Study of Fault Diagnosis for Rolling Bearing Based on GA-BP Algorithm
Yue Wen,Minping Jia,Cheng Luo
DOI: https://doi.org/10.2991/amcce-17.2017.137
2017-01-01
Abstract:With the development of modern technology, the diagnosis technology for mechanical equipment has attached more attentions. This paper mainly presents the fault diagnosing process for rolling bearing by using the hybrid GA-BP algorithm, which combines genetic algorithm (GA) with error back-propagation algorithm (BP). The weights and thresholds of BP neural network are optimized by using genetic algorithm firstly and then assigned to the BP neural network. With the using of proposed approach, the diagnosis performance of BP neural network can be improved. In this paper, the optimal nonlinear mapping relationship between fault types and fault symptoms of rolling bearings is obtained and then realize failure recognition of rolling bearings though the hybrid GA-BP algorithm. Finally, the results of the test experiment verifies the efficiency and accuracy of the proposed GA-BP algorithm. Introduction As an important economic loss and damage for the country, the failure of mechanical equipment, especially the damage of key mechanical component has been brought more attention. As a result, the mechanical equipment fault diagnosis technology is extremely important. Update to 2016, the failure of rotating machinery caused by rolling bearing faults accounted for 30%. Therefore, the fault diagnosis for rolling bearings is essential. In recent years, artificial intelligence has been developed and artificial neural network is widely applied in many fields [1] including fault diagnosis as an original modeling technique. Artificial neural network is a mathematical model that imitates animals’ neural network behavior characteristics to realize distributed parallel processing of information. Among several kinds of artificial neural network models, the application of BP neural network with a single hidden layer accounts for 90%. However, the disadvantage of the conventional BP algorithm is that it is easy to fall into the local minimum and the convergence speed of the network weights and thresholds is too slow. GA algorithm has a strong ability of macroscopic search and can find the global optimal solution with large probability. The hybrid algorithm combines GA and BP algorithm provides a new approach for the fault diagnosis. Theories of BP Neural Network and Genetic Algorithm Artificial neural network (ANN) is a kind of information processing system, and the neuron is the basic component unit of neural network. Multiple neurons interconnected together, resulting in a neural network structure of adaptive nonlinear dynamic system. Artificial neural network is generally composed of input layers, hidden layers and output layers. In practice, the artificial neural network containing only one hidden layer is called BP neural network. A typical BP neural network model is shown in figure 1 below. 2nd International Conference on Automation, Mechanical Control and Computational Engineering (AMCCE 2017) Copyright © 2017, the Authors. Published by Atlantis Press. This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/). Advances in Engineering Research, volume 118