Research on Error Tracing of Synchronizer Gear Hub Based on Variable Step Fa-Rbf Neural Network

Abstract

The synchronizer gear hub has a complex structure and involves a wide variety of processing equipment. It is difficult to trace the reason for the abnormal fluctuation of the processing quality by simply detecting the data of local processing conditions. Based on the analysis of the machining error of gear hub, an error tracing method based on the radial basis function (RBF) neural network which optimized by variable step firefly algorithm (FA) was proposed. The FA-RBF error tracing model was built based on the machining error, gear hub quality feature information and the error influencing factor, and the relevant parameters of the neural network were optimized through the variable step firefly algorithm. The verification experiment was done by 164 groups quality data collected from a gear hub production line, 159 groups of the data were used to train the model and 5 groups of the data were used to test it. The experimental results show that the support degrees of the corresponding error types of the 5 groups test data are 0.882, 0.889, 0.98, 0.895 and 0.768. It means that the error types can be accurately traced by the proposed error tracing method. The method can effectively identify the error source of abnormal fluctuation of machining error during the machining process, and it can provide a basis for error compensation.