Research On Intelligent Fault Diagnosis Method Of Rotating Machinery

With the development and progress of modern industrial technology,rotating machinery is playing a more and more important role in the production of enterprises.At the same time,the rotating machinery is developing towards the trend of large-scale,automation and intelligence,and the mechanical structures also become more and more sophisticated.While these new developments of rotating machinery help to improve the production efficiency and reduce the production cost,they raise more requirements and difficulties to ensure the reliable operation and fault maintenance of rotating machinery.Hence,how to diagnose the fault of rotating machinery in time and accurately becomes an important research problem.The fault diagnosis of rotating machinery has a broad application prospect in enterprise production system.The vibration mechanism of rotating machinery is complex,and the induced vibration signal is of nonlinearity and non-stationarity,and the characteristic signals of failure contain numerous strong noises,which makes it difficult to extract fault features effectively and even affects the accurate diagnosis of faults.In the diagnosis of rotating machinery,fault feature extraction and pattern recognition are the key.The rapid development of modern signal processing technology and artificial intelligence technology provides some new technical approaches for fault diagnosis of rotating machinery.In the dissertation,modern signal processing and artificial intelligence are used as tools,and taking the key components(i.e.,bearings and gears)of rotating machinery as the research objects.In summary,the study of the work involves the following main contents:1.The vibration signal generated by the rotating machinery is not stable and irregular,and hence it is difficult to directly diagnose the fault through the vibration signal waveform analysis.A fault diagnosis method based on signal mode decomposition is studied.First,the vibration signal is decomposed into several intrinsic mode functions,and then the fault features are extracted from each intrinsic mode function,and the eigenvector set is constructed.Then the feature dimension reduction is carried out.Finally,the naive bayes classifier is used for fault identification.CWRU bearing data set is used in the simulation experiment,and the performance of fault diagnosis is good.2.Aiming at the problem that the parameter value setting of support vector machine is not reasonable,which affects the accuracy of fault diagnosis,the fault diagnosis of rotating machinery based on parameter optimized support vector machine is studied.First,the parameters of support vector machine are optimized by two methods:(1)the parameters of support vector machine are optimized by quantum genetic algorithm,and(2)the parameters of support vector machine are optimized by improved quantum particle swarm optimization algorithm.Then the training data set is trained to optimize the support vector machine.Finally,the test data set is input into the support vector machine model for fault identification.The experimental results show that the proposed method has a high accuracy of fault diagnosis.3.Aiming at the parameter setting problem of the relevance vector machine(RVM),the fault diagnosis method of rotating machinery based on bat algorithm optimization relevance vector machine is studied.First,bat algorithm is used to optimize kernel function parameters of the relevance vector machine.Then the relevance vector machine model is trained.Finally,the trained relevance vector machine is used for fault identification.Simulation experiments were conducted to compare BA-RVM fault diagnosis method with SVM method and RVM method.The experimental results show that the accuracy of BA-RVM fault diagnosis method is higher than that of SVM method and RVM method.4.In view of the complex symptoms of rotating machinery,and the vibration signals obtained from the measurement of rotating machinery are affected by uncertain factors,thus affecting the accuracy of the diagnosis results,an information fusion fault diagnosis method based on improved evidence theory is studied.Two improved evidence synthesis methods were studied:(1)evidence synthesis method based on Tanimoto similarity measure and information entropy,and(2)evidence synthesis method based on static discounting factor and weight coefficient.These two methods are applied to the fault diagnosis of rotating machinery.First,the fault features are extracted from the mechanical vibration signals to form the feature vector set.Then support vector machine and K nearest neighbor algorithm are used to obtain the basic probability assignment value respectively.Finally,the improved evidence synthesis method is used to fuse these evidences,so as to form the final diagnosis results.