基于模式记忆并行蚁群算法的输电网规划

该文讨论了目前大规模输电网规划求解中常见的无法完成对解空间的充分搜索，从而难于求得全局最优解的问题；给出了泛函形式的输电网规划模型，并重点分析了输电网规划解的模式。在分析了传统蚁群算法易产生未成熟收敛现象及其原因的基础上，设计了一种基于模式记忆的并行蚁群算法，该算法通过模式记忆实现了解空间分解，能够有效地识别、记忆和跳出局部最优解；通过局部细化搜索进一步加强了局部搜索能力；通过并行计算提高了计算速度。某实际77节点的算例分析表明了该算法具有很高的计算效率和优秀的局部、全局收敛性，有效克服了现代启发式算法在求解输电网规划问题时存在的效率不高及未成熟收敛等现象。

TRANSMISSION NETWORK PLANNING BASED ON SCHEMA RECORDING PARALLEL ANT COLONY ALGORITHM

While the large-scale transmission networks planning problems are dealt with, usually, the solution space is hard to be searched thoroughly and the global optimal solution is also hardly possible to be found. In this paper, a functional model of transmission network planning is proposed, and the schema of its solution is studied in detail. After the main disadvantage of being inclined to premature convergence in traditional ant colony algorithm is discussed, a schema recording parallel ant colony algorithm is proposed. It can partition the solution space through schema recording, and can identify, record and jump away from the local optimal solution. A local search is proposed to search the sub-space more thoroughly and a reasonable parallel strategy is proposed to enhance the computing speed. The simulation results of a practical 77-bus system show that this algorithm has high computing efficiency and good local & global convergence. This algorithm can avoid the problems of low computing efficiency and premature convergence in normal meta-heuristic algorithms.