Multi-sensor High Frequency Weld Control Based on Particle Filtering and Fuzzy Fusion

Abstract

For power control problem in high frequency weld with tube, a control algorithm based on multi-sensor particle filtering and fuzzy fusion is proposed here. For adjust weld power, the signals from the tube walking speed sensor and lossless detection sensor are monitored and fused. Not only the relation between the target control function and the tube walking speed or the weld wound is nonlinear but also the state equation of the speed or the weld wound is nonlinear, the sequential importance resampling Bayesian filtering is used to track and predict the speed and wound. These predictive signals are considered the input of the fuzzy neural network to predict the weld state. The weld power is adjusted according to the weld state prediction so the melted weld or lack weld is avoided. Finally a simulation example is given and a good result is obtained