【作者】 李丹；

【导师】 黄亚峰；

【作者基本信息】 东北电力大学 ， 工程硕士（专业学位）， 2020， 硕士

【摘要】 我国的东北、华北和西北地区电源结构单一,大多以热电机组为主,同时风电大量并网,使电网难以调峰。热电机组受“以热定电”工况约束,缺乏其他灵活调节能源设备,风电具有间歇性、随机性和与电负荷相反的峰谷特性,这些因素造成了风电高峰期调峰困难,弃风现象较为严重。计及热网热能传输动态特性以及储热装置储热的电-热综合能源系统将电力系统、供热系统耦合在一起,形成了多维能源互联的能源形式,源网荷协调参与到日前优化调度中,对于解耦热电机组“以热定电”的约束,促进清洁能源的高效发展有着重要意义。首先,电能和热能作为电-热综合能源系统的主要能源输出,受到多方面因素的影响呈现一定波动性和随机性,为了分析电热负荷存在的变化规律,准确地预测电热负荷,本文提出了一种基于GA-BP神经网络算法的电热负荷预测。首先考虑了各影响因素数据的波动以及获取难易,确定电热负荷的输入输出变量,在此基础上构建了 BP神经网络模型并描述了遗传算法的构成与设计,通过仿真算例,与传统的BP神经网络算法的电热负荷预测进行对比分析,验证GA-BP神经网络算法预测电热负荷的精度以及可信度。其次,在实现基于GA-BP神经网络算法的电热负荷预测的基础上,介绍了供热系统典型结构,并在分析实际热网结构、运行特点基础上,建立供热系统模型。着重分析采用节点法来描述供热管网热能传输的动态过程,建立考虑热传输延时以及热能损耗的热能传输动态模型。根据供热系统模型,仿照电力系统潮流计算,设置实例仿真计算供热一次网典型管道出口温度以及换热首站供回水温度。对比预测的热负荷值证明热能传输动态模型搭建的正确性与可行性,并对考虑热能传输动态特性与储热能够使热能可跨时段转移进行了分析研究。最后,将热网作为可调度资源,提出四种计及热网热能传输动态特性及储热装置储热的电-热综合能源系统日前优化调度方案,实现对热网虚拟储能的调度利用。其中通过最优分配能源设备出力,使总运行成本最低并加入弃风惩罚作为目标函数。通过实际电-热综合能源系统作为算例进行算例分析,对比基于GA-BP神经网络算法的电热负荷预测值以及风电预测量,分析计算不同优化调度方案通过改变各能源设备出力带来的风电消纳率与经济性的结果。更多还原

【Abstract】 China’s northeast,north and northwest regions have a single power supply structure,most of which are dominated by thermal power units.At the same time,a large amount of wind power is connected to the grid,making it difficult to adjust the peak power grid.The thermal power generating unit is constrained by the "power subject to heat" operating conditions,lacking other flexible energy-regulating equipment,and wind power has intermittent,random and peak-valley characteristics opposite to the electrical load.These factors make it difficult to adjust the peak wind power peak period The phenomenon is more serious.Taking into account the dynamic characteristics of heat energy transmission in the heat network and the heat-storage device’s heat-storage device,the electricity-heat Integrated energy system couples the power system and the heating system together to form a multi-dimensional energy interconnected energy form.Among them,it is of great significance to the constraint of decoupling thermal power units to"power subject to heat" and promote the efficient development of clean energy.At first,electrical energy and thermal energy,as the main energy output of the electricity-heat Integrated energy system,are affected by various factors and present a certain degree of volatility and randomness.In order to analyze the changes in the electric heating load and accurately predict the electric heating load,this paper proposes a Electric heating load forecast based on GA-BP neural network algorithm.First of all,considering the fluctuation of the data of various influencing factors and the difficulty of obtaining data,the input and output variables of the electric heating load are determined.On this basis,the BP neural network model is constructed and the composition and design of the genetic algorithm are described.Through simulation examples,the traditional The electric heating load forecast of BP neural network algorithm is compared and analyzed to verify the accuracy and credibility of the GA-BP neural network algorithm for forecasting electric heating load.Secondly,on the basis of realizing the electric heating load forecast based on GA-BP neural network algorithm,the typical structure of the heating system is introduced,and the heating network model is established on the basis of analyzing the actual heating network structure and operating characteristics.It focuses on the analysis of the node method to describe the dynamic process of thermal energy transmission in the heating pipe network,and establishes a thermal energy transmission dynamic model that takes into account thermal transmission delay and thermal energy loss.According to the heat network model,modeled on the power system power flow calculation,an example is set to simulate the typical pipe outlet temperature of the heating primary network and the return water temperature of the first heat exchange station.Comparing the predicted heat load values proves the correctness and feasibility of building a dynamic model of thermal energy transmission,and analyzes and studies the thermal energy transfer characteristics and heat storage can make thermal energy transfer across time periods.Finally,using the heat network as a schedulable resource,four types of optimal scheduling schemes for the electricity-heat Integrated energy system that take into account the dynamic characteristics of heat energy transmission in the heat network and the heat storage of the heat storage device are proposed to realize the scheduling and utilization of the virtual energy storage of the heat network.Among them,through the optimal allocation of energy equipment output,the total operating cost is the lowest and the wind curtailment penalty is added as the objective function.The actual electricity-heat Integrated energy system is used as an example to analyze the calculation example,compare the predicted value of the electric heating load and the predicted amount of wind power based on the GA-BP neural network algorithm,and analyze and calculate the wind power brought by different optimized dispatching schemes by changing the output of each energy equipment Results of consumption rate and economy.更多还原