电动汽车行驶路径优化及其价格响应特性分析
发布时间：2019-01-14 13:09
【摘要】：随着我国经济的快速发展以及交通基础设施的不断完善,区域内物资和人员的流通日益频繁,物流业逐渐成为社会经济的重要组成部分。配送是物流业的终端,配送环节的车辆路径优化是提高运输速度、降低运输成本的重要手段。目前,普通燃料汽车是物流配送的主要工具,但随着石油危机、能源安全和环境保护等问题的凸显,电动汽车作为新型经济环保的交通工具,有取代普通燃料汽车的趋势,进而引出了电动汽车行驶路径优化问题。电动汽车使用电能,与普通燃料汽车在能量补给方式和使用特性上有明显区别,使得电动汽车行驶路径优化问题比传统车辆路径问题更加复杂。因此,对该问题的研究在理论和现实应用方面都有重要意义。首先,构建了固定电价下的单辆电动汽车行驶路径优化模型。该模型同时考虑了电动汽车在配送过程中的快速充电行为,载货量对单位里程耗电量的影响以及快速充电对电池寿命的损耗,在满足路径约束和电池容量约束条件下实现配送成本最小。由电动汽车在同一充电站多次快速充电所带来的模型表述困难,通过引入充电站虚拟节点加以解决。模型采用遗传算法求解,在种群初始化阶段构建染色体标准化操作,使得构造的染色体适应于遗传算法的交叉操作。算法使用MATLAB编程,32节点配送系统的数值仿真显示了电动汽车在使用成本方面的优势,验证了模型的有效性。其次,构建了考虑分时充电成本的多辆电动汽车行驶路径优化模型。该模型同时考虑了电动汽车的快速充电行为,返回配送中心的常规充电优化以及快速充电对电池寿命的损耗,在满足路径约束、时间约束、载货量约束和电池容量约束条件下,实现配送成本最小。构建学习型单亲遗传算法对模型进行求解,该算法在种群初始化阶段制定了初始化规则,克服了多约束条件下初始可行解生成困难的问题；在基因变异阶段构建充电站节点删除算子和节点添加算子,保证算法的全局收敛性；构建包含了精英个体知识和专家经验知识的知识模型,通过个体对知识的学习,提升算法求解效能。基于62节点配送系统和112节点配送系统进行数值仿真,与遗传算法和禁忌算法的优化结果比较,显示出学习型单亲遗传算法在求解速度、求解质量和稳定性方面的优势。基于62节点配送系统,对比分析了电动汽车的配送成本和行驶路径对分时电价、固定电价和折扣电价等三种电价的响应特性。基于一小型配电网,研究了三种电价下较大规模电动汽车的充电负荷对配电系统用电负荷的影响。
[Abstract]:With the rapid development of China's economy and the continuous improvement of transportation infrastructure, the circulation of goods and personnel in the region is becoming more and more frequent, and the logistics industry has gradually become an important part of the social economy. Distribution is the terminal of logistics industry. Vehicle routing optimization is an important means to improve transportation speed and reduce transportation cost. At present, ordinary fuel vehicles are the main tools of logistics and distribution. However, with the oil crisis, energy security and environmental protection, electric vehicles as a new type of economic and environmental protection vehicles, there is a trend to replace ordinary fuel vehicles. Furthermore, the optimization problem of electric vehicle driving path is introduced. The use of electric energy in electric vehicles is obviously different from that of ordinary fuel vehicles in the way of energy supply and performance, which makes the optimization problem of electric vehicles' driving path more complex than the traditional vehicle's path problem. Therefore, the study of this problem is of great significance both in theory and in practice. Firstly, a single electric vehicle (EV) driving path optimization model at fixed electricity price is constructed. The model also takes into account the fast charging behavior of electric vehicles during distribution, the influence of loading capacity on power consumption per mileage, and the loss of battery life caused by rapid charging. The delivery cost is minimized under the condition of satisfying path constraints and battery capacity constraints. It is difficult to describe the model caused by electric vehicle charging at the same charging station many times quickly, which is solved by introducing the virtual node of charging station. The model is solved by genetic algorithm, and the standardized operation of chromosome is constructed in the initial stage of population, which makes the constructed chromosome adapt to the crossover operation of genetic algorithm. The algorithm is programmed with MATLAB, and the numerical simulation of 32-node distribution system shows the advantages of the electric vehicle in the cost of use, and verifies the validity of the model. Secondly, a multi-vehicle driving path optimization model considering time-sharing charging cost is constructed. The model also takes into account the fast charging behavior of electric vehicles, the conventional charging optimization of return distribution center and the loss of battery life due to rapid charging, under the condition of satisfying path constraints, time constraints, load constraints and battery capacity constraints. The cost of distribution is minimized. A learning parthenogenetic genetic algorithm is constructed to solve the model. The algorithm formulates initialization rules in the initial stage of population initialization, which overcomes the difficulty of generating initial feasible solutions under the condition of multiple constraints. In the phase of gene mutation, the node deletion operator and node addition operator are constructed to ensure the global convergence of the algorithm. A knowledge model including elite individual knowledge and expert experiential knowledge is constructed to improve the efficiency of the algorithm through individual learning of knowledge. Compared with the optimization results of genetic algorithm and Tabu algorithm, the numerical simulation based on 62 node distribution system and 112 node distribution system shows the advantages of learning single parent genetic algorithm in solving speed, solution quality and stability. Based on the 62 node distribution system, the response characteristics of the distribution cost and the driving path of the electric vehicle to the time-sharing price, the fixed price and the discount price are compared and analyzed. Based on a small distribution network, the influence of charging load of large scale electric vehicle under three kinds of electricity price on the load of distribution system is studied.
【学位授予单位】：长沙理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U492.22
本文编号：2408724
[Abstract]:With the rapid development of China's economy and the continuous improvement of transportation infrastructure, the circulation of goods and personnel in the region is becoming more and more frequent, and the logistics industry has gradually become an important part of the social economy. Distribution is the terminal of logistics industry. Vehicle routing optimization is an important means to improve transportation speed and reduce transportation cost. At present, ordinary fuel vehicles are the main tools of logistics and distribution. However, with the oil crisis, energy security and environmental protection, electric vehicles as a new type of economic and environmental protection vehicles, there is a trend to replace ordinary fuel vehicles. Furthermore, the optimization problem of electric vehicle driving path is introduced. The use of electric energy in electric vehicles is obviously different from that of ordinary fuel vehicles in the way of energy supply and performance, which makes the optimization problem of electric vehicles' driving path more complex than the traditional vehicle's path problem. Therefore, the study of this problem is of great significance both in theory and in practice. Firstly, a single electric vehicle (EV) driving path optimization model at fixed electricity price is constructed. The model also takes into account the fast charging behavior of electric vehicles during distribution, the influence of loading capacity on power consumption per mileage, and the loss of battery life caused by rapid charging. The delivery cost is minimized under the condition of satisfying path constraints and battery capacity constraints. It is difficult to describe the model caused by electric vehicle charging at the same charging station many times quickly, which is solved by introducing the virtual node of charging station. The model is solved by genetic algorithm, and the standardized operation of chromosome is constructed in the initial stage of population, which makes the constructed chromosome adapt to the crossover operation of genetic algorithm. The algorithm is programmed with MATLAB, and the numerical simulation of 32-node distribution system shows the advantages of the electric vehicle in the cost of use, and verifies the validity of the model. Secondly, a multi-vehicle driving path optimization model considering time-sharing charging cost is constructed. The model also takes into account the fast charging behavior of electric vehicles, the conventional charging optimization of return distribution center and the loss of battery life due to rapid charging, under the condition of satisfying path constraints, time constraints, load constraints and battery capacity constraints. The cost of distribution is minimized. A learning parthenogenetic genetic algorithm is constructed to solve the model. The algorithm formulates initialization rules in the initial stage of population initialization, which overcomes the difficulty of generating initial feasible solutions under the condition of multiple constraints. In the phase of gene mutation, the node deletion operator and node addition operator are constructed to ensure the global convergence of the algorithm. A knowledge model including elite individual knowledge and expert experiential knowledge is constructed to improve the efficiency of the algorithm through individual learning of knowledge. Compared with the optimization results of genetic algorithm and Tabu algorithm, the numerical simulation based on 62 node distribution system and 112 node distribution system shows the advantages of learning single parent genetic algorithm in solving speed, solution quality and stability. Based on the 62 node distribution system, the response characteristics of the distribution cost and the driving path of the electric vehicle to the time-sharing price, the fixed price and the discount price are compared and analyzed. Based on a small distribution network, the influence of charging load of large scale electric vehicle under three kinds of electricity price on the load of distribution system is studied.
【学位授予单位】：长沙理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U492.22
【参考文献】
相关期刊论文 前1条
1 曹二保;赖明勇;张汉江;;模糊需求车辆路径问题研究[J];系统工程;2007年11期
，本文编号：2408724
本文链接：https://www.wllwen.com/kejilunwen/jiaotonggongchenglunwen/2408724.html
