基于非合作博弈的能源互联网多元主体行为分析方法

发布时间：2018-01-04 20:05

本文关键词：基于非合作博弈的能源互联网多元主体行为分析方法　出处：《华北电力大学(北京)》2017年博士论文　论文类型：学位论文

【摘要】：随着能源互联网的建设推进,可再生能源广泛接入的特征将日趋明显,电力系统的结构逐渐由纵向垂直向扁平化型式进行转变,互联网思维的引入在促进能源高效流动的同时,也将给电力系统的稳定运行带来一系列的挑战。而当系统中决策主体的属性趋于对等放开,使用传统的优化理论开展研究将存在一定的局限性,亟需一种新的数学工具来对能源互联网中新型多元主体之间的互动问题进行建模分析。本论文基于非合作博弈理论,分别面向电能的数据感知、供需交易和优化管理三个领域,研究新兴多元主体在有效机制设计下各自的利益优化策略以及最终达到的系统均衡状态。主要的创新性研究工作如下:(1)针对数据感知无线传感网的能效提升问题,提出了多传感节点间的功率控制非合作博弈模型,利用Hessian矩阵的负定性质证明了所提模型纳什均衡点的存在,基于注水算法提出了一种迭代式的分布式功率分配方法,最大程度地提高了传感节点到数据收集节点(Sink节点)的链路传输能效。(2)针对分散自治的电能交易商业运营模式,提出了适应新型源网荷主体的售电竞争无限策略式非合作博弈模型,构造了包含激励和惩罚约束的售电主体非线性效用函数,在此情况下证明了所提博弈模型中纯策略纳什均衡点的存在性,通过最优反应算法获得了每个售电主体在市场达到均衡时的售电策略,仿真结果验证了算法的有效性。(3)针对需求响应中用户主动参与模式,设计了一项基于动态实时电价的负荷转移机制,允许智能用户自主参与并调整用电需求,并利用虚拟行动算法,求解出用户间在调峰指标分配非合作博弈模型中混合策略的(近似)纳什均衡点,并给出了适应规模主体的虚拟行动改进算法和聚类分析算法,揭示了用户在需求响应项目中参与的随机性并实现了概率量化预测。(4)针对源荷互联互通的新能源消纳模式,基于动态进化博弈理论设计了用户群面向微源群的分布式电能请求方法,使用复制动态方程捕捉了用户通过将分散式购电策略中各个体依次进化来竞争微网利用系数的过程,在解决供需两侧随机性电能匹配问题的同时,提升了用户需求实时满意度和微源的经济收益。论文的研究结果表明,应用非合作博弈理论对能源互联网中多元主体的互动行为进行研究分析,既能够实现对系统均衡状态的提前预测,也可以为系统运行机制的优化完善提供有效反馈。
[Abstract]:With the construction of energy Internet, the characteristics of extensive access to renewable energy will become more and more obvious. The structure of power system is gradually changing from vertical to flat. The introduction of Internet thinking will not only promote the efficient flow of energy, but also bring a series of challenges to the stable operation of power system. There will be some limitations in using the traditional optimization theory to carry out the research. A new mathematical tool is urgently needed to model and analyze the interaction between the new multi-agent in the energy Internet. Based on the non-cooperative game theory, this paper is based on the data perception of electric energy. Supply and demand transaction and management optimization. This paper studies the interest optimization strategy and the system equilibrium state of the emerging multi-agent under the effective mechanism design. The main innovative research work is as follows: 1). To improve the energy efficiency of data-aware wireless sensor networks. A non-cooperative game model of power control between multi-sensor nodes is proposed. The existence of Nash equilibrium point of the proposed model is proved by the negative definite property of Hessian matrix. An iterative distributed power allocation method based on water injection algorithm is proposed. The link transmission efficiency of sensor node to data collection node (Sink node) is improved to the maximum extent. An infinite strategy non-cooperative game model for power sale competition is proposed, and a nonlinear utility function of the agent is constructed, which includes incentive and penalty constraints. In this case, the existence of pure strategy Nash equilibrium point in the proposed game model is proved, and the optimal reaction algorithm is used to obtain the power sale strategy of each power seller when the market reaches equilibrium. Simulation results verify the effectiveness of the algorithm. (3) A load transfer mechanism based on dynamic real-time pricing is designed for the user active participation mode in demand response. Intelligent users are allowed to participate in and adjust the demand of electricity independently, and using virtual action algorithm to solve the (approximate) Nash equilibrium point of mixed strategy in the non-cooperative game model of peak-shaving index allocation among users. An improved virtual action algorithm and a clustering analysis algorithm are presented. The randomness of user participation in the demand response project and the realization of probabilistic quantitative prediction. 4) the new energy consumption model for source load interconnection is revealed. Based on the dynamic evolutionary game theory, a distributed power request method for micro-source group is designed. The replication dynamic equation is used to capture the process of users competing for the utilization coefficient of microgrid by evolution of each individual in the decentralized power purchase strategy, which solves the problem of stochastic power matching on both sides of supply and demand at the same time. The research results show that the non-cooperative game theory is used to analyze the interaction behavior of multi-agent in the energy Internet. It can not only predict the equilibrium state of the system in advance, but also provide effective feedback for the optimization and improvement of the system operation mechanism.
【学位授予单位】：华北电力大学(北京)
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TK01

【相似文献】