Fault Diagnosis of Rolling Bearing Based on Gradient Adaptive Corrected Logistic Regression Model
In order to accurately analyze the high-dimensional signal data,reduce the information loss caused by feature selection,and correctly evaluate the performance degradation state of rolling bearing,a fault diagnosis method based on gradient adaptive corrected(GAC)Logistic regression model is proposed,and the upper bound of convergence of the GAC algorithm is found.Firstly,typical characteristic parameters of bearing detec-tion signal are extracted as model variables;Secondly,denoise and normalize the signal characteristic data;Fi-nally,the GAC-Logistic regression model is established on the premise of maintaining the data dimension,and the performance state of rolling bearing is evaluated.The experimental results show the proposed method can improve the efficiency of model construction and the accuracy of bearing state discrimination,and has good ro-bustness,which can effectively reduce the influence of random signal fluctuation on bearing state evaluation.The average fitting accuracy and average verification accuracy of GAC-Logistic regression model reaches 99.08％and 98.17％,respectively.
NETL
NSTL
万方数据
