| Research Abstract |
The objective of this research is the following two subjects.
(A) To establish a solution algorithm of the distribution loss minimization problem by the genetic algorithm through creating the string structure and the fitness function, etc.
(B) To improve the algorithm so as to get more accurate solutions for large scale problems. The research results are summarized as follows.
(1) The solution algorithm of the distribution loss minimization problem by the genetic algorithm is developed.
(2) For a small scale problem, if the number of string bits is not so large, an accurate (near optimal) solution can be available through the developed algorithm. The computation time is 1/6 times smaller than that by simulated annealing method. For the second objective, (B), how the string structure, fitness function, crossover rate, mutation rate, number of crossover points, artificial crossovers, initial string patterns, etc. affect the calculation results is investigated. The investigated results are as follows.
(3) The optimal crossover rate and mutation rate exist for each specific problem. Ordinarily, they are 0.6-0.8 and one bit for 1 - 2 string(s) respectively.
(4) For a large system, an accurate calculation result can be available only when effective (good for loss minimum structure) schema exist in the initial strings.
(5) Other artificial operations do not affect the result significantly. From the above result, it can be seen that objectives (A) and (B) are achieved. The above results are published in the papers shown in the next page.
|
