Fault Diagnosis of Precision Screw Pump Based on Multi-Scale Convolutional Neural Networks

In the stator coating process of new energy automobile motor manufacturing, precision screw pumps have become the key equipment due to high-efficiency transportation and good adaptability, but they are prone to failures in long-term operation, which affects production efficiency and motor performance. Traditional fault diagnosis relies on manual labour, which is inefficient and inaccurate. In this paper, the three-axis vibration signals of the screw pump are collected, and the signal processing methods such as <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{3}\boldsymbol{\sigma}$</tex>, LSTM and robust normalization are combined with BP neural network, CNN and multi-scale convolutional neural network MSCNN, respectively, to construct a fault diagnosis model of precision screw pump based on MSCNN, with a classification accuracy of up to 92%, which realizes the timely and accurate diagnosis of the faults of the precision screw pump and improves the stability and overall performance of the stator coating process.