配电网络重构的研究

STUDY ON ALGORITHMS OF DISTRIBUTION NETWORK RECONFIGURATION

总结了求解配电网重构的各种方法。指出由于组合数学的特性 ,数学优化理论不适用于配电网重构 ;基于模拟退火方法的配电网重构算法可以获得全局最优解 ,但存在算法依赖参数和计算量大的缺点 ;基于人工神经网络的算法 ,其精度取决于样本 ,获得完整样本困难 ,而且训练样本的时间较长 ;遗传算法的很多特点适于求解配电网重构的问题 ,如果能结合配电网重构的特点对算法收敛性进一步研究 ,提高其速度 ,则将在配电网重构中得到更好的应用 ;模糊数学和专家系统必须依赖于其他技术的发展 ;最优流模式和基于开关交换的算法不能保证得到全局最优解 ,但与启发式规则结合后 ,可以较快地得到满意的结果 ,是目前解决配电网重构的有效算法。

Existing approaches to solve distribution network reconfiguration problem are summarized.Mathematical optimal approaches are not suitable for solving this problem.Algorithms based on simulated annealing can work out optimal solution,however,they are very time-consuming and their performances depend on initial parameters.The algorithms based on artificial neural network can find out optimal distribution configuration quickly,however,their results depend on the training sets and the total training sets are very difficult to obtain,further the training set is very time-consuming.Genetic algorithms are suitable for solving distribution network reconfiguration problem,and its wide applications to solve this problem will be expected if its convergence is improved.The results of the algorithms based on expert system or fuzzy mathematics are dependent on other techniques' development.The solutions worked out using the algorithms based on optimal flow pattern or switch exchange method may not be optimal or near optimal.But the algorithms combined with heuristic rules can reduce the computation time and may be most effective for solving distribution network reconfiguration problem so far.