Convolution Diagnosis Model of Centrifugal Pump Based on Fractal Dimension

Abstract

Nowadays, more and more centrifugal pumps are used in the industrial domain. Normal operation of centrifugal pumps is also the key to maintain stable operation of production and process. So it is very important to monitor the operation status of centrifugal pumps real time on-line. In order to diagnose the fault of centrifugal pumps, we proposed a Convolutional Neural Network (CNN) fault diagnosis model, based on Fractal Dimension (FD) features. Firstly, we collected the vibration signal of centrifugal pumps and preprocessed them. Then we used the Empirical Mode Decomposition (EMD) to calculate the fractal dimension and the fractal matrix of the signal. Finally, we constructed the CNN model to diagnose and classify the fault of centrifugal pumps. The experiment shows that the accuracy of the proposed model can reach over 90%.