电网可靠性评估中负荷相关性的广义交叉熵模型

Dependence Modeling of Correlated Bus Loads Based on Generalized Cross Entropy Method for Composite Power System Reliability Evaluation

节点负荷相关性的概率密度估计模型是实现负荷相关性变化对电网可靠性影响研究的关键环节。该文从模型稀疏性和模型准确性兼顾的角度,提出多维节点负荷联合概率密度估计的改进广义交叉熵模型(generalized cross entropy,GCE)。该文以广义交叉熵公设为基础,构建概率密度函数的有约束泛函极值问题及其对偶优化问题,并将对偶问题转换成核函数权重的二次规划问题,从而实现概率密度的准确估计及其稀疏性的自适应优化。此外,针对传统GCE模型在权重优化求解中存在的问题进行改进,进一步提高了密度估计的准确性。通过将改进GCE模型与传统GCE、非参数密度估计和半参数密度估计在稀疏性和准确性上的理论分析和算例比较,以及对RBTS、IEEE-RTS79、IEEE-RTS96测试系统的评估分析,验证了改进GCE模型的有效性。

The dependence modeling of bus loads based on probability density estimation technique is the key to research the impacts of correlated bus loads on the power system reliability.From the view of model sparsity and accuracy,an improved generalized cross entropy(GCE)method is proposed to estimate the joint PDF of nodal loads.First,the constrained functional extreme problem based on GCE postulate and the corresponding dual problem are presented to achieve the joint PDF.Moreover,the dual problem is converted into a quadratic programming problem for weights optimization of the kernel functions in the estimated joint PDF.Furthermore,the problem existing in the traditional GCE method related with the solution of the quadratic programming is pointed out,and then a significant improvement is made to achieve more accurate density estimation.The comparisons of the sparsity and accuracy among the improved GCE,conventional GCE,non-parametric and semi-parametric density estimation are conducted,and the reliability of RBTS,IEEE-RTS79 and IEEE-RTS96 is evaluated to validate the improved GCE.