The Research On Automatic Classification Of Electrodes Based On Shape Retrieval

The EDM process is an important way of the mould's processing, this process needs a large number of electrodes. To shorten the electrode's processing time, there are some enterprises to adopt template technology, which reduced the heavy manual labor to some extent and also reduced the errors in NC programming. However the template for process planning finished manually often is not able to fully utilize the process knowledge and to accurately load the template .The author have established an classification system for electrodes based on UG4.0 and V C++2005, which can classify of the group of electrodes, as well as single electrode, so that the clustering and classification problems of electrode are resolved. For this purpose,it is needed to extract suitable features of electrodes to express their shapes. Based on the structural features of electrodes, the knowledge of NC machining and the template of the electrode, 5 algorithm are proposed, i.e. slice, D2 shape distribution, thickness histogram, the radius of 3D moment and the radius of 3D surface moment. Four algorithms are finally adopted through computer testing and application tryal. The author have finished three Clustering algorithm, i.e. slice, the most near rule of test method and the smallest maximum distance algorithm. To slove the single electrode classification problems, the method of one by one comparison and the method of cluster center comparison are utilized.This system can not only cluster the electrodes according to their process features, utilize electrode processing knowledge, and create process templates, but also classify sigle electrode so that to choose the electrode's template automatically based on its category. Thus, automated NC programming of electrodes are realise, and the reuse of process knowledge achieved.