Abstract: To effectively identify the chatter in milling which had great influence on the quality of workpiece,the chatter recognition model based on Estimation of Signal Parameters via Rotational Invariance Techniques (ESPRIT) and Hidden Markov Model (HHM) was established.Frequency spectrum of cutting force signals which were collected through fixed cycle was estimated by using the method of ESPRIT.The cutting state in the process was determined by spectrum characteristic,and cutting force signals were labeled after the filtering of spindle rotation cycle.HHM parameter was trained by cutting force signals which had class labels,and the parameters of this chatter recognition model were obtained.To realize chatter recognition and monitoring of cutting state,PC based real-time acquisition and processing of cutting force signals were developed and integrated into Open Modular Architecture Controller (OMAC) of machine tool.The result of machining experiment showed that CNC system which integrated the function of measuring force could accurately determine the cutting state in the process of machining based on established model.

Key words: chatter recognition, hidden Markov model, spectrum estimation, open modular architecture controller

摘要： 为有效辨识铣削过程中对工件表面质量影响较大的颤振并确定其具体发生时刻,建立了基于旋转不变技术估计信号参数(ESPRIT)和隐马尔可夫模型(HMM)的铣削颤振辨识模型。建模过程中先利用ESPRIT算法分别对以固定周期采集的切削力信号进行频谱估计。根据频谱特征将经过主轴旋转周期滤波后的切削力信号进行分类,利用具有类标签的力信号训练HMM参数,获得颤振辨识模型的参数。为实现铣削颤振辨识及切削状态监测,开发了基于“PC+FIFO高速数据采集”的切削力信号实时采集与处理功能,建立了该功能与开放式模块化结构数控系统的数据传输机制。实验结果表明,辨识系统可根据建立的颤振辨识模型准确判断加工过程中各时刻的切削状态,实现了铣削颤振的有效辨识。

关键词: 颤振辨识, 隐马尔可夫模型, 频谱估计, 开放式模块化结构数控系统