论文部分内容阅读
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。文中描述了一个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。通过实验 ,该系统在正确识别的前提下 ,降低了整个 Micro- EDM系统的成本 ,提高了检测的可靠性 ,体现了多传感器信息融合的优越性。
In Micro-EDM, conventional methods for detecting discharge conditions, such as voltage waveform sampling and discharge delay, are no longer suitable due to the high frequency of processing signals and the distortion of the processing waveform. The use of multi-sensor information fusion target recognition, can avoid the limitations of a single sensor to reduce the impact of the uncertainty of the sensor. In this paper, a model-based multi-sensor system for target recognition and classification is described. The system uses a decision-making-oriented approach to fuzzy neural network as its information fusion tools. Through experiments, under the premise of correct identification, the system reduces the cost of the whole Micro-EDM system, improves the reliability of detection, and reflects the superiority of multi-sensor information fusion.