Research On Fault Diagnosis Of Rotary Kiln Based On Bayesian Network Algorithm

As one of the core equipment in the new dry process cement production line,the operation status of the rotary kiln has a great impact on the quality,output and energy consumption of cement,so it is of great significance to complete the fault diagnosis of the rotary kiln efficiently and accurately.However,the fault diagnosis of rotary kiln still has the problems of modeling difficulties and poor accuracy of modeling parameters,so this paper uses the improved Bayesian network structure learning algorithm to establish the rotary kiln network model,and the parameter learning algorithm optimizes the parameters of the rotary kiln network and realizes the diagnosis of the fault nodes of the rotary kiln.This project focuses on the fault diagnosis modeling of rotary kiln based on Bayesian network algorithm,and the specific research content is as follows.Firstly,aiming at the problem that Bayesian network structure learning algorithms are prone to fall into local optimum,a hybrid simplified particle swarm algorithm is proposed to optimize Bayesian network structure learning algorithms.The algorithm uses the maximum support tree strategy to constrain the search space,the climbing algorithm constructs the initial population,and the paraparticle mitigation strategy and the conditional crossover and mutation strategy are iteratively optimized,so as to obtain the optimal Bayesian network structure,which is then used to solve the problem of difficult fault diagnosis network modeling of rotary kiln.Secondly,aiming at the problem of low parameter accuracy in Bayesian network parameter learning algorithm,a differential algorithm based on prior conditions is proposed to optimize the Bayesian network parameter learning algorithm.The algorithm takes the monotonicity of parameters and uniform distribution of parameters as a priori conditions,encodes the probability distribution of network nodes,and iteratively optimizes by using the crossover,mutation,natural selection and other operations in the improved difference algorithm,so as to improve the accuracy of Bayesian network parameter learning,which is then used to solve the problem of poor accuracy of network modeling parameters in fault diagnosis of rotary kiln.Finally,the improved Bayesian network structure learning algorithm and parameter learning algorithm are used to establish a network model for potential fault nodes of rotary kiln,and the variable elimination method is used to learn the fault nodes by reasoning,and the fault diagnosis of the rotary kiln is realized according to the reasoning results,and some real data collected by the Jidong cement plant are used to verify the feasibility of the algorithm for actual fault diagnosis.