The prediction model and experimental research of grinding surface roughness based on AE signal

This paper is based on the investigation of the relationship between the processing parameters and the characteristic parameters of acoustic emission signal (AE signal) including RMS value, ringing count, and signal spectrum during the grinding of several difficult-to-machine metallic materials; the variation of AE signal characteristic parameters and spectrum with the parameters of grinding depth a(p), grinding wheel velocity v(s), and feed velocity v(w) was analyzed, then the corresponding relationship between acoustic emission signal characteristic parameters and machining surface roughness was given. On this basis, the multi-information fusion algorithm based on BP neural network was used to reasonably fuse various characteristic parameters of AE signals, then predict and recognize the surface roughness of grinding workpieces. Finally, the established model was optimized by using genetic algorithm, which significantly improved the prediction accuracy and provided a reliable prediction model for the grinding of difficult-to-machine alloys, providing a feasible method for predicting surface roughness for practical production.

Automated summary

This paper investigates the relationship between processing parameters and characteristic parameters of acoustic emission signals during the grinding of difficult-to-machine metallic materials. The study analyzes the variation of characteristic parameters and spectrum of AE signals with grinding depth, grinding wheel velocity, and feed velocity, and establishes a corresponding relationship between AE signal parameters and machining surface roughness. A multi-information fusion algorithm based on BP neural network is used to predict and recognize surface roughness, and the model is further optimized using genetic algorithm to improve prediction accuracy.