Research On Intelligent Health Diagnosis System For Aluminum Electrolysis Cells

The Aluminum Electrolysis Industry (AEI) is a complicated industrial processwith high nonlinearity, strong interfered and uncertainty, lots of unknown factors,little detectable and usable information. All these disadvantages make it a bigobstacle for realization of the health diagnosis. So the Country's 863 Plan take it as animportant task in Industry Automation Field.In the dissertation, based on the field research, the technics process and faultmode of Aluminum Electrolysis Producing is introduced and analyzed in detail. Justlike the Anode Effect (AE),the cause of fault is intricate and the appearance isvarious. It is reasonable to combine multi diagnosis methods together, and utilize thecompensation of each method to improve the diagnosis effect. On such basis, ablackboard model structure is proposed and realized, which is suitable for themulti-expert cooperating diagnosis strategy.The correct rate of health diagnosis could be improved obviously if Fuzzy Expertmethod is combined with Nerve Network method. But Nerve Network method hasnot been mature enough in application for AEI health diagnosis. So it is mainlyintroduced how to use Fuzzy Expert method fault diagnosis in dissertation.Base on the feature of AEI and demand of 863 Plan, the Intelligent HealthDiagnosis System for Aluminum Electrolysis Cells (IHDSAEC) was designed anddeveloped independently. In dissertation, the software's task, total design, functiondesign, structure design, and main methods of system are introduced in detail,including the computing method of various parameters, the prediction method of AEand other fault.By the simulation with real data from factory, IHDSAEC make a good result forreal time alarm and classic fault diagnosis, especially for AE prediction. It will behelpful for reducing energy cost and increasing the benefit of factory.