Partial Differential Equations(PDEs)are model candidates of soft sensing for aero-engine health management units.The existing Physics-Informed Neural Networks(PINNs)have made achievements.However,unmeasurable aero-eng...Partial Differential Equations(PDEs)are model candidates of soft sensing for aero-engine health management units.The existing Physics-Informed Neural Networks(PINNs)have made achievements.However,unmeasurable aero-engine driving sources lead to unknown PDE driving terms,which weaken PINNs feasibility.To this end,Physically Informed Hierarchical Learning followed by Recurrent-Prediction Term(PIHL-RPT)is proposed.First,PIHL is proposed for learning nonhomogeneous PDE solutions,in which two networks NetU and NetG are constructed.NetU is for learning solutions satisfying PDEs;NetG is for learning driving terms to regularize NetU training.Then,we propose a hierarchical learning strategy to optimize and couple NetU and NetG,which are integrated into a data-physics-hybrid loss function.Besides,we prove PIHL-RPT can iteratively generate a series of networks converging to a function,which can approximate a solution to well-posed PDE.Furthermore,RPT is proposed for prediction improvement of PIHL,in which network NetU-RP is constructed to compensate for information loss caused by data sampling and driving sources’immeasurability.Finally,artificial datasets and practical vibration process datasets from our wear experiment platform are used to verify the feasibility and effectiveness of PIHL-RPT based soft sensing.Meanwhile,comparisons with relevant methods,discussions,and PIHL-RPT based health monitoring example are given.展开更多
Aiming at the limitations of traditional thermal model and intelligent model, a new hybrid model is established for soft sensing of the molten steel temperature in LF. Firstly, a thermal model based on energy conserva...Aiming at the limitations of traditional thermal model and intelligent model, a new hybrid model is established for soft sensing of the molten steel temperature in LF. Firstly, a thermal model based on energy conservation is described; and then, an improved intelligent model based on process data is presented by ensemble ELM （extreme learning machine） for predicting the molten steel temperature in LF. Secondly, the self-adaptive data fusion is pro- posed as a hybrid modeling method to combine the thermal model with the intelligent model. The new hybrid model could complement mutual advantage of two models by combination. It can overcome the shortcoming of parameters obtained on-line hardly in a thermal model and the disadvantage of lacking the analysis of ladle furnace metallurgical process in an intelligent model. The new hybrid model is applied to a 300 t LF in Baoshan Iron and Steel Co Ltd for predicting the molten steel temperature. The experiments demonstrate that the hybrid model has good generalization performance and high accuracy.展开更多
Local learning based soft sensing methods succeed in coping with time-varying characteristics of processes as well as nonlinearities in industrial plants. In this paper, a local partial least squares based soft sensin...Local learning based soft sensing methods succeed in coping with time-varying characteristics of processes as well as nonlinearities in industrial plants. In this paper, a local partial least squares based soft sensing method for multi-output processes is proposed to accomplish process states division and local model adaptation,which are two key steps in development of local learning based soft sensors. An adaptive way of partitioning process states without redundancy is proposed based on F-test, where unique local time regions are extracted.Subsequently, a novel anti-over-fitting criterion is proposed for online local model adaptation which simultaneously considers the relationship between process variables and the information in labeled and unlabeled samples. Case study is carried out on two chemical processes and simulation results illustrate the superiorities of the proposed method from several aspects.展开更多
Soft sensor is widely used in industrial process control. It plays animportant role to improve the quality of product and assure safety in production. The core of softsensor is to construct soft sensing model. A new s...Soft sensor is widely used in industrial process control. It plays animportant role to improve the quality of product and assure safety in production. The core of softsensor is to construct soft sensing model. A new soft sensing modeling method based on supportvector machine (SVM) is proposed. SVM is a new machine learning method based on statistical learningtheory and is powerful for the problem characterized by small sample, nonlinearity, high dimensionand local minima. The proposed methods are applied to the estimation of frozen point of light dieseloil in distillation column. The estimated outputs of soft sensing model based on SVM match the realvalues of frozen point and follow varying trend of frozen point very well. Experiment results showthat SVM provides a new effective method for soft sensing modeling and has promising application inindustrial process applications.展开更多
A novel adaptive subspace ensemble slow feature regression model was developed for soft sensing application.Compared to traditional single models and random subspace models,the proposed method is improved in three asp...A novel adaptive subspace ensemble slow feature regression model was developed for soft sensing application.Compared to traditional single models and random subspace models,the proposed method is improved in three aspects.Firstly,sub-datasets are constructed through slow feature directions and variables in each subdatasets are selected according to the output related importance index.Then,an adaptive slow feature regression is presented for sub-models.Finally,a Bayesian inference strategy based on a slow feature analysis process that monitors statistics is developed for probabilistic combination.Two industrial examples were used to evaluate the proposed method.展开更多
Aiming at the problem of soft sensing modeling for chemical process with strong nonlinearity and complexity,a soft sensing modeling method based on kernel-based orthogonal projections to latent structures(K-OPLS)is pr...Aiming at the problem of soft sensing modeling for chemical process with strong nonlinearity and complexity,a soft sensing modeling method based on kernel-based orthogonal projections to latent structures(K-OPLS)is proposed.Orthogonal projections to latent structures(O-PLS)is a general linear multi-variable data modeling method.It can eliminate systematic variations from descriptive variables(input)that are orthogonal to response variables(output).In the framework of O-PLS model,K-OPLS method maps descriptive variables to high-dimensional feature space by using“kernel technique”to calculate predictive components and response-orthogonal components in the model.Therefore,the K-OPLS method gives the non-linear relationship between the descriptor and the response variables,which improves the performance of the model and enhances the interpretability of the model to a certain extent.To verify the validity of K-OPLS method,it was applied to soft sensing modeling of component content of debutane tower base butane(C4),the quality index of the key product output for industrial fluidized catalytic cracking unit(FCCU)and H 2S and SO 2 concentration in sulfur recovery unit(SRU).Compared with support vector machines(SVM),least-squares support-vector machine(LS-SVM),support vector machine with principal component analysis(PCA-SVM),extreme learning machine(ELM),kernel based extreme learning machine(KELM)and kernel based extreme learning machine with principal component analysis(PCA-KELM)methods under the same conditions,the experimental results show that the K-OPLS method has superior modeling accuracy and good model generalization ability.展开更多
Accurately soft sensing of the mechanical properties of hot-rolled strips is essential to ensure product quality,optimize production,and reduce costs.However,it faces the difficulty caused by limited labeled samples,f...Accurately soft sensing of the mechanical properties of hot-rolled strips is essential to ensure product quality,optimize production,and reduce costs.However,it faces the difficulty caused by limited labeled samples,for which co-training based semi-supervised learning offers a potential solution.So in this paper,a novel soft sensing method for mechanical properties based on improved co-training(ICO)is proposed.Compared with the existing co-training framework,the proposed ICO introduces improvements from the aspects of multiple view partition,confidence estimation,and pseudo-label assignment.Specifically,(ⅰ)in the stage of multiple view partition,ICO integrates metallurgical mechanisms of hot rolling processes and statistical mutual information to achieve a balance between view sufficiency and independence,which improves model performance and interpretability;(ⅱ)in the stage of confidence estimation,ICO evaluates the confidence of unlabeled samples at the cluster level rather than at the level of a single sample,which facilitates the exploration of sample distribution and the selection of representative samples;(ⅲ)in the pseudo-label assignment stage,ICO adopts a safe pseudo-label algorithm(which is called SAFER by its author and originally used for each single sample)to assign pseudo-labels for cluster of samples with the highest confidence determined in the previous step stage,to take advantage of the merit of handling unlabeled samples at the cluster level mentioned above on one hand,and the merit of SAFER in enhancing the quality of pseudo-labels on the other hand.The proposed soft sensing method effectively predicts mechanical properties on the real hot rolling dataset,achieving approximately 5%improvement in R~2 compared to traditional supervised learning.展开更多
Precise, real-time measurements of overflow particle size distributions in hydrocyclones are necessary for accurate control of the comminution circuits. Soft sensing measurements provide real-time, flexible, and low-c...Precise, real-time measurements of overflow particle size distributions in hydrocyclones are necessary for accurate control of the comminution circuits. Soft sensing measurements provide real-time, flexible, and low-cost measurements appropriate for the overflow particle size distributions in hydrocyclones. Three soft sensing methods were investigated for measuring the overflow particle size distributions in hydrocyclones. Simulations show that these methods have various advantages and disadvantages. Optimal Bayesian estimation fusion was then used to combine three methods with the fusion parameters determined according to the performance of each method with validation samples. The combined method compensates for the disadvantages of each method for more precise measurements. Simulations using real operating data show that the absolute root mean square measurement error of the combined method was always about 2% and the method provides the necessary accuracy for beneficiation plants.展开更多
基金supported in part by the National Science and Technology Major Project of China(No.2019-I-0019-0018)the National Natural Science Foundation of China(Nos.61890920,61890921,12302065 and 12172073).
文摘Partial Differential Equations(PDEs)are model candidates of soft sensing for aero-engine health management units.The existing Physics-Informed Neural Networks(PINNs)have made achievements.However,unmeasurable aero-engine driving sources lead to unknown PDE driving terms,which weaken PINNs feasibility.To this end,Physically Informed Hierarchical Learning followed by Recurrent-Prediction Term(PIHL-RPT)is proposed.First,PIHL is proposed for learning nonhomogeneous PDE solutions,in which two networks NetU and NetG are constructed.NetU is for learning solutions satisfying PDEs;NetG is for learning driving terms to regularize NetU training.Then,we propose a hierarchical learning strategy to optimize and couple NetU and NetG,which are integrated into a data-physics-hybrid loss function.Besides,we prove PIHL-RPT can iteratively generate a series of networks converging to a function,which can approximate a solution to well-posed PDE.Furthermore,RPT is proposed for prediction improvement of PIHL,in which network NetU-RP is constructed to compensate for information loss caused by data sampling and driving sources’immeasurability.Finally,artificial datasets and practical vibration process datasets from our wear experiment platform are used to verify the feasibility and effectiveness of PIHL-RPT based soft sensing.Meanwhile,comparisons with relevant methods,discussions,and PIHL-RPT based health monitoring example are given.
基金Item Sponsored by National Natural Science Foundation of China (50474086,60843007)
文摘Aiming at the limitations of traditional thermal model and intelligent model, a new hybrid model is established for soft sensing of the molten steel temperature in LF. Firstly, a thermal model based on energy conservation is described; and then, an improved intelligent model based on process data is presented by ensemble ELM （extreme learning machine） for predicting the molten steel temperature in LF. Secondly, the self-adaptive data fusion is pro- posed as a hybrid modeling method to combine the thermal model with the intelligent model. The new hybrid model could complement mutual advantage of two models by combination. It can overcome the shortcoming of parameters obtained on-line hardly in a thermal model and the disadvantage of lacking the analysis of ladle furnace metallurgical process in an intelligent model. The new hybrid model is applied to a 300 t LF in Baoshan Iron and Steel Co Ltd for predicting the molten steel temperature. The experiments demonstrate that the hybrid model has good generalization performance and high accuracy.
基金Supported by the National Natural Science Foundation of China(61273160)the Fundamental Research Funds for the Central Universities(14CX06067A,13CX05021A)
文摘Local learning based soft sensing methods succeed in coping with time-varying characteristics of processes as well as nonlinearities in industrial plants. In this paper, a local partial least squares based soft sensing method for multi-output processes is proposed to accomplish process states division and local model adaptation,which are two key steps in development of local learning based soft sensors. An adaptive way of partitioning process states without redundancy is proposed based on F-test, where unique local time regions are extracted.Subsequently, a novel anti-over-fitting criterion is proposed for online local model adaptation which simultaneously considers the relationship between process variables and the information in labeled and unlabeled samples. Case study is carried out on two chemical processes and simulation results illustrate the superiorities of the proposed method from several aspects.
基金This project is supported by Special Foundation for Major State Basic Research of China (No.G1998030415).
文摘Soft sensor is widely used in industrial process control. It plays animportant role to improve the quality of product and assure safety in production. The core of softsensor is to construct soft sensing model. A new soft sensing modeling method based on supportvector machine (SVM) is proposed. SVM is a new machine learning method based on statistical learningtheory and is powerful for the problem characterized by small sample, nonlinearity, high dimensionand local minima. The proposed methods are applied to the estimation of frozen point of light dieseloil in distillation column. The estimated outputs of soft sensing model based on SVM match the realvalues of frozen point and follow varying trend of frozen point very well. Experiment results showthat SVM provides a new effective method for soft sensing modeling and has promising application inindustrial process applications.
基金the support from the National Natural Science Foundation of China(No.21676086).
文摘A novel adaptive subspace ensemble slow feature regression model was developed for soft sensing application.Compared to traditional single models and random subspace models,the proposed method is improved in three aspects.Firstly,sub-datasets are constructed through slow feature directions and variables in each subdatasets are selected according to the output related importance index.Then,an adaptive slow feature regression is presented for sub-models.Finally,a Bayesian inference strategy based on a slow feature analysis process that monitors statistics is developed for probabilistic combination.Two industrial examples were used to evaluate the proposed method.
基金National Natural Science Foundation of China(No.51467008)。
文摘Aiming at the problem of soft sensing modeling for chemical process with strong nonlinearity and complexity,a soft sensing modeling method based on kernel-based orthogonal projections to latent structures(K-OPLS)is proposed.Orthogonal projections to latent structures(O-PLS)is a general linear multi-variable data modeling method.It can eliminate systematic variations from descriptive variables(input)that are orthogonal to response variables(output).In the framework of O-PLS model,K-OPLS method maps descriptive variables to high-dimensional feature space by using“kernel technique”to calculate predictive components and response-orthogonal components in the model.Therefore,the K-OPLS method gives the non-linear relationship between the descriptor and the response variables,which improves the performance of the model and enhances the interpretability of the model to a certain extent.To verify the validity of K-OPLS method,it was applied to soft sensing modeling of component content of debutane tower base butane(C4),the quality index of the key product output for industrial fluidized catalytic cracking unit(FCCU)and H 2S and SO 2 concentration in sulfur recovery unit(SRU).Compared with support vector machines(SVM),least-squares support-vector machine(LS-SVM),support vector machine with principal component analysis(PCA-SVM),extreme learning machine(ELM),kernel based extreme learning machine(KELM)and kernel based extreme learning machine with principal component analysis(PCA-KELM)methods under the same conditions,the experimental results show that the K-OPLS method has superior modeling accuracy and good model generalization ability.
基金supported in part by National Key Research&Development Program of China(2021YFB3301200)in part by the National Natural Science Foundation of China(61933015)。
文摘Accurately soft sensing of the mechanical properties of hot-rolled strips is essential to ensure product quality,optimize production,and reduce costs.However,it faces the difficulty caused by limited labeled samples,for which co-training based semi-supervised learning offers a potential solution.So in this paper,a novel soft sensing method for mechanical properties based on improved co-training(ICO)is proposed.Compared with the existing co-training framework,the proposed ICO introduces improvements from the aspects of multiple view partition,confidence estimation,and pseudo-label assignment.Specifically,(ⅰ)in the stage of multiple view partition,ICO integrates metallurgical mechanisms of hot rolling processes and statistical mutual information to achieve a balance between view sufficiency and independence,which improves model performance and interpretability;(ⅱ)in the stage of confidence estimation,ICO evaluates the confidence of unlabeled samples at the cluster level rather than at the level of a single sample,which facilitates the exploration of sample distribution and the selection of representative samples;(ⅲ)in the pseudo-label assignment stage,ICO adopts a safe pseudo-label algorithm(which is called SAFER by its author and originally used for each single sample)to assign pseudo-labels for cluster of samples with the highest confidence determined in the previous step stage,to take advantage of the merit of handling unlabeled samples at the cluster level mentioned above on one hand,and the merit of SAFER in enhancing the quality of pseudo-labels on the other hand.The proposed soft sensing method effectively predicts mechanical properties on the real hot rolling dataset,achieving approximately 5%improvement in R~2 compared to traditional supervised learning.
基金the Key Technologies Research and Development Program of the Eleventh Five-Year Plan of China (No. 2006AA060206)
文摘Precise, real-time measurements of overflow particle size distributions in hydrocyclones are necessary for accurate control of the comminution circuits. Soft sensing measurements provide real-time, flexible, and low-cost measurements appropriate for the overflow particle size distributions in hydrocyclones. Three soft sensing methods were investigated for measuring the overflow particle size distributions in hydrocyclones. Simulations show that these methods have various advantages and disadvantages. Optimal Bayesian estimation fusion was then used to combine three methods with the fusion parameters determined according to the performance of each method with validation samples. The combined method compensates for the disadvantages of each method for more precise measurements. Simulations using real operating data show that the absolute root mean square measurement error of the combined method was always about 2% and the method provides the necessary accuracy for beneficiation plants.
