Analysis of Microgap Electrostatic Discharge Parameters With Algorithms of Neural Network and Wavelet Transform

Abstract

Special relationship exists between environmental conditions and discharge characteristic parameters in microgap electrostatic discharge (ESD) events. Potential relations between input and output of neural network can be explored if taken discharge environmental factors as neural network input. The characteristic parameters of discharge results are affected by environmental conditions, and hence, discharge parameters can be described with an output of neural network. Circumstances factors effect on discharge parameters in microgap ESD result was analyzed with two algorithms of neural network wavelet transform combined with Kalman filter. Nonlinear relationship between circumstances conditions and discharge result effect was a feature in microgap ESD events. Strong positive relationship existed between discharge parameters and circumstances factors of electrode moving speed, gas pressure, and temperature. Characteristic parameters measured in real ESD experiment were compared to predictive parameters of calculation result from neural network algorithm. The analysis of accuracies was given on the prediction of discharge process trend compared to discharge current data measured in experiment. Noise in discharge current waveforms can be suppressed effectively with the method of wavelet transform combined with Kalman filter.