Research On The Fault Diagnosis Method Of Plasma Cutting Machine Based On Bayesian Network And ISSA-ELM

As an indispensable equipment in the steel cutting process of ships,the production process will be seriously affected in case of failure of plasma cutting machine.Therefore,it is of great importance to carry out fault diagnosis research on plasma cutting machine to ensure the stable operation of the equipment.In this thesis,a fault diagnosis method based on Bayesian network and ISSA-ELM is proposed for the problem of complex structure of plasma cutting machine and the difficulty of fault identification,in which Bayesian network is responsible for locating the cause of plasma cutting machine faults and ISSA-ELM is responsible for the diagnosis of specific fault types.The details of the study are as follows:(1)First,the structural composition of the plasma cutting machine was analyzed,the fault mechanism of the plasma cutting machine was studied,its common fault types and causes were summarized,and the fault tree of the plasma cutting machine was constructed;second,the fault tree of the plasma cutting machine was transformed into a fault Bayesian network according to the transformation relationship between the fault tree and the Bayesian network;finally,the nodes in the Bayesian network were assigned according to the expert experience and the maintenance records Finally,the nodes in the Bayesian network were assigned values based on expert experience and maintenance records to complete the construction of the Bayesian network and to locate the cause of the plasma cutter fault.(2)To address the problem of difficult feature extraction of plasma cutting machine fault signals,a feature extraction method based on Improved complete ensemble empirical mode decomposition with adaptive noise(ICEEMDAN)and entropy features was proposed.The method of feature extraction.ICEEMDAN was used to decompose the fault signal and obtain a series of Intrinsic mode functions(IMFs).The Spearman correlation coefficient of each IMF was used as the evaluation criterion to remove the invalid IMFs and keep the valid ones,and the fuzzy entropy and alignment entropy of the valid IMFs are calculated and used as the fault feature quantity.(3)For the problem of low diagnostic accuracy of specific fault types,an improved sparrow search algorithm(ISSA)was proposed to optimize the extreme learning machine(ELM).The population initialization method of the sparrow search algorithm was improved by introducing Iterative chaos mapping and Fuch chaos mapping to increase the population diversity;meanwhile,the adaptive dynamic factor and Levy flight strategy were used to improve the individual position update method to increase the model convergence speed.The ISSA algorithm obtained from the improvement was applied to the process of finding the optimal ELM weights and thresholds to complete the construction of ISSA-ELM model and realize the diagnosis of specific fault types.(4)Based on the above research,a plasma cutting machine fault diagnosis system was developed,various functional modules were designed,and a plasma cutting machine fault database was established.Based on the Bayesian network and ISSA-ELM algorithm,the localization of the causes of plasma cutting machine faults and the diagnosis of specific fault types were realized.