Stochastic multiobjective generation dispatch of combined heat and power systems

Abstract

A new stochastic formulation of multiobjective optimisation for combined heat and power system dispatch is presented. Both the power and heat demands are treated as random variables. The three objective functions to be minimised are: the total generation cost, the expected power generation deviation and the expected heat generation deviation. As these objectives are in conflict with each other, only a non-inferior solution can be reached. The goal attainment method is used to solve this multiobjective optimisation problem by assuming that the decision-maker (DM) has desired a certain goal for each of these objectives. Very often, the DM finds it difficult to define the relative importance or a merit function among improving individual objectives and reducing constraint violations. A fuzzy decision index is proposed here to provide the best-compromise solution among all objectives and constraints, and a means of maximising the most underachieved objective. A genetic algorithm approach is also proposed to help the DM to search for such a solution mathematically. Application results of the method to a seven-generator sample system are presented.