Quantitative Prediction of Axis Alignment Based on Magnetic Flux Leakage Detection and GA-BP Algorithm

In solving the problem of shaft alignment detection, traditional shaft alignment methods such as the gauge method, double table method, and laser method can detect shaft alignment in engineering. However, there are still problems with low accuracy and cumbersome detection steps in traditional shaft alignment detection. In order to improve the accuracy of axis alignment detection and simplify the tedious steps of axis alignment detection, we propose a new method based on the combination of magnetic flux leakage detection and artificial neural network. Firstly, by using the grey correlation analysis method in data feature correlation, the features that have a significant impact on the leakage magnetic flux data of the axial alignment model are selected, and a negative feedback neural network model is established for prediction. In order to solve the problem of traditional BP neural networks randomly selecting thresholds and easily falling into local optima, we introduced a genetic algorithm to optimize the random threshold of BP neural networks and improve the prediction accuracy of nonlinear problems and the global search ability of the algorithm. The experimental results show that our proposed method can significantly improve the detection accuracy of axis alignment and effectively achieve quantitative detection of axis alignment based on magnetic flux leakage detection.