多能源复合型电动汽车充换储放电站的能量管理技术研究

发布时间：2018-06-27 02:09

本文选题：电动汽车充换储放电站 + 光伏发电预测　；参考：《华中科技大学》2014年博士论文

【摘要】：由于全世界都面临着环境和能源危机,可再生能源和电动汽车受到了世界各国的高度重视。电动汽车充电基础设施影响了电动汽车推广和发展,多能源复合型电动汽车充换储放电站(Electric Vehicle Battery Charge-Swap-Storge-Discharge Power Station, EV-BCSSDPS)作为重要的充电基础设施,它将充电站、换电站和储能电站的功能融合到一起。多能源复合型EV-BCSSDPS不仅可为电动汽车提供快速便捷的换电服务,还可为电动汽车提供清洁的充电能源,另外还可利用梯次储能作为后备电源。随着电池技术的进一步发展和智能电网的建设,EV-BCSSDPS作为智能电网的组成部分,合理利用其储能特性将对平抑电网负荷波动、接纳间歇性可再生能源及提高电网运行效率起到重大作用。本文以多能源复合的EV-BCSSDPS为对象,对其能量预测和能量管理的基本分析理论和设计方法进行了深入研究,包括有光伏发电预测、充电负荷需求预测、经济优化运行管理和成本收益分析。准确的预知光伏系统的输出功率,对EV-BCSSDPS在未来时段内动力电池、梯次储能的充放电及与电网交易有着非常重要的意义。针对分布式发电的随机性问题,本文建立了光伏系统短期发电预测系统框架。首先从理论和数据上分析了气象因素与光伏发电量之间的相关性,并给出了基于距离分析方法的相关性计算准则,考虑了国内太阳辐射站点稀少且预报能力较低的特点,确定了以气温和湿度作为基于神经网络预测模型的输入因子。并给出了含隐含层节点的网络结构设计方法,建立了基于反传播(Back Propagation, BP)神经网络的短期无辐照度输出功率的预报模型,给出了定量评估模型精度的准则。此外为增强模型对天气突变的适应能力,由云量预报信息对天气类型聚类识别,采用自组织特征映射(Self-organizing Feature Map, SOM)方法聚类天气类型,继而对各天气类型采用相应的预测网络,避免单神经网络的过拟合问题。针对电动汽车换电需求的随机性问题,准确预知电动汽车的换电及充电需求,开展EV-BCSSDPS充电负荷特性研究对于EV-BCSSDPS内动力电池的有序、经济充电,电网安全运行,站内其它微电源和电网的经济调度等都有重要意义。本文分析了某BCSSDPS的运营基础数据,利用BP及RBF神经网络和随机建模方法建立了逐小时换电车辆数模型,基于此建立了电池箱充电起始时刻模型；此外,提出利用行驶里程作为电池充电量需求的度量标准,分析了车辆行驶里程多样性的特征,建立了基于高斯混合模型(Gaussian Mixture Model, GMM)的行驶里程模型,间接得出电池初始荷电状态(Initial State-of-Charge, SOC0);另外,电池与充电机的特性决定了电池充电功率与充电时长。因此综合考虑上述因素,建立了EV-BCSSDPS的充电负荷模型,并给出充电负荷功率计算及预测流程,并编写了充电负荷预报软件,实现了EV-BCSSDPS充电负荷需求的预报功能。最后基于非参数核密度估计方法分析了电动汽车充电负荷预报的不确定性。针对光伏出力和电动汽车充换电需求的随机性会对EV-BCSSDPS运营带来不利的影响,本文建立了多能源复合型EV-BCSSDPS的能量管理模型和经济化调度策略。本文对EV-BCSSDPS内微电源如动力电池、梯次储能、光伏、非充电负荷等进行了分析和建模,优化模型中充分考虑了电动汽车换电需求、动力电池充电需求、电池均衡使用约束、功率平衡约束、梯次储能功率约束等,并在光伏发电预报和电动汽车充换电需求预报的基础上,建立了涵盖充电成本最小、系统负荷波动最小及兼顾二者的EV-BCSSDPS能量调度优化模型。所建立的混合整数规划和二次规划模型可利用CPLEX求解,可给出包括每组卸载动力电池、梯次储能、电网、光伏内的出力组合。将仿真结果与无序充电方案对比,定量评估了不同优化策略的能量管理模型对充电成本和系统负荷波动的影响。由于实际运行中,充换电需求预报会有误差,本文也定量评估了换电需求预报误差对EV-BCSSDPS优化运行的影响。 EV-BCSSDPS是我国电动汽车应用的重要充电基础设施之一,得到了广泛关注和示范推广。本文探讨了EV-BCSSDPS的组成结构和运营模式,选择以电池租赁运营模式的EV-BCSSDPS为研究对象,考虑了其投资成本、运营以及维护费用、人工薪酬等成本及充换电服务等收益,并给出评估EV-BCSSDPS成本效益模型的评价指标。建立了基于GUI的EV-BCSSDPS全寿命周期成本收益分析软件,并对该模型进行敏感性分析,得出影响EV-BCSSDPS收益的关键因素序列。该模型及分析结果为EV-BCSSDPS商业化运行提供了一些评估依据。
[Abstract]:As the world is facing the environmental and energy crisis, renewable and electric vehicles have been highly valued by all countries in the world. The electric vehicle charging infrastructure has affected the promotion and development of electric vehicles, and the Electric Vehicle Battery Charge-Swap-Storge-Discharge Power Station EV-BCSSDPS), as an important charging infrastructure, combines the functions of the charging station, the power station and the energy storage power station. The multi energy composite EV-BCSSDPS can not only provide fast and convenient switching services for electric vehicles, but also provide a clean charging energy source for electric vehicles. In addition, the cascade energy storage can be used as a backup power supply. With the further development of battery technology and the construction of smart grid, EV-BCSSDPS as a component of the smart grid, the rational use of its energy storage characteristics will play an important role in reducing the load fluctuation of the power grid, accepting intermittent renewable energy and improving the efficiency of the power grid operation. The basic analysis theory and design method of measurement and energy management are studied in depth, including photovoltaic power generation forecast, charge load demand forecast, economic optimization operation management and cost income analysis.
The accurate prediction of the output power of the photovoltaic system is of great significance to the charging and discharging of the power battery, the cascade energy storage and the power grid transaction in the future period of time. Aiming at the randomness of the distributed generation, the framework of the short-term generation forecasting system for the photovoltaic system is set up in this paper. First, the meteorology is analyzed in theory and data from the theory and data. The correlation between the factors and the photovoltaic power generation is given, and the correlation calculation criterion based on the distance analysis method is given. Considering the rare and low prediction ability of the domestic solar radiation stations, the temperature and humidity as the input factor based on the neural network prediction model are determined, and the network structure containing the hidden layer nodes is given. The prediction model of short term irradiance output power based on Back Propagation (BP) neural network is established, and the criterion for quantitative evaluation of the model accuracy is given. In addition, the adaptability of the model to weather mutation, the clustering recognition of cloudiness forecast information to the sky gas type, and the self organizing feature mapping (Self-or) are used. Ganizing Feature Map (SOM) method clustering weather types, and then adopt corresponding prediction network for different weather types to avoid over fitting of single neural network.
In view of the randomness of the demand for electric vehicles, it is accurate to predict the demand for electric vehicles switching and charging. It is of great significance to carry out the study of the EV-BCSSDPS charging load characteristics for the order of the power battery in the EV-BCSSDPS, the economic charging, the safe operation of the power grid, the other micro power supply and the economic dispatch of the power grid in the station. This paper analyzes a BCS SDPS's basic operation data, using BP and RBF neural network and random modeling method to establish an hour Model for the number of tramcars. Based on this, the starting time model of battery tank charging is established. In addition, the characteristics of the range diversity of vehicle driving range are analyzed by using the driving mileage as the measurement standard of the battery charge demand. The initial charge state of the battery (Initial State-of-Charge, SOC0) is indirectly obtained by the driving mileage model of the Gauss Mixture Model (GMM Model, GMM). In addition, the characteristics of the battery and the charger determine the battery charging power and the length of the charge. Therefore, the charging load model of EV-BCSSDPS is established, and the charge load model of EV-BCSSDPS is established, and the charge load model of EV-BCSSDPS is established. The charging load power calculation and prediction process are calculated, and the charging load forecasting software is written. The forecast function of the EV-BCSSDPS charging load demand is realized. Finally, the uncertainty of the electric vehicle charging load forecasting is analyzed based on the non parameter kernel density estimation method.
In this paper, the energy management model of multi energy complex EV-BCSSDPS and the economic scheduling strategy are established in this paper. This paper analyzes and builds the micro power supply in EV-BCSSDPS, such as power battery, cascade energy storage, photovoltaic, non charge load, etc. in this paper, aiming at the adverse effects of the photovoltaic power and the randomness of the electric vehicle charging and switching demand on the operation of EV-BCSSDPS. In the model, the optimization model takes full account of the electric vehicle switching demand, the power battery charging demand, the battery balance constraint, the power balance constraint, the cascade energy storage power constraints and so on. On the basis of the photovoltaic power forecast and the electric vehicle replacement demand forecast, the minimum charge cost, the minimum load fluctuation of the system and the balance of two are established. The EV-BCSSDPS energy scheduling optimization model. The mixed integer programming and the two programming model can be solved by CPLEX. The combination of power battery, cascade energy storage, power grid and PV can be given. The simulation results are compared with the disordered charging scheme, and the energy management model of different optimization strategies is evaluated by the fixed quantity. The effect of charging cost and system load fluctuation. Due to the error of the demand forecast for charge transfer in actual operation, this paper also quantitatively evaluates the influence of the demand forecast error on the optimized operation of EV-BCSSDPS.
EV-BCSSDPS is one of the important charging infrastructure for electric vehicles in China, and it has received extensive attention and demonstration and promotion. This paper discusses the structure and operation mode of EV-BCSSDPS, chooses the EV-BCSSDPS of battery leasing operation mode as the research object, and takes into account the cost of investment, operation and maintenance, artificial compensation and so on. And the benefit of charge transfer service, and the evaluation index of EV-BCSSDPS cost benefit model is given. A GUI based EV-BCSSDPS full life cycle cost benefit analysis software is set up, and the sensitivity analysis of the model is carried out to get the key factors affecting the EV-BCSSDPS income. The model and the analysis result are commercial operation of EV-BCSSDPS. Some basis for evaluation is provided.
【学位授予单位】：华中科技大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TM615;U469.72

【引证文献】