Research on a new hybrid intelligent fault diagnosis method and its application

Zhenhua Wang, Zhentao Liu, Xueyan Lan, Jian Liu, Shaowei Wang, Yangming Wu, Yanbing Xue
2016 International Journal of Smart Home  
In order to overcome the shortcomings of slow convergence speed and easy falling into the local minimum values of the BP neural network, an improved particle swarm optimization(PSO) algorithm is proposed to optimize the redial basic function (RBF) neural network, in order to propose a new hybrid intelligent fault diagnosis(IMPSO-RBFNN) method. In the IMPSO-RBFNN method, the adaptive dynamic adjusting strategy is used to control the inertia weight of the PSO algorithm in order to an improved
more » ... icle swarm optimization(IMPSO) algorithm. Then the IMPSO algorithm is selected to optimize the parameters of RBF neural network by encoding the particle and continuous iteration of the IMPSO algorithm in order to obtain the optimal combination values of the parameters of RBF neural network. The optimal combination values are regarded as the values of these parameters of the RBFNN for constructing the final IMPSO-RBFNN method. In order to test the effectiveness of the proposed IMPSO-RBFNN method, the data from bearing data center of CWRU is selected in this paper. The experiment results show that the IMPSO algorithm can effectively optimize the weights of RBFNN, the IMPSO-RBFNN method can accurately realize high precision fault diagnosis of rolling bearing. In this paper, an improved particle swarm optimization algorithm and RBF neural network are integrated in order to propose a new hybrid intelligent fault diagnosis(IMPSO-RBFNN) method. In the IMPSO-RBFNN method, the adaptive dynamic adjusting strategy is used to adjust the inertia weight of the PSO algorithm. Then the IMPSO algorithm is selected to optimize the parameters of RBF neural network by encoding the particle and continuous iteration of the IMPSO algorithm in order to construct the optimal IMPSO-RBFNN method. The deep groove ball bearing of JEM SKF from Bearing Data Center of Case Western Reserve University is used to test the validity of the IMPSO-RBFNN-FD method. The experiment results show that the IMPSO algorithm can effectively optimize the weights of RBFNN, the IMPSO-RBFNN method can obtain the higher fault diagnosis correctness rate for rolling bearing than the RBFNN-FD method and PSO-RBFNN-FD method. Authors Zhenhua Wang, male,born in 1990, College of electrical and information, Dalian Jiaotong University, 2015 graduate students, member of the Chinese electronic society, the main research direction for the development of embedded hardware and fault detection. Zhentao Liu, male, 2015 graduate students at the school of electrical and information, Dalian Jiaotong University, the main research directions for the development of embedded software and fault detection. Xueyan Lan, female, College of electrical and information, Dalian Jiaotong University, 2015 graduate students, the main research direction for traffic control intelligent and fault detection. Jian Liu, male, 2015 graduate students at the school of electrical and information, Dalian Jiaotong University, the main research direction for the traffic information control and communication. Shaowei Wang, male, 2015 graduate students at the school of electrical and information, Dalian Jiaotong University, the main research direction for the embedded development and control. Yangming Wu, male, 2015 graduate students at the school of electrical and information, Dalian Jiaotong University, the main research direction for the embedded development and control. Xue Yanbing, female, born in 1973, Professor of electrical and Information College, Dalian Jiaotong University, Ph.D., master's tutor, the main research direction for the semiconductor gas sensor.
doi:10.14257/ijsh.2016.10.7.14 fatcat:c2fum6gidjazljviuv3hbo6jze