Research On The Model Of Fault Diagnosis And Maintenance Decision-Making Based On The Hybrid Intelligence With Its Practice

Along with the development of computer technique and artificial intelligence, fault diagnosis and maintenance decision-making technique has been developed to a new level — intelligence fault diagnosis and maintenance decision-making. The intelligence diagnosis technique has been obtained at some aspects successfully as a result of the development of more than ten years, but it is not good at adaptability, learning ability, knowledge acquirements and the processing to inaccurate information. However, it is very difficult to solve these problems only depending on a single intelligent technique. Moreover, maintenance decision-making lacks the scientific evaluation method, and the intelligence level of decision-making is very low. So, it is of important theory significance and real significance how to integrate with many kinds of intelligence technique and study a hybrid intelligent model for diagnosis and maintenance decision, in order to stir the intelligent and scientific levels in diagnosis and maintenance decision-making. This paper will be inquired into and will study at the aspects that raise the intelligence level of diagnosis and maintenance decision. It will chiefly include: establishing a kind of hybrid intelligence model facing diagnosis and maintenance decision tasks, studying expressing methods and the processing techniques in diagnosing and maintenance decision-making, probing into the diagnosis method of incremental learning, and setting up the scientific system for maintenance decision-making, and etc.The research thread of this paper centers on the modeling of intelligent diagnosis and maintenance decision. Following are the major research contents.Firstly, the research actual status about the intelligent diagnosis and maintenance decision both inside and outside the nation has been summarized, and the existing problems have been pointed out.Secondly, the basic concepts of the hybrid intelligent system have been systematically analyzed. And then, the basic three problems that need to study on the hybrid intelligent system have been pointed out, which are the model of mixing intelligence technique, the knowledge expression and the inference technique. The five basic requirements that the diagnosis and maintenance decision system should be had have been studied, and in accordance with these 5 requirements, a kind ofhybrid intelligent model based on multi-states mixing mode have been set up.The third, two kinds of foundation but critical intelligence in the hybridintelligent model have been thoroughly studied------the fuzzy knowledge processingand the neural network technique. Aiming at inaccuracy existing in diagnosis and maintenance decision, a double-fuzzy knowledge expressing has been put forward, and the corresponding fuzzy inference method has been studied in detail which includes a generalized fuzzy matching arithmetic and a strategy clearing up conflict based on the matching degree and the belief degree. Then, a generalized fuzzy knowledge processor system platform has been developed. In order to solve the problems that generic ANN is lack of incremental learning and easily to make a wrong or a leak diagnosis, the paper has put forward a composite radial basis function network classifier oriented the diagnosis task, and studied the realization algorithm of this classifier in detail. The distinguished feature of this classifier lies in incremental learning, therefor, it can detect and recognize a novel fault.The fourth, based on the maintenance technology and the decision analysis theory, the paper has researched into maintenance scheme's evaluation and decision-making. By analyzing maintenance decision's course and effecting factors, the fifth chapter has set up synthetic appraisal objects, and studied the mathematical model of synthetic evaluation. Lastly, taking metal structure's evaluation for example, it has established a comprehensive appraisal system for maintenance scheme based on knowledge, and studied its structure and its components.The 5th, aiming at forecasting of loading ability in the faulty diagnosis oi structure, a non-parameter forecasting method based on the RBF network has been proposed in the 6th chapter. Taking Metal Structure as a object, the 6th chapter has researched into the modeling of loading capability and the forecasting steps.Finally, on the basis of the theoretical research results of this paper , the 7"chapter has designed and developed a practical system------" A decision supponsystem for faulty diagnosis and maintenance of metal structure of the port crane based on hybrid intelligence model".