ECONOMIC LOAD DISPATCH FOR PIECEWISE QUADRATIC COST FUNCTION USING HOPFIELD NEURAL-NETWORK

This paper presents a new method to solve the problem of economic powe r dispatch with piecewise quadratic cost function using the Hopfield n eural network. Traditionally one convex cost function for each generat or is assumed. However, it is more realistic to represent the cost fun ction as a piecewise quadratic function rather than one convex functio n. In this study, multiple intersecting cost functions are used for ea ch unit. Through case studies, we have shown the possibility of the ap plication of the Hopfield neural network to the ELD problem with gener al nonconvex cost functions. The proposed approach is much simpler and the results are very close to those of the numerical method.