Novel Fuzzy Neural Network Sliding Mode Control of Active Power Filter

Abstract

A novel adaptive sliding mode control using long and short-term memory fuzzy neural network (ASMCLSTMFNN) is proposed to improve the dynamic performance of active power filter (APF). Combining the fuzzy neural structure (FNN) and long and short-term memory (LSTM) mechanisms, a long and short-term memory fuzzy neural network (LSTMFNN) is proposed to estimate the unknown parts of the system caused by parameter variances and external disturbances in the APF system. Experimental results show that the proposed ASMCLSTMFNN strategy is effective and has better steady state and dynamic performance than the adaptive sliding mode control using recurrent fuzzy neural network.