Deep Learning has been widely used to model soft sensors in modern industrial processes with nonlinear variables and uncertainty.Due to the outstanding ability for high-level feature extraction,stacked autoencoder(SAE...Deep Learning has been widely used to model soft sensors in modern industrial processes with nonlinear variables and uncertainty.Due to the outstanding ability for high-level feature extraction,stacked autoencoder(SAE)has been widely used to improve the model accuracy of soft sensors.However,with the increase of network layers,SAE may encounter serious information loss issues,which affect the modeling performance of soft sensors.Besides,there are typically very few labeled samples in the data set,which brings challenges to traditional neural networks to solve.In this paper,a multi-scale feature fused stacked autoencoder(MFF-SAE)is suggested for feature representation related to hierarchical output,where stacked autoencoder,mutual information(MI)and multi-scale feature fusion(MFF)strategies are integrated.Based on correlation analysis between output and input variables,critical hidden variables are extracted from the original variables in each autoencoder's input layer,which are correspondingly given varying weights.Besides,an integration strategy based on multi-scale feature fusion is adopted to mitigate the impact of information loss with the deepening of the network layers.Then,the MFF-SAE method is designed and stacked to form deep networks.Two practical industrial processes are utilized to evaluate the performance of MFF-SAE.Results from simulations indicate that in comparison to other cutting-edge techniques,the proposed method may considerably enhance the accuracy of soft sensor modeling,where the suggested method reduces the root mean square error(RMSE)by 71.8%,17.1%and 64.7%,15.1%,respectively.展开更多
Coking at the fractionating tower bottom and the decant oil circulation system disrupts the heat balance,leading to unplanned shutdown and destroying the long period stable operation of the Fluid Catalytic Cracking Un...Coking at the fractionating tower bottom and the decant oil circulation system disrupts the heat balance,leading to unplanned shutdown and destroying the long period stable operation of the Fluid Catalytic Cracking Unit(FCCU).The FCCU operates through interconnected subsystems,generating high-dimensional,nonlinear,and non-stationary data characterized by spatiotemporally correlated.The decant oil solid content is the crucial indicator for monitoring catalyst loss from the reactor-regenerator system and coking risk tendency at the fractionating tower bottom that relies on sampling and laboratory testing,which is lagging responsiveness and labor-intensive.Developing the online decant oil solid content soft sensor using industrial data to support operators in conducting predictive maintenance is essential.Therefore,this paper proposes a hybrid deep learning framework for soft sensor development that combines spatiotemporal pattern extraction with interpretability,enabling accurate risk identification in dynamic operational conditions.This framework employs a Filter-Wrapper method for dimensionality reduction,followed by a 2D Convolutional Neural Network(2DCNN)for extracting spatial patterns,and a Bidirectional Gated Recurrent Unit(BiGRU)for capturing long-term temporal dependencies,with an Attention Mechanism(AM)to highlight critical features adaptively.The integration of SHapley Additive exPlanations(SHAP),Complementary Ensemble Empirical Mode Decomposition with Adaptive Noise(CEEMDAN),2DCNN,and expert knowledge precisely quantifies feature contributions and decomposes signals,significantly enhancing the practicality of risk identification.Applied to a China refinery with processing capacity of 2.80×10^(6) t/a,the soft sensor achieved the R^(2) value of 0.93 and five-level risk identification accuracy of 96.42%.These results demonstrate the framework's accuracy,robustness,and suitability for complex industrial scenarios,advancing risk visualization and management.展开更多
In order to solve the problem of the invalidation of thermal parameters andoptimal running, we present an efficient soft sensor approach based on sparse online Gaussianprocesses( GP), which is based on a combination o...In order to solve the problem of the invalidation of thermal parameters andoptimal running, we present an efficient soft sensor approach based on sparse online Gaussianprocesses( GP), which is based on a combination of a Bayesian online algorithm together with asequential construction of a relevant subsample of the data to specify the prediction of the GPmodel. By an appealing parameterization and projection techniques that use the reproducing kernelHubert space (RKHS) norm, recursions for the effective parameters and a sparse Gaussianapproximation of the posterior process are obtained. The sparse representation of Gaussian processesmakes the GP-based soft sensor practical in a large dataset and real-time application. And theproposed thermalparameter soft sensor is of importance for the economical running of the powerplant.展开更多
Presented is a multiple model soft sensing method based on Affinity Propagation （AP）, Gaussian process （GP） and Bayesian committee machine （BCM）. AP clustering arithmetic is used to cluster training samples acco...Presented is a multiple model soft sensing method based on Affinity Propagation （AP）, Gaussian process （GP） and Bayesian committee machine （BCM）. AP clustering arithmetic is used to cluster training samples according to their operating points. Then, the sub-models are estimated by Gaussian Process Regression （GPR）. Finally, in order to get a global probabilistic prediction, Bayesian committee mactnne is used to combine the outputs of the sub-estimators. The proposed method has been applied to predict the light naphtha end point in hydrocracker fractionators. Practical applications indicate that it is useful for the online prediction of quality monitoring in chemical processes.展开更多
Multi-model approach can significantly improve the prediction performance of soft sensors in the process with multiple operational conditions.However,traditional clustering algorithms may result in overlapping phenome...Multi-model approach can significantly improve the prediction performance of soft sensors in the process with multiple operational conditions.However,traditional clustering algorithms may result in overlapping phenomenon in subclasses,so that edge classes and outliers cannot be effectively dealt with and the modeling result is not satisfactory.In order to solve these problems,a new feature extraction method based on weighted kernel Fisher criterion is presented to improve the clustering accuracy,in which feature mapping is adopted to bring the edge classes and outliers closer to other normal subclasses.Furthermore,the classified data are used to develop a multiple model based on support vector machine.The proposed method is applied to a bisphenol A production process for prediction of the quality index.The simulation results demonstrate its ability in improving the data classification and the prediction performance of the soft sensor.展开更多
The ammonia synthesis reactor is the core unit in the whole ammonia synthesis production. The ammonia concentration at the ammonia converter outlet is a significant process variable, which reflects directly the produc...The ammonia synthesis reactor is the core unit in the whole ammonia synthesis production. The ammonia concentration at the ammonia converter outlet is a significant process variable, which reflects directly the production efficiency. However, it is hard to be measured reliably online in real applications. In this paper, a soft sensor based on BP neural network (BPNN) is applied to estimate the ammonia concentration. A modified group search optimization with nearest neighborhood (GSO-NH) is proposed to optimize the weights and thresholds of BPNN. GSO-NH is integrated with BPNN to build a soft sensor model. Finally, the soft sensor model based on BPNN and GSO-NH (GSO-NH-NN) is used to infer the outlet ammonia concentration in a real-world application. Three other modeling methods are applied for comparison with GSO-NH-NN. The results show that the soft sensor based on GSO-NH-NN has a good prediction performance with high accuracy. Moreover, the GSO-NH-NN also provides good generalization ability to other modeling problems in ammonia synthesis production.展开更多
A novel data-driven, soft sensor based on support vector regression （SVR） integrated with a data compression technique was developed to predict the product quality for the hydrodesulfurization （HDS） process. A wid...A novel data-driven, soft sensor based on support vector regression （SVR） integrated with a data compression technique was developed to predict the product quality for the hydrodesulfurization （HDS） process. A wide range of experimental data was taken from a HDS setup to train and test the SVR model. Hyper-parameter tuning is one of the main challenges to improve predictive accuracy of the SVR model. Therefore, a hybrid approach using a combination of genetic algorithm （GA） and sequential quadratic programming （SQP） methods （GA-SQP） was developed. Performance of different optimization algorithms including GA-SQP, GA, pattern search （PS）, and grid search （GS） indicated that the best average absolute relative error （AARE）, squared correlation coefficient （R2）, and computation time （CT） （AARE = 0.0745, R2 = 0.997 and CT = 56 s） was accomplished by the hybrid algorithm. Moreover, to reduce the CT and improve the accuracy of the SVR model, the vector quantization （VQ） technique was used. The results also showed that the VQ technique can decrease the training time and improve prediction performance of the SVR model. The proposed method can provide a robust, soft sensor in a wide range of sulfur contents with good accuracy.展开更多
The processes of building dynamic and static relationships between secondary and primary variables are usually integrated in most of nonlinear dynamic soft sensor models. However, such integration limits the estimatio...The processes of building dynamic and static relationships between secondary and primary variables are usually integrated in most of nonlinear dynamic soft sensor models. However, such integration limits the estimation accuracy of soft sensor models. Wiener model effectively describes dynamic and static characteristics of a system with the structure of dynamic and static submodels in cascade. We propose a soft sensor model derived from Wiener model structure, which is an extension of Wiener model. Dynamic and static relationships between secondary and primary variables are built respectively to describe the dynamic and static characteristics of system. The feasibility of this model is verified. Then the expression of discrete model is derived for soft sensor system. Conjugate gradient algorithm is applied to identify the dynamic and static model parameters alternately. Corresponding update method for soft sensor system is also given. Case studies confirm the effectiveness of the proposed model, alternate identification algorithm, and update method.展开更多
A novel method for developing a reliable data driven soft sensor to improve the prediction accuracy of sulfur content in hydrodesulfurization(HDS) process was proposed. Therefore, an integrated approach using support ...A novel method for developing a reliable data driven soft sensor to improve the prediction accuracy of sulfur content in hydrodesulfurization(HDS) process was proposed. Therefore, an integrated approach using support vector regression(SVR) based on wavelet transform(WT) and principal component analysis(PCA) was used. Experimental data from the HDS setup were employed to validate the proposed model. The results reveal that the integrated WT-PCA with SVR model was able to increase the prediction accuracy of SVR model. Implementation of the proposed model delivers the best satisfactory predicting performance(EAARE=0.058 and R2=0.97) in comparison with SVR. The obtained results indicate that the proposed model is more reliable and more precise than the multiple linear regression(MLR), SVR and PCA-SVR.展开更多
Chemical processes are usually nonlinear singular systems.In this study,a soft sensor using nonlinear singular state observer is established for unknown inputs and uncertain model parameters in chemical processes,whic...Chemical processes are usually nonlinear singular systems.In this study,a soft sensor using nonlinear singular state observer is established for unknown inputs and uncertain model parameters in chemical processes,which are augmented as state variables.Based on the observability of the singular system,this paper presents a simplified observability criterion under certain conditions for unknown inputs and uncertain model parameters.When the observability is satisfied,the unknown inputs and the uncertain model parameters are estimated online by the soft sensor using augmented nonlinear singular state observer.The riser reactor of fluid catalytic cracking unit is used as an example for analysis and simulation.With the catalyst circulation rate as the only unknown input without model error,one temperature sensor at the riser reactor outlet will ensure the correct estimation for the catalyst circulation rate.However,when uncertain model parameters also exist,additional temperature sensors must be used to ensure correct estimation for unknown inputs and uncertain model parameters of chemical processes.展开更多
In order to meet the demand of online optimal running, a novel soft sensor modeling approach based on Gaussian processes was proposed. The approach is moderately simple to implement and use without loss of performance...In order to meet the demand of online optimal running, a novel soft sensor modeling approach based on Gaussian processes was proposed. The approach is moderately simple to implement and use without loss of performance. It is trained by optimizing the hyperparameters using the scaled conjugate gradient algorithm with the squared exponential covariance function employed. Experimental simulations show that the soft sensor modeling approach has the advantage via a real-world example in a refinery. Meanwhile, the method opens new possibilities for application of kernel methods to potential fields.展开更多
There is a certain extent of measuring error for each pyrometer in plate production process,which has a great impact on process control system.A soft sensor technique is used to calculate the confidence interval of me...There is a certain extent of measuring error for each pyrometer in plate production process,which has a great impact on process control system.A soft sensor technique is used to calculate the confidence interval of measuring temperature data from pyrometers and an inference estimator is designed by partial least square method,then the pyrometer with highest confidence is considered as the reference one,with which inference estimation for measured data on-line of other pyrometers is started and the measuring error is conformed.From actual application it can be found that this method can improve the measuring accuracy by 3%-5%,and has high field application value.展开更多
The mechanical properties of hot rolled strip are the key index of product quality,and the soft sensing of them is an important decision basis for the control and optimization of hot rolling process.To solve the probl...The mechanical properties of hot rolled strip are the key index of product quality,and the soft sensing of them is an important decision basis for the control and optimization of hot rolling process.To solve the problem that it is difficult to measure the mechanical properties of hot rolled strip in time and accurately,a soft sensor based on ensemble local modeling was proposed.Firstly,outliers of process data are removed by local outlier factor.After standardization and transformation,normal data that can be used in the model are obtained.Next,in order to avoid redundant variables participating in modeling and reducing performance of models,feature selection was applied combing the mechanism of hot rolling process and mutual information among variables.Then,features of samples were extracted by supervised local preserving projection,and a prediction model was constructed by Gaussian process regression based on just-in-time learning(JITL).Other JITL-based models,such as support vector regression and gradient boosting regression tree models,keep all variables and make up for the lost information during dimension reduction.Finally,the soft sensor was developed by integrating individual models through stacking method.Superiority and reliability of proposed soft sensors were verified by actual process data from a real hot rolling process.展开更多
To explore the problems of monitoring chemical processes with large numbers of input parameters, a method based on Auto-associative Hierarchical Neural Network(AHNN) is proposed. AHNN focuses on dealing with datasets ...To explore the problems of monitoring chemical processes with large numbers of input parameters, a method based on Auto-associative Hierarchical Neural Network(AHNN) is proposed. AHNN focuses on dealing with datasets in high-dimension. AHNNs consist of two parts: groups of subnets based on well trained Autoassociative Neural Networks(AANNs) and a main net. The subnets play an important role on the performance of AHNN. A simple but effective method of designing the subnets is developed in this paper. In this method,the subnets are designed according to the classification of the data attributes. For getting the classification, an effective method called Extension Data Attributes Classification(EDAC) is adopted. Soft sensor using AHNN based on EDAC(EDAC-AHNN) is introduced. As a case study, the production data of Purified Terephthalic Acid(PTA) solvent system are selected to examine the proposed model. The results of the EDAC-AHNN model are compared with the experimental data extracted from the literature, which shows the efficiency of the proposed model.展开更多
This work proposes a soft sensor based on a phenomenological model for online estimation of the density and viscosity of a slurry flowing through a pipe-and-fittings assembly(PFA). The model is developed considering t...This work proposes a soft sensor based on a phenomenological model for online estimation of the density and viscosity of a slurry flowing through a pipe-and-fittings assembly(PFA). The model is developed considering the conservation principle applied to mass and momentum transfer and considering frictional energy losses to include the variables directly affecting slurry properties. A reported proposal for state observers with unknown inputs is used to develop the first block of the observer structure. The second block is constructed with two options for evaluating slurry viscosity, generating two possible estimator structures, which are tested using real data. A comparison between them indicates different uses and capabilities according to available process information.展开更多
Soft sensor is attractive in dealing with online product quality measurement by virtue of other easily measured variables. In AMOCO PTA (purified terephthalic acid) production process, the unavailability of real-time ...Soft sensor is attractive in dealing with online product quality measurement by virtue of other easily measured variables. In AMOCO PTA (purified terephthalic acid) production process, the unavailability of real-time measurement of 4-CBA makes it impossible for timely adjustment and thereby influences the product quality and the plant economy benefit. In this paper, a kind of FCMAC (fuzzy cerebellar model articulation controller) method is presented to solve the online measurement problem. Different from the conventional CMAC (cerebellar model articulation controller) networks, which has inferior smoothing ability because of its table look-up based technology. Integrating fuzzy model into CMAC networks, it becomes more accurate in functional mapping without weakening its generalization ability. Numerical example and industrial application results show the method proposed here is satisfactory and feasible.展开更多
In soft sensor field, just-in-time learning(JITL) is an effective approach to model nonlinear and time varying processes. However, most similarity criterions in JITL are computed in the input space only while ignoring...In soft sensor field, just-in-time learning(JITL) is an effective approach to model nonlinear and time varying processes. However, most similarity criterions in JITL are computed in the input space only while ignoring important output information, which may lead to inaccurate construction of relevant sample set. To solve this problem, we propose a novel supervised feature extraction method suitable for the regression problem called supervised local and non-local structure preserving projections(SLNSPP), in which both input and output information can be easily and effectively incorporated through a newly defined similarity index. The SLNSPP can not only retain the virtue of locality preserving projections but also prevent faraway points from nearing after projection,which endues SLNSPP with powerful discriminating ability. Such two good properties of SLNSPP are desirable for JITL as they are expected to enhance the accuracy of similar sample selection. Consequently, we present a SLNSPP-JITL framework for developing adaptive soft sensor, including a sparse learning strategy to limit the scale and update the frequency of database. Finally, two case studies are conducted with benchmark datasets to evaluate the performance of the proposed schemes. The results demonstrate the effectiveness of LNSPP and SLNSPP.展开更多
An improved immune algorithm is proposed in this paper. The problems, such as convergence speed and optimization precision, existing in the basic immune algorithm are well addressed. Besides, a fuzzy adaptive method i...An improved immune algorithm is proposed in this paper. The problems, such as convergence speed and optimization precision, existing in the basic immune algorithm are well addressed. Besides, a fuzzy adaptive method is presented by using the fuzzy system to realize the adaptive selection of two key parameters （possibility of crossover and mutation）. By comparing and analyzing the results of several benchmark functions, the performance of fuzzy immune algorithm （FIA） is approved. Not only the difficulty of parameters selection is relieved, but also the precision and stability are improved. At last, the FIA is ap- plied to optimization of the structure and parameters in radial basis function neural network （RBFNN） based on an orthogonal sequential method. And the availability of algorithm is proved by applying RBFNN in modeling in soft sensor of solvent tower.展开更多
Data-driven soft sensor is an effective solution to provide rapid and reliable estimations for key quality variables online. The secondary variables affect the primary variable in considerably different speed, and sof...Data-driven soft sensor is an effective solution to provide rapid and reliable estimations for key quality variables online. The secondary variables affect the primary variable in considerably different speed, and soft sensor systems exhibit multi-dynamic characteristics. Thus, the first contribution is improving the model in the previous study with multi-time-constant. The characteristics-separation-based model will be identified in substep way,and the stochastic Newton recursive(SNR) algorithm is adopted. Considering the dual-rate characteristics of soft sensor systems, the proposed model cannot be identified directly. Thus, two auxiliary models are first proposed to offer the intersample estimations at each update period, based on which the improved algorithm(DAM-SNR) is derived. These two auxiliary models function in switching mechanism which has been illustrated in detail. This algorithm serves for the identification of the proposed model together with the SNR algorithm, and the identification procedure is then presented. Finally, the laboratorial case confirms the effectiveness of the proposed soft sensor model and the algorithms.展开更多
In cases of measurement,like the beverage discrimination or waste water monitoring,since sensors with direct measuring for specific variables cannot be practically applied,the interest in soft sensor technology,which ...In cases of measurement,like the beverage discrimination or waste water monitoring,since sensors with direct measuring for specific variables cannot be practically applied,the interest in soft sensor technology,which is based on the acquisition and analysis for a lot of information variables,has rapidly increased.This article presents a novel soft sensor which activates a voltage signal between two specially-designed electrodes,inside the Chinese tea solution,and measures the produced current.The proposed sensor discriminated four different kinds of tea,and the dissolution processes of two of them are described as curves on a bidimensional plane.The results indicate that the proposed approach can also be adapted in characterization or discrimination for other solutions.展开更多
基金supported by the National Key Research and Development Program of China(2023YFB3307800)National Natural Science Foundation of China(62394343,62373155)+2 种基金Major Science and Technology Project of Xinjiang(No.2022A01006-4)State Key Laboratory of Industrial Control Technology,China(Grant No.ICT2024A26)Fundamental Research Funds for the Central Universities.
文摘Deep Learning has been widely used to model soft sensors in modern industrial processes with nonlinear variables and uncertainty.Due to the outstanding ability for high-level feature extraction,stacked autoencoder(SAE)has been widely used to improve the model accuracy of soft sensors.However,with the increase of network layers,SAE may encounter serious information loss issues,which affect the modeling performance of soft sensors.Besides,there are typically very few labeled samples in the data set,which brings challenges to traditional neural networks to solve.In this paper,a multi-scale feature fused stacked autoencoder(MFF-SAE)is suggested for feature representation related to hierarchical output,where stacked autoencoder,mutual information(MI)and multi-scale feature fusion(MFF)strategies are integrated.Based on correlation analysis between output and input variables,critical hidden variables are extracted from the original variables in each autoencoder's input layer,which are correspondingly given varying weights.Besides,an integration strategy based on multi-scale feature fusion is adopted to mitigate the impact of information loss with the deepening of the network layers.Then,the MFF-SAE method is designed and stacked to form deep networks.Two practical industrial processes are utilized to evaluate the performance of MFF-SAE.Results from simulations indicate that in comparison to other cutting-edge techniques,the proposed method may considerably enhance the accuracy of soft sensor modeling,where the suggested method reduces the root mean square error(RMSE)by 71.8%,17.1%and 64.7%,15.1%,respectively.
基金supported by the Innovative Research Group Project of the National Natural Science Foundation of China(22021004)Sinopec Major Science and Technology Projects(321123-1)。
文摘Coking at the fractionating tower bottom and the decant oil circulation system disrupts the heat balance,leading to unplanned shutdown and destroying the long period stable operation of the Fluid Catalytic Cracking Unit(FCCU).The FCCU operates through interconnected subsystems,generating high-dimensional,nonlinear,and non-stationary data characterized by spatiotemporally correlated.The decant oil solid content is the crucial indicator for monitoring catalyst loss from the reactor-regenerator system and coking risk tendency at the fractionating tower bottom that relies on sampling and laboratory testing,which is lagging responsiveness and labor-intensive.Developing the online decant oil solid content soft sensor using industrial data to support operators in conducting predictive maintenance is essential.Therefore,this paper proposes a hybrid deep learning framework for soft sensor development that combines spatiotemporal pattern extraction with interpretability,enabling accurate risk identification in dynamic operational conditions.This framework employs a Filter-Wrapper method for dimensionality reduction,followed by a 2D Convolutional Neural Network(2DCNN)for extracting spatial patterns,and a Bidirectional Gated Recurrent Unit(BiGRU)for capturing long-term temporal dependencies,with an Attention Mechanism(AM)to highlight critical features adaptively.The integration of SHapley Additive exPlanations(SHAP),Complementary Ensemble Empirical Mode Decomposition with Adaptive Noise(CEEMDAN),2DCNN,and expert knowledge precisely quantifies feature contributions and decomposes signals,significantly enhancing the practicality of risk identification.Applied to a China refinery with processing capacity of 2.80×10^(6) t/a,the soft sensor achieved the R^(2) value of 0.93 and five-level risk identification accuracy of 96.42%.These results demonstrate the framework's accuracy,robustness,and suitability for complex industrial scenarios,advancing risk visualization and management.
文摘In order to solve the problem of the invalidation of thermal parameters andoptimal running, we present an efficient soft sensor approach based on sparse online Gaussianprocesses( GP), which is based on a combination of a Bayesian online algorithm together with asequential construction of a relevant subsample of the data to specify the prediction of the GPmodel. By an appealing parameterization and projection techniques that use the reproducing kernelHubert space (RKHS) norm, recursions for the effective parameters and a sparse Gaussianapproximation of the posterior process are obtained. The sparse representation of Gaussian processesmakes the GP-based soft sensor practical in a large dataset and real-time application. And theproposed thermalparameter soft sensor is of importance for the economical running of the powerplant.
基金Supported by the National High Technology Research and Development Program of China (2006AA040309)National BasicResearch Program of China (2007CB714000)
文摘Presented is a multiple model soft sensing method based on Affinity Propagation （AP）, Gaussian process （GP） and Bayesian committee machine （BCM）. AP clustering arithmetic is used to cluster training samples according to their operating points. Then, the sub-models are estimated by Gaussian Process Regression （GPR）. Finally, in order to get a global probabilistic prediction, Bayesian committee mactnne is used to combine the outputs of the sub-estimators. The proposed method has been applied to predict the light naphtha end point in hydrocracker fractionators. Practical applications indicate that it is useful for the online prediction of quality monitoring in chemical processes.
基金Supported by the National Natural Science Foundation of China(61273070)the Foundation of Priority Academic Program Development of Jiangsu Higher Education Institutions
文摘Multi-model approach can significantly improve the prediction performance of soft sensors in the process with multiple operational conditions.However,traditional clustering algorithms may result in overlapping phenomenon in subclasses,so that edge classes and outliers cannot be effectively dealt with and the modeling result is not satisfactory.In order to solve these problems,a new feature extraction method based on weighted kernel Fisher criterion is presented to improve the clustering accuracy,in which feature mapping is adopted to bring the edge classes and outliers closer to other normal subclasses.Furthermore,the classified data are used to develop a multiple model based on support vector machine.The proposed method is applied to a bisphenol A production process for prediction of the quality index.The simulation results demonstrate its ability in improving the data classification and the prediction performance of the soft sensor.
基金Supported by the National Natural Science Foundation of China (61074079)Shanghai Leading Academic Discipline Project(B504)+1 种基金Specialized Research Fund for the Doctoral Program of Higher Education of China (20100074120010)the Natural Science Foundation of Shanghai City (11ZR1409700)
文摘The ammonia synthesis reactor is the core unit in the whole ammonia synthesis production. The ammonia concentration at the ammonia converter outlet is a significant process variable, which reflects directly the production efficiency. However, it is hard to be measured reliably online in real applications. In this paper, a soft sensor based on BP neural network (BPNN) is applied to estimate the ammonia concentration. A modified group search optimization with nearest neighborhood (GSO-NH) is proposed to optimize the weights and thresholds of BPNN. GSO-NH is integrated with BPNN to build a soft sensor model. Finally, the soft sensor model based on BPNN and GSO-NH (GSO-NH-NN) is used to infer the outlet ammonia concentration in a real-world application. Three other modeling methods are applied for comparison with GSO-NH-NN. The results show that the soft sensor based on GSO-NH-NN has a good prediction performance with high accuracy. Moreover, the GSO-NH-NN also provides good generalization ability to other modeling problems in ammonia synthesis production.
文摘A novel data-driven, soft sensor based on support vector regression （SVR） integrated with a data compression technique was developed to predict the product quality for the hydrodesulfurization （HDS） process. A wide range of experimental data was taken from a HDS setup to train and test the SVR model. Hyper-parameter tuning is one of the main challenges to improve predictive accuracy of the SVR model. Therefore, a hybrid approach using a combination of genetic algorithm （GA） and sequential quadratic programming （SQP） methods （GA-SQP） was developed. Performance of different optimization algorithms including GA-SQP, GA, pattern search （PS）, and grid search （GS） indicated that the best average absolute relative error （AARE）, squared correlation coefficient （R2）, and computation time （CT） （AARE = 0.0745, R2 = 0.997 and CT = 56 s） was accomplished by the hybrid algorithm. Moreover, to reduce the CT and improve the accuracy of the SVR model, the vector quantization （VQ） technique was used. The results also showed that the VQ technique can decrease the training time and improve prediction performance of the SVR model. The proposed method can provide a robust, soft sensor in a wide range of sulfur contents with good accuracy.
基金Supported by the National Natural Science Foundation of China(61104218,21006127)the National Basic Research Program of China(2012CB720500)the Science Foundation of China University of Petroleum(YJRC-2013-12)
文摘The processes of building dynamic and static relationships between secondary and primary variables are usually integrated in most of nonlinear dynamic soft sensor models. However, such integration limits the estimation accuracy of soft sensor models. Wiener model effectively describes dynamic and static characteristics of a system with the structure of dynamic and static submodels in cascade. We propose a soft sensor model derived from Wiener model structure, which is an extension of Wiener model. Dynamic and static relationships between secondary and primary variables are built respectively to describe the dynamic and static characteristics of system. The feasibility of this model is verified. Then the expression of discrete model is derived for soft sensor system. Conjugate gradient algorithm is applied to identify the dynamic and static model parameters alternately. Corresponding update method for soft sensor system is also given. Case studies confirm the effectiveness of the proposed model, alternate identification algorithm, and update method.
文摘A novel method for developing a reliable data driven soft sensor to improve the prediction accuracy of sulfur content in hydrodesulfurization(HDS) process was proposed. Therefore, an integrated approach using support vector regression(SVR) based on wavelet transform(WT) and principal component analysis(PCA) was used. Experimental data from the HDS setup were employed to validate the proposed model. The results reveal that the integrated WT-PCA with SVR model was able to increase the prediction accuracy of SVR model. Implementation of the proposed model delivers the best satisfactory predicting performance(EAARE=0.058 and R2=0.97) in comparison with SVR. The obtained results indicate that the proposed model is more reliable and more precise than the multiple linear regression(MLR), SVR and PCA-SVR.
基金Supported by the National Natural Science Foundation of China （21006127）, the National Basic Research Program of China （2012CB720500） and the Science Foundation of China University of Petroleum, Beijing （KYJJ2012-05-28）.
文摘Chemical processes are usually nonlinear singular systems.In this study,a soft sensor using nonlinear singular state observer is established for unknown inputs and uncertain model parameters in chemical processes,which are augmented as state variables.Based on the observability of the singular system,this paper presents a simplified observability criterion under certain conditions for unknown inputs and uncertain model parameters.When the observability is satisfied,the unknown inputs and the uncertain model parameters are estimated online by the soft sensor using augmented nonlinear singular state observer.The riser reactor of fluid catalytic cracking unit is used as an example for analysis and simulation.With the catalyst circulation rate as the only unknown input without model error,one temperature sensor at the riser reactor outlet will ensure the correct estimation for the catalyst circulation rate.However,when uncertain model parameters also exist,additional temperature sensors must be used to ensure correct estimation for unknown inputs and uncertain model parameters of chemical processes.
文摘In order to meet the demand of online optimal running, a novel soft sensor modeling approach based on Gaussian processes was proposed. The approach is moderately simple to implement and use without loss of performance. It is trained by optimizing the hyperparameters using the scaled conjugate gradient algorithm with the squared exponential covariance function employed. Experimental simulations show that the soft sensor modeling approach has the advantage via a real-world example in a refinery. Meanwhile, the method opens new possibilities for application of kernel methods to potential fields.
基金Item Sponsored by National Natural Science Foundation of China(50104004)
文摘There is a certain extent of measuring error for each pyrometer in plate production process,which has a great impact on process control system.A soft sensor technique is used to calculate the confidence interval of measuring temperature data from pyrometers and an inference estimator is designed by partial least square method,then the pyrometer with highest confidence is considered as the reference one,with which inference estimation for measured data on-line of other pyrometers is started and the measuring error is conformed.From actual application it can be found that this method can improve the measuring accuracy by 3%-5%,and has high field application value.
基金the National Natural Science Foundation of China(NSFC)under Grants 61773053,61873024Fundamental Research Funds for the China Central Universities of USTB(FRF-TP-19-049A1Z)the National Key R&D Program of China(No.2017YFB0306403).
文摘The mechanical properties of hot rolled strip are the key index of product quality,and the soft sensing of them is an important decision basis for the control and optimization of hot rolling process.To solve the problem that it is difficult to measure the mechanical properties of hot rolled strip in time and accurately,a soft sensor based on ensemble local modeling was proposed.Firstly,outliers of process data are removed by local outlier factor.After standardization and transformation,normal data that can be used in the model are obtained.Next,in order to avoid redundant variables participating in modeling and reducing performance of models,feature selection was applied combing the mechanism of hot rolling process and mutual information among variables.Then,features of samples were extracted by supervised local preserving projection,and a prediction model was constructed by Gaussian process regression based on just-in-time learning(JITL).Other JITL-based models,such as support vector regression and gradient boosting regression tree models,keep all variables and make up for the lost information during dimension reduction.Finally,the soft sensor was developed by integrating individual models through stacking method.Superiority and reliability of proposed soft sensors were verified by actual process data from a real hot rolling process.
基金Supported by the National Natural Science Foundation of China(61074153)
文摘To explore the problems of monitoring chemical processes with large numbers of input parameters, a method based on Auto-associative Hierarchical Neural Network(AHNN) is proposed. AHNN focuses on dealing with datasets in high-dimension. AHNNs consist of two parts: groups of subnets based on well trained Autoassociative Neural Networks(AANNs) and a main net. The subnets play an important role on the performance of AHNN. A simple but effective method of designing the subnets is developed in this paper. In this method,the subnets are designed according to the classification of the data attributes. For getting the classification, an effective method called Extension Data Attributes Classification(EDAC) is adopted. Soft sensor using AHNN based on EDAC(EDAC-AHNN) is introduced. As a case study, the production data of Purified Terephthalic Acid(PTA) solvent system are selected to examine the proposed model. The results of the EDAC-AHNN model are compared with the experimental data extracted from the literature, which shows the efficiency of the proposed model.
基金Colciencias and SUMICOL(Suministros de Colombia S.A.)for their support and financing for this project
文摘This work proposes a soft sensor based on a phenomenological model for online estimation of the density and viscosity of a slurry flowing through a pipe-and-fittings assembly(PFA). The model is developed considering the conservation principle applied to mass and momentum transfer and considering frictional energy losses to include the variables directly affecting slurry properties. A reported proposal for state observers with unknown inputs is used to develop the first block of the observer structure. The second block is constructed with two options for evaluating slurry viscosity, generating two possible estimator structures, which are tested using real data. A comparison between them indicates different uses and capabilities according to available process information.
基金Supported by the special Funds for Major State Basic Research Program of China (973 Program) (No. 2002CB312200) the 863 Hi-Tech. Research and Development Program of China (No. 2001AA413130, No.2002AA412110)the Key Technologies R&D Programme of China (No. 2001BA201A04).
文摘Soft sensor is attractive in dealing with online product quality measurement by virtue of other easily measured variables. In AMOCO PTA (purified terephthalic acid) production process, the unavailability of real-time measurement of 4-CBA makes it impossible for timely adjustment and thereby influences the product quality and the plant economy benefit. In this paper, a kind of FCMAC (fuzzy cerebellar model articulation controller) method is presented to solve the online measurement problem. Different from the conventional CMAC (cerebellar model articulation controller) networks, which has inferior smoothing ability because of its table look-up based technology. Integrating fuzzy model into CMAC networks, it becomes more accurate in functional mapping without weakening its generalization ability. Numerical example and industrial application results show the method proposed here is satisfactory and feasible.
基金Supported by the National Natural Science Foundation of China(61273160)the Fundamental Research Funds for the Central Universities(14CX06067A,13CX05021A)
文摘In soft sensor field, just-in-time learning(JITL) is an effective approach to model nonlinear and time varying processes. However, most similarity criterions in JITL are computed in the input space only while ignoring important output information, which may lead to inaccurate construction of relevant sample set. To solve this problem, we propose a novel supervised feature extraction method suitable for the regression problem called supervised local and non-local structure preserving projections(SLNSPP), in which both input and output information can be easily and effectively incorporated through a newly defined similarity index. The SLNSPP can not only retain the virtue of locality preserving projections but also prevent faraway points from nearing after projection,which endues SLNSPP with powerful discriminating ability. Such two good properties of SLNSPP are desirable for JITL as they are expected to enhance the accuracy of similar sample selection. Consequently, we present a SLNSPP-JITL framework for developing adaptive soft sensor, including a sparse learning strategy to limit the scale and update the frequency of database. Finally, two case studies are conducted with benchmark datasets to evaluate the performance of the proposed schemes. The results demonstrate the effectiveness of LNSPP and SLNSPP.
文摘An improved immune algorithm is proposed in this paper. The problems, such as convergence speed and optimization precision, existing in the basic immune algorithm are well addressed. Besides, a fuzzy adaptive method is presented by using the fuzzy system to realize the adaptive selection of two key parameters （possibility of crossover and mutation）. By comparing and analyzing the results of several benchmark functions, the performance of fuzzy immune algorithm （FIA） is approved. Not only the difficulty of parameters selection is relieved, but also the precision and stability are improved. At last, the FIA is ap- plied to optimization of the structure and parameters in radial basis function neural network （RBFNN） based on an orthogonal sequential method. And the availability of algorithm is proved by applying RBFNN in modeling in soft sensor of solvent tower.
基金Supported by the Scientific Research Foundation of Shandong University of Science and Technology for Recruited Talents(2016RCJJ046)the National Basic Research Program of China(2012CB720500)
文摘Data-driven soft sensor is an effective solution to provide rapid and reliable estimations for key quality variables online. The secondary variables affect the primary variable in considerably different speed, and soft sensor systems exhibit multi-dynamic characteristics. Thus, the first contribution is improving the model in the previous study with multi-time-constant. The characteristics-separation-based model will be identified in substep way,and the stochastic Newton recursive(SNR) algorithm is adopted. Considering the dual-rate characteristics of soft sensor systems, the proposed model cannot be identified directly. Thus, two auxiliary models are first proposed to offer the intersample estimations at each update period, based on which the improved algorithm(DAM-SNR) is derived. These two auxiliary models function in switching mechanism which has been illustrated in detail. This algorithm serves for the identification of the proposed model together with the SNR algorithm, and the identification procedure is then presented. Finally, the laboratorial case confirms the effectiveness of the proposed soft sensor model and the algorithms.
文摘In cases of measurement,like the beverage discrimination or waste water monitoring,since sensors with direct measuring for specific variables cannot be practically applied,the interest in soft sensor technology,which is based on the acquisition and analysis for a lot of information variables,has rapidly increased.This article presents a novel soft sensor which activates a voltage signal between two specially-designed electrodes,inside the Chinese tea solution,and measures the produced current.The proposed sensor discriminated four different kinds of tea,and the dissolution processes of two of them are described as curves on a bidimensional plane.The results indicate that the proposed approach can also be adapted in characterization or discrimination for other solutions.
