基于量化正交免疫克隆粒子群算法的水电站水库优化调度研究

Orthogonal immune clone particle swarm optimization algorithm and its application in hydropower station reservoir optimal operation

【目的】解决传统粒子群算法在求解水库优化调度问题中存在的早熟、收敛速度慢和易陷入局部最优的问题。【方法】基于抗体克隆选择学说理论,提出了一种量化正交免疫克隆粒子群算法(OICPSO/Q)。采用正交交叉策略来增强子代个体解分布的均匀性;通过接种疫苗和计算亲合度等操作,对算法的进化过程进行有目的、有选择地指导,使得算法快速收敛,同时保持一定的多样性,抑制了早熟现象。提出一种自学习算子,避免个体邻域内最优解的丢失。建立了基于量化正交免疫克隆粒子群算法的水库优化调度数学模型,并给出其具体的求解步骤。最后应用该方法与标准粒子群算法(SPSO)及动态规划方法进行比较。【结果】与SPSO算法和动态规划方法计算结果相比,OICPSO/Q算法计算时间明显降低,但发电量明显增加,说明OICPSO/Q算法可提高解的精度,加快其收敛速度,其性能优于标准粒子群算法和动态规划方法。【结论】OICPSO/Q算法为求解水库优化调度问题提供了一条新的有效求解途径。

[Objective] The study was done in order to overcome the disadvantages of classical particle swarm optimization algorithm used to solve reservoir optimal operation problems, such as premature convergence,poor convergence,and easy to fall in local optima. [Method] Based on the binary particle swarm optimization and clone selection theory,an orthogonal immune clone particle swarm algorithm with quantization （OICPSO/Q） was presented in this paper. The orthogonal crossover strategy was used in immune gene operation to increase the solution＇s uniformity. The improved method adopted the operation of vaccine inoculation to accelerate the optimization searching speed. And a self-learning operator was presented to avoid better solution around individuals losing. The mathematical model and the procedures for solving the optimized reservoir operation optimization by using OICPSO/Q were proposed in detail. The performance of OICPSO/Q was compared with the standard particle swarm optimization （SPSO） and dynamic programming method. [Result] Study results show that this method has better performance in stabilization and convergence than SPSO and dynamic programming method. [Conclusion] OICPSO/Q algorithm will be a new and valid method for reservoir operation optimization.