Research On Hydraulic Pump Fault Diagnosis Method Based On Big Data Cloud Platform And Deep Learning Network CNN

With the development of science and technology,hydraulic equipment is developing in the direction of intelligence,large-scale and systematization.In the era of big data,hydraulic equipment has many monitoring points,high sensor sampling frequency,long equipment life cycle,and large amount of operating data.Fault diagnosis technology is gradually transitioning from focusing on traditional knowledge-driven methods to focusing on data-driven methods.How to automatically mine features from massive data and ensure the correct rate of fault diagnosis has become the top priority of fault diagnosis.Deep learning is good at mining features from massive data,and has good generalization ability,which has attracted more and more researchers’ attention.And Convolutional Neural Network(CNN)is a classic and widely used structure in deep learning.Therefore,this paper proposes a new convolutional neural network "WCNN(Convolutional Neural Networks with Wide First-layer Kernel)" based on the convolutional neural network.The first-layer wide kernel can automatically learn fault-oriented features,eliminating the time-consuming and laborious feature extraction process of traditional fault diagnosis.The size of the remaining convolution kernels are all 3×1 small convolution kernels,which avoids too many convolution kernel parameters and effectively suppresses the problem of network overfitting.Taking hydraulic pumps as the research object,the original vibration signal can be well diagnosed.The recognition rate of the model on the hydraulic pump data is over 99%.Moreover,in the case of introducing noise,it is found that the first-layer wide convolution kernel also has good noise suppression.The WCNN model is compared with the traditional feature extraction fault diagnosis algorithms FFT-SVM,FFT-KNN,FFT-BP and FFT-DNN under the condition of changing the signal-to-noise ratio.The results show that the constructed WCNN has the highest diagnostic accuracy rate.This article describes the WCNN algorithm in detail,introduces the methods used in the training model to prevent overfitting,explores the influence of different parameters on the network model WCNN,and analyzes the performance of the WCNN model in a noisy environment.The hydraulic pump fault data is collected through the hydraulic pump fault simulation experiment platform,and the experimental data is processed using the WCNN model.Compared with the results obtained by other algorithms,the model diagnosis results obtained are greatly improved.Finally,based on the Huawei Cloud Model Arts cloud service platform,the model is deployed to the cloud to realize the fault diagnosis of the hydraulic pump based on the cloud platform.