Research On Multilevel And Multiple Faults Diagnosis Method For Aluminum Electrolysis Process

Aluminum electrolysis process is complex,the electrolysis process by the temperature field,magnetic field,electric field interaction interference.Therefore,the difficulty of control,Fault occurred frequently,Some failure in the event,causing serious economic losses,Seriously affecting the quality and yield of aluminum.Therefore,the effective diagnosis of aluminum electrolysis process failure,Timely processing failure,is particularly important,To improve the yield and quality of green is of great significance.This paper presents a new method for fault diagnosis and prediction of aluminum electrolysis,Design of aluminum electrolysis fault diagnosis and forecasting system,The development of aluminum electrolysis process management platform,So that aluminum electrolytic failure to be real-time monitoring.Specific work as follows:Firstly,the aluminum electrolysis process of an aluminum plant as the research background,The development status of aluminum industry and aluminum electrolysis fault diagnosis at home and abroad are expounded.The significance of aluminum electrolysis fault diagnosis is summarized This paper discusses the existing problems in the diagnosis of aluminum electrolysis and a new fault forecasting method is proposed.Second,the model based fault diagnosis method is studied.Combine system identification with parameter estimation,and verify its effectiveness,through the model of the method can quickly diagnose the occurrence of aluminum electrolysis failure.Thirdly,the principle and method of neural network application in fault diagnosis are expounded,analyze its effectiveness and necessity.by selecting the feature quantity that reflects the anode effect,the data were processed by nonlinear principal component analysis.in order to achieve the purpose of reducing the dimension and decoupling.The training process of neural network is optimized by genetic algorithm.Finally,the feasibility of multi-level and multi-fault diagnosis method is verified by simulation experimentFourthly,in order to improve the real-time and accuracy of fault diagnosis,according to the characteristics of faults in aluminum electrolysis process.A new fusion fault diagnosis method based on model diagnosis and intelligent diagnosis is proposed,the system identification is combined with parameter estimation and fuzzy neural network fault diagnosis,Through the fault diagnosis,fault classification process,achieve the objective of the identification of aluminum electrolytic fault types and fault prediction.In this paper,we use fault diagnosis model to judge the fault,achieve fault classification by neural network,multi-layer fault Integrated diagnosis mode through fault recognition and Optimization of fault Type by Neural Network not only to improve the real-time fault diagnosis,and improve the accuracy of the system fault diagnosisThe fifth article uses a model fault diagnosis,Multi-layer Fault Diagnosis Model for Neural Network Fault Classification,fault Type Recognition and Optimization by Neural Network,the validity of the fault diagnosis method is verified by simulation,proof of practice,not only improve the real-time fault diagnosis,but also improve the accuracy of system fault diagnosisSixthly,the development of aluminum electrolysis process control system fault diagnosis and management platform,to achieve optimal management,real-time fault alarm,real-time query function...