Interval Type II Fuzzy Neural Network Super-Twisting Sliding Mode Control and Application

Abstract

This paper provides a Self-adjustment Interval Type II Neural Network (SAIT2NN) method combined with Super-Twisting sliding mode control (STSMC), aiming at solving the harmonic current problem and improving power quality in an active power filter (APF) system. The designed neural network takes the advantages of the interval type II fuzzy system to improve the learning ability for complex object, thereby dealing with the nonlinear changes caused by the APF system well. At the same time, the neural network adopts a signal classification and selecting mechanism to raise learning speed while minimizing signal loss to ensure correctness. The eigenvalues used to evaluate the fuzzed signal are also part of the system signal, ensuring functionality and reducing structural complexity. In order to achieve a theoretically tight fit, the online learning algorithms of the SAIT2NN are developed from the Lyapunov method. In order to make up for the chattering problem caused by the inherent error of the neural network and delay of the system, STSMC is used as a compensation mechanism to further improve the accuracy. Experimental study and comparisons are provided to prove the effectiveness and superiority of the performance.

Keywords

• Fuzzy neural network (FNN)
• active power filter
• self-adjustment interval type II neural network
• super-twisting sliding mode control (STSMC)