面向主动配电网的分布式能源多目标鲁棒经济优化调度

Due to the wide access to a large number of distributed energy resource, active power distribution system becomes more complex and has more flexible network structure, and most of them are intermittent energy resource, system also has strong random features, which brings new challenges to distributed energy resource optimization scheduling under the active power distribution network. In consideration with the multi-objective, multi-constrained, high-dimensional nonlinear and stochastic characteristics of active distribution network system, this project takes the load and generation forecast as the starting point, forecasts preliminarily in a long time scale with ARMA method, and corrects its forecast results with proposed empirical mode decomposition based Kalman filter correction method in short time scale. On this basis, it can obtain uncertain set based on the Lindeberg-Levy central limit theorem, and establishes robust economic multi-objective optimization scheduling model of distributed energy resources, combined with a two-stage zero-sum game theory, culture differential evolutionary algorithm is used to solve them under Soyster robust optimization framework. According to evaluate the degree of potential risk of the scheme set, comparative potential energy based multi-attribute decision making risk assessment with fuzzy trade-off method is proposed combined with WAA operator theory to provide a stable sort of scheme set, which supports for distributed energy resource optimization scheduling under active distribution network.

由于大量分布式能源的广泛接入，主动配电网系统更加复杂、网架结构更加灵活，且因大部分分布式能源为间歇式能源，系统又呈现出强随机的特征，给主动配电网下的分布式能源优化调度带来了新的挑战。本项目针对主动配电网系统多目标、多约束、高维非线性以及随机性强等特点，以主动配电网负荷和发电预测为出发点，运用ARMA自回归方法在长时间尺度进行初步预测，在短时间尺度提出基于经验模式分解的卡尔曼滤波校正方法对其预测结果进行校正；在此基础上，基于Lindeberg-Levy中心极限定理推求其不确定集合，建立分布式能源多目标鲁棒经济优化调度模型，结合两阶段零和博弈理论，采用Soyster鲁棒优化框架下的文化微分进化算法对其进行求解；根据方案集可能造成的潜在风险程度，结合WAA算子理论，提出基于相对优势可能势的模糊多属性折衷型风险决策评价方法，推求出各方案的稳定排序，为主动配电网分布式能源优化调度提供可靠的决策支持。

本项目针对主动配电网中分布式式能源优化存在的多目标、多约束、高维非线性以及随机性强等特点，以负荷和发电预测为出发点，基于多时间尺度预测框架，采用长时间尺度预测和短时间尺度校正的方法，结合环境气象数据信息，对负荷和发电出力进行准确预测；在此基础上，分析分布式能源出力的概率分布特性，推求出系统模型的不确定性集合，建立分布式能源多目标鲁棒经济优化调度模型，重点分析其在最坏情况下的调度情景，采用文化微分进化算法对其进行求解，推求出多目标优化Pareto方案集；根据其方案集可能造成的潜在风险程度，推求出各方案的稳定排序，为分布式能源优化调度提供可靠的决策支持。