Gear Fault Intelligent Recognition Based On Neural Network

In the context of global information,industry is developing rapidly,and the development direction of rotating machinery equipment tends to be complex structure,process automation and intelligent operation.Gears are widely used in modern equipment,such as aviation,mining,agricultural machinery,military equipment and metallurgical machinery,due to their large carrying capacity and high transmission efficiency.Gear is a transmission component with a high failure rate during equipment operation.Gear failure will directly affect the reliable operation of the whole machine,reduce the efficiency and accuracy of the entire equipment,and even cause irreversible serious consequences.Therefore,it is of great significance to carry out gear fault diagnosis and identification.Acoustic emission detection is a non-destructive testing technology.Acoustic emission technology can not only perform dynamic and real-time monitoring,but also obtain defect information.Compared with other detection methods,especially compared with traditional vibration signal analysis methods,its sensitivity is higher.In order to solve the problem of difficult feature extraction of acoustic emission signal,avoid relying on traditional time-frequency analysis method for gear fault diagnosis,and improve the accuracy of identification,this paper introduces deep learning based on acoustic emission detection technology,and adopts acoustic emission detection technology.A method of combining emission technology with neural networks.First by acoustic emission equipment acquisition of various fault type gear of the acoustic emission signal,then the collected signal to make use of principal component analysis feature extraction and dimensionality reduction,will signal the reasonable division of the neural network training set and testing set construction of probabilistic neural network and generalized regression neural fault recognition model as the core,and finally optimize the network to join the Harris hawk algorithm,According to the recognition accuracy,the most suitable gear fault recognition model is built,and the intelligent recognition of gear fault is realized.The first is to build a gear fault identification model based on a probabilistic neural network,input the processed data into the neural network,and obtain an iterative curve and confusion matrix.The results of confusion matrix show that the probabilistic neural network gear fault identification model based on Harris Hawks optimization has a recognition accuracy of 98.08%.Compared with the gear fault identification model of probabilistic neural network without optimization,the accuracy is increased by 7.38 percentage points.Secondly,a gear fault identification model based on generalized regression neural network is built to compare and verify the advantages of the two models in different states.The results show that the accuracy of the generalized regression neural network gear fault identification model based on Harris Hawks optimization is 92.88%,which is 5.2 percentage points lower than that of the probabilistic neural network gear fault identification model based on Harris Hawks optimization.For small sample fault identification,the accuracy of generalized regression neural network will be higher.