Assessment of dynamical properties in EDM process¡ªdetecting signature of latent change to deleterious process in advance

详细信息    查看全文

• 作者：Ming Zhou ; Fuzhu Han ; Yongxain Wang and Isago Soichiro
• 关键词：EDM ; Discrimination of discharging pulses ; Cross ; prediction errors ; Non ; stationarity
• 刊名：The International Journal of Advanced Manufacturing Technology
• 出版年：2009
• 出版时间：September, 2009
• 年：2009
• 卷：44
• 期：3-4
• 页码：283-292
• 全文大小：775.2 KB

文摘

Time series data of gap state were often used as feedback signal in EDM control systems. An effective way to quantify gap state in machining was developed in this paper. Based on a time series of gap states recording a machining process, the entire process was partitioned into three parts, transient process, efficient machining process, and deleterious process. It is expected that efficient machining process should be maintained long enough in consideration of machining efficiency and avoiding workpiece surface damage. In this case, a signature of structural change from efficient machining process to deleterious process has to be detected or detected in advance. It has been found that linear analysis methods failed the task. In this paper, a nonlinear analysis method, cross-prediction error, was employed to track this structural change in efficient machining process in two tests. In the first test, the efficient machining process was roughly split into 30 segments. By using the cross-prediction-error method in this test, the non-stationarity, a signature of structural change from stationary process in efficient machining process to non-stationary process, was successfully revealed in the last segment of the process. In another test, efficient machining process was even more finely split into 70 segments, and by using the cross-prediction-error method again it is found that the detected non-stationarity in the process was in the last segment but one. The second test not only proves the effectiveness of the method, but also brings forward the detected non-stationarity more than 10s. The detected non-stationarity in advance, thus, is supposed to be used for avoiding the occurrence of abnormal deleterious process and maintaining a persisted efficient machining process.