A progressive mapping method for classifying the discharging states in micro-electrical discharge machining

摘要

High frequency and weak energy of micro-electrical discharge machining (MEDM) cause the waveforms of voltage and current to be highly distorted, and thus indistinguishable from the commonly used EDM-discriminating methods. Therefore, a new progressive mapping method is presented and modes for accomplishment of three mappings are also developed. Fuzzy rules are used to combine the complementary signals with voltage and current, and then a scalar in a range representing a state of the sampled point through the first mapping is deduced. A learning vector quantification (LVQ) neural network is adopted to convert this scalar to the corresponding state vector. After a series of pulses are mapped to the state vectors, a summation of all the vectors is conducted. Then normalization of the summation vector is followed. The ratios in the vector clarify the discharging pulses through the third mapping, judging mode. Experimental results are presented to verify the effectiveness of this discharging pulses discriminator for MEDM.