基于改进离散粒子群算法的电力系统机组组合问题

Unit Commitment Based on Improved Discrete Particle Swarm Optimization

提出一种新的离散粒子群算法。结合改进的自学习策略优化粒子群算法适用于求解电力系统中的机组组合(unit commitment,UC)问题。算法将UC问题分解为具有整型变量和连续变量的2个优化子问题,采用离散粒子群优化和原对偶内点法相结合的双层嵌套方法对外层机组启、停状态变量和内层机组功率经济分配子问题进行交替迭代优化求解。在处理约束问题时采用修正法代替传统的罚因子法,提高了解的质量。以10×2台机组组成的2个测试系统为算例,通过与其他算法结果进行比较分析,证明了该方法的可行性和有效性。仿真结果表明,该方法解决UC问题具有求解精度高和收敛速度快的优势。

An improved discrete particle swarm optimization （PSO） algorithm is proposed. The optimized PSO combined with improved self-learning strategy is suitable to solve unit commitment （UC） problem in power system. The proposed algorithm divides UC problem to two optimization subproblems, i.e., the subproblem with integer variables and the subproblem with continuous variables, and the former is suitable to solve UC state problem and the latter is suitable to solve economic dispatching problem. Adopting the two-layer embedded method that combines discrete PSO with primal-dual interior point method, the starting/stop state variables of outer layer units and economic power allocation of inner layer units are solved by alternate iteration optimization. During the processing of constraints, traditional penalty factor method is replaced by modification algorithm to improve the quality Of solution. Taking a testing system composed by ten units as simulation example, simulation results show that the proposed method is feasible and effective, besides, simulation results also show that the proposed method possesses the superiority in both solution accuracy and convergence speed than other algorithms.