Tool wear monitoring method based on AMIDBOAB and imbalanced data optimization

Development of precise models for monitoring tool wear faces challenges due to imbalance of experimental data. To address the issues of data imbalance and low monitoring accuracy in various tool wear stages of CNC machine tools, an AMIDBOAB tool wear monitoring method based on imbalanced data optimization is proposed. The multi-scale and multi-domain value feature extraction method, the maximum relevance minimum redundancy feature selection method, and the adaptive gravitation mixing sampling algorithm are integrated to optimize the dataset. Introducing circle chaotic mapping, Levy flight strategy, and adaptive variable inertia weight, a multiple improvement dung beetle optimizer algorithm is proposed to optimize the parameters of the base classifier in the adaptive boosting model. Experimental validation is carried out on the PHM2010 public dataset and a self-built ceramic machining dataset. The results indicate that the method significantly enhances the performance of the classification algorithm, achieving a tool wear monitoring accuracy of 95.06%.