Publication:
Evaluation on training algorithms of back propagation neural network for a solar photovoltaic based DSTATCOM system
Evaluation on training algorithms of back propagation neural network for a solar photovoltaic based DSTATCOM system
No Thumbnail Available
Date
2019-01-01
Authors
Baharudin N.H.
Mansur T.M.N.T.
Ali R.
Misrun M.I.
Journal Title
Journal ISSN
Volume Title
Publisher
Research Projects
Organizational Units
Journal Issue
Abstract
This chapter discusses evaluation on the Back Propagation Neural Network (BPNN) control algorithm based on Fast Fourier Transform (FFT) control algorithm with different BPNN training algorithms for Distribution Static Compensator (DSTATCOM) with integrated solar photovoltaic system. Furthermore, the comparison is performed with different weight or bias training functions such as supervised and unsupervised. Each training algorithms have been utilized to investigate its performance in generating the target pattern for harmonic elimination in term of accuracy, learning epochs and training time. The performance of the BPNN training algorithms is determined by calculating the error between the target and output pattern using Mean Squared Error (MSE). The lower value of the MSE shows the higher accuracy of the output pattern according to the target pattern given. Number of iterations (epochs) and training time are evaluated to investigate the performance of different BPNN training algorithms on DSTATCOM for harmonic reduction under nonlinear load condition.