用于Micro-EDM放电状态分类的多传感器集成与信息融合系统

在Ｍｉｃｒｏ－ＥＤＭ（微细电火花加工）中，由于加工信号的频率高、加工波形的畸变，使得常规的用于放电状态检测的方法，如电压波形采样、放电延时等，已不再适用。利用多传感器的信息融合进行目标识别，可以避免单一传感器的局限性，减少各传感器不确定性的影响。文中描述了一人用于目标识别与分类的基于模型的多传感器系统。该系统选用以决策层为主的方法，以模糊神经网络作为其信息融合的工具。体现了多传感器信息融合优越性。

Multi-Sensor Integration and Data Fusion System A pplied to Discharge Condition Recognition and Classification in Micro-EDM

Discharge condition detection methods, such as discharg e time delay, voltage wave sampling, used in the conventional EDM (electric disch arge manufacturing) are not suitable for the Micro-EDM due to the high frequenc y and the distortion of the voltage wave shape. In order to avoid the limitation of a single sensor and to reduce the negative effect caused by the uncertainty of individual sensors, the data fusion of a multi-sensor system is used to acqu ire the relevant knowledge of the target. In this paper, a multi-sensor system is described, which is based on the model tool and applied to target recognition a nd classification. In the data fusion process, a fuzzy neural network (FNN) is selected and used for the data fusion at the report level. Experimental results show that the method works correctly, the cost decreases and the relia b ility of the recognition increases. The superiority of this method has been show n.