基于充电需求预测与OD流量生成的电动汽车充电站选址优化

Optimization of Electric Vehicle Charging Station Location Based on Charging Demand Prediction and OD Flow Generation

摘要: 针对当前电动汽车充电站选址规划依赖历史起点–终点(OD)出行数据的局限性，本文提出一种基于充电需求预测的出行流量生成新方法，并以此驱动充电站布局优化。研究首先构建了融合里程焦虑量化、初始电量分布与马尔可夫链目的地选择的充电需求生成机制，通过蒙特卡洛模拟得到充电需求的时空分布；然后将预测的充电需求作为出行流量生成的源头，通过引入广义出行成本(涵盖行驶时间、充电次数与路径偏差)作为阻抗，改进双约束重力模型，生成出行流量的OD流量矩阵。在此基础上，建立以最大化捕获流量为目标的充电站选址模型，并设计贪婪–遗传混合启发式算法进行求解。以淄博市张店区为例的案例研究表明，该方法能够从充电行为机理直接生成出行流量，有效识别需求热点与关键节点，且选址效益呈现显著边际递减规律，为充电基础设施的精准、分期规划提供了从需求预测到流量生成的一体化决策框架。

Abstract: To address the limitations of current Electric Vehicle (EV) charging station location planning that relies on historical Origin-Destination (OD) travel data, this paper proposes a novel travel flow generation method based on charging demand prediction and uses it to drive the optimization of charging station layout. First, a charging demand generation mechanism integrating range anxiety quantification, initial battery state-of-charge distribution, and Markov chain-based destination selection is constructed, and the spatiotemporal distribution of charging demand is obtained through Monte Carlo simulation. Then, the predicted charging demand is taken as the source for travel flow generation. By introducing generalized travel cost (including travel time, charging frequency, and route deviation) as impedance, an improved double-constrained gravity model is used to generate an OD matrix of travel flow. On this basis, a flow-capturing location model aimed at maximizing captured flow is established and solved using a greedy–genetic hybrid heuristic algorithm. A case study of Zhangdian District, Zibo City shows that the proposed method can directly generate travel flow from charging behavior mechanisms, effectively identify demand hotspots and key nodes, and exhibit a clear diminishing marginal benefit in site selection. This provides an integrated decision-making framework from demand prediction to flow generation for precise and phased planning of charging infrastructure.