INVESTIGATION OF THE PROPERTIES OF SOFT ALGORITHM FUZZY INFERENCE

Abstract

The article is proposes a new approach of decision-making, based on the use of soft arithmetic operations in the structure of the fuzzy system output. The need to develop this method proved the lack of information signs at the defuzzification traditional systems fuzzy inference that use hard statements of finding the minimum and maximum. This circumstance greatly increases the parameter of the mean-square deviations (RMSE) in traditional systems of fuzzy inference. Therefore, as an evaluation criterion developed mild algorithm-fuzzy inference in the article to use this parameter, the best solution is achieved when the minimum value. As a comparative evaluation, listed the results, obtained at modeling decision-making process, based as multiple regression. So and models of fuzzy inference. To improve the reliability received the results, numerical modeling work soft algorithm fuzzy inference was performed to triangular and sigmoidal membership functions. The cited numerical simulation shows the effectiveness of the proposed approach, at that minimization of parameter RMSE is achieved without the use of adaptive neuro-fuzzy inference systems. Experimental studies of control cutting speed on the machine CNC using fuzzy logic have shown that the stability of the fuzzy system is increased using in the structure of the fuzzy inference soft arithmetic operations.