The Research On Prediction System For Anode Effect

The electrolytic cell is the main production equipment in the aluminum production process,and it is also the main object of analysis and monitoring to the anode effect prediction system. The electrolytic cell worked normal or not is not only related to the normal state of aluminum in the economic and technical indicators, but also related to the life of electrolytic cell and the normal production of aluminum. Aluminum reduction cell is a nonlinear, multi-coupled, time varying delay system of industrial process. Therefore, in aluminum electrolysis process, the groove is characterized by complex and volatile situation, the types of failure are many and difficult to detect. Such as:anode effect, cold tank, hot tank, anode lesions and unstable ducts. Some fault Occurred and could affect the security of the entire aluminum production, resulting in huge economic losses. Thus, prompt and correct prediction of aluminum tank failure is an important guarantee of normal production. Therefore, In this paper, we have researched the anode effect prediction and developed the Anode effect prediction software system.First, the background of the aluminum, the production principle and the Development of prediction was introduced.Second,the failure mode of aluminum cell are described. The system starting from the tank resistance signal and set a neural network model. Elman neural network is a local memory unit and local feedback connections forward neural network. His feedback connections from its hidden layer output to its input contact. Elman neural network will be applied to anode effect prediction of aluminum cell. Using MATLAB write the corresponding program, the ultimate success of the corresponding prediction of failure to achieve the desired requirements.Using Viusal Basic.NET and SQL Server 2000 designed this system. This paper describes the development of software and development environment.After the final test can be found on the network, the network can accurately identify all the fault type. Compared with BP network, Elman network, in terms of training or the running time on the number of greatly increased...