面向智能电网中能量管理的分布式算法研究

发布时间：2017-12-28 17:32

本文关键词：面向智能电网中能量管理的分布式算法研究　出处：《浙江大学》2017年博士论文　论文类型：学位论文

【摘要】：智能电网是在传统物理电网的基础上,覆盖了通信网络,融合了大量分布式能源、柔性负载、先进的传感测量技术和控制手段的新一代电网,又称"电网2.0"。与传统的电力系统一样,智能电网中的最优能量管理具有十分重要的地位。最优的能量管理不但可以产生巨大的经济效益,提高智能电网的经济效率,同时也有助于智能电网的稳定运行。在智能电网中,由于传统中心化算法往往会受限于有限的通信和计算能力而导致不可行,因此需要设计分布式算法,使得各个节点仅进行局部计算和邻居间通信即可合作地实现全局目标。本文主要研究了智能电网中能量管理问题的分布式算法,具体的贡献有如下几方面:1.简要介绍了智能电网、分布式算法、能量管理等概念及其背景和国内外研究现状。2.研究了静态机组负荷经济分配问题,提出了分布式二分算法,使得智能电网中的各个节点能够分布式地找到全局最优解,实现单一时刻的发电总成本最小化。该算法基于类协同算法,利用二分查找的思想使得各个节点渐进地找到最优拉格朗日乘子以及最优解。该算法对通信网络的要求较低,不仅适用于连通的无向图,也适用于强连通的有向图。3.研究了含有储能设备的动态机组负荷经济分配问题,其中储能设备既可以实现能量倒卖,也能为电网提供旋转备用服务。针对该问题,本文提出了基于ADMM思想的分布式算法,能够实现多时刻累积的发电总成本最小化。通过把问题转化为ADMM的形式,该问题被分解为各节点的局部带约束优化问题和领导者节点的无约束优化问题,从而迭代地实现分布式最优求解。4.研究了电动汽车的最优充放电规划问题,即在满足各个电动车用电需求的前提下,最优地规划各个时刻的充电或放电功率,使得总负载曲线尽可能地平坦。其中,电动汽车不仅可以通过充电来实现"填谷",还可以在用电高峰时利用逆变器将剩余电能回馈到电网中。由于本文考虑了充放电过程中的能量损耗,因此该问题不可避免地被描述为一个混合离散优化问题。为了解决这一难题,本文利用夜间基础负载曲线的特征作为假设,将原问题合理地近似为一个凸优化问题,提出了一种分布式算法。5.最后,本文研究了利用温控负载(如空调)进行需求侧响应的问题,并提出了一种分布式控制算法,使得多个空调在共同消耗给定电能的基础上,实现温度舒适度意义上的电能公平分配。
[Abstract]:Smart grid is a new generation of power grid which is based on the traditional physical grid, which covers the communication network, and integrates a large number of distributed energy, flexible load, advanced sensing and measuring technology and control means. It is also known as the "power grid 2". As with the traditional power system, the optimal energy management in the smart grid is of great importance. The optimal energy management can not only produce huge economic benefits, improve the economic efficiency of the smart grid, but also help the stable operation of the smart grid. In smart grid, because traditional centering algorithms are often limited by limited communication and computing power, it is not feasible. Therefore, we need to design distributed algorithm, so that each node can achieve global goals cooperatively by only local computation and neighbor communication. In this paper, the distributed algorithm of energy management in smart grid is mainly studied. The contributions are as follows: 1., the concept, background and research status of smart grid, distributed algorithm and energy management are briefly introduced. 2., the problem of static unit load economic allocation is studied. A distributed two point algorithm is proposed, which enables all nodes in smart grid to find the global optimal solution in a distributed way and minimize the total cost of generating at a single time. The algorithm is based on the class cooperative algorithm, using the idea of two point search to make each node find the optimal Lagrange multiplier and the optimal solution. The algorithm is less required for the communication network, not only for the connected undirected graph, but also for the strong connected digraph. 3., the problem of dynamic unit load economic allocation with energy storage equipment is studied, in which energy storage equipment can not only realize energy reselling, but also provide spares reserve service for power grid. In order to solve this problem, a distributed algorithm based on ADMM is proposed to minimize the total generation cost of generation. By transforming the problem into the form of ADMM, the problem is decomposed into the local constrained optimization problem of each node and the unconstrained optimization problem of leader nodes, so as to achieve the optimal solution in an iterative way. 4., the optimal charging and discharging planning problem of electric vehicle is studied. That is, under the premise of meeting the electric demand of each electric vehicle, the charging or discharging power of each time is optimized, so that the total load curve is as flat as possible. Among them, the electric vehicle can not only "fill the valley" by charging, but also can use the inverter to feed the residual energy into the power grid at the peak of electricity use. Because the energy loss in the charge discharge process is considered in this paper, the problem is inevitably described as a mixed discrete optimization problem. In order to solve this problem, we use the characteristics of night load curve as a hypothesis to reasonably approximate the original problem to a convex optimization problem, and propose a distributed algorithm. 5., in the end, this paper studies the demand side response using temperature control load (such as air conditioning), and proposes a distributed control algorithm, which enables multiple air-conditioners to achieve fair distribution of electricity in the sense of temperature comfort on the basis of common consumption of a given electric energy.
【学位授予单位】：浙江大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TM76

【相似文献】