Research On Fault Diagnosis Of Rotating Machinery Based On Ontology

With the development of information technology,intelligent fault diagnosis technology has been paid more and more attention.Reasoning method based on device fault domain ontology and SWRL rules is a commonly used intelligent fault diagnosis technology.High quality device fault domain ontology requires rich and accurate relations between instances.The unstructured text of fault domain contains a lot of knowledge,which can meet the requirement of instances relation extraction.However,the traditional neural network based relation extraction methods may mix up closely-related relations.Meanwhiles,because the distances between instances are far apart in the equipment failure domain texts,there exists difficulty in the task of instances relation extraction.Moreover,due to the existence of environmental interference and artificial error in judgment,it is difficult to obtain complete and accurate fault features of some underlying faults.As a result,the diagnostic result(the fault cause)is uncertain,so the uncertainty of the fault causes needs to be measured.However,the fault diagnosis method based on ontology and SWRL rules is not good at dealing with the uncertainty problem,and it is difficult to measure the uncertainty of the fault causes effectively,so it is difficult to obtain reliable results.To solve the above problems,this paper proposes a fault diagnosis method based on ontology,SWRL rules and Bayesian network.The main work of this paper is as follows:(1)This paper proposes a method to extract equipment failure domain ontology instance relations based on a hierarchical structure.First,by identifying the classes of the instances,extract relation between instances preliminarily.Second,classifiers based on joint self-attention based Bi LSTM model are used to classify relations between instances accurately,and the relation extraction is completed by statistical method.Experimental results show that compared with the existing neural network-based methods,this method achieves better results and can obtain rich and accurate instances relations.(2)This paper uses a method based on SWRL rules and equipment fault ontology to infer the fault occurring in the equipment.A method based on Bayesian Network can obtain the probability of the underlying fault causes.The fault cause with maximum probability will be selected to obtain the corresponding maintenance strategies through the reasoning based on SWRL.In this paper,the fault ontology of a rotating machinery test platform is builded,and the Bayesian Network of a gear box fault knowledge is constructed.In the process of constructing fault ontology of rotating machinery test platform,the effect of instance relations extraction method based on hierarchical structure is tested,and it will be compared with the traditional instance relation extraction method based on neural network.The effect of the proposed fault diagnosis method is verified by gear fault cases obtained from the test platform.