Combined Economic and Emission Dispatch Using Improved Backpropagation Neural Network

Abstract

The purpose of this paper is to design, train, and test an improved Backpropagation Neural Network (BPNN) for application to the problem of combined economic and emission dispatch. Focus is on the reduction of single pollutant nitrogen oxide, NO x ; transmission losses are included. The equality constraint of power balance and inequality generator capacity constraints are considered. The total load supplied is the input to the neural network. The thermal generator outputs and total system transmission loss are considered as outputs of the neural network. The program for Optimization Technique using the Quick Method is developed in Matlab, and the program for an improved BPNN is developed using the Matlab Neural Network Toolbox. Performance of an improved BPNN is compared with the Quick Method, and it is observed that the proposed neural network technique is very fast and predicts accurate results while satisfying inequality generator capacity constraints at various load levels. It also offers a considerable saving in computer memory. The proposed BPNN technique has been demonstrated through a sample system consisting of six thermal generators.