Coal Mine Motor Fault Diagnosis System Based On Particle Swarm Optimization Capsule Network

As an important equipment and the main power source in the coal mine,the motor has a wide range of applications in various occasions.Coal industry as a special industry,its reliable operation of the motor is almost harsh,if the malfunction of the motor damaged electrical equipment light,affect production progress,heavy will endanger the life and property safety of coal mining enterprises,cause irreparable loss,so how to quickly and accurately to the coal mine motor fault diagnosis and predicting maintenance,has very important practical significance.Taking induction motor as an example,this paper expounds several main types of induction motor faults,analyzes the common fault principles such as stator fault,rotor fault,eccentricity and so on in detail,and summarizes its fault characteristic signals,which lays a theoretical foundation for the principle and judgment method of mine motor fault diagnosis.In the fault signal processing part,considering that the fault signals inevitably contain noise,which affects the extraction of fault signals,the wavelet decomposition method is adopted to de-noise the fault signals,improve the readability of data,and retain most effective data information.Wavelet packet is used to decompose the fault signal in frequency domain,and the frequency band energy distribution diagram of the signal is obtained.The energy of different frequency bands is used as the feature vector to form the data samples for the training and test of the diagnostic model.In the part of fault diagnosis model,an improved capsule network based on particle swarm optimization is proposed for fault diagnosis and predictive maintenance model of coal mine motor.In order to effectively enhance the poor optimization ability of particle swarm optimization algorithm and solve the problem of slow convergence speed of particle swarm optimization algorithm,the optimization ability of improved particle swarm optimization algorithm is greatly improved by improving inertia weight factor,learning factor and position iteration formula.The improved capsule network algorithm adopts the strategy of small size convolution kernel and multiple convolution to extract the features of fault signals.The addition of pooling layer reduces the dimension of data,reduces the data burden of the model,and greatly improves the training speed of the model.The experimental results show that the proposed algorithm has high calculation precision and fast convergence speed,and can optimize the improved capsule network algorithm accurately,which greatly improves the accuracy of fault diagnosis of coal mine motor.Figure 21 table 6 reference 67...