Research On Fault Diagnosis Method Of Asynchronous Motor Based On Deep Learning

With the development and progress of science and technology,the automation-degree of motors used in various fields is gradually improved,and the structure of motors is gradually becoming more and more complicated.Because many motors are in a long-term continuous operation and maintaining a highintensity working state,so the motors will inevitably produce a variety of different faults.This paper takes asynchronous motor as the research object to study two methods of fault diagnosis for asynchronous motors based on deep learning.Firstly,this paper uses a kind of fault diagnosis method of asynchronous motor based on deep belief network.It is based on the deep belief network and building a fault diagnosis model applied to asynchronous motors.Studied the internal structure of the network,the initialization steps of the fault diagnosis model and the methods commonly used to train the model.And the experiment shows that this model has some disadvantages,for example,its internal arithmetic is very complicated and the ability of fault data feature extraction is insufficient.Then,for the disadvantages of the deep belief network,I put forward a fault diagnosis method of asynchronous motor based on the stacked auto encoder network,studied its internal structure and process of derivation in detail.Built a fault diagnosis model combining the stacked auto encoder network and Softmax classifier,studied the internal structure of the network,the initialization steps of the fault diagnosis model and the methods commonly used to train the model.Compared this fault diagnosis model with the model based on the deep belief network,and verified the method of fault diagnosis for asynchronous based on the stacked auto encoder network has many advantages by experiments.Finally,the fault diagnosis model based on deep belief network and the fault diagnosis model based on stacked auto encoder network with Softmax regression classifier are compared.The experimental results show that the fault diagnosis method based on stacked auto encoder network can achieve higher diagnosis accuracy.Compared with the traditional fault diagnosis method,the deep learning method in this paper has advantages in the fault diagnosis of asynchronous motor,which shows that the deep learning method is suitable for the fault diagnosis of asynchronous motor.