微铣削刀具磨损状态监测方法研究

为提高金属微铣削过程中刀具磨损状态在线监测系统的预测效率与精度,提出一种基于线性判别分析与改进型BP神经网络模型识别刀具磨损的方法;该方法通过传感器与数据采集系统采集微铣削过程振动信号,提取其时域和频域特征并通过线性判别方法进行降维约简;将降维后的特征输入经灰狼优化改进的BP神经网络模型,从而实现微铣刀磨损状态特征的分类;结果表明,提出的微铣刀在线监测方法能够准确识别微铣刀的各种磨损状态;此外,和其它分类算法相比,提出的基于灰狼优化算法的BP神经网络模型在分类精度和计算效率方面具有综合优势;这对实际生产过程中微铣刀的磨损状态监测具有非常重要的实际意义.

Research on Monitoring Method of Wear State of Micro-milling Tool

In order to improve the prediction efficiency and accuracy of the online tool wear monitoring system in the metal micro-milling process, a method based on linear discriminant analysis and improved BP neural network model to identify tool wear is proposed. This method collects the vibration signal of the micro-milling process through a sensor and a data acquisition system, extracts its time-domain and frequency-domain features, and performs dimensionality reduction through a linear discrimination method. The dimensionality-reduced features are input into the BP neural network model optimized and improved by the gray wolf, so as to realize the classification of the characteristics of the wear state of the micro-milling cutter. The results show that the proposed online monitoring method for micro-milling cutters can accurately identify various wear states of micro-milling cutters. In addition, compared with other classification algorithms, the proposed BP neural network model based on gray wolf optimization algorithm has comprehensive advantages in classification accuracy and computational efficiency. This has very important practical significance for monitoring the wear status of micro-milling cutters in the actual production process.