SOLVING THE THERMAL ECONOMIC DISPATCH PROBLEM USING ARTIFICIAL NEURAL NETWORKS

The problem of economic dispatch (generation) is considered as one of the optimization problems, and many different methods have been used to solve it. In this study, the problem is solved using artificial neural networks (ANNs) which are considered as a branch of the artificial intelligence. The tool used in the study is the ANNs toolbox in MATLAB, which contains advanced capabilities. Through the toolbox, many options can be tested to reach the optimal solution to the problem. ANNs need to be trained on samples of inputs-outputs (solution) of problems before they can solve them on their own. During this study computer programs have been written to solve the thermal dispatch problem using a traditional mathematical method; the -iteration method. Through it, training samples are provided to the ANN. From the results of the study it has been proved that the solutions of the problem obtained using the trained ANN almost match with the solutions obtained using the -iteration method. The goal of using ANNs is to solve the thermal of economic dispatch problem in a faster time in the case of power systems that contain a large number of generating units. Keywords: Thermal Economic Dispatch, Optimal Generation, Lambda Iteration, Multi-Layer Feed-Forward Artificial Neural Networks, Error back propagation algorithm