Fault classification of meta-action unit using CEEMDAN double-layer decomposition and COA-SVM

To diagnose the meta-action unit of CNC machine tools efficiently and accurately, a fault diagnosis model of a meta-action unit based on envelope entropy double-layer decomposition and an improved Support Vector Machine (SVM) was established. First, using the worm rotation element action unit as the research object, the Complementary Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) algorithm and envelope entropy analysis were combined to decompose the vibration signal in a double-layer manner, and the IMF component was reconstructed according to the Pearson correlation coefficient. Second, the time-domain and frequency-domain features of the reconstructed signal were extracted, and the model assigned weights to each feature using the ReliefF algorithm and rounded off the features that were less than the weight threshold. Third, the model is optimized by the Coati Optimization Algorithm (COA) to optimize the fault classification performance of the SVM to achieve efficient and accurate fault diagnosis of the meta-action unit of the model. In the experiment, the proposed fault diagnosis model based on envelope entropy double-layer decomposition and COA-SVM was verified, and the results showed that the average diagnostic accuracy of the model reached 97.7962%, which can be used for efficient and accurate fault diagnosis.