Research On Tool Wear Monitoring Technology For Rotary Ultrasonic Machining Of Hard And Brittle Materials

Since the special characteristics of hard and brittle materials, making the process has high cutting force and low materials removal rate during rotary ultrasonic machining, causing tool wears faster. And Irregular morphology changes of tool caused by tool wear, thus affecting the shape accuracy and surface quality of the workpiece. So carry out tool wear monitoring during rotary ultrasonic machining of hard and brittle materials for improving machining quality and efficiency of workpiece and realizing autonomy process is extremely important.This article used multi-sensor fusion technology to build tool wear monitoring system, signals used here including vibration and cutting force, and set up experiments internet. Studied tool wear form throughout the life cycle during rotary ultrasonic machining, observed influences of tool wear on workpiece quality, and divided tool wear into several stages. Designed sanity experiments to acquire vibration and force signals, and conducted noise reduction and smoothing to these signals.Extracted features from vibration and force signals using time domain, frequency domain and wavelet package analysis methods, these feature have some relationships with tool wear, then feature selection method was used to reduce feature dimension.Feature selection method takes in-class scatter, between-class scatter and sensitivity of the features to process parameters into account, result show that this method is very effective.For the shortcomings of BP neural network, adopting genetic algorithm to optimize initial weights and thresholds, using BP algorithm to fine-tune weights and thresholds,and adopting structure learning method based on fuzzy reasoning to optimize network structure. Then adopted improved GA-BP neural network proposed in this article to identify tool wear during rotary ultrasonic machining, results show it has a lot of merits compared to conventional BP neural network.A tool wear condition monitoring system was built adopting virtual instrument technology, and the software of this system was written based on Labview, which consists of three modules: multi-sensor signal acquisition and storage, offline monitoring system parameters adjustment and online tool wear monitoring. Then,through simulating experiments tested accuracy and reliability of the system.