An improved self-organizing mapping neural network and its application in fault diagnosis of CNC machine tool servo drive system

Abstract

Computer numerical control (CNC)-based systems are key functional components of industrial manufacturing installations, and the servo drive system is the main functional component of CNC systems. The complex working environment of industrial facilities will lead to the frequent failure of servo drive systems, and effective fault diagnosis measures are important to ensure the normal operation of CNC machine tools. In this paper, the application of fault diagnosis methods in servo drive systems is considered, and a method suitable for high-dimensional data of CNC systems is presented. Using data collected by a physical system and related indicators, the technique can be used to identify hidden fault characteristics in the data and to diagnose the fault types. The core of this method is the self-organizing map neural network, which uses unsupervised competitive learning to cluster data with different characteristics, find the winning neurons, and diagnose the fault. The introduction of feature standardization in the map’s initialization stage can accelerate the model’s training convergence speed and reduce the feature weight deviation. At the same time, principal component analysis is introduced to balance the influence of different feature scales, enhance the features of fault data, reduce the data dimensionality, and improve the interpretability of the model. A comparison with the conventional algorithm and testing various fault datasets shows that the proposed method exhibits improved performance when processing high-dimensional data and its enhancement fault recognition effect is verified.