Minor component analysis based anti-hebbian neural network scheme of decoupled voltage and frequency controller (DVFC) for nanohydro system

摘要

This paper presents a Minor Component Analysis Based Anti-Hebbian Neural Network (MCA based AHNN) algorithm with adaptive step size convergence parameter for decoupled voltage and frequency control of single phase standalone Nano-hydro generation systems. In the proposed control scheme the adaptive step size convergence (ASSC) parameter significantly improves the convergence property of the algorithm in dynamic conditions like load perturbation or change in input mechanical power fed to generator. Since the proposed MCA based AHNN algorithm takes the input data samples in an adaptive manner using stochastic approximation technique, therefore it provides a very efficient and simple scheme for real time implementation of DVFC of single phase standalone Nanohydro generation system based on single phase SEIG. The proposed scheme is found highly suitable for adaptive parameter estimation problems in highly non-linear single phase Nanohydro power generation systems where the multiple system variables varies in a highly non-linear manner under varying load conditions. The memory requirement for the implementation of this proposed algorithm is significantly reduced. The proposed scheme is similar to a local gradient decent algorithm with the linearized gradient vector of instantaneous estimation of total least square function of fundamental components of load current.