多阶段输电网络最优规划的并行蚁群算法

PARALLEL ANT COLONY ALGORITHM FOR THE MULTISTAGE TRANSMISSION NETWORK OPTIMAL PLANNING

多阶段输电网络最优规划是一个复杂的非线性组合优化问题,难以采用传统的数学优化方法求解。蚁群算法是近年来出现的用于解决组合优化问题的一种高效的内启发式搜索技术,但存在着未成熟收敛问题。文中给出了多阶段输电网络最优规划的数学模型及其解的向量形式;详细分析了传统蚁群算法的未成熟收敛现象及其原因;提出一种并行蚁群算法并用于求解多阶段输电网络最优规划问题。并行蚁群算法无需初始可行解,能很好地协调局部搜索与全局搜索,在加快计算速度的同时有效地避免了因参数设置、种群规模等不同而引起的未成熟收敛。对实际算例的计算结果表明,该方法具有很高的计算效率和良好的全局收敛性。

Multistage transmission network optimal planning (MTNOP) is a complex nonlinear combinatorial optimization problem, which cannot be well solved by traditional optimization methods. Ant colony algorithm (ACA), a recently emerged meta-heuristic method, is highly efficient for quickly finding high quality solutions to combinatorial optimization problems. But, it is inclined to premature convergence problems. In this paper, firstly, the mathematical model of MTNOP problem is introduced and the vector of its solution is also proposed. Then, after discussing the premature convergence of ACA in detail, a parallel ant colony algorithm (PACA) is introduced to solve the MTNOP problem. In PACA, no initial feasible solutions are needed, and the local search and global search are harmonized effectively. Compared with ACA, PACA can not only enhance the computation speed, but also solve the premature convergence caused by the inappropriate set of parameters and colony size. The simulation results for an 87-bus sample system, although preliminary, show its advantage on computation speed and convergence.