基于Hopfield神经网络的供电网黑启动分区方案计算方法

Calculation Method of Black-start Zone Partitioning Scheme for Power Supply Network Based on Hopfield Neural Network

针对电网大面积停电后黑启动分区划分问题,提出一种以分区联络线之和最少及总体操作最优为目标的数学模型。通过定义最小拓扑关系矩阵、节点关系矩阵和分区状态矩阵实现了对数学模型的解析描述,将对数学模型的求解转化为0-1优化问题。针对0-1优化问题多为NP-hard问题的现状,提出利用Hopfield神经网络进行求解。构造了计及各项约束条件的神经网络能量函数,给出了对应的求解方法。利用IEEE-39节点系统对所提模型及方法进行了验证,通过对分区评价指标的定量分析表明所提的模型正确,求解方法可行、有效。

Targeting the problem of black-start zone partitioning after large-scale power outage in power grid,the paper proposes a mathematical model to minimize the sum of partition contact lines and optimize the overall operation.The analytical description of the mathematical model is realized by defining the minimum topological relation matrix,node relation matrix and partition state matrix,and the mathematical model solving is transformed into a 0-1 optimization problem.Because the 0-1 optimization problems are mostly NP-hard problems,Hopfield neural network is proposed to solve the problems.The neural network energy function considering various constraints is constructed,and the corresponding solution method is given.The proposed model and method are verified by the IEEE39-node system.The quantitative analysis of the zoning evaluation index shows that the proposed model is correct and the solution method is feasible and effective.