Design And Simulation Of The Extended Tennessee Eastman Process Based On The Intelligent Manufacturing Background

Process industry is one of the most important parts of the manufacturing industry,which dominates the national economy.Recently,with the fast development of equipment interconnection,data sharing,storage and computing technology,intelligent manufacturing is considered as the fourth industrial revolution in the global manufacturing industry.Under the background of intelligent manufacturing,the management and technology of the process industry needs an urgent reform,and its optimized operation can enhance the production efficiency,reduce the manual decision-making,and improve the comprehensive competitiveness of enterprises.In the exist literature,the researches on intelligent manufacturing based process industry are mainly focused on theoretical analysis,while the practical application in actual process industrial system is still absent.In this paper,a new process industry model is designed and simulated by extending a typical process industrial simulation process(TE process)under the background of intelligent manufacturing.Moreover,feature selection and process monitoring strategies are carried out based on this process.The main work and contributions lie in the following aspects:(1)The background of the intelligent manufacturing in the field of process industry is reviewed.To overcome the drawback of the traditional TE simulation process,such as single production variety and lack of effective management,an extended TE process is proposed according to the characteristics of the horizontal integration and the vertical business penetration of the intelligent manufacturing chain.Based on the multi-level modeling framework,the traditional TE process is extended with respect to horizontal extension,and longitudinal aggregation and disaggregation.Compared with the traditional TE process,the extended TE process is closer to the actual production.Moreover,its production capacity and manufacturing execution capability are greatly improved.(2)The detailed design and realization of the extended TE process is introduced.For the longitudinal extension of the TE process,the workflow model and function module of manufacturing execution system are defined based on the multi-agent simulation.The multi-agent simulation includes the cooperation among the upper layers of each agent,and the underlying process simulation.Moreover,cross-layer simulation is also realized through the data transmission between these two layers.At last,the simulation results are provided to illustrate the cases with different levels of normal operations and fault conditions.(3)To remedy the problem of the increase of the state variables and the decline of computational complexity in the monitor process of the extended TE process system,a feature selection method based on multi-core learning is proposed to select the useful features.By transforming the feature selection problem into an optimization problem under the multi-core learning framework,the problem of"monotonicity" in the traditional feature selection method can be avoided.Finally,the experimental results show that the proposed feature selection method has a superior feature selection performance.(4)A hierarchical process monitoring method is designed to monitor the extended TE process.The extended TE process is systematically decomposed into different levels,and the principal components are extracted for each decomposition block.Different levels of statistical process monitoring model are established by using Bayesian fusion,and the fault variables can be analyzed based on them.Through the experimental validation on the extended TE process simulation datasets,the performance of this hierarchical process monitoring method is demonstrated.Finally,the work of this paper is summarized,and the prospects and the potential research directions of this research field are also provided.