An Application of Gaussian Machine in Neural Network to the Unit Commitment Problem

The unit commitment problem is to determine an optimal schedule of what thermal units must he started or shut off to meet the anticipated demand; it can he formulated as complicated mixed integer programming prohlem with a number of equality and inequality constraints. This problem needs huge calculations for the large power system. There are various methods considering applications of neural network which have inherent characteristics of being able to find good solutions quickly. In this paper, the authors present a neural network technique for determining the on/off state and also MW power output of each unit. Neural network of this method is formulated by Gaussian machine, The Gaussian machine have advantage of heing able to escape from a local minima by adding random noise of Gaussian distribution to input of neuron. Therefore, we can obtain the global minima independent on initial values. Furthermore, we also investigate an application of the simulated annealing and inequality neuron. The effectiveness of the proposed method is clarified by numerical examples. It examines with the LR method.