Industrial data mining usually deals with data from different sources.These heterogeneous datasets describe the same object in different views.However,samples from some of the datasets may be lost.Then the remaining s...Industrial data mining usually deals with data from different sources.These heterogeneous datasets describe the same object in different views.However,samples from some of the datasets may be lost.Then the remaining samples do not correspond one-to-one correctly.Mismatched datasets caused by missing samples make the industrial data unavailable for further machine learning.In order to align the mismatched samples,this article presents a cooperative iteration matching method(CIMM)based on the modified dynamic time warping(DTW).The proposed method regards the sequentially accumulated industrial data as the time series.Mismatched samples are aligned by the DTW.In addition,dynamic constraints are applied to the warping distance of the DTW process to make the alignment more efficient.Then a series of models are trained with the cumulated samples iteratively.Several groups of numerical experiments on different missing patterns and missing locations are designed and analyzed to prove the effectiveness and the applicability of the proposed method.展开更多
ZTE Corporation announced on August 25,2009 that sales of its Data Card have topped 7 million in first half of 2009,representing an increase of 366% compared to the same period last year,the fastest growth amongst all
Artificial intelligence is a new technological science that researches and develops theories,methods,technologies and application systems for simulating,extending and expanding human intelligence.It simulates certain ...Artificial intelligence is a new technological science that researches and develops theories,methods,technologies and application systems for simulating,extending and expanding human intelligence.It simulates certain human thought processes and intelligent behaviors(such as learning,reasoning,thinking,planning,etc.),and produces a new type of intelligent machine that can respond in a similar way to human intelligence.In the past 30 years,it has achieved rapid development in various industries and related disciplines such as manufacturing,medical care,finance,and transportation.展开更多
Blast furnace (BF) ironmaking is the most typical “black box” process, and its complexity and uncertainty bring forth great challenges for furnace condition judgment and BF operation. Rich data resources for BF iron...Blast furnace (BF) ironmaking is the most typical “black box” process, and its complexity and uncertainty bring forth great challenges for furnace condition judgment and BF operation. Rich data resources for BF ironmaking are available, and the rapid development of data science and intelligent technology will provide an effective means to solve the uncertainty problem in the BF ironmaking process. This work focused on the application of artificial intelligence technology in BF ironmaking. The current intelligent BF ironmaking technology was summarized and analyzed from five aspects. These aspects include BF data management, the analyses of time delay and correlation, the prediction of BF key variables, the evaluation of BF status, and the multi-objective intelligent optimization of BF operations. Solutions and suggestions were offered for the problems in the current progress, and some outlooks for future prospects and technological breakthroughs were added. To effectively improve the BF data quality, we comprehensively considered the data problems and the characteristics of algorithms and selected the data processing method scientifically. For analyzing important BF characteristics, the effect of the delay was eliminated to ensure an accurate logical relationship between the BF parameters and economic indicators. As for BF parameter prediction and BF status evaluation,a BF intelligence model that integrates data information and process mechanism was built to effectively achieve the accurate prediction of BF key indexes and the scientific evaluation of BF status. During the optimization of BF parameters, low risk, low cost, and high return were used as the optimization criteria, and while pursuing the optimization effect, the feasibility and site operation cost were considered comprehensively.This work will help increase the process operator’s overall awareness and understanding of intelligent BF technology. Additionally, combining big data technology with the process will improve the practicality of data models in actual production and promote the application of intelligent technology in BF ironmaking.展开更多
A successful mechanical property data-driven prediction model is the core of the optimal design of hot rolling process for hot-rolled strips. However, the original industrial data, usually unbalanced, are inevitably m...A successful mechanical property data-driven prediction model is the core of the optimal design of hot rolling process for hot-rolled strips. However, the original industrial data, usually unbalanced, are inevitably mixed with fluctuant and abnormal values. Models established on the basis of the data without data processing can cause misleading results, which cannot be used for the optimal design of hot rolling process. Thus, a method of industrial data processing of C-Mn steel was proposed based on the data analysis. The Bayesian neural network was employed to establish the reliable mechanical property prediction models for the optimal design of hot rolling process. By using the multi-objective optimization algorithm and considering the individual requirements of costumers and the constraints of the equipment, the optimal design of hot rolling process was successfully applied to the rolling process design for Q345B steel with 0.017% Nb and 0.046% Ti content removed. The optimal process design results were in good agreement with the industrial trials results, which verify the effectiveness of the optimal design of hot rolling process.展开更多
To evaluate and improve the real-time performance of Ethernet for plant automation（EPA） industrial Ethernet,the real-time performance of EPA periodic data transmission was theoretically and experimentally studied.By...To evaluate and improve the real-time performance of Ethernet for plant automation（EPA） industrial Ethernet,the real-time performance of EPA periodic data transmission was theoretically and experimentally studied.By analyzing information transmission regularity and EPA deterministic scheduling mechanism,periodic messages were categorized as different modes according to their entering-queue time.The scheduling characteristics and delivery time of each mode and their interacting relations were studied,during which the models of real-time performance of periodic information transmission in EPA system were established.On this basis,an experimental platform is developed to test the delivery time of periodic messages transmission in EPA system.According to the analysis and the experiment,the main factors that limit the real-time performance of EPA periodic data transmission and the improvement methods were proposed.展开更多
To solve the problems in the quality control and improvement of coiled tubing steel strips production, such as scattered and inefficient production data, difficult performance fluctuation factor analysis, complex mult...To solve the problems in the quality control and improvement of coiled tubing steel strips production, such as scattered and inefficient production data, difficult performance fluctuation factor analysis, complex multivariate statistical analysis, and low accuracy and difficulty in mechanical property prediction, an industrial data analysis platform for coiled tubing steel strips production has been preliminarily developed.As the premise and foundation of analysis, industrial data collection, storage, and utilization are realized by using multiple big data technologies.With Django as the agile development framework, data visualization and comprehensive analyses are achieved.The platform has functions including overview survey, stability analysis, comprehensive analysis(such as exploratory data analysis, correlation analysis, and multivariate statistics),precise steel strength prediction, and skin-passing process recommendation.The platform is helpful for production overviewing and prompt responding, laying a foundation for an in-depth understanding of product characteristics and improving product performance stability.展开更多
By the end of 2013, Chinese large-scaled commercial banks have basically completed the construction of IT banking system, in order to provide technical guarantee of deepening business operation and operation managemen...By the end of 2013, Chinese large-scaled commercial banks have basically completed the construction of IT banking system, in order to provide technical guarantee of deepening business operation and operation management reform. This indicates that our national banking industry information technology has been in a new level. This paper, based on the operation principle of commercial bank safety, liquidity and profitability, makes research on improving commercial banking profit data timeliness, accuracy, integrity and realizing daily accounting profit. According to survey, the four big banks have proposed the prospect of 2013 full scope implementing daily accounting profit. Thus, this paper, based on the profit accounting status of four big banks of data center, analyzes the feasibility of daily accounting profit and puts forward the relevant solutions.展开更多
According to statistics of Printing and Printing Equipment Industries Association of China (PEIAC), the total output value of printing industry of China in 2007 reached 440 billion RMB , the total output value of prin...According to statistics of Printing and Printing Equipment Industries Association of China (PEIAC), the total output value of printing industry of China in 2007 reached 440 billion RMB , the total output value of printing equipment was展开更多
As industrial production progresses toward digitalization,massive amounts of data have been collected,transmitted,and stored,with characteristics of large-scale,high-dimensional,heterogeneous,and spatiotemporal dynami...As industrial production progresses toward digitalization,massive amounts of data have been collected,transmitted,and stored,with characteristics of large-scale,high-dimensional,heterogeneous,and spatiotemporal dynamics.The high complexity of industrial big data poses challenges for the practical decision-making of domain experts,leading to ever-increasing needs for integrating computational intelligence with human perception into traditional data analysis.Industrial big data visualization integrates theoretical methods and practical technologies from multiple disciplines,including data mining,information visualization,computer graphics,and human-computer interaction,providing a highly effective manner for understanding and exploring the complex industrial processes.This review summarizes the state-of-the-art approaches,characterizes them with six visualization methods,and categorizes them based on analytical tasks and applications.Furthermore,key research challenges and potential future directions are identified.展开更多
The launching of CBERS-01(China Brazil Earth Resource Satellite)in 1999,China’s first land observation satellite,signifies an unprecedented milestone in Chinese satellite remote sensing history.Since then,a large num...The launching of CBERS-01(China Brazil Earth Resource Satellite)in 1999,China’s first land observation satellite,signifies an unprecedented milestone in Chinese satellite remote sensing history.Since then,a large number of applications have been developed that drew upon solely CBERS-01 and other Chinese land observation satellites.The application development evolves from one satellite to multiple satellites,from one series of satellites to multiple series,from scientific research to industrial applications.Six aspects of the Chinese land observation satellite program are discussed in this paper:development status,data sharing and distribution,satellite calibration,industrial data applications,future prospects,and conclusion.展开更多
A recommender system(RS)relying on latent factor analysis usually adopts stochastic gradient descent(SGD)as its learning algorithm.However,owing to its serial mechanism,an SGD algorithm suffers from low efficiency and...A recommender system(RS)relying on latent factor analysis usually adopts stochastic gradient descent(SGD)as its learning algorithm.However,owing to its serial mechanism,an SGD algorithm suffers from low efficiency and scalability when handling large-scale industrial problems.Aiming at addressing this issue,this study proposes a momentum-incorporated parallel stochastic gradient descent(MPSGD)algorithm,whose main idea is two-fold:a)implementing parallelization via a novel datasplitting strategy,and b)accelerating convergence rate by integrating momentum effects into its training process.With it,an MPSGD-based latent factor(MLF)model is achieved,which is capable of performing efficient and high-quality recommendations.Experimental results on four high-dimensional and sparse matrices generated by industrial RS indicate that owing to an MPSGD algorithm,an MLF model outperforms the existing state-of-the-art ones in both computational efficiency and scalability.展开更多
This research develops a new mathematical modeling method by combining industrial big data and process mechanism analysis under the framework of generalized additive models(GAM)to generate a practical model with gener...This research develops a new mathematical modeling method by combining industrial big data and process mechanism analysis under the framework of generalized additive models(GAM)to generate a practical model with generalization and precision.Specifically,the proposed modeling method includes the following steps.Firstly,the influence factors are screened using mechanism knowledge and data-mining methods.Secondly,the unary GAM without interactions including cleaning the data,building the sub-models,and verifying the sub-models.Subsequently,the interactions between the various factors are explored,and the binary GAM with interactions is constructed.The relationships among the sub-models are analyzed,and the integrated model is built.Finally,based on the proposed modeling method,two prediction models of mechanical property and deformation resistance for hot-rolled strips are established.Industrial actual data verification demonstrates that the new models have good prediction precision,and the mean absolute percentage errors of tensile strength,yield strength and deformation resistance are 2.54%,3.34%and 6.53%,respectively.And experimental results suggest that the proposed method offers a new approach to industrial process modeling.展开更多
In order to improve the gas turbine engine health monitoring capability, using multiple detector fusion method in the monitoring system of gas turbine data monitor. Multi detector frame fusion includes point bias anom...In order to improve the gas turbine engine health monitoring capability, using multiple detector fusion method in the monitoring system of gas turbine data monitor. Multi detector frame fusion includes point bias anomaly detector, contextual bias anomaly detector and collective bias anomaly detector, common to analyze the new arrival data, and the possible abnormal state to vote and weighted statistics as a result output. The experimental results show the method can effectively detect the mutation phenomenon, relatively slow changes and abnormal behavior discordant to the conditions. The framework applied to the gas turbine engine can effectively enhance the health diagnosis ability, will be highly applied for real industry.展开更多
Monitoring of industrial plant performance and detection on flaws is important to the successful operation on industrial production units. Malfunctioning equipment can greatly impact plant performance by reducing the ...Monitoring of industrial plant performance and detection on flaws is important to the successful operation on industrial production units. Malfunctioning equipment can greatly impact plant performance by reducing the efficiency and increasing the production cost. Phenomenological equations cannot properly describe industrial processes. Thus, it is necessary to develop new equations for model industrial operations. The purpose of this study is to develop an empirical model for industrial demethanizer tower which is malfunctioning due to an error in the design in one of its plates. A nonlinear statistical model was designed to predict the pressure variation in the column, and consequently, the flooding conditions. This model was validated using industrial data to predict the maximum loads in the column.展开更多
There is a growing demand for time series data analysis in industry areas.Apache loTDB is a time series database designed for the Internet of Things(loT)with enhanced storage and I/O performance.With User-Defined Func...There is a growing demand for time series data analysis in industry areas.Apache loTDB is a time series database designed for the Internet of Things(loT)with enhanced storage and I/O performance.With User-Defined Functions(UDF)provided,computation for time series can be executed on Apache loTDB directly.To satisfy most of the common requirements in industrial time series analysis,we create a UDF library,loTDQ,on Apache loTDB.This library integrates stream computation functions on data quality analysis,data profiling,anomaly detection,data repairing,etc.loTDQ enables users to conduct a wide range of analyses,such as monitoring,error diagnosis,equipment reliability analysis.It provides a framework for users to examine loT time series with data quality problems.Experiments show that loTDQ keeps the same level of performance compared to mainstream alternatives,and shortens I/O consumption for Apache loTDB users.展开更多
Recently developed fault classification methods for industrial processes are mainly data-driven.Notably,models based on deep neural networks have significantly improved fault classification accuracy owing to the inclu...Recently developed fault classification methods for industrial processes are mainly data-driven.Notably,models based on deep neural networks have significantly improved fault classification accuracy owing to the inclusion of a large number of data patterns.However,these data-driven models are vulnerable to adversarial attacks;thus,small perturbations on the samples can cause the models to provide incorrect fault predictions.Several recent studies have demonstrated the vulnerability of machine learning methods and the existence of adversarial samples.This paper proposes a black-box attack method with an extreme constraint for a safe-critical industrial fault classification system:Only one variable can be perturbed to craft adversarial samples.Moreover,to hide the adversarial samples in the visualization space,a Jacobian matrix is used to guide the perturbed variable selection,making the adversarial samples in the dimensional reduction space invisible to the human eye.Using the one-variable attack(OVA)method,we explore the vulnerability of industrial variables and fault types,which can help understand the geometric characteristics of fault classification systems.Based on the attack method,a corresponding adversarial training defense method is also proposed,which efficiently defends against an OVA and improves the prediction accuracy of the classifiers.In experiments,the proposed method was tested on two datasets from the Tennessee–Eastman process(TEP)and steel plates(SP).We explore the vulnerability and correlation within variables and faults and verify the effectiveness of OVAs and defenses for various classifiers and datasets.For industrial fault classification systems,the attack success rate of our method is close to(on TEP)or even higher than(on SP)the current most effective first-order white-box attack method,which requires perturbation of all variables.展开更多
The industrial Internet has germinated with the integration of the traditional industry and information technologies.An identifier is the identification of an object in the industrial Internet.The identifier technolog...The industrial Internet has germinated with the integration of the traditional industry and information technologies.An identifier is the identification of an object in the industrial Internet.The identifier technology is a method to validate the identification of an object and trace it.The identifier is a bridge to connect information islands in the industry,as well as the data basis for building a technology application ecosystem based on identifier resolution.We propose three practical applications and application scenarios of the industrial Internet identifier in this paper.Future applications of identifier resolution in the industrial Internet field are also presented.展开更多
Industrial big data was usually multi-source, heterogeneous, and deeply intertwined. It had a wide range of data sources, high data dimensions, and strong data correlation. In order to effectively analyze and process ...Industrial big data was usually multi-source, heterogeneous, and deeply intertwined. It had a wide range of data sources, high data dimensions, and strong data correlation. In order to effectively analyze and process streaming industrial big data generated by edge computing, it was very important to provide an effective real-time incremental data method. However, in the process of incremental processing, industrial big data incremental computing faced the challenges of dimensional disaster, repeated calculations, and the explosion of intermediate results. Therefore, in order to solve the above problems effectively, a QR-based tensor-train(TT) decomposition(TTD) method and a QR-based incremental TTD(QRITTD) method were proposed. This algorithm combined the incremental QR-based decomposition algorithm with an approximate singular value decomposition(SVD) algorithm and had good scalability. In addition, the computational complexity, space complexity, and approximation error analysis were analyzed in detail. The effectiveness of the three algorithms of QRITTD, non-incremental TTD(NITTD), and TT rank-1(TTr1) SVD(TTr1 SVD)were verified by comparison. Experimental results show that the SVD QRITTD method has better performance under the premise of ensuring the same tensor size.展开更多
基金the Key National Natural Science Foundation of China(No.U1864211)the National Natural Science Foundation of China(No.11772191)the Natural Science Foundation of Shanghai(No.21ZR1431500)。
文摘Industrial data mining usually deals with data from different sources.These heterogeneous datasets describe the same object in different views.However,samples from some of the datasets may be lost.Then the remaining samples do not correspond one-to-one correctly.Mismatched datasets caused by missing samples make the industrial data unavailable for further machine learning.In order to align the mismatched samples,this article presents a cooperative iteration matching method(CIMM)based on the modified dynamic time warping(DTW).The proposed method regards the sequentially accumulated industrial data as the time series.Mismatched samples are aligned by the DTW.In addition,dynamic constraints are applied to the warping distance of the DTW process to make the alignment more efficient.Then a series of models are trained with the cumulated samples iteratively.Several groups of numerical experiments on different missing patterns and missing locations are designed and analyzed to prove the effectiveness and the applicability of the proposed method.
文摘ZTE Corporation announced on August 25,2009 that sales of its Data Card have topped 7 million in first half of 2009,representing an increase of 366% compared to the same period last year,the fastest growth amongst all
文摘Artificial intelligence is a new technological science that researches and develops theories,methods,technologies and application systems for simulating,extending and expanding human intelligence.It simulates certain human thought processes and intelligent behaviors(such as learning,reasoning,thinking,planning,etc.),and produces a new type of intelligent machine that can respond in a similar way to human intelligence.In the past 30 years,it has achieved rapid development in various industries and related disciplines such as manufacturing,medical care,finance,and transportation.
基金financially supported by the General Program of the National Natural Science Foundation of China(No.52274326)the Fundamental Research Funds for the Central Universities (Nos.2125018 and 2225008)China Baowu Low Carbon Metallurgy Innovation Foundation(BWLCF202109)。
文摘Blast furnace (BF) ironmaking is the most typical “black box” process, and its complexity and uncertainty bring forth great challenges for furnace condition judgment and BF operation. Rich data resources for BF ironmaking are available, and the rapid development of data science and intelligent technology will provide an effective means to solve the uncertainty problem in the BF ironmaking process. This work focused on the application of artificial intelligence technology in BF ironmaking. The current intelligent BF ironmaking technology was summarized and analyzed from five aspects. These aspects include BF data management, the analyses of time delay and correlation, the prediction of BF key variables, the evaluation of BF status, and the multi-objective intelligent optimization of BF operations. Solutions and suggestions were offered for the problems in the current progress, and some outlooks for future prospects and technological breakthroughs were added. To effectively improve the BF data quality, we comprehensively considered the data problems and the characteristics of algorithms and selected the data processing method scientifically. For analyzing important BF characteristics, the effect of the delay was eliminated to ensure an accurate logical relationship between the BF parameters and economic indicators. As for BF parameter prediction and BF status evaluation,a BF intelligence model that integrates data information and process mechanism was built to effectively achieve the accurate prediction of BF key indexes and the scientific evaluation of BF status. During the optimization of BF parameters, low risk, low cost, and high return were used as the optimization criteria, and while pursuing the optimization effect, the feasibility and site operation cost were considered comprehensively.This work will help increase the process operator’s overall awareness and understanding of intelligent BF technology. Additionally, combining big data technology with the process will improve the practicality of data models in actual production and promote the application of intelligent technology in BF ironmaking.
文摘A successful mechanical property data-driven prediction model is the core of the optimal design of hot rolling process for hot-rolled strips. However, the original industrial data, usually unbalanced, are inevitably mixed with fluctuant and abnormal values. Models established on the basis of the data without data processing can cause misleading results, which cannot be used for the optimal design of hot rolling process. Thus, a method of industrial data processing of C-Mn steel was proposed based on the data analysis. The Bayesian neural network was employed to establish the reliable mechanical property prediction models for the optimal design of hot rolling process. By using the multi-objective optimization algorithm and considering the individual requirements of costumers and the constraints of the equipment, the optimal design of hot rolling process was successfully applied to the rolling process design for Q345B steel with 0.017% Nb and 0.046% Ti content removed. The optimal process design results were in good agreement with the industrial trials results, which verify the effectiveness of the optimal design of hot rolling process.
基金Supported by the National High Technology Research and Development Program of China (2006AA040301-4,2007AA041301-6)
文摘To evaluate and improve the real-time performance of Ethernet for plant automation（EPA） industrial Ethernet,the real-time performance of EPA periodic data transmission was theoretically and experimentally studied.By analyzing information transmission regularity and EPA deterministic scheduling mechanism,periodic messages were categorized as different modes according to their entering-queue time.The scheduling characteristics and delivery time of each mode and their interacting relations were studied,during which the models of real-time performance of periodic information transmission in EPA system were established.On this basis,an experimental platform is developed to test the delivery time of periodic messages transmission in EPA system.According to the analysis and the experiment,the main factors that limit the real-time performance of EPA periodic data transmission and the improvement methods were proposed.
文摘To solve the problems in the quality control and improvement of coiled tubing steel strips production, such as scattered and inefficient production data, difficult performance fluctuation factor analysis, complex multivariate statistical analysis, and low accuracy and difficulty in mechanical property prediction, an industrial data analysis platform for coiled tubing steel strips production has been preliminarily developed.As the premise and foundation of analysis, industrial data collection, storage, and utilization are realized by using multiple big data technologies.With Django as the agile development framework, data visualization and comprehensive analyses are achieved.The platform has functions including overview survey, stability analysis, comprehensive analysis(such as exploratory data analysis, correlation analysis, and multivariate statistics),precise steel strength prediction, and skin-passing process recommendation.The platform is helpful for production overviewing and prompt responding, laying a foundation for an in-depth understanding of product characteristics and improving product performance stability.
文摘By the end of 2013, Chinese large-scaled commercial banks have basically completed the construction of IT banking system, in order to provide technical guarantee of deepening business operation and operation management reform. This indicates that our national banking industry information technology has been in a new level. This paper, based on the operation principle of commercial bank safety, liquidity and profitability, makes research on improving commercial banking profit data timeliness, accuracy, integrity and realizing daily accounting profit. According to survey, the four big banks have proposed the prospect of 2013 full scope implementing daily accounting profit. Thus, this paper, based on the profit accounting status of four big banks of data center, analyzes the feasibility of daily accounting profit and puts forward the relevant solutions.
文摘According to statistics of Printing and Printing Equipment Industries Association of China (PEIAC), the total output value of printing industry of China in 2007 reached 440 billion RMB , the total output value of printing equipment was
基金supported in part by the National Key Research and Development Plan Project(2022YFB3304700)in part by the Xinliao Talent Program of Liaoning Province(XLYC2202002).
文摘As industrial production progresses toward digitalization,massive amounts of data have been collected,transmitted,and stored,with characteristics of large-scale,high-dimensional,heterogeneous,and spatiotemporal dynamics.The high complexity of industrial big data poses challenges for the practical decision-making of domain experts,leading to ever-increasing needs for integrating computational intelligence with human perception into traditional data analysis.Industrial big data visualization integrates theoretical methods and practical technologies from multiple disciplines,including data mining,information visualization,computer graphics,and human-computer interaction,providing a highly effective manner for understanding and exploring the complex industrial processes.This review summarizes the state-of-the-art approaches,characterizes them with six visualization methods,and categorizes them based on analytical tasks and applications.Furthermore,key research challenges and potential future directions are identified.
基金supported in part by the National Basic Research Program of China(973 Program,Nos.2014CB744201 and 2012CB719902)the Program for New Century Excellent Talents in University+2 种基金the National High Technology Research and Development Program of China(No.2011AA120203)the National Natural Science Foundation of China(No.41371430)the Program for Changjiang Scholars and Innovative Research Team in University under Grant IRT1278.
文摘The launching of CBERS-01(China Brazil Earth Resource Satellite)in 1999,China’s first land observation satellite,signifies an unprecedented milestone in Chinese satellite remote sensing history.Since then,a large number of applications have been developed that drew upon solely CBERS-01 and other Chinese land observation satellites.The application development evolves from one satellite to multiple satellites,from one series of satellites to multiple series,from scientific research to industrial applications.Six aspects of the Chinese land observation satellite program are discussed in this paper:development status,data sharing and distribution,satellite calibration,industrial data applications,future prospects,and conclusion.
基金supported in part by the National Natural Science Foundation of China(61772493)the Deanship of Scientific Research(DSR)at King Abdulaziz University(RG-48-135-40)+1 种基金Guangdong Province Universities and College Pearl River Scholar Funded Scheme(2019)the Natural Science Foundation of Chongqing(cstc2019jcyjjqX0013)。
文摘A recommender system(RS)relying on latent factor analysis usually adopts stochastic gradient descent(SGD)as its learning algorithm.However,owing to its serial mechanism,an SGD algorithm suffers from low efficiency and scalability when handling large-scale industrial problems.Aiming at addressing this issue,this study proposes a momentum-incorporated parallel stochastic gradient descent(MPSGD)algorithm,whose main idea is two-fold:a)implementing parallelization via a novel datasplitting strategy,and b)accelerating convergence rate by integrating momentum effects into its training process.With it,an MPSGD-based latent factor(MLF)model is achieved,which is capable of performing efficient and high-quality recommendations.Experimental results on four high-dimensional and sparse matrices generated by industrial RS indicate that owing to an MPSGD algorithm,an MLF model outperforms the existing state-of-the-art ones in both computational efficiency and scalability.
基金Project(51774219)supported by the National Natural Science Foundation of China
文摘This research develops a new mathematical modeling method by combining industrial big data and process mechanism analysis under the framework of generalized additive models(GAM)to generate a practical model with generalization and precision.Specifically,the proposed modeling method includes the following steps.Firstly,the influence factors are screened using mechanism knowledge and data-mining methods.Secondly,the unary GAM without interactions including cleaning the data,building the sub-models,and verifying the sub-models.Subsequently,the interactions between the various factors are explored,and the binary GAM with interactions is constructed.The relationships among the sub-models are analyzed,and the integrated model is built.Finally,based on the proposed modeling method,two prediction models of mechanical property and deformation resistance for hot-rolled strips are established.Industrial actual data verification demonstrates that the new models have good prediction precision,and the mean absolute percentage errors of tensile strength,yield strength and deformation resistance are 2.54%,3.34%and 6.53%,respectively.And experimental results suggest that the proposed method offers a new approach to industrial process modeling.
基金Sponsored by the National Science Fundation for Distinguished Young Scholars of China(Grant No.50925625)
文摘In order to improve the gas turbine engine health monitoring capability, using multiple detector fusion method in the monitoring system of gas turbine data monitor. Multi detector frame fusion includes point bias anomaly detector, contextual bias anomaly detector and collective bias anomaly detector, common to analyze the new arrival data, and the possible abnormal state to vote and weighted statistics as a result output. The experimental results show the method can effectively detect the mutation phenomenon, relatively slow changes and abnormal behavior discordant to the conditions. The framework applied to the gas turbine engine can effectively enhance the health diagnosis ability, will be highly applied for real industry.
文摘Monitoring of industrial plant performance and detection on flaws is important to the successful operation on industrial production units. Malfunctioning equipment can greatly impact plant performance by reducing the efficiency and increasing the production cost. Phenomenological equations cannot properly describe industrial processes. Thus, it is necessary to develop new equations for model industrial operations. The purpose of this study is to develop an empirical model for industrial demethanizer tower which is malfunctioning due to an error in the design in one of its plates. A nonlinear statistical model was designed to predict the pressure variation in the column, and consequently, the flooding conditions. This model was validated using industrial data to predict the maximum loads in the column.
文摘There is a growing demand for time series data analysis in industry areas.Apache loTDB is a time series database designed for the Internet of Things(loT)with enhanced storage and I/O performance.With User-Defined Functions(UDF)provided,computation for time series can be executed on Apache loTDB directly.To satisfy most of the common requirements in industrial time series analysis,we create a UDF library,loTDQ,on Apache loTDB.This library integrates stream computation functions on data quality analysis,data profiling,anomaly detection,data repairing,etc.loTDQ enables users to conduct a wide range of analyses,such as monitoring,error diagnosis,equipment reliability analysis.It provides a framework for users to examine loT time series with data quality problems.Experiments show that loTDQ keeps the same level of performance compared to mainstream alternatives,and shortens I/O consumption for Apache loTDB users.
基金This work was supported in part by the National Natural Science Foundation of China(NSFC)(92167106,62103362,and 61833014)the Natural Science Foundation of Zhejiang Province(LR18F030001).
文摘Recently developed fault classification methods for industrial processes are mainly data-driven.Notably,models based on deep neural networks have significantly improved fault classification accuracy owing to the inclusion of a large number of data patterns.However,these data-driven models are vulnerable to adversarial attacks;thus,small perturbations on the samples can cause the models to provide incorrect fault predictions.Several recent studies have demonstrated the vulnerability of machine learning methods and the existence of adversarial samples.This paper proposes a black-box attack method with an extreme constraint for a safe-critical industrial fault classification system:Only one variable can be perturbed to craft adversarial samples.Moreover,to hide the adversarial samples in the visualization space,a Jacobian matrix is used to guide the perturbed variable selection,making the adversarial samples in the dimensional reduction space invisible to the human eye.Using the one-variable attack(OVA)method,we explore the vulnerability of industrial variables and fault types,which can help understand the geometric characteristics of fault classification systems.Based on the attack method,a corresponding adversarial training defense method is also proposed,which efficiently defends against an OVA and improves the prediction accuracy of the classifiers.In experiments,the proposed method was tested on two datasets from the Tennessee–Eastman process(TEP)and steel plates(SP).We explore the vulnerability and correlation within variables and faults and verify the effectiveness of OVAs and defenses for various classifiers and datasets.For industrial fault classification systems,the attack success rate of our method is close to(on TEP)or even higher than(on SP)the current most effective first-order white-box attack method,which requires perturbation of all variables.
文摘The industrial Internet has germinated with the integration of the traditional industry and information technologies.An identifier is the identification of an object in the industrial Internet.The identifier technology is a method to validate the identification of an object and trace it.The identifier is a bridge to connect information islands in the industry,as well as the data basis for building a technology application ecosystem based on identifier resolution.We propose three practical applications and application scenarios of the industrial Internet identifier in this paper.Future applications of identifier resolution in the industrial Internet field are also presented.
基金supported by the Science and Technology Project in Shaanxi Province of China (2019ZDLGY07-08)the Natural Science Foundation Research Program of Shaanxi Province, China。
文摘Industrial big data was usually multi-source, heterogeneous, and deeply intertwined. It had a wide range of data sources, high data dimensions, and strong data correlation. In order to effectively analyze and process streaming industrial big data generated by edge computing, it was very important to provide an effective real-time incremental data method. However, in the process of incremental processing, industrial big data incremental computing faced the challenges of dimensional disaster, repeated calculations, and the explosion of intermediate results. Therefore, in order to solve the above problems effectively, a QR-based tensor-train(TT) decomposition(TTD) method and a QR-based incremental TTD(QRITTD) method were proposed. This algorithm combined the incremental QR-based decomposition algorithm with an approximate singular value decomposition(SVD) algorithm and had good scalability. In addition, the computational complexity, space complexity, and approximation error analysis were analyzed in detail. The effectiveness of the three algorithms of QRITTD, non-incremental TTD(NITTD), and TT rank-1(TTr1) SVD(TTr1 SVD)were verified by comparison. Experimental results show that the SVD QRITTD method has better performance under the premise of ensuring the same tensor size.
