考虑交流潮流安全约束的随机确定性耦合电力系统机组组合模型及数值求解方法研究

Research on the accurate day-ahead dispatching and decision method is urgently needed for the power system in China under the circumstances of large scale wind power integration and electricity market reform. This project aims to put forward a modelling and numerical solution for unit commitment of coupled deterministic and random power system considering the AC power flow security constraints. First construct a random unit commitment model by AC power flow model, which is described by security constraints considering the uncertainty of wind power output. Since the calculation quantity increases sharply since the model includes random variables, so this project first study the constraint softening method based on the rapid identification of security constraints and then an improved random constrained ordinal optimization is put forward to solve the model above. The proposed method in this project has considered the uncertainty of wind power while using more precise AC power flow model to describe the security constraints, which can enhance the effectiveness of the electricity market decisions, reduce the voltage limit risk and improve the economy of system operating..Related research of this project may provide a more efficient and accurate strategy for day-ahead electricity market after large-scale wind power integration, which is of important theoretical significance and application value.

研究精细化的日前调度决策方法是目前风电大规模接入和电力市场化改革背景下我国电力系统发展的迫切需求。本项目拟提出一种考虑交流潮流安全约束的随机确定性耦合电力系统机组组合模型及其数值求解方法。考虑风电出力不确定性，利用交流潮流模型描述网络安全约束，构建融入交流潮流安全约束的不确定性机组组合模型；针对模型在不确定性环境下计算量激增的求解难题，首先研究基于安全约束快速辨识的“约束软化”方法，然后在此基础上提出一种改进的随机约束序优化方法对模型进行求解。本项目所提方法在考虑风电不确定性的同时还采用更为精确的交流潮流模型描述系统网络安全约束条件，可以有效提升电力市场决策的有效性，降低电压越限风险，提升系统运行经济性。

研究具有高适应性、高精度的机组组合决策对编制发电计划和电力系统的经济运行具有重要的实际意义。本项目的主要成果包括：考虑到新能源并网对电力系统调度带来的影响，提出基于自适应多变量非参数核密度估计的多风电场联合概率密度建模方法和综合考虑负荷、风电以及光伏出力不确定性及概率相关性的日前鲁棒调度方法，在此基础上，建立考虑交流潮流安全约束的含风电安全约束机组组合模型，并基于随机约束序优化进行求解，从而全面评估风电接入后带来的无功问题；针对基于物理模型驱动的机组组合决策方法适用性较低，本项目通过构建基于长短时记忆神经网络的数据驱动模型求解安全约束机组组合问题，提出一种具有自我进化和自我学习能力的机组组合决策方法；另外针对不确定性安全约束机组组合问题融入了大量离散变量和无效安全约束，本项目基于改进约束序优化算法和数据驱动思路，构建了离散变量辨识模型和无效安全约束辨识模型，进一步提升求解效率。本项目研究为机组组合问题的建模与求解提供了新思路，对提升系统的经济运行具有重要意义。