An Application of Neural Network to Economic Load Dispatching Using Parallel Processors

The economic load dispatching (ELD) among thermal units can be formulated as one of the constrained optimization problems in power system operation. Since an optimization is required under severe constraints, all constraints cannot be taken into account. Therefore, at present stage, an automation is performed considering several constraints and, for the time interval where some constraints are not satisfied, a manual operation is performed for ELD based on the empirical knowledge of operators. This paper presents an application of artificial neural network to ELD. The method uses one of the probabilistic neuron model, called Gaussian Machine and several constraints are handled by heuristic methods. The method can solve a local minimum problem, initial condition problem of neurons, and parameter tuning problem. In this paper, parallel processors are used for inplementation of the system. The processors are high performance microprocessors that support parallel processing through on-chip hardware. The number of processors is examined for detecting appropriate parallel computation