Study On Process State Fault Diagnosis Method Based On Aluminum Electrolysis Knowledge

Aluminum electrolysis industry is a complex process industry,its production process has a high degree of non-linearity,large lag and many other adverse factors.The occurrence of failure will have a great impact on the whole electrolysis series and various production technical indicators,resulting in the decline of aluminum output and quality,and need to consume a large amount of energy consumption.Therefore,the fault prediction and diagnosis can be effectively carried out in the process of aluminum electrolysis production,so as to improve production efficiency and save electric energy.Based on the actual production of an aluminum plant in Guangxi,this paper studies the process state fault diagnosis method based on aluminum electrolysis knowledge.(1)Through the research and research of this topic,the study of the aluminum electrolytic process state fault scheme,that is,based on the aluminum electrolytic process knowledge combined with data-driven method to solve the aluminum electrolytic fault problem.Outliers are eliminated and correlation analysis is carried out on the collected data,so as to eliminate problems caused by different dimensions in subsequent calculation.(2)The parameters in the electrolytic cell will have characteristic changes when faults occur.According to this theory,the correlation parameters of the electrolytic cell were analyzed,and the current efficiency and electrolytic temperature were taken as the performance indexes to evaluate the condition of the cell.The evaluation model of electrolytic cell was established by the improved twin support vector machine.The simulation results show that the soft sensor model established in this paper has a high classification performance and can be used to evaluate the groove condition.(3)The failure data in the process of aluminum electrolysis production belongs to the characteristics of the "small sample,poor information",to solve this problem,this section generate against network is proposed based on the expansion of aluminum electrolytic process data,designed to increase the number of aluminum electrolytic process failure data and the diversity of samples,experimental data foundation for the follow-up.(4)On the basis of the above research,the principal component analysis method is firstly used to monitor the main parameters in the production process of aluminum electrolysis to determine whether there is a tendency of failure;When the fault is judged,combining with the knowledge of aluminum electrolysis production process,the Bayesian network is used to determine the type of fault,and the fault is transmitted to the field operators.