The widespread adoption of tunnel boring machines(TBMs)has led to an increased focus on disc cutter wear,including both normal and abnormal types,for efficient and safe TBM excavation.However,abnormal wear has yet to ...The widespread adoption of tunnel boring machines(TBMs)has led to an increased focus on disc cutter wear,including both normal and abnormal types,for efficient and safe TBM excavation.However,abnormal wear has yet to be thoroughly investigated,primarily due to the complexity of considering mixed ground conditions and the imbalance in the number of instances between the two types of wear.This study developed a prediction model for abnormal TBM disc cutter wear,considering mixed ground conditions,by employing interpretable machine learning with data augmentation.An equivalent elastic modulus was used to consider the characteristics of mixed ground conditions,and wear data was obtained from 65 cutterhead intervention(CHI)reports covering both mixed ground and hard rock sections.With a balanced training dataset obtained by data augmentation,an extreme gradient boosting(XGB)model delivered acceptable results with an accuracy of 0.94,an F1-score of 0.808,and a recall of 0.8.In addition,the accuracy for each individual disc cutter exhibited low variability.When employing data augmentation,a significant improvement in recall was observed compared to when it was not used,although the difference in accuracy and F1-score was marginal.The subsequent model interpretation revealed the chamber pressure,cutter installation radius,and torque as significant contributors.Specifically,a threshold in chamber pressure was observed,which could induce abnormal wear.The study also explored how elevated values of these influential contributors correlate with abnormal wear.The proposed model offers a valuable tool for planning the replacement of abnormally worn disc cutters,enhancing the safety and efficiency of TBM operations.展开更多
A new method of detecting abnormal sounding data based on LS-SVM is presented.The theorem proves that the trend surface filter is the especial result of LS-SVM.In order to depict the relationship of trend surface filt...A new method of detecting abnormal sounding data based on LS-SVM is presented.The theorem proves that the trend surface filter is the especial result of LS-SVM.In order to depict the relationship of trend surface filter and LS-SVM,a contrast is given.The example shows that abnormal sounding data could be detected effectively by LS-SVM when the training samples and kernel function are reasonable.展开更多
The parameters of abnormal data are defined and their influence on the original data and Fourier Algorithm is studied. A formula is proposed to quantify how the abnormal data influences the amplitude calculated by Fou...The parameters of abnormal data are defined and their influence on the original data and Fourier Algorithm is studied. A formula is proposed to quantify how the abnormal data influences the amplitude calculated by Fourier Algorithm. Two simulation models are established in Matlab to study the influence of abnormal data on relay protection. The simulation results show that the abnormal data can make distance protection extend the fault’s influence and make over current protection start by error.展开更多
High-order asymmetric flatness defects resulting from the abnormal state of roll system are the main issue of precision rolling mill in the manufacturing process of high-strength thin strip.Due to the difficulty of mo...High-order asymmetric flatness defects resulting from the abnormal state of roll system are the main issue of precision rolling mill in the manufacturing process of high-strength thin strip.Due to the difficulty of monitoring and adjusting the abnormal state,the spatial state of roll system cannot be controlled by traditional methods.It is difficult to fundamentally improve these high-order asymmetric flatness defects.Therefore,a digital twin model of flatness control process for S6-high rolling mill was established,which could be used to analyze the influence of the abnormal state on the flatness control characteristic and propose improvement strategies.The internal relationship between the force state of side support roll system and the abnormal state of roll system was proposed.The XGBoost algorithm model was established to analyze the contribution degree of the side support roll system force to the flatness characteristic quantity.The abnormal state of roll system in the S6-high rolling mill can be diagnosed by analyzing the flatness characteristic difference between flatness value of the rolled strip and calculated characteristic value of finite element simulation.The flatness optimization model of the gray wolf optimization–long short-term memory non-dominated sorting whale optimization algorithm(GWO-LSTM-NSWOA)was established,and the decision-making selection was made from the Pareto frontier based on the flatness requirements of cold rolling to regulate the abnormal state of the roll system.The results indicate that the contribution degree of the force of the side support roll system to the flatness characteristics is more than 25%,which is the main influence of high-order asymmetric flatness defect.The performance of the GWO-LSTM flatness feature prediction model has clear advantages over back propagation and LSTM.The practical applications show that optimizing the force of side support roll system can reduce the high point of high-strength strip flatness from 13.2 to 6 IU and decrease the percentage of low-strength strip flatness defects from 1.6%to 1.2%.This optimization greatly reduced the proportion of flatness defects,improved the accuracy level of flatness control of precision rolling mill,and provided a guarantee for the stable production of thin strip.展开更多
This paper deals with an important subject of rejecting the abnormal dataintelligently in the dynamic data system.Based on the principle of nearest neighbor of fuz-zy mathematics,an approach of mathematically abstract...This paper deals with an important subject of rejecting the abnormal dataintelligently in the dynamic data system.Based on the principle of nearest neighbor of fuz-zy mathematics,an approach of mathematically abstracting the human thinking and phys-ical practice knowledge is discussed,a new method of automatic rejection of abnormal da-ta is then proposed.The experimental results show that the method is available to the practice.展开更多
A compound relay protection operation criterion is proposed based on Kirchhoff’s Current Law and abnormal data detecting algorithm. The abnormal data detecting algorithm are proposed after deep research on the abnorm...A compound relay protection operation criterion is proposed based on Kirchhoff’s Current Law and abnormal data detecting algorithm. The abnormal data detecting algorithm are proposed after deep research on the abnormal data properties. The current transformer status monitoring system and current phase angle detecting system is introduced. A simulation model containing different power sources and loads is established in Matlab. The simulation results show that this compound criterion can work quickly and reliably in all conditions.展开更多
With the increasing number of digital devices generating a vast amount of video data,the recognition of abnormal image patterns has become more important.Accordingly,it is necessary to develop a method that achieves t...With the increasing number of digital devices generating a vast amount of video data,the recognition of abnormal image patterns has become more important.Accordingly,it is necessary to develop a method that achieves this task using object and behavior information within video data.Existing methods for detecting abnormal behaviors only focus on simple motions,therefore they cannot determine the overall behavior occurring throughout a video.In this study,an abnormal behavior detection method that uses deep learning(DL)-based video-data structuring is proposed.Objects and motions are first extracted from continuous images by combining existing DL-based image analysis models.The weight of the continuous data pattern is then analyzed through data structuring to classify the overall video.The performance of the proposed method was evaluated using varying parameter settings,such as the size of the action clip and interval between action clips.The model achieved an accuracy of 0.9817,indicating excellent performance.Therefore,we conclude that the proposed data structuring method is useful in detecting and classifying abnormal behaviors.展开更多
As cyber attacks increase in volume and complexity,it becomes more and more difficult for existing analytical tools to detect previously unseen malware.This paper proposes a cooperative framework to leverage the robus...As cyber attacks increase in volume and complexity,it becomes more and more difficult for existing analytical tools to detect previously unseen malware.This paper proposes a cooperative framework to leverage the robustness of big data analytics and the power of ensemble learning techniques to detect the abnormal behavior.In addition to this proposal,we implement a large scale network abnormal traffic behavior detection system performed by the framework.The proposed model detects the abnormal behavior from large scale network traffic data using a combination of a balanced decomposition algorithm and an ensemble SVM.First,the collected dataset is divided into k subsets based on the similarity between patterns using a parallel map reduce k-means algorithm.Then,patterns are randomly selected from each cluster and balanced training sub datasets are formed.Next,the subsets are fed into the mappers to build an SVM model.The construction of the ensemble is achieved in the reduce phase.The proposed structure closely delivers a high accuracy as the number of iterations increases.Experimental results show a promising gain in detection rate and false alarm compared with other existing models.展开更多
With the changes in educational models,applying computer algorithms and artificial intelligence technologies to data analysis in universities has become a research hotspot in the field of intelligent education.In resp...With the changes in educational models,applying computer algorithms and artificial intelligence technologies to data analysis in universities has become a research hotspot in the field of intelligent education.In response to the increasing amount of student data in universities,this study proposes to use an optimized isolated forest algorithm for recognizing features to detect abnormal student behavior concealed in big data for educational management.Firstly,it uses a logistic regression algorithm to update the calculation method of isolated forest weights and then uses residual statistics to eliminate redundant forests.Finally,it utilizes discrete particle swarm optimization to optimize the isolated forest algorithm.On this basis,improvements have also been made to the traditional gated loop unit network.It merges the two improved algorithm models and builds an anomaly detection model for collecting college student education data.The experiment shows that the optimized isolated forest algorithm has a recognition accuracy of 0.986 and a training time of 1s.The recognition accuracy of the improved gated loop unit network is 0.965,and the training time is 0.16s.In summary,the constructed model can effectively identify abnormal data of college students,thereby helping educators to detect students’problems in time and helping students to improve their learning status.展开更多
Online monitoring methods have been widely used in many major devices, however the normal and abnormal states of equipment are estimated mainly based on the monitoring results whether monitored parameters exceed the s...Online monitoring methods have been widely used in many major devices, however the normal and abnormal states of equipment are estimated mainly based on the monitoring results whether monitored parameters exceed the setting thresholds. Using these monitoring methods may cause serious false positive or false negative results. In order to precisely monitor the state of equipment, the problem of abnormality degree detection without fault sample is studied with a new detection method called negative potential field group detectors（NPFG-detectors）. This method achieves the quantitative expression of abnormality degree and provides the better detection results compared with other methods. In the process of Iris data set simulation, the new algorithm obtains the successful results in abnormal detection. The detection rates for 3 types of Iris data set respectively reach 100%, 91.6%, and 95.24% with 50% training samples. The problem of Bearing abnormality degree detection via an abnormality degree curve is successfully solved.展开更多
In the study of the application of MODIS satellite remote sensing data to earthquake prediction, the paper puts forward for the first time a quantitative method to estimate the ratio for the pixels with abnormal brigh...In the study of the application of MODIS satellite remote sensing data to earthquake prediction, the paper puts forward for the first time a quantitative method to estimate the ratio for the pixels with abnormal brightness temperature (BT) increase and a preliminary scheme for cloud removal. The principle is that, firstly, the cloudless data observed by the same satellite at the same period of time but in different days (usually 1 day to 3 days) are mosaiced to get high ratio of clear sky, and then the BT variation curve and mean square difference (MSD) of each pixel are calculated with the data from the covered area to determine daily whether the BT data of the day is normal or not at a certain pixel by using double the MSD as the criterion. The ratio for the pixels with abnormal BT increase can be calculated by dividing the total number of abnormal pixels with the total pixels of the whole area. Analysis on a series of recent earthquakes in the Taiwan Region shows that the ratio for pixels with abnormal BT increase, which normally undulates around zero, has a sudden enhancement 1 day to 20 days before medium-strong earthquakes. It is expected that a new method for identifying earthquake auspice could be found through special studies in regions with frequent seismic activity by analyzing the change of the ratio for the pixels with abnormal BT increase from MODIS satellite remote sensing infrared (IR) information from which the effect of clouds has been removed to a certain extent.展开更多
The noise data in vertical component records of 85 seismic stations in Fujian Province during 2012 is used as the research object in this paper. The noise data is divided into fiveminute segments to calculate the powe...The noise data in vertical component records of 85 seismic stations in Fujian Province during 2012 is used as the research object in this paper. The noise data is divided into fiveminute segments to calculate the power spectra. The high reference line and low reference line of station are then identified by drawing a probability density function graph( PDF)using the power spectral probability density function. Moreover, according to the anomalies of PDF graphs in 85 seismic stations,the abnormal noise is divided into four categories: dropped packet, low noise, high noise, and median noise anomalies.Afterwards,four selection methods are found by the high or low noise reference line of the stations,and the system of real-time monitoring of seismic noise is formed by combining the four selection methods. Noise records of 85 seismic stations in Fujian Province in July2013 are selected for verification,and the results show that the anomalous noise-recognition system could reach a 90% success rate at most stations and the effect of selection are very good. Therefore,it could be applied to the seismic noise real-time monitoring in stations.展开更多
电网系统运行大安全管理数据规模较大,传统方法难以准确且快速地挖掘出异常数据。因此,文章提出了基于大语言模型(Large Language Model, LLM)的电网系统运行大安全管理异常数据挖掘。该方法在清洗并脱敏处理原始电网系统运行大安全管...电网系统运行大安全管理数据规模较大,传统方法难以准确且快速地挖掘出异常数据。因此,文章提出了基于大语言模型(Large Language Model, LLM)的电网系统运行大安全管理异常数据挖掘。该方法在清洗并脱敏处理原始电网系统运行大安全管理数据后,形成结构化的电网系统运行大安全管理数据集。通过构建LLM,该方法将结构化数据集输入模型,经训练后输出异常数据挖掘结果。实验分析表明,该方法数据挖掘结果的相对平方根误差仅为0.47%,交叉熵损失收敛值低至2.2146,显著优于传统聚类与小波方法,可为电网安全运行提供高效、可靠的异常监测支持。展开更多
基金support of the“National R&D Project for Smart Construction Technology (Grant No.RS-2020-KA157074)”funded by the Korea Agency for Infrastructure Technology Advancement under the Ministry of Land,Infrastructure and Transport,and managed by the Korea Expressway Corporation.
文摘The widespread adoption of tunnel boring machines(TBMs)has led to an increased focus on disc cutter wear,including both normal and abnormal types,for efficient and safe TBM excavation.However,abnormal wear has yet to be thoroughly investigated,primarily due to the complexity of considering mixed ground conditions and the imbalance in the number of instances between the two types of wear.This study developed a prediction model for abnormal TBM disc cutter wear,considering mixed ground conditions,by employing interpretable machine learning with data augmentation.An equivalent elastic modulus was used to consider the characteristics of mixed ground conditions,and wear data was obtained from 65 cutterhead intervention(CHI)reports covering both mixed ground and hard rock sections.With a balanced training dataset obtained by data augmentation,an extreme gradient boosting(XGB)model delivered acceptable results with an accuracy of 0.94,an F1-score of 0.808,and a recall of 0.8.In addition,the accuracy for each individual disc cutter exhibited low variability.When employing data augmentation,a significant improvement in recall was observed compared to when it was not used,although the difference in accuracy and F1-score was marginal.The subsequent model interpretation revealed the chamber pressure,cutter installation radius,and torque as significant contributors.Specifically,a threshold in chamber pressure was observed,which could induce abnormal wear.The study also explored how elevated values of these influential contributors correlate with abnormal wear.The proposed model offers a valuable tool for planning the replacement of abnormally worn disc cutters,enhancing the safety and efficiency of TBM operations.
基金The National High-Tech Research and Development Program of China (863 Program) under contract No.2007AA12Z326the National Natural Science Foundation of China under contract Nos 40974010 and 40971306
文摘A new method of detecting abnormal sounding data based on LS-SVM is presented.The theorem proves that the trend surface filter is the especial result of LS-SVM.In order to depict the relationship of trend surface filter and LS-SVM,a contrast is given.The example shows that abnormal sounding data could be detected effectively by LS-SVM when the training samples and kernel function are reasonable.
文摘The parameters of abnormal data are defined and their influence on the original data and Fourier Algorithm is studied. A formula is proposed to quantify how the abnormal data influences the amplitude calculated by Fourier Algorithm. Two simulation models are established in Matlab to study the influence of abnormal data on relay protection. The simulation results show that the abnormal data can make distance protection extend the fault’s influence and make over current protection start by error.
基金financially supported by the National Key Research and Development Program of China(Grant No.2023YFB3812602).
文摘High-order asymmetric flatness defects resulting from the abnormal state of roll system are the main issue of precision rolling mill in the manufacturing process of high-strength thin strip.Due to the difficulty of monitoring and adjusting the abnormal state,the spatial state of roll system cannot be controlled by traditional methods.It is difficult to fundamentally improve these high-order asymmetric flatness defects.Therefore,a digital twin model of flatness control process for S6-high rolling mill was established,which could be used to analyze the influence of the abnormal state on the flatness control characteristic and propose improvement strategies.The internal relationship between the force state of side support roll system and the abnormal state of roll system was proposed.The XGBoost algorithm model was established to analyze the contribution degree of the side support roll system force to the flatness characteristic quantity.The abnormal state of roll system in the S6-high rolling mill can be diagnosed by analyzing the flatness characteristic difference between flatness value of the rolled strip and calculated characteristic value of finite element simulation.The flatness optimization model of the gray wolf optimization–long short-term memory non-dominated sorting whale optimization algorithm(GWO-LSTM-NSWOA)was established,and the decision-making selection was made from the Pareto frontier based on the flatness requirements of cold rolling to regulate the abnormal state of the roll system.The results indicate that the contribution degree of the force of the side support roll system to the flatness characteristics is more than 25%,which is the main influence of high-order asymmetric flatness defect.The performance of the GWO-LSTM flatness feature prediction model has clear advantages over back propagation and LSTM.The practical applications show that optimizing the force of side support roll system can reduce the high point of high-strength strip flatness from 13.2 to 6 IU and decrease the percentage of low-strength strip flatness defects from 1.6%to 1.2%.This optimization greatly reduced the proportion of flatness defects,improved the accuracy level of flatness control of precision rolling mill,and provided a guarantee for the stable production of thin strip.
文摘This paper deals with an important subject of rejecting the abnormal dataintelligently in the dynamic data system.Based on the principle of nearest neighbor of fuz-zy mathematics,an approach of mathematically abstracting the human thinking and phys-ical practice knowledge is discussed,a new method of automatic rejection of abnormal da-ta is then proposed.The experimental results show that the method is available to the practice.
文摘A compound relay protection operation criterion is proposed based on Kirchhoff’s Current Law and abnormal data detecting algorithm. The abnormal data detecting algorithm are proposed after deep research on the abnormal data properties. The current transformer status monitoring system and current phase angle detecting system is introduced. A simulation model containing different power sources and loads is established in Matlab. The simulation results show that this compound criterion can work quickly and reliably in all conditions.
基金supported by Basic Science Research Program through the NationalResearch Foundation of Korea (NRF)funded by the Ministry of Education (2020R1A6A1A03040583).
文摘With the increasing number of digital devices generating a vast amount of video data,the recognition of abnormal image patterns has become more important.Accordingly,it is necessary to develop a method that achieves this task using object and behavior information within video data.Existing methods for detecting abnormal behaviors only focus on simple motions,therefore they cannot determine the overall behavior occurring throughout a video.In this study,an abnormal behavior detection method that uses deep learning(DL)-based video-data structuring is proposed.Objects and motions are first extracted from continuous images by combining existing DL-based image analysis models.The weight of the continuous data pattern is then analyzed through data structuring to classify the overall video.The performance of the proposed method was evaluated using varying parameter settings,such as the size of the action clip and interval between action clips.The model achieved an accuracy of 0.9817,indicating excellent performance.Therefore,we conclude that the proposed data structuring method is useful in detecting and classifying abnormal behaviors.
文摘As cyber attacks increase in volume and complexity,it becomes more and more difficult for existing analytical tools to detect previously unseen malware.This paper proposes a cooperative framework to leverage the robustness of big data analytics and the power of ensemble learning techniques to detect the abnormal behavior.In addition to this proposal,we implement a large scale network abnormal traffic behavior detection system performed by the framework.The proposed model detects the abnormal behavior from large scale network traffic data using a combination of a balanced decomposition algorithm and an ensemble SVM.First,the collected dataset is divided into k subsets based on the similarity between patterns using a parallel map reduce k-means algorithm.Then,patterns are randomly selected from each cluster and balanced training sub datasets are formed.Next,the subsets are fed into the mappers to build an SVM model.The construction of the ensemble is achieved in the reduce phase.The proposed structure closely delivers a high accuracy as the number of iterations increases.Experimental results show a promising gain in detection rate and false alarm compared with other existing models.
文摘With the changes in educational models,applying computer algorithms and artificial intelligence technologies to data analysis in universities has become a research hotspot in the field of intelligent education.In response to the increasing amount of student data in universities,this study proposes to use an optimized isolated forest algorithm for recognizing features to detect abnormal student behavior concealed in big data for educational management.Firstly,it uses a logistic regression algorithm to update the calculation method of isolated forest weights and then uses residual statistics to eliminate redundant forests.Finally,it utilizes discrete particle swarm optimization to optimize the isolated forest algorithm.On this basis,improvements have also been made to the traditional gated loop unit network.It merges the two improved algorithm models and builds an anomaly detection model for collecting college student education data.The experiment shows that the optimized isolated forest algorithm has a recognition accuracy of 0.986 and a training time of 1s.The recognition accuracy of the improved gated loop unit network is 0.965,and the training time is 0.16s.In summary,the constructed model can effectively identify abnormal data of college students,thereby helping educators to detect students’problems in time and helping students to improve their learning status.
基金Supported by National Natural Science Foundation of China(Grant No.51175316)Specialized Research Fund for the Doctoral Program of Higher Education,China(Grant No.20103108110006)Basic Research Project of Shanghai Science and Technology Commission,China(Grant No.11JC1404100)
文摘Online monitoring methods have been widely used in many major devices, however the normal and abnormal states of equipment are estimated mainly based on the monitoring results whether monitored parameters exceed the setting thresholds. Using these monitoring methods may cause serious false positive or false negative results. In order to precisely monitor the state of equipment, the problem of abnormality degree detection without fault sample is studied with a new detection method called negative potential field group detectors（NPFG-detectors）. This method achieves the quantitative expression of abnormality degree and provides the better detection results compared with other methods. In the process of Iris data set simulation, the new algorithm obtains the successful results in abnormal detection. The detection rates for 3 types of Iris data set respectively reach 100%, 91.6%, and 95.24% with 50% training samples. The problem of Bearing abnormality degree detection via an abnormality degree curve is successfully solved.
基金This project was sponsored by the Science and Technology Programof Sichuan Province (05SG031-004)
文摘In the study of the application of MODIS satellite remote sensing data to earthquake prediction, the paper puts forward for the first time a quantitative method to estimate the ratio for the pixels with abnormal brightness temperature (BT) increase and a preliminary scheme for cloud removal. The principle is that, firstly, the cloudless data observed by the same satellite at the same period of time but in different days (usually 1 day to 3 days) are mosaiced to get high ratio of clear sky, and then the BT variation curve and mean square difference (MSD) of each pixel are calculated with the data from the covered area to determine daily whether the BT data of the day is normal or not at a certain pixel by using double the MSD as the criterion. The ratio for the pixels with abnormal BT increase can be calculated by dividing the total number of abnormal pixels with the total pixels of the whole area. Analysis on a series of recent earthquakes in the Taiwan Region shows that the ratio for pixels with abnormal BT increase, which normally undulates around zero, has a sudden enhancement 1 day to 20 days before medium-strong earthquakes. It is expected that a new method for identifying earthquake auspice could be found through special studies in regions with frequent seismic activity by analyzing the change of the ratio for the pixels with abnormal BT increase from MODIS satellite remote sensing infrared (IR) information from which the effect of clouds has been removed to a certain extent.
基金sponsored by the National Key Technology R&D Program of China(2009BAK55B00)the Earthquake Industry Research Project(201508012)
文摘The noise data in vertical component records of 85 seismic stations in Fujian Province during 2012 is used as the research object in this paper. The noise data is divided into fiveminute segments to calculate the power spectra. The high reference line and low reference line of station are then identified by drawing a probability density function graph( PDF)using the power spectral probability density function. Moreover, according to the anomalies of PDF graphs in 85 seismic stations,the abnormal noise is divided into four categories: dropped packet, low noise, high noise, and median noise anomalies.Afterwards,four selection methods are found by the high or low noise reference line of the stations,and the system of real-time monitoring of seismic noise is formed by combining the four selection methods. Noise records of 85 seismic stations in Fujian Province in July2013 are selected for verification,and the results show that the anomalous noise-recognition system could reach a 90% success rate at most stations and the effect of selection are very good. Therefore,it could be applied to the seismic noise real-time monitoring in stations.
文摘电网系统运行大安全管理数据规模较大,传统方法难以准确且快速地挖掘出异常数据。因此,文章提出了基于大语言模型(Large Language Model, LLM)的电网系统运行大安全管理异常数据挖掘。该方法在清洗并脱敏处理原始电网系统运行大安全管理数据后,形成结构化的电网系统运行大安全管理数据集。通过构建LLM,该方法将结构化数据集输入模型,经训练后输出异常数据挖掘结果。实验分析表明,该方法数据挖掘结果的相对平方根误差仅为0.47%,交叉熵损失收敛值低至2.2146,显著优于传统聚类与小波方法,可为电网安全运行提供高效、可靠的异常监测支持。
