An Algorithm for Unit Commitment Based on Hopfield Neural Network

Abstract

This paper presents an algorithm, which is based on a Hopfield neural network, for determining unit commitment. By constructing an appropriate energy function, a single layer Hopfield neural network can solve the problem of assigning output power of generators at any given time. Based on this single layer Hopfield neural network, a multi-layer Hopfield neural network is presented. The multi-layer Hopfield neural network can solve the problem of power system unit commitment. The energy functions of single layer and multi-layer Hopfield neural network and the corresponding algorithm are given in the paper. The restricted conditions of the balance between power supply and demand, maximum and minimum outputs of power plants are considered in the energy function. So is the speed of propulsion and decreasing power of generators. An example shows that the result obtained by Hopfield neural network is somewhat similar to that obtained by genetic algorithm, but the calculation time is much shorter.