基于DAEM算法的风力发电并网系统概率潮流计算

Probabilistic load flow calculation based on DAEM algorithm for wind-farm integration system

针对基于加权高斯混合分布(WGMD,weighted Gaussian mixture distribution)构建风电场概率模型的方法中,EM(expectation maximization)算法由于其固有缺陷导致整个模型拟合精度降低的问题,提出基于DAEM(deterministic annealing expectation maximization)算法的风电场概率建模方法,并结合马尔科夫链蒙特卡洛(MCMC,Markov Chain Monte Carlo)模拟法进行风力发电并网系统概率潮流计算.DAEM算法通过引入退火机制,避免了在模型参数最大似然估计时,EM算法容易陷入局部最优的问题,使得风电场模型更加准确.在接有风电场的IEEE39节点系统中进行概率潮流计算,计算结果证明了所提算法的精确性和有效性.

Aimed at the problem happened to the probability model of wind-farm built up with weighted Gaussian mixture distribution（WGMD）algorithm that the EM（expectation maximization）algorithm will lead entire model fitting precision to lower due to its inherent defect,a method for probability modeling of wind-farm is proposed based on DAEM（deterministic annealing expectation maximization）algorithm and integrated into MCMC（Markov Chain Monte Carlo）simulation method to conduct the computation of probabilistic load flow of wind farm integration system.When the maximum likelihood estimation of wind farm modeling parameters is being made,the DAEM algorithm by means of introducing an annealing mechanism,can avoid such problem that EM algorithm would easily lead to converging to local optimum,in order to make the model of windfarm even more accurate.The probability flow calculation is performed in IEEE 39 bus system connected with the wind-farm,and it is verified by the calculation result that the proposed algorithm will be accurate and valid.