Artificial neural network application to economic operation of all thermal power plants

Presented in this paper is the application of artificial neural networks (ANN) to the economic operation of all thermal power plants. Training data are obtained, in practice, from a database of power system economic operation. In this paper, the training data are obtained for a simulated example, via running an approximate dispatch program, taking into account the power network transmission losses (B coefficients), as well as the generator limits. The simulated example comprises three thermal units supplying the power to a variable load. The proposed ANN algorithm is tested for two groups of data; while the first group represents about 30% of the whole training data set, the second group is a complete set of new data. Effects of the number of hidden neurons and the type of the selected transfer function on the performance of the neural network are investigated. Simulated results are reported, which form the conclusions presented at the end of the paper.