基于混合域相对特征和FOA⁃XGBoost滚动轴承退化评估

Aiming at deficiency in modeling rolling bearing degradation assessment using multi-domain features, a prediction method is proposed based on the fruit fly optimization algorithm (FOA) and extreme gradient boosting (XGBoost) to estimate the rolling bearing condition. Firstly, the relative feature set in mixed domain is constructed according to the multi-dimensional characteristic parameters such as time domain, frequency domain and time-frequency domain of the whole life cycle of rolling bearings. Then, the relative root mean square value is used to determine the corresponding parameters of bearing degradation, and then the XGBoost model is trained by using mixed domain features, and the parameters of the degradation evaluation model are optimized by using FOA algorithm. The experimental results show that the performance of the degraded evaluation model is 27.15% and 34.96% higher than that of the commonly used support vector regression (SVR) model on the two data sets, respectively. The proposed method can accurately and effectively evaluate the bearing degraded condition.