Research Of Motor Fault Diagnosis Method Based On Bayesian Network

Along with the development of science and technology, electrical equipment is widely used in various fields, therefore, it also put forward higher to motor fault diagnosis technology. In the field of the motor fault diagnosis, many scholars use wavelet analysis, artificial neural network, the expert system for the motor fault diagnosis system, but because the motor structure is complex, and the relationship of each parts is quite closely, therefore, motor fault diagnosis system usually exist unascertained information, and how to properly and accurately deal with these uncertain information, it is required to constantly perfect and improve in the field of the motor fault diagnosis.For the uncertain information in the field of the motor fault diagnosis system, it uses D-S evidence theory of the information fusion technology to analyze and fusion processing the collected motor fault characteristics signal. D-S evidence theory provides a strong method for the uncertain information expression and synthesis. And then, it uses the Bayesian network to diagnose the signal of motor fault features, until the motor fault is reasoned. Bayesian network is a directed acyclic graph, it can use a conditional probability distribution directly express the dependencies between variables. At the same time, it use Bayesian network that is based on the Bayesian method for motor fault diagnosis, so that it can make the result of the diagnostic reasoning more accurately and reasonably.It provides a new method which is the combining of Bayesian network and decision tree, namely the decision tree-Bayesian network. This method is based on the respective characteristics of Bayesian network and decision tree for the uncertain information in the field of motor fault diagnosis system. This paper expounds the basic ideas and relevant algorithm of the decision tree and Bayesian network, and then through the analysis and summary of the decision tree and Bayesian network, it advances the transformation algorithm of the decision tree-Bayesian network model, and finally through the clique tree reasoning algorithm of Bayesian network, it can be quickly and accurately reached the final motor fault diagnosis results. It can solve the uncertain information of the motor fault diagnosis system effectively; meanwhile, it can be obtained more accurate and efficient diagnostic results.