粒子群算法及其在汽轮机调节系统参数辨识中的应用

Particle Swarm Algorithm and Its Application in Parameter Identification of Steam Turbine Governing System

汽轮发电机组调速系统参数辨识工作量大,用最小二乘法等常规辨识方法进行参数辨识时算法计算时间长,将标准粒子群算法引入到汽轮发电机组调速系统参数辨识工作中,并对仿真结果和现场实测数据进行了对比,结果发现相比于最小二乘法,该算法在对模型非线性环节进行辨识时计算速度更快,虽然种群大小及迭代代数对辨识结果会有一定的影响,但总体来说辨识结果较为准确可靠,能够满足电力系统稳定性分析的要求。

Turbine governor system parameter identification is a heavy workload, and the conventional parame- ters estimation methods （such as the least square method） usually cost most computing time. In this paper, the standard particle swarm algorithm is introduced to the turbine model parameters identification. The feasibility and advantage of the PSO algorithm is verified by simulation and experiment. The results show that the algorithm has a fast calculation speed. Although the population size and the iteration number of the swarm do have a certain impact on the results in the identification of nonlinear elements, the estimation results are still accurate and reliable. Generally speaking, the PSO algorithm can meet the requirements of the stability analysis of power systems.