Study On Multi-parameter Optimization Control Of Chlor-alkali Production Process

The chlor-alkali industry belongs to high energy consumption industry and is also an important pillar industry in national life.In order to reduce the energy consumption in the production process,it is very important to optimize control of the chlor-alkali electrolysis process.Because of the complexity of the electrolysis process and the variety of chemical properties and variables,establishing a chlor-alkali multi-parameter optimal control system to achieve the purpose of energy saving and consumption reduction is the emphases of the research.In this paper,it analyzed the influencing factors of the current efficiency and the direct current consumption in the process of the chlor-alkali electrolysis,and the control conditions which were suitable for the production range at first.Then,using the principal nuclear element method analyzed the production data that three main influencing factors were obtained.Because the structure of the mechanism model was very complicated,could also not achieve the ideal control effect,that the Elman neural network was chosen to build the model.In order to improve the control accuracy,after the establishment of the neural network model,the BP neural network was chosen as the controller of the control system.However,the BP neural network had some shortcomings.Through the analysis and research,along to cover the shortages,then the improved quantum multiobjective evolutionary algorithm was chosen to optimize the weights and thresholds of BP neural networks.In the improvement of quantum multiobjective evolutionary algorithm,it used the undominated sorting operator and the selection strategy that based on Pareto individual gravity.In order to make the improved quantum multi-objective evolutionary algorithm had better optimization effect on BP neural network controller,it proposed the algorithm fusion method and designed the concrete fusion method that made the quantum multiobjective evolutionary algorithm sufficient optimization the weights and thresholds of BP neural network controllers.Over time,as some related factors such as production equipment would also change in the actual production process,so that the neural network control system trained in advance was no longer suitable for the current production equipment.In order to solve those problems,the online correction strategy of the control system was proposed.Firstly,it selected the production data and added to the database,then judged the confidence level of the control system,and designed the online switching of the dual control system at once.Then the neural network in the control system could be trained offline on the premise of ensuring the stable operation of the system.Ultimately,through computer simulation,it compared study the results of the production process of artificial parameter design,the off-line chlor-alkali multi-parameter optimization system,and online chlor-alkali multi-parameter optimization system.The results verified that under the same conditions,it was effective in the chlor-alkali multi-parameter optimization system and on-line correction.Figure 28;Table 5;Reference 43.