Research On Safety Operation Control For Industrial Process Based On Bayesian Network

The safety control problem for the abnormity in the development of industrial automation has attracted extent attention.It has an important significance to make sure the performance of production and the product quality.The abnormity often happens in the industrial process because of the change of raw material characteristics and the improper operation.The abnormity will result in the serious financial losses or even safety threat.Due to the poor production circumstance,simple basic automation facilities etc.,the most safety control decisions are made by the manual way based on the experience,memory and knowledge of operators.However,the different operators own different knowledge degree,the decisions from the operators are always qualitative and the ability of operators to deal with the multi-source information is limited.These problems make the decision inaccurate.Therefore,it owns an important theoretical significance and the value of practical application to make effective safety control scheme based on the characteristics of abnormity to ensure that the process runs well.Due to the influences of external environment,raw material characteristics and operation condition,the practical process includes a variety of uncertainty information.The precision of sensors is affected by the limitations in themselves and circumstances.Compared with the available data information of normal cor-dition,the abnormal data information is very limited because of the limited abnormities in the historical operation.The traditional data-based modeling methods are not able to obtain the accurate decisions.But the expert knowledge and operation experience can be collected for the safety control of abnormity.It is an important problem to solve in this paper that how to integrate the expert knowledge,operation experience and limited abnormal data information to establish the effective safety control model.Bayesian network(BN)is an effective tool for the integration of expert knowledge and data information,which owns strengths on the expression and reasoning of uncertainty information.Based on the analysis of safety control problem for the abnormity in the practical industrial process,this paper explores the safety control method based on BN to solve the practical problem.(1)For the safety operation control problem of complex industrial process,this paper proposes the safety operation control modeling method and online control strategy based on BN.In this process,the form of structure,the parameters learning scheme,the choice of evidence and inference method are determined.The proposed method is used to establish the safety control model for the thickening process of gold hydrometallurgy.This paper summarizes the common abnormities,collects the related variables and analyzes the causes,phenomena and corresponding solutions for this process.Based on the expert knowledge,operator experience and data information of abnormity,the proposed method is used to establish the safety control model for this process.After receiving the online evidences,the model can provide real-time decision support when the abnormity happens.(2)This paper proposes the scheme of group decision making(GDM)based on the intuitionistic fuzzy set(IFS)for the relationship determination between the BN nodes.By analyzing the alternative relationships between the BN nodes,the relationship determination problem is transformed into the GDM problem.For this problem,the group expert knowledge expression,aggregation and decision-making methods are proposed.For the different alternative information,the choice of aggregation operators and the influence of hesitancy degree on the aggregation result of inconsistent information are analyzed.(3)For the BN structure learning,this paper proposes a new BN structure learning method by integrating expert knowledge and data information.The group expert knowledge expression and aggregation based on the IFS are used to determine the relationship between the BN nodes,and the BN priori structure matrix can be obtained.The improved score searched method is used to obtain the optimal structure under the constraints from the priori structure matrix.This method improves the efficiency and accuracy of BN structure learning.(4)For the BN updating learning,this paper proposes the new updating learning strategy by integrating the expert knowledge and the data information under the new environment.To make the established model own the ability to adapt to the change of environment,the parameter updating learning and the structure updating learning methods are proposed to ensure the performance of decision.By searching for the unconformable nodes in the new environment as the target nodes,only the partial structure which does not adapt to the change of environment is changed.The expert knowledge and the data information under the new environment are utilized.At the same time,the useful information in the established model is reserved.The proposed method improves the efficiency of BN updating learning.(5)When the data information of target domain is very limited,it is difficult to establish the accurate model to analyze the target problem.A new BN transfer learning strategy is developed to establish the target model,which includes the structure transfer learning method and parameters transfer learning method.For the structure transfer learning,by integrating the common structural information of multiple sources and the useful information of target,the final structure of target is determined.For the parameters transfer learning,by distinguishing the similarities of multiple sources,the parameters of target are obtained by the fusion algorithm.The proposed transfer learning method improves the accuracy of target modeling.The proposed safety operation control scheme based on BN is applied in the thickening process of gold hydrometallurgy to establish the safety control model.The simulation results demonstrate that the proposed method is effective.The proposed approach in this paper provides the thought of solving the practical problems in the industrial process in some degree.It owns the value of theoretical research and the significance of practical application.