PLS Based Operating Optimality Assessment Method For Complex Industrial Processes

With the development of science and industrial technologies,production processes and production technologies have had great improvements and become more and more automated and complicated.Subsequently,the field environment of industrial processes has become increasingly complex.Because of disturbance,noise and other uncertainties,industrial processes are not likely to be maintained on the optimal point obtained by theories and there is no doubt the process operating performance will decline as time goes by.In order to decrease the production cost,reduce the emission and energy consumption,improve production efficiency and maintain the competitiveness of enterprises,it is great significance to propose and realize the method of online operating optimality assessment for the industrial production process.In this thesis,the comprehensive economic index(CEI)of industrial production process is used as the standard for operating optimality assessment,and the comprehensive economic index(CEI)prediction based operating optimality assessment method and non-optimal cause identification method are proposed for multimode continuous processes to solve the problem of online operating optimality assessment and non-optimal cause identification.The evaluation method considers the operability of the process and the reliability of the evaluation index and it mainly predicts and evaluates for complex production process.During the online assessment,the CEI is obtained by multiple partial least squares modeling strategy that is based on just-in-time learning(JITL-MPLS).The Optimality Index and Non-optimal Index are constructed based on the predicted CEI.Finally the method depends on evaluation strategy to achieve online operating optimality assessment.When the operating performance is non-optimal,this method takes advantage of the contribution plot based on virtual scale factor to get the change rate of contribution rate of the process variables to evaluation index and considers the mechanistic knowledge of process to trace the deep reasons causing the non-optimal performance.Ultimately,through deep analysis of the impact of high copper wet metallurgical processes critical factor,combined with field measured variables,practical operability and predicted dependent variable CEI,this thesis choose a few variables which are targeted and useful for modeling and evaluation.Finally,the feasibility and efficiency of the proposed strategies are demonstrated through the high copper hydrometallurgical process.