UNIT COMMITMENT BASED ON GENETIC ALGORITHMS

How to solve unit commitment (UC) and load dispatch of power system by genetic algorithms (GAs) is researched. Using binary coding GAs to solve UC the amount of calculation and employed ram will be greatly increased and the classical GAs does not possess the ability of asymptotic convergence. For these problems a coding scheme is used in which the binary encoding and floating numbers are combined and according to this feature the corresponding genetic operators are designed. For the randomness of classical GA which appears in calculation process, the contraction mapping GA is applied to make the calculation asymptotically convergent. The results of calculation examples show that the proposed GAs is symptotically convergent, comparing with the algorithm of binary coding this algorithm needs less calculation time and less ram to be employed and the relevant information of UC can be led into more easily.