness exercise that can be seen in many gyms.For athletes,it is very important to perform pull-ups correctly and scientifically.The pull-up scoring method designed in this paper can score ...Pull-ups are a very common fitness exercise that can be seen in many gyms.For athletes,it is very important to perform pull-ups correctly and scientifically.The pull-up scoring method designed in this paper can score the quality of pull-up movement scientifically and objectively,and provide guidance to help athletes better complete the pull-up movement.In this method,the OpenPose algorithm is used to identify the coordinates of skeleton points,and then the coordinate data are processed by a Kalman filter to obtain coordinates closer to the true values.Finally,the filtered data are input into the scoring algorithm designed based on the fuzzy comprehensive evaluation algorithm,and the results of the pull-up quality score and the corresponding guidance are obtained.展开更多
This paper investigates the consensus tracking control problem for high order nonlinear multi-agent systems subject to non-affine faults,partial measurable states,uncertain control coefficients,and unknown external di...This paper investigates the consensus tracking control problem for high order nonlinear multi-agent systems subject to non-affine faults,partial measurable states,uncertain control coefficients,and unknown external disturbances.Under the directed topology conditions,an observer-based finite-time control strategy based on adaptive backstepping and is proposed,in which a neural network-based state observer is employed to approximate the unmeasurable system state variables.To address the complexity explosion problem associated with the backstepping method,a finite-time command filter is incorporated,with error compensation signals designed to mitigate the filter-induced errors.Additionally,the Butterworth low-pass filter is introduced to avoid the algebraic ring problem in the design of the controller.The finite-time stability of the closed-loop system is rigorously analyzed with the finite-time Lyapunov stability criterion,validating that all closed-loop signals of the system remain bounded within a finite time.Finally,the effectiveness of the proposed control strategy is verified through a simulation example.展开更多
This article presents an adaptive intelligent control strategy applied to a lumped-parameter evaporator model,i.e.,a simplified dynamic representation treating the evaporator as a single thermal node with uniform temp...This article presents an adaptive intelligent control strategy applied to a lumped-parameter evaporator model,i.e.,a simplified dynamic representation treating the evaporator as a single thermal node with uniform temperature distribution,suitable for control design due to its balance between physical fidelity and computational simplicity.The controller uses a wavelet-based neural proportional,integral,derivative(PID)controller with IIR filtering(infinite impulse response).The dynamic model captures the essential heat and mass transfer phenomena through a nonlinear energy balance,where the cooling capacity“Qevap”is expressed as a non-linear function of the compressor frequency and the temperature difference,specifically,Q_(evap)=k_(1)u(T_(in)−T_(e))with u as compressor frequency,Te evaporator temperature,and Tin inlet fluid temperature.The operating conditions of the system,in general terms,focus on the following variables,the overall thermal capacity is 1000 J/K,typical for small-capacity heat exchangers,The mass flow is 0.05 kg/s,typical for secondary liquid cooling circuits,the overall loss coefficient of 50 W/K that corresponds to small evaporators with partial insulation,the temperatures(inlet)of 10℃and the temperature of environment of 25℃,thermal load of 200 W that corresponds to a small-scaled air conditioning applications.To handle system nonlinearities and improve control performance,aMorlet wavelet-based neural network(Wavenet)is used to dynamically adjust the PID gains online.An IIR filter is incorporated to smooth the adaptive gains,improving stability and reducing oscillations.In contrast to prior wavelet-or neural-adaptive PID controllers in HVAC applications,which typically adjust gains without explicit filtering or not tailored to evaporator dynamics,this work introduces the first PID–Wavenet scheme augmented with an IIR-based stabilization layer,specifically designed to address the combined challenges of nonlinear evaporator behavior,gain oscillation,and real-time implementability.The proposed controller(PID-Wavenet+IIR)is implemented and validated inMATLAB/Simulink,demonstrating superior performance compared to a conventional PID tuned using Simulink’s auto-tuning function.Key results include a reduction in settling time from 13.3 to 8.2 s,a reduction in overshoot from 3.5%to 0.8%,a reduction in steady-state error from 0.12℃ to 0.02℃and a 13%reduction in energy overall consumption.The controller also exhibits greater robustness and adaptability under varying thermal loads.This explicit integration of wavelet-driven adaptation with IIR-filtered gain shaping constitutes the main methodological contribution and novelty of the work.These findings validate the effectiveness of the wavelet-based adaptive approach for advanced thermal management in refrigeration and HVAC systems,with potential applications in controlling variable-speed compressors,liquid chillers,and compact cooling units.展开更多
Using a recognition model of atmospheric gravity waves(AGWs),we identified 519 AGW events from the OH airglow images observed at the Dandong and Lhasa stations from 2015 to 2017.The 317 AGW events detected at the Dand...Using a recognition model of atmospheric gravity waves(AGWs),we identified 519 AGW events from the OH airglow images observed at the Dandong and Lhasa stations from 2015 to 2017.The 317 AGW events detected at the Dandong station have wavelengths ranging from 30 to 60 km,periods from 14 to 20 min,horizontal speeds from 30 to 60 m/s,and relative intensities from 0.4%to 0.6%,respectively.The parameters of 202 events recorded at the Lhasa station mainly vary within 15-35 km in horizontal wavelength,4-6 min in period,40-100 m/s in horizontal velocity,and 0.1%-0.3%in relative intensity.The occurrence rate peaks in winter and summer at Dandong and the peak in summer are absent at Lhasa because of the lack of convective weather.The seasonal propagation directions of the waves are influenced by both the wind field-filtering effect and the distribution of wave sources.In spring,because of the southeastward background wind field,fewer southeastward events are observed at the Dandong station.The situation at the Lhasa station is similar.In summer,both the Lhasa and Dandong stations are dominated by northeastward AGWs,which can be attributed to the southwestward wind.In autumn,ray-tracing results show that the events at Dandong mainly originate from wind shear,whereas the events at the Lhasa station are triggered by convective weather.The location of the wave sources determines the trend of the propagation directions at the Dandong and Lhasa stations in autumn.In winter,because of the eastward wind,more events are propagating to the southwest at the Dandong station.展开更多
Recommendation systems have become indispensable for providing tailored suggestions and capturing evolving user preferences based on interaction histories.The collaborative filtering(CF)model,which depends exclusively...Recommendation systems have become indispensable for providing tailored suggestions and capturing evolving user preferences based on interaction histories.The collaborative filtering(CF)model,which depends exclusively on user-item interactions,commonly encounters challenges,including the cold-start problem and an inability to effectively capture the sequential and temporal characteristics of user behavior.This paper introduces a personalized recommendation system that combines deep learning techniques with Bayesian Personalized Ranking(BPR)optimization to address these limitations.With the strong support of Long Short-Term Memory(LSTM)networks,we apply it to identify sequential dependencies of user behavior and then incorporate an attention mechanism to improve the prioritization of relevant items,thereby enhancing recommendations based on the hybrid feedback of the user and its interaction patterns.The proposed system is empirically evaluated using publicly available datasets from movie and music,and we evaluate the performance against standard recommendation models,including Popularity,BPR,ItemKNN,FPMC,LightGCN,GRU4Rec,NARM,SASRec,and BERT4Rec.The results demonstrate that our proposed framework consistently achieves high outcomes in terms of HitRate,NDCG,MRR,and Precision at K=100,with scores of(0.6763,0.1892,0.0796,0.0068)on MovieLens-100K,(0.6826,0.1920,0.0813,0.0068)on MovieLens-1M,and(0.7937,0.3701,0.2756,0.0078)on Last.fm.The results show an average improvement of around 15%across all metrics compared to existing sequence models,proving that our framework ranks and recommends items more accurately.展开更多
In the field of intelligent air combat,real-time and accurate recognition of within-visual-range(WVR)maneuver actions serves as the foundational cornerstone for constructing autonomous decision-making systems.However,...In the field of intelligent air combat,real-time and accurate recognition of within-visual-range(WVR)maneuver actions serves as the foundational cornerstone for constructing autonomous decision-making systems.However,existing methods face two major challenges:traditional feature engineering suffers from insufficient effective dimensionality in the feature space due to kinematic coupling,making it difficult to distinguish essential differences between maneuvers,while end-to-end deep learning models lack controllability in implicit feature learning and fail to model high-order long-range temporal dependencies.This paper proposes a trajectory feature pre-extraction method based on a Long-range Masked Autoencoder(LMAE),incorporating three key innovations:(1)Random Fragment High-ratio Masking(RFH-Mask),which enforces the model to learn long-range temporal correlations by masking 80%of trajectory data while retaining continuous fragments;(2)Kalman Filter-Guided Objective Function(KFG-OF),integrating trajectory continuity constraints to align the feature space with kinematic principles;and(3)Two-stage Decoupled Architecture,enabling efficient and controllable feature learning through unsupervised pre-training and frozen-feature transfer.Experimental results demonstrate that LMAE significantly improves the average recognition accuracy for 20-class maneuvers compared to traditional end-to-end models,while significantly accelerating convergence speed.The contributions of this work lie in:introducing high-masking-rate autoencoders into low-informationdensity trajectory analysis,proposing a feature engineering framework with enhanced controllability and efficiency,and providing a novel technical pathway for intelligent air combat decision-making systems.展开更多
Optical coherence tomography(OCT),particularly Swept-Source OCT,is widely employed in medical diagnostics and industrial inspections owing to its high-resolution imaging capabilities.However,Swept-Source OCT 3D imagin...Optical coherence tomography(OCT),particularly Swept-Source OCT,is widely employed in medical diagnostics and industrial inspections owing to its high-resolution imaging capabilities.However,Swept-Source OCT 3D imaging often suffers from stripe artifacts caused by unstable light sources,system noise,and environmental interference,posing challenges to real-time processing of large-scale datasets.To address this issue,this study introduces a real-time reconstruction system that integrates stripe-artifact suppression and parallel computing using a graphics processing unit.This approach employs a frequency-domain filtering algorithm with adaptive anti-suppression parameters,dynamically adjusted through an image quality evaluation function and optimized using a convolutional neural network for complex frequency-domain feature learning.Additionally,a graphics processing unit integrated 3D reconstruction framework is developed,enhancing data processing throughput and real-time performance via a dual-queue decoupling mechanism.Experimental results demonstrate significant improvements in structural similarity(0.92),peak signal-to-noise ratio(31.62 dB),and stripe suppression ratio(15.73 dB)compared with existing methods.On the RTX 4090 platform,the proposed system achieved an end-to-end delay of 94.36 milliseconds,a frame rate of 10.3 frames per second,and a throughput of 121.5 million voxels per second,effectively suppressing artifacts while preserving image details and enhancing real-time 3D reconstruction performance.展开更多
To advance the theoretical understanding,technological development,and field application of electric charge induction for monitoring rock deformation and failure,this study investigates the induced electric charge gen...To advance the theoretical understanding,technological development,and field application of electric charge induction for monitoring rock deformation and failure,this study investigates the induced electric charge generated during the deformation and failure of igneous rocks.The charge originates mainly from a combination of electrical polarization and triboelectric effects.Through laboratory experiments,we analyzed the time-frequency evolution of induced electric charge signals and identified relevant monitoring parameters.An online downhole electric charge induction monitoring system was developed and validated in the field.Experimental results show that the dominant frequency range of induced electric charge signals generated during igneous rock deformation and failure lies between 0 and 23 Hz,and a low-pass finite impulse response(FIR)filter effectively suppresses noise.Optimal sensor distances for monitoring cubic and cylindrical specimens were determined to be 17 mm and 13 mm,respectively.We proposed early warning indicators,including the maximum absolute value of the induced electric charge,the arithmetic mean value,the distribution dispersion coefficient,and the cumulative sum value.In field application,time-domain curves and spatial distribution charts of these warning indicators correspond well with changes in abutment stress ahead of the mining face,offering indirect insights into local stress evolution.This research provides technical and equipment support for the application of electric charge induction technology to monitoring and early warning of coal bursts.展开更多
The surge of large-scale models in recent years has led to breakthroughs in numerous fields,but it has also introduced higher computational costs and more complex network architectures.These increasingly large and int...The surge of large-scale models in recent years has led to breakthroughs in numerous fields,but it has also introduced higher computational costs and more complex network architectures.These increasingly large and intricate networks pose challenges for deployment and execution while also exacerbating the issue of network over-parameterization.To address this issue,various network compression techniques have been developed,such as network pruning.A typical pruning algorithm follows a three-step pipeline involving training,pruning,and retraining.Existing methods often directly set the pruned filters to zero during retraining,significantly reducing the parameter space.However,this direct pruning strategy frequently results in irreversible information loss.In the early stages of training,a network still contains much uncertainty,and evaluating filter importance may not be sufficiently rigorous.To manage the pruning process effectively,this paper proposes a flexible neural network pruning algorithm based on the logistic growth differential equation,considering the characteristics of network training.Unlike other pruning algorithms that directly reduce filter weights,this algorithm introduces a three-stage adaptive weight decay strategy inspired by the logistic growth differential equation.It employs a gentle decay rate in the initial training stage,a rapid decay rate during the intermediate stage,and a slower decay rate in the network convergence stage.Additionally,the decay rate is adjusted adaptively based on the filter weights at each stage.By controlling the adaptive decay rate at each stage,the pruning of neural network filters can be effectively managed.In experiments conducted on the CIFAR-10 and ILSVRC-2012 datasets,the pruning of neural networks significantly reduces the floating-point operations while maintaining the same pruning rate.Specifically,when implementing a 30%pruning rate on the ResNet-110 network,the pruned neural network not only decreases floating-point operations by 40.8%but also enhances the classification accuracy by 0.49%compared to the original network.展开更多
基金supported by the National Natural Science Foundation of China(Nos.51308008,51179099)SA Water Visiting International Academics,Graduates,Researchers and Affiliates(VIAGRA) FundUni SA Visiting Researcher Fund
文摘Minimizing particles in water is a key goal for improving drinking water quality and safety.The media filtration process,as the last step of the solid–liquid separation process,is largely influenced by the characteristics of flocs,which are formed and controlled within the coagulation process.In a laboratory-based study,the impacts of the physical characteristics of flocs formed using aluminum sulfate on the filtration treatment of two comparative water samples were investigated using a photometric dispersion analyzer and a filterability apparatus.In general,the optimum dosage for maximizing filterability was higher than that for minimizing turbidity under neutral p H conditions.For a monomeric aluminum-based coagulant,the charge neutralization mechanism produced better floc characteristics,including floc growth speed and size,than the sweep flocculation mechanism.In addition,the charge neutralization mechanism showed better performance compared to sweep flocculation in terms of DOC removal and floc filterability improvement for both waters,and showed superiority in turbidity removal only when the raw water had high turbidity.For the different mechanisms,the ways that floc characteristics impacted on floc filterability also differed.The low variation in floc size distribution obtained under the charge neutralization mechanism resulted in the flocs being amenable to removal by filtration processes.For the sweep flocculation mechanism,increasing the floc size improved the settling ability of flocs,resulting in higher filter efficiency.
文摘The effectiveness of deep-bed filtration with respect to suspension formed during the preceding processes is evaluated by the test of filterability. The properties and concentration of the suspension being filtered are affected by the efficiency of the preceding aggregation and separation processes. The basic principles of the test of filterability are based on the mechanistic model of filtration. The equations in the mathematical model of the mechanistic conception of filtration are derived from the theory of filtration. The arrangement of the pilot filtration plant for the determination of filterability of flocculent suspension is presented in this paper. The test of filterability is carried out with a thin-layer filter element. The design of a filter element arrangement and its installation are also disclosed in this paper. The inter-dependence of the coefficient of filtration efficiency on the specific volume of intercepted suspensions, filter media grain sizes and different filtration velocities are graphically presented. In addition, the effect of the filter bed clogging resulting from the properties of different suspensions on the head loss generated, the length of filtration cycle and quality of filtrate produced are also shown in this paper.
文摘The Savitzky-Golay(SG)filter,which employs polynomial least-squares approximations to smooth data and estimate derivatives,is widely used for processing noisy data.However,noise suppression by the SG filter is recognized to be limited at data boundaries and high frequencies,which can significantly reduce the signal-to-noise ratio(SNR).To solve this problem,a novel method synergistically integrating Principal Component Analysis(PCA)with SG filtering is proposed in this paper.This approach avoids the is-sue of excessive smoothing associated with larger window sizes.The proposed PCA-SG filtering algorithm was applied to a CO gas sensing system based on Cavity Ring-Down Spectroscopy(CRDS).The perform-ance of the PCA-SG filtering algorithm is demonstrated through comparison with Moving Average Filtering(MAF),Wavelet Transformation(WT),Kalman Filtering(KF),and the SG filter.The results demonstrate that the proposed algorithm exhibits superior noise reduction capabilities compared to the other algorithms evaluated.The SNR of the ring-down signal was improved from 11.8612 dB to 29.0913 dB,and the stand-ard deviation of the extracted ring-down time constant was reduced from 0.037μs to 0.018μs.These results confirm that the proposed PCA-SG filtering algorithm effectively improves the smoothness of the ring-down curve data,demonstrating its feasibility.
基金supported by the National Natural Science Foundation of China(No.62371263)the Postgraduate Research&Practice Innovation Program of Jiangsu Province(No.SJCK25_1995).
文摘1 Introduction Recently,the increasing demand for advanced telecommunication systems has spurred extensive research into bandpass filters(BPFs),with particular emphasis on miniaturization,reduction of insertion loss(IL),and enhancement of upper stopband rejection(Huang et al.,2021;Snyder et al.,2021;Lin et al.,2023;Zeng et al.,2023).
基金supported by the Royal Academy of Engineering,UK,in the scheme of Distinguished International Associate(DIA-2424-5-134).
文摘Accurate estimation of the State of Charge(SOC),State of Health(SOH),and Terminal Resistance(TR)is crucial for the effective operation of Battery Management Systems(BMS)in lithium-ion batteries.This study conducts a comprehensive comparative analysis of four Kalman filter variants Extended Kalman Filter(EKF),Extended Kalman-Bucy Filter(EKBF),Unscented Kalman Filter(UKF),and Unscented Kalman-Bucy Filter(UKBF)under varying battery parameter conditions.These include temperature fluctuation,self-discharge,current direction,cell capacity,process noise,and measurement noise.Our findings reveal significant variations in the performance of SOC and SOH predictions across filters,emphasizing that UKF demonstrates superior robustness to noise,while EKF performs better under accurate system dynamics.The study underscores the need for adaptive filtering strategies that can dynamically adjust to evolving battery parameters,thereby enhancing BMS reliability and extending battery lifespan.
基金Supported by the Guangxi Special Program for Technological Innovation Guidance(No.GuiKeAC25069006).
文摘In this study,a fifth-degree cubature particle filter(5CPF)is proposed to address the limited estimation accuracy in traditional particle filter algorithms for bearings-only tracking(BOT).This algorithm calculates the recommended density function by introducing a fifth-degree cubature Kalman filter algorithm to guide particle sampling,which effectively alleviates the problem of particle degradation and significantly improves the estimation accuracy of the filter.However,the 5CPF algorithm exhibits high computational complexity,particularly in scenarios with a large number of particles.Therefore,we propose the extended Kalman filter(EKF)-5CPF algorithm,which employs an EKF to replace the time update step for each particle in the 5CPF.This enhances the algorithm’s real-time capability while maintaining the high precision advantage of the 5CPF algorithm.In addition,we construct bearing-only dual-station and single-motion station target tracking systems,and the filtering performances of 5CPF and EKF-5CPF algorithms under different conditions are analyzed.The results show that both the 5CPF algorithm and EKF-5CPF have strong robustness and can adapt to different noise environments.Furthermore,both algorithms significantly outperform traditional nonlinear filtering algorithms in terms of convergence speed,tracking accuracy,and overall stability.
基金supported by the Program of the National Natural Science Foundation of China(22238002)the Fundamental Research Funds for the Central Universities(DUT22-LAB610)Research and Innovation Team Project of Dalian University of Technology(DUT2022TB10).
文摘Color filters are essential components for optical modulation.However,conventional filters are restricted to operating exclusively in either reflective or transmissive mode.Furthermore,they suffer from limited UV and thermal stability,low color purity,and exhibit identical coloration on both surfaces.Herein,we propose a novel design strategy for trans-reflective color filters by integrating the absorptive properties of dye-doped polysulfone(PSU)with the diffractive capabilities of photonic crystals.This composite filter achieved broad-spectrum transmission with deep color outputs—yellow(0.410,0.510),magenta(0.446,0.231),and cyan(0.201,0.425)—closely aligned with standard color space coordinates.By tuning the refractive index of CeO_(2)@SiO_(2)nanoparticles to match dye-based PSU matrix,the transmittance of filters exceeded 70%.Moreover,dye-mediated absorption reduces the scattering light,thereby enhancing reflection color purity(full width at half maxima(FWHM)=25 nm)and producing vibrant blue,green,and red hues.The incorporation of UV-absorbing CeO_(2)@SiO_(2)nanoparticles effectively mitigated dye photodegradation,yielding exceptional UV stability(ΔT<2%under prolonged UV exposure).The filters also exhibited outstanding thermal stability(ΔT<1%after 30 min heat treatment at 230°C).This work establishes a robust materials design framework for multifunctional optical filters,advancing the development of highfidelity dual-mode color systems for next-generation display technologies.
基金funded by the Fundamental Research Funds for the Central Universities(Grant No.106-YDZX2025022)the Startup Foundation of New Professor at Nanjing Agricultural University(Grant No.106-804005)the“Qing Lan Project”of Jiangsu Higher Education Institutions.
文摘Understanding fish movement trajectories in aquaculture is essential for practical applications,such as disease warning,feeding optimization,and breeding management.These trajectories reveal key information about the fish’s behavior,health,and environmental adaptability.However,when multi-object tracking(MOT)algorithms are applied to the high-density aquaculture environment,occlusion and overlapping among fish may result in missed detections,false detections,and identity switching problems,which limit the tracking accuracy.To address these issues,this paper proposes FishTracker,a MOT algorithm,by utilizing a Tracking-by-Detection framework.First,the neck part of the YOLOv8 model is enhanced by introducing a Multi-Scale Dilated Attention(MSDA)module to improve object localization and classification confidence.Second,an Adaptive Kalman Filter(AKF)is employed in the tracking phase to dynamically adjust motion prediction parameters,thereby overcoming target adhesion and nonlinear motion in complex scenarios.Experimental results show that FishTracker achieves a multi-object tracking accuracy(MOTA)of 93.22% and 87.24% in bright and dark illumination conditions,respectively.Further validation in a real aquaculture scenario reveal that FishTracker achieves aMOTA of 76.70%,which is 5.34% higher than the baselinemodel.The higher order tracking accuracy(HOTA)reaches 50.5%,which is 3.4% higher than the benchmark.In conclusion,FishTracker can provide reliable technical support for accurate tracking and behavioral analysis of high-density fish populations.
基金funded by the Directorate General of Research and Development,Ministry of Higher Education,Science and Technology of the Republic of Indonesia,with grant number 2.6.63/UN32.14.1/LT/2025.
文摘Recommendation systems are an integral and indispensable part of every digital platform,as they can suggest content or items to users based on their respective needs.Collaborative filtering is a technique often used in various studies,which produces recommendations by analyzing similarities between users and items based on their behavior.Although often used,traditional collaborative filtering techniques still face the main challenge of sparsity.Sparsity problems occur when the data in the system is sparse,meaning that only a portion of users provide feedback on some items,resulting in inaccurate recommendations generated by the system.To overcome this problem,we developed aHybrid Collaborative Filtering model based onMatrix Factorization andGradient Boosting(HCF-MFGB),a new hybrid approach.Our proposed model integrates SVD++,the XGBoost ensemble learning algorithm,and utilizes user demographic data and meta items.We utilize information,both explicitly and implicitly,to learn user preference patterns using SVD++.The XGBoost algorithm is used to create hundreds of decision trees incrementally,thereby improving model accuracy.Meanwhile,user demographic and meta-item data are clustered using the K-Means Clustering algorithm to capture similarities in user and item characteristics.This combination is designed to improve rating prediction accuracy by reducing reliance on minimal explicit rating data,while addressing sparsity issues in movie recommendation systems.The results of experiments on the MovieLens 100K,MovieLens 1M,and CiaoDVD datasets show significant improvements,outperforming various other baselinemodels in terms of RMSE and MAE.On theMovieLens 100K dataset,the HCF-MFGB model obtained an RMSE value of 0.853 and an MAE value of 0.674.On theMovieLens 1M dataset,the HCF-MFGB model obtained an RMSE value of 0.763 and an MAE value of 0.61.On the CiaoDCD dataset,the HCF-MFGB model achieved an RMSE value of 0.718 and an MAE value of 0.495.These results confirm a significant improvement in movie recommendation accuracy with the proposed approach.
基金supported by the National Natural Science Foundation of China(No.51939009)Shenzhen Science and Technology Program(Nos.JCYJ20241202125905008 and GXWD20201231165807007-20200810165349001).
文摘A trace analytical method based on solid-phase extraction gas chromatography-tandem mass spectrometry(SPE–GC–MS/MS)was developed for the rapid detection of 256 semi-volatile organic compounds(SVOCs),including 25 polycyclic aromatic hydrocarbons(PAHs),70 polychlorinated biphenyls(PCBs),123 pesticides,20 phthalate esters(PAEs),4 organophosphate esters(OPEs),9 synthetic musks(SMs),and 5 UV filters(UVs)in water.No-tably,this method provided a decent linearity of calibration standards(R^(2)>0.999),excellent method limits of quantification(MLOQs)(0.12–11.41 ng/L),satisfactory matrix spiking recovery rates(60.4%–126%),and high precision(intra-day relative standard deviations(RSDs):1.0%–10.0%,inter-day RSDs:3.0%–15.0%,and inter-week RSDs:3.4%–15.7%),making it suitable for trace-level studies.Statistical analysis revealed that SVOCs with higher volatility exhibited enhanced recovery rates.Validation of the methodology involved analyzing SVOCs in real spring water and river water samples.Twenty-seven SVOCs were detected in spring water and 58 in river water,with an average concentration of 631.73 and 16,095 ng/L,respectively.Among the detected SVOCs,PAEs constituted the predominant proportion.This study underscored the presence of SVOCs contamination specifi-cally within the spring water,although SVOCs concentrations in river water were significantly greater than those found in spring water.In summary,this sensitive method based on SPE–GC–MS/MS was successfully developed and validated for the rapid analysis of a diverse array of 256 SVOCs at trace levels in water,including not only the traditional highly valued PAHs,PCBs,pesticides,and PAEs,but also the emerging OPEs,UVs,and SMs.
文摘Pull-ups are a very common fitness exercise that can be seen in many gyms.For athletes,it is very important to perform pull-ups correctly and scientifically.The pull-up scoring method designed in this paper can score the quality of pull-up movement scientifically and objectively,and provide guidance to help athletes better complete the pull-up movement.In this method,the OpenPose algorithm is used to identify the coordinates of skeleton points,and then the coordinate data are processed by a Kalman filter to obtain coordinates closer to the true values.Finally,the filtered data are input into the scoring algorithm designed based on the fuzzy comprehensive evaluation algorithm,and the results of the pull-up quality score and the corresponding guidance are obtained.
基金supported in part by the Beijing Natural Science Foundation under Grant 4252050in part by the National Science Fund for Distinguished Young Scholars under Grant 62425304in part by the Basic Science Center Programs of NSFC under Grant 62088101.
文摘This paper investigates the consensus tracking control problem for high order nonlinear multi-agent systems subject to non-affine faults,partial measurable states,uncertain control coefficients,and unknown external disturbances.Under the directed topology conditions,an observer-based finite-time control strategy based on adaptive backstepping and is proposed,in which a neural network-based state observer is employed to approximate the unmeasurable system state variables.To address the complexity explosion problem associated with the backstepping method,a finite-time command filter is incorporated,with error compensation signals designed to mitigate the filter-induced errors.Additionally,the Butterworth low-pass filter is introduced to avoid the algebraic ring problem in the design of the controller.The finite-time stability of the closed-loop system is rigorously analyzed with the finite-time Lyapunov stability criterion,validating that all closed-loop signals of the system remain bounded within a finite time.Finally,the effectiveness of the proposed control strategy is verified through a simulation example.
文摘This article presents an adaptive intelligent control strategy applied to a lumped-parameter evaporator model,i.e.,a simplified dynamic representation treating the evaporator as a single thermal node with uniform temperature distribution,suitable for control design due to its balance between physical fidelity and computational simplicity.The controller uses a wavelet-based neural proportional,integral,derivative(PID)controller with IIR filtering(infinite impulse response).The dynamic model captures the essential heat and mass transfer phenomena through a nonlinear energy balance,where the cooling capacity“Qevap”is expressed as a non-linear function of the compressor frequency and the temperature difference,specifically,Q_(evap)=k_(1)u(T_(in)−T_(e))with u as compressor frequency,Te evaporator temperature,and Tin inlet fluid temperature.The operating conditions of the system,in general terms,focus on the following variables,the overall thermal capacity is 1000 J/K,typical for small-capacity heat exchangers,The mass flow is 0.05 kg/s,typical for secondary liquid cooling circuits,the overall loss coefficient of 50 W/K that corresponds to small evaporators with partial insulation,the temperatures(inlet)of 10℃and the temperature of environment of 25℃,thermal load of 200 W that corresponds to a small-scaled air conditioning applications.To handle system nonlinearities and improve control performance,aMorlet wavelet-based neural network(Wavenet)is used to dynamically adjust the PID gains online.An IIR filter is incorporated to smooth the adaptive gains,improving stability and reducing oscillations.In contrast to prior wavelet-or neural-adaptive PID controllers in HVAC applications,which typically adjust gains without explicit filtering or not tailored to evaporator dynamics,this work introduces the first PID–Wavenet scheme augmented with an IIR-based stabilization layer,specifically designed to address the combined challenges of nonlinear evaporator behavior,gain oscillation,and real-time implementability.The proposed controller(PID-Wavenet+IIR)is implemented and validated inMATLAB/Simulink,demonstrating superior performance compared to a conventional PID tuned using Simulink’s auto-tuning function.Key results include a reduction in settling time from 13.3 to 8.2 s,a reduction in overshoot from 3.5%to 0.8%,a reduction in steady-state error from 0.12℃ to 0.02℃and a 13%reduction in energy overall consumption.The controller also exhibits greater robustness and adaptability under varying thermal loads.This explicit integration of wavelet-driven adaptation with IIR-filtered gain shaping constitutes the main methodological contribution and novelty of the work.These findings validate the effectiveness of the wavelet-based adaptive approach for advanced thermal management in refrigeration and HVAC systems,with potential applications in controlling variable-speed compressors,liquid chillers,and compact cooling units.
基金supported by the National Key Research and Development Program of China(Grant No.2022YFF0711402)the Specialized Research Fund for State Key Laboratories。
文摘Using a recognition model of atmospheric gravity waves(AGWs),we identified 519 AGW events from the OH airglow images observed at the Dandong and Lhasa stations from 2015 to 2017.The 317 AGW events detected at the Dandong station have wavelengths ranging from 30 to 60 km,periods from 14 to 20 min,horizontal speeds from 30 to 60 m/s,and relative intensities from 0.4%to 0.6%,respectively.The parameters of 202 events recorded at the Lhasa station mainly vary within 15-35 km in horizontal wavelength,4-6 min in period,40-100 m/s in horizontal velocity,and 0.1%-0.3%in relative intensity.The occurrence rate peaks in winter and summer at Dandong and the peak in summer are absent at Lhasa because of the lack of convective weather.The seasonal propagation directions of the waves are influenced by both the wind field-filtering effect and the distribution of wave sources.In spring,because of the southeastward background wind field,fewer southeastward events are observed at the Dandong station.The situation at the Lhasa station is similar.In summer,both the Lhasa and Dandong stations are dominated by northeastward AGWs,which can be attributed to the southwestward wind.In autumn,ray-tracing results show that the events at Dandong mainly originate from wind shear,whereas the events at the Lhasa station are triggered by convective weather.The location of the wave sources determines the trend of the propagation directions at the Dandong and Lhasa stations in autumn.In winter,because of the eastward wind,more events are propagating to the southwest at the Dandong station.
基金funded by Soonchunhyang University,Grant Number 20250029。
文摘Recommendation systems have become indispensable for providing tailored suggestions and capturing evolving user preferences based on interaction histories.The collaborative filtering(CF)model,which depends exclusively on user-item interactions,commonly encounters challenges,including the cold-start problem and an inability to effectively capture the sequential and temporal characteristics of user behavior.This paper introduces a personalized recommendation system that combines deep learning techniques with Bayesian Personalized Ranking(BPR)optimization to address these limitations.With the strong support of Long Short-Term Memory(LSTM)networks,we apply it to identify sequential dependencies of user behavior and then incorporate an attention mechanism to improve the prioritization of relevant items,thereby enhancing recommendations based on the hybrid feedback of the user and its interaction patterns.The proposed system is empirically evaluated using publicly available datasets from movie and music,and we evaluate the performance against standard recommendation models,including Popularity,BPR,ItemKNN,FPMC,LightGCN,GRU4Rec,NARM,SASRec,and BERT4Rec.The results demonstrate that our proposed framework consistently achieves high outcomes in terms of HitRate,NDCG,MRR,and Precision at K=100,with scores of(0.6763,0.1892,0.0796,0.0068)on MovieLens-100K,(0.6826,0.1920,0.0813,0.0068)on MovieLens-1M,and(0.7937,0.3701,0.2756,0.0078)on Last.fm.The results show an average improvement of around 15%across all metrics compared to existing sequence models,proving that our framework ranks and recommends items more accurately.
文摘In the field of intelligent air combat,real-time and accurate recognition of within-visual-range(WVR)maneuver actions serves as the foundational cornerstone for constructing autonomous decision-making systems.However,existing methods face two major challenges:traditional feature engineering suffers from insufficient effective dimensionality in the feature space due to kinematic coupling,making it difficult to distinguish essential differences between maneuvers,while end-to-end deep learning models lack controllability in implicit feature learning and fail to model high-order long-range temporal dependencies.This paper proposes a trajectory feature pre-extraction method based on a Long-range Masked Autoencoder(LMAE),incorporating three key innovations:(1)Random Fragment High-ratio Masking(RFH-Mask),which enforces the model to learn long-range temporal correlations by masking 80%of trajectory data while retaining continuous fragments;(2)Kalman Filter-Guided Objective Function(KFG-OF),integrating trajectory continuity constraints to align the feature space with kinematic principles;and(3)Two-stage Decoupled Architecture,enabling efficient and controllable feature learning through unsupervised pre-training and frozen-feature transfer.Experimental results demonstrate that LMAE significantly improves the average recognition accuracy for 20-class maneuvers compared to traditional end-to-end models,while significantly accelerating convergence speed.The contributions of this work lie in:introducing high-masking-rate autoencoders into low-informationdensity trajectory analysis,proposing a feature engineering framework with enhanced controllability and efficiency,and providing a novel technical pathway for intelligent air combat decision-making systems.
文摘Optical coherence tomography(OCT),particularly Swept-Source OCT,is widely employed in medical diagnostics and industrial inspections owing to its high-resolution imaging capabilities.However,Swept-Source OCT 3D imaging often suffers from stripe artifacts caused by unstable light sources,system noise,and environmental interference,posing challenges to real-time processing of large-scale datasets.To address this issue,this study introduces a real-time reconstruction system that integrates stripe-artifact suppression and parallel computing using a graphics processing unit.This approach employs a frequency-domain filtering algorithm with adaptive anti-suppression parameters,dynamically adjusted through an image quality evaluation function and optimized using a convolutional neural network for complex frequency-domain feature learning.Additionally,a graphics processing unit integrated 3D reconstruction framework is developed,enhancing data processing throughput and real-time performance via a dual-queue decoupling mechanism.Experimental results demonstrate significant improvements in structural similarity(0.92),peak signal-to-noise ratio(31.62 dB),and stripe suppression ratio(15.73 dB)compared with existing methods.On the RTX 4090 platform,the proposed system achieved an end-to-end delay of 94.36 milliseconds,a frame rate of 10.3 frames per second,and a throughput of 121.5 million voxels per second,effectively suppressing artifacts while preserving image details and enhancing real-time 3D reconstruction performance.
基金supported by the National Key Research and Development Project of the National Natural Science Foundation of China(Grant No.2022YFC3004605)the National Natural Science Foundation of China Youth Science Fund(Grant No.52104087).
文摘To advance the theoretical understanding,technological development,and field application of electric charge induction for monitoring rock deformation and failure,this study investigates the induced electric charge generated during the deformation and failure of igneous rocks.The charge originates mainly from a combination of electrical polarization and triboelectric effects.Through laboratory experiments,we analyzed the time-frequency evolution of induced electric charge signals and identified relevant monitoring parameters.An online downhole electric charge induction monitoring system was developed and validated in the field.Experimental results show that the dominant frequency range of induced electric charge signals generated during igneous rock deformation and failure lies between 0 and 23 Hz,and a low-pass finite impulse response(FIR)filter effectively suppresses noise.Optimal sensor distances for monitoring cubic and cylindrical specimens were determined to be 17 mm and 13 mm,respectively.We proposed early warning indicators,including the maximum absolute value of the induced electric charge,the arithmetic mean value,the distribution dispersion coefficient,and the cumulative sum value.In field application,time-domain curves and spatial distribution charts of these warning indicators correspond well with changes in abutment stress ahead of the mining face,offering indirect insights into local stress evolution.This research provides technical and equipment support for the application of electric charge induction technology to monitoring and early warning of coal bursts.
基金supported by the National Natural Science Foundation of China under Grant No.62172132.
文摘The surge of large-scale models in recent years has led to breakthroughs in numerous fields,but it has also introduced higher computational costs and more complex network architectures.These increasingly large and intricate networks pose challenges for deployment and execution while also exacerbating the issue of network over-parameterization.To address this issue,various network compression techniques have been developed,such as network pruning.A typical pruning algorithm follows a three-step pipeline involving training,pruning,and retraining.Existing methods often directly set the pruned filters to zero during retraining,significantly reducing the parameter space.However,this direct pruning strategy frequently results in irreversible information loss.In the early stages of training,a network still contains much uncertainty,and evaluating filter importance may not be sufficiently rigorous.To manage the pruning process effectively,this paper proposes a flexible neural network pruning algorithm based on the logistic growth differential equation,considering the characteristics of network training.Unlike other pruning algorithms that directly reduce filter weights,this algorithm introduces a three-stage adaptive weight decay strategy inspired by the logistic growth differential equation.It employs a gentle decay rate in the initial training stage,a rapid decay rate during the intermediate stage,and a slower decay rate in the network convergence stage.Additionally,the decay rate is adjusted adaptively based on the filter weights at each stage.By controlling the adaptive decay rate at each stage,the pruning of neural network filters can be effectively managed.In experiments conducted on the CIFAR-10 and ILSVRC-2012 datasets,the pruning of neural networks significantly reduces the floating-point operations while maintaining the same pruning rate.Specifically,when implementing a 30%pruning rate on the ResNet-110 network,the pruned neural network not only decreases floating-point operations by 40.8%but also enhances the classification accuracy by 0.49%compared to the original network.
