Adaptive neuro-fuzzy nonsingular terminal attractor and its application in high-performance AC power conditioners

Abstract

This paper describes an adaptive neuro-fuzzy (ANF) nonsingular terminal attractor (NTA) that can control output voltage in an AC power conditioner. The NTA experiences the convergence of the system trajectory to zero within finite time, and it can avoid singularity problems. However, when the estimated value exceeds or falls short of the uncertain system boundary, chattering or steady-state error problems occur. Therefore, the ANF technique is introduced as a computationally rapid and mathematically simple means of reducing the chatter and steady-state errors occurring in NTA if the estimated value of the uncertain system boundary is not met. The Lyapunov theory proves that the proposed AC power conditioner is stable and convergent. Associating NTA with the ANF technique provides the AC power conditioner with low distortion and fast transience in the presence of various loading. Experimental results from a 1 kW prototype confirm the mathematical analyses and simulations. Because the proposed AC power conditioner offers significant advantages over the classic terminal attractor (TA)-controlled AC power conditioner in terms of complexity and ease of implementation, this paper represents a useful reference for designers of artificial intelligence.