基于随机优化的工业区智能电网调度

发布时间：2018-06-23 16:15

本文选题：马尔可夫决策过程(MDP) + 在线学习　；参考：《吉林大学》2017年硕士论文

【摘要】：目前,随着科技的日益发展,智能电网也逐渐受到用户的关注和重视,而新型能源的开发与应用也是标志科技创新发展的重要依据.由于天气气候的变化,新型电能产能量带有的未知性,结合用户日常用电的不规律性:建立带有未知随机意义下的分布式电能调度系统,变得更有现实意义:也成为了建立智能电网的热点.本文主要针对工业区传统能源和新型能源的产能,利用工业电车的蓄电池作为剩余电能的临时存储点,建立智能电网进行电能调度,使得从电力公司输入的电能费用达到最少.采用马尔科夫决策过程(MDP)对带有未知随机初值的电能产值、消费、存储进行建模.提出一个在线学习控制方法能够适应系统信息的动态增长并且使得模型收敛到最优状态.这样,当算法收敛到最优状态时所对应的控制策略便是使得电能消费最少的最优策略.
[Abstract]:At present, with the development of science and technology, smart grid has been paid more and more attention by users, and the development and application of new energy is also the important basis to mark the development of science and technology innovation. Due to the change of weather and climate, the unknown energy of the new power generation, combined with the irregularity of users' daily electricity consumption, a distributed power dispatching system with unknown random sense is established. Becoming more realistic: it has also become a hot spot in the establishment of smart grids. In this paper, aiming at the capacity of traditional and new energy in industrial area, the battery of industrial tram is used as the temporary storage point of residual energy, and the smart grid is established for power dispatching, which makes the input cost of electric energy from power company reach the minimum. Markov decision process (MDP) is used to model the output value, consumption and storage of electric energy with unknown random initial values. An online learning control method is proposed, which can adapt to the dynamic growth of system information and make the model converge to the optimal state. Thus, when the algorithm converges to the optimal state, the corresponding control strategy is the optimal strategy to minimize the power consumption.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM73

【相似文献】