A Gear-box Fault Diagnosis Method Based On Improved Fish-swarm Optimization Algorithm To Optimize The Neural Network

In recent years,China has been developing rapidly in the field of mechanical manufacturing.The successful completion of one piece of superengineering,such as the humaling tunnel,the Rejuvenation high speed train and the hong kong-zhuhai-macao bridge,shows the growing strength of China's manufacturing industry.The mechanical equipment plays an extremely crucial role.The gearbox is the most commonly used and the most important transmission part in the actual industrial equipment,and its ability to maintain a healthy state during the working process will become an important part of the actual industrial production.Therefore,the research of gearbox fault diagnosis technology has very real sense.BP neural network is a kind of intelligent fault pattern recognition technology,which has strong nonlinear processing power,but also has the disadvantage of being prone to local extremum value.This paper discussed an improved fish-swarm algorithm,and the initial weight and threshold of BP network are optimized by using the algorithm's excellent global optimization capability.The ADAFSA-BP network model is constructed by combining their advantages,and the feasibility of the model is verified by the fault diagnosis of gearbox.Firstly,this paper introduces the research background and significance of the project,included the present situation of research on fault diagnosis of gearbox,neural network and fish-swarm algorithm,as well as the current development of intelligent diagnosis technology.Then the paper analyzed the common failure modes and vibration mechanism of gearbox,and introduced some time-frequency domain analysis methods.The fish swarm algorithm and its improvement mode are discussed in detail,included parameter setting and behavior rule description in the algorithm.The improved fish swarm algorithm is combined with BP neural network to optimize the initial weight and threshold.And the ADAFSA-BP network model is constructed and simulated in MATLAB,which proves the feasibility of the model in pattern recognition.Finally,this paper set up the gearbox fault diagnosis experiment platform,and collected the vibration signals of the JZQ250 gear box.The ADAFSA-BP network model is used to identify the fault state of gearbox.The experimental data prove that the ADAFSA-BP network model shows better diagnostic ability and provides a new intelligent identification method for gearbox fault diagnosis.