Fault Diagnosis Of Motor Production Line Based On GA-BP Neural Network

Due to the complexity relationship between various of machinery and electronic equipment, the equipment reliability and maintainability are decreased, which often exit in hanging type air-conditioning fan motor automatic production line. Once equipment failures occurrence, even sector failures, will make the whole production line to shut down, and result in great economic loss. This even threaten the personal safety of workers. Thus, it can be seen that the fault diagnosis of production equipment has irreplaceable status. The fault diagnosis technology must be studied systematically and scientifically. Not only be guided by the original sensory experience, but also guided by the scientific methods to serve for practice. Therefore, it is significance to study the fault diagnosis technology of the equipment.The automatic production line of the electric machine is composed of mechanical, electric, pneumatic components and thus the symptoms of the production line are diverse and causes more trouble. And the nature of the fault is often different, which leads to the failure of the motor automatic production line has the characteristics of hierarchy, randomness, correlation, coupling and so on. According to the above fault features, this article will use the method of BP neural network to diagnose the fault of the production line. Simultaneously, the weights and thresholds are optimized by using the genetic algorithm of network. Finally, through the data fusion method, the fault information decision of each diagnosis sub network is fused.According to the actual situation of the production line, the automatic pressure stator station minimum system with high frequency of failure is chosen as the research object in this paper. Building fault diagnosis subnet A1, the fault diagnosis of the X shaft electric cylinder system is carried out. At the same time, considering the coupling of the equipment, fault diagnosis subnet A2 of stator motor feeding system is built, and the subnet A3 of the stator Y shaft cylinder system is set up. Using genetic algorithm, BP neural network initial weights and thresholds are optimized. And the data fusion technology is used to fuse the fault information of the three fault diagnosis sub networks.The results of sub network A1 showed that:the BP neural network and genetic algorithm optimization of the BP neural network, respectively, to achieve convergence in the 170 step and the 34 step, the convergence rate is greatly improved; meanwhile, it can be seen from the error curve that the GA-BP network error curve is smaller, and the operation process is more stable; test results of 30 sets data showed that the accuracy of fault was 93.3 percent. more fault diagnosis information is given with fusion subnet, more than any others. Thus, the diagnosis is more realistic and the accuracy of fault diagnosis is improved.