【作者】 韩文博；

【导师】 吴杰康；

【作者基本信息】 广东工业大学 ， 电气工程， 2016， 硕士

【摘要】 随着科技的发展和城市配套设施的不断完善,原有的配电网规划方法已经无法适应与日俱增的新兴电力负荷。为更好的应对具有不同特点的各种新型负荷,需要将负荷细分,为不同类型的负荷制定不同的配电网规划准则。本文针对公路网无线充电负荷特点,在保证配网运行可靠性和安全性的基础上,建立了以建设和运行成本最低为目标的配网规划模型,为未来公路网无线充电负荷接入电网提供技术参考。本文中公路网负荷指的就是铺设了电动汽车动态无线充电系统的公路,在大量电动汽车进行动态充电时所产生的负荷。公路网负荷预测是进行电网规划建设以及运行调度控制的基础。电动汽车动态无线充电的时空分布具有很强的随机性,本文在大量文献的基础上,根据公路网无线充电系统的工作原理和工作场景,得出大量电动汽车接入充电时的负荷预测模型,并分析模型中的各种参数：对不同用途的电动汽车行驶特征进行归纳总结,利用蒙特卡洛法抽样计算得出其不同时刻的接入概率,并用对数正态概率分布函数和三参数威布尔函数进行拟合,使用Greenshields--LN模型求取平均车速,确定车流量和平均车速的模型,对电动车渗透率、气象条件、车道分担率等参数进行标定,最后通过使用蒙特卡洛模拟方法进行大量的抽样仿真,计算得出单位路段实际日负荷的均值和方差。通过对比城市地铁配电措施和普通住宅小区配电方案,结合《国家电网公司城市配电网技术导则》。根据公路网负荷无法一次性完全接入已有配电网、且负荷呈带状分布等实际情况,本文建议：“公路网负荷优先考虑接入已有配电网,并根据实际情况微调,最后为剩余负荷新建专供变电站”。为了使公路网负荷以最优的方式接入已有配电网,本文提出“电网找负荷,以点寻线”的新思路,以电网节点为源头寻找连接公路网的最短路径,用Floyd算法结合迷宫算法并引入竞争机制,解决了实际电网建设时公路电源接入点以及最优路径的选取问题。通过使用IEEE33节点网络,仿真计算了公路网负荷接入配电网时,负荷的随机性对电网电能质量的影响。本文最后提出了一种面向公路网负荷的配电网新建变电站多源连续选址定容模型。介绍了一种新的变电站选址定容智能优化算法：纵横交叉算法(CSO)。根据约束条件和目标函数以及相关信息,设置CSO参数并进行求解,最后与PSO、GA算法进行了对比。结果表明,纵横交叉算法具有更强的全局寻优搜索性能和更高的收敛速率,而且丰富了种群的多样性,极大的改善了传统智能算法后期收敛速度慢、易陷入局部最优解等缺点。寻优过程中可快速令变电站费用、网损费用等经济投资费用收敛至最优,因此使用本文提供的模型和算法得到的新建变电站多源选址定容方案具有较好的工程应用价值。更多还原

【Abstract】 With the development of science and technology and the continuous improvement of the urban infrastructure, the original distribution network planning method unable to adapt to the day growing emerging load.In order to better deal with various new types of load with different characteristics, we need to segment the load, formulating different distribution network planning criteria for different types of load.According to the features of road network wireless charging, based on the operating reliability and security of distribution network, established the distribution network planning model with the objective of construction cost and operate cost minimum, it can provide technical information for road network wireless charging load integration into power grid in the future.In this paper, the road network load refers to the load produced by dynamic charging of a large number of electric vehicles with the road which has installed the dynamic wireless charging system of electric vehicles.Highway network load forecasting is based on power grid planning and scheduling control. The dynamic wireless charging load of EVs is fairly random on time and space scale. On the basis of large amounts of literature, according to the working principle and different scenes of the highway network dynamic wireless charging system, charging load calculation formula is acquired when a large number of electric vehicles access to power grid and parameters of model are found. By analyzing the traveling characteristics of EVs with different functions, Monte Carlo sampling method is used to generalize the probability of starting charging on different periods and the logarithmic normal probability distribution function and the three parameter Weibull function are used to fit the probabilistic model. Meanwhile, Greenshields-LN model is proposed to acquire the correlation between traffic flow and average speed to simulate traffic flow and average speed model. The penetration of electric vehicles, weather conditions and lane sharing rate are calibrated. Finally by using Monte Carlo sampling, the mean and variance of daily charging load per road unit are calculated.By comparing the urban subway distribution measures and ordinary residential distribution scheme, combined with the State Grid Corporation of the Urban Distribution Network Technical Guidelines, and according to the features that road network load cannot one-time complete access to the existing distribution network and the load was zonal distribution and other characteristics, this paper suggests:"road network load priority access to the existing grid, and fine-tuning according to the actual situation, and finally the remaining load access to the new substation".In order to make the road load in the optimal way to access the existing distribution network, we proposed the new ideas of "Grid find the load, through the point connect a line", using the grid node as the source to find the shortest path connecting the road network, choosing the hybrid Floyd and maze algorithm with the introduction of competition mechanism, this method can solve the problem of selecting the road load access point and optimal path when constructing the actual grid. By using IEEE33 node network system, this paper had simulated when the road load access to the distribution network the randomness of the traffic flow impact on the power quality, and analysis the influence of daily road load on the daily distribution network load curve.In this paper, we use the multi-source and continuous selecting location model to analysis the problem of new constructing substation selecting location and confirming capacity. A novel intelligent optimization algorithm called Crisscross optimization algorithm (CSO) is proposed to solve this problem. Also, according to the actual situation of load, this paper carries out the division of the road load unit, extracted the size of load unit and location parameters, and then input these information into the multi-objective transformer substation model. Basing the constraints and objective function and associated message to set the parameters of CSO algorithm to solve this problem, the simulation results ware compared with the results obtained by PSO and GA. The results show that CSO not only has stronger global search ability, faster convergence rate, but also enriches the diversity of the population, greatly improve the shortcomings of low convergence rate at the late period and easy fall into local optimal solution of ordinary intelligent algorithm. The optimization process can quickly make substation cost, network loss cost and other economic investment spending converge to the optimal solution, obtained results has good engineering application value.更多还原