基于RVM和全纯嵌入法的考虑多风电场出力相关性的电-热互联概率能流计算方法

Probabilistic energy flow calculation method based on RVM and holomorphic embedding method considering correlation of multiple wind farms

随着风电接入规模的不断攀升,多风场之间的出力相关性会对含风电机组的综合能源系统(Integrated Energy System,IES)的安全稳定运行产生越来越显著的影响。同时,考虑相关性的系统概率能流计算有助于合理地进行资源分配和优化运行方式,保证系统的稳定性。文章基于相关向量机(Relevance Vector Machine,RVM)方法,提出了一种多维风电场风速的联合分布计算方法,该方法采用核函数有效避免了参数估计带来的二次误差,从而提高了计算精度。在此基础上,提出了一种基于全纯嵌入法(Holomorphic Embedding Method,HEM)的新半不变量法用于电-热联合网络的概率能流计算,相较于传统的牛顿-拉夫逊法(Newton-Raphson,NR),该算法具有更好的收敛性和计算精度。最后,采用经过修改的IEEE 33节点系统结合热力网络系统进行算例分析,进一步验证了所提算法的准确性和适用性。

With the large-scale operation of wind power,the correlation of wind speed of multi-dimensional wind farm will affect the safe and stable operation of integrated energy system with new energy generation equipment.The calculation of probabilistic energy flow considering the correlation is helpful to the reasonable allocation of resources and the optimization of scheduling operation mode to ensure the stability of the system.Therefore,this paper proposes a joint distribution calculation method of multi-dimensional wind farm wind speed based on relevance vector machine(RVM).This method uses kernel function,which can effectively avoid the secondary error caused by parameter estimation,so as to improve the calculation accuracy.On this basis,a new semi invariant method based on holomorphic embedding method(HEM)is proposed to calculate the probabilistic energy flow of combined electric and thermal networks.Compared with the traditional Newton Raphson method,the algorithm has better convergence and accuracy.Finally,a modified IEEE 33 bus system combined with thermal network system is used to verify the accuracy and applicability of the proposed algorithm.