Fuzzy Petri Nets Fault Diagnosis Techniques And Learning Capability Studying

With the rapid development of modern industry and science technology, the industrial sector run safely to improve economic efficiency , as well as the country's economic strength, have very important significance. Therefore, it is necessary to research new effective and accurate method of fault diagnosis. The present dissertation uses Petri nets as a tool to study the fault diagnosis. The principal research results and contents can be outlined as following:First, introduced the fault diagnosis expert system and Petri nets and fuzzy Petri nets, introduced a fuzzy Petri net algorithm.Secondly, based on the reverse and forward reasoning a fuzzy Petri nets diagnosis is studied, used fuzzy Petri nets reverse search algorithm to model reduction, then used forward fuzzy reasoning algorithms to calculate enable the fault diagnosis to achieve precise credible. The result of a relevant example has shown that algorithm is correct.Finally, weighted fuzzy Petri nets learning ability is studied. Introduced weighted fuzzy production rules and reasoning of several confidence. Weighted fuzzy Petri nets model (WFPN) is defined. For the functions is continuous, used the sigmoid function to built max and min functions. Learned algorithm for strict hierarchical WFPN model based on neural network weights and thresholds. Used weighted fuzzy Petri nets with neural networks such as BP (Back Propagation) network to study the WFPN that is difficult for hierarchical. Introduce a variable rate to speed up the algorithm convergence rate. The result of the example has shown that algorithm is correct.