论文部分内容阅读
The functional relationship of approximation accuracy and number of fuzzy sets is used to find the rational balance point between the control accuracy and the control cost of fuzzy systems. This approach efficiently eliminates the drawback of rapid control cost increase caused by blind increase of fuzzy set number in practical engineering. The sufficient conditions for T-S fuzzy systems as universal approximators are derived. A special T-S fuzzy system that satisfied these conditions is analyzed, and the simulation results show that when the number of fuzzy sets is increased moderately, the model parameters training epochs can be effectually decreased while the model accuracy improved significantly. A practical welding power source controlled by a T-S fuzzy system is developed with satisfactory experimental results.