New Algorithm Optimizes EV Charging Station Placement and Capacity
As the global push toward sustainable transportation accelerates, electric vehicles (EVs) have emerged as a cornerstone of the future mobility landscape. With governments worldwide setting ambitious targets for carbon neutrality, the transition from internal combustion engine vehicles to EVs is no longer a distant vision but an ongoing reality. In China, this shift is particularly pronounced, driven by national policies aimed at achieving carbon peak by 2030 and carbon neutrality by 2060. The country’s EV market has experienced exponential growth in recent years, with projections indicating that by 2025, there will be 25 million EVs on Chinese roads, supported by a charging infrastructure aiming for a vehicle-to-charger ratio of 2:1. However, realizing this goal requires more than just scaling up the number of charging stations—it demands intelligent, data-driven planning that balances economic feasibility with user convenience.
Despite significant investments in charging infrastructure, a persistent challenge remains: the mismatch between supply and demand. The phenomenon of “cars without chargers” or “chargers without cars” highlights inefficiencies in current deployment strategies. In some areas, charging stations sit underutilized, representing wasted capital and operational costs. In others, drivers face long wait times and range anxiety due to insufficient or poorly located charging options. This imbalance not only frustrates users but also hinders the broader adoption of EVs. The root of the problem lies in the complexity of charging station planning, which involves determining not only where to build stations but also how large they should be—a process known as site selection and capacity determination.
Traditional approaches to this problem have often prioritized a single objective, such as minimizing construction costs or maximizing coverage. While these methods have contributed to the expansion of charging networks, they frequently fail to account for the competing interests of different stakeholders. For charging station operators, the primary concern is return on investment and operational efficiency. For EV drivers, the focus is on accessibility, short travel distances to stations, and minimal waiting times. A truly effective solution must reconcile these often conflicting goals, ensuring that the charging network is both economically viable and user-friendly.
Recognizing this need, a team of researchers from Fujian Normal University has developed a novel approach to EV charging station planning that integrates economic and user-centric considerations into a unified optimization framework. Their work, published in the Journal of Fujian Normal University (Natural Science Edition), introduces an advanced algorithm designed to simultaneously optimize station location and capacity. The method, named Adaptive Particle Swarm Optimization Genetic Algorithm (APSOGA), represents a significant advancement in the field, offering a more holistic and efficient solution to one of the most pressing challenges in the EV ecosystem.
The research team, led by Huang Ziqing, Lin Bing, Lu Yu, Liu Dui, and Wang Mingfen, began by addressing the foundational element of any planning model: demand prediction. Accurate forecasting of EV charging behavior is essential for determining the number and size of stations required. The team analyzed daily driving patterns of EVs, leveraging statistical models to predict the spatial and temporal distribution of charging demand. By understanding where and when drivers are likely to need charging, the researchers were able to establish a realistic range for the number of stations needed in a given area. This data-driven approach ensures that the planning process is grounded in real-world usage patterns rather than arbitrary assumptions.
Building on this demand forecast, the team constructed a comprehensive optimization model that considers both the costs incurred by charging station operators and the costs experienced by EV users. The operator’s costs include capital expenditures for land acquisition, construction, and equipment, as well as ongoing operational and maintenance expenses. These are balanced against the user’s costs, which are primarily composed of the time and energy spent traveling to a charging station (detour cost) and the time spent waiting for a charger to become available (waiting cost). By combining these two cost components into a single objective function, the model seeks to minimize the total societal cost of the charging network.
The heart of the researchers’ innovation lies in the APSOGA algorithm, which combines the strengths of two well-known optimization techniques: Particle Swarm Optimization (PSO) and Genetic Algorithms (GA). PSO is inspired by the social behavior of birds flocking or fish schooling, where individual agents (particles) move through a solution space, guided by their own best-known position and the best-known position of the entire swarm. While PSO is efficient and easy to implement, it can sometimes converge too quickly to a suboptimal solution, getting stuck in local minima. GA, on the other hand, mimics the process of natural selection, using operations such as crossover and mutation to generate new solutions. GA is excellent at exploring a wide solution space but can be slower to converge.
APSOGA intelligently merges these two approaches. It uses the particle update mechanism of PSO as its core, but enhances it with the crossover and mutation operators from GA. This hybridization allows the algorithm to maintain a diverse population of potential solutions, reducing the risk of premature convergence. Additionally, the researchers introduced adaptive mechanisms that dynamically adjust key parameters, such as the inertia weight, based on the progress of the optimization. In the early stages, the algorithm favors exploration, searching broadly across the solution space. As it progresses, it shifts toward exploitation, refining the best solutions found so far. This balance between exploration and exploitation is crucial for finding high-quality solutions in complex, multi-dimensional problems like charging station planning.
To validate their method, the researchers conducted a case study in Xiamen, a major city in southeastern China. They selected a representative region and mapped its road network, identifying 40 candidate sites for charging stations. Using real-world data on EV ownership and driving patterns, they simulated the performance of their APSOGA algorithm against several established methods, including traditional GA, standard PSO, Cuckoo Search (CS), and a hybrid CS-PSO algorithm. The results were compelling.
The APSOGA-based planning solution achieved a comprehensive cost reduction of 7.19% to 14.37% compared to the other methods. This reduction stems from a more efficient allocation of resources—building the right number of stations in the right places with the optimal number of chargers. Crucially, this cost efficiency did not come at the expense of user experience. On the contrary, the utilization rate of the charging stations in the APSOGA plan increased by 14.28% to 29.03%. Higher utilization indicates that the stations are being used more effectively, reducing both idle capacity (which wastes money) and congestion (which frustrates users).
A detailed analysis of the cost components revealed the nuanced advantages of the APSOGA approach. While some competing methods, like the pure GA, achieved lower waiting times by deploying a large number of chargers, this strategy led to a massive increase in construction and operational costs. The GA-based plan, for instance, required 125 chargers, compared to APSOGA’s 85. This over-provisioning resulted in significant underutilization, with many chargers sitting idle. Conversely, other methods that minimized construction costs ended up with stations that were too small, leading to long queues and high user waiting costs. APSOGA struck the ideal balance, minimizing the total cost by carefully calibrating station size and location to match predicted demand.
The final optimized plan for Xiamen recommended building six charging stations at specific candidate sites, equipped with a total of 85 chargers. The distribution of chargers across the stations was not uniform; it was carefully tailored to the local demand density. For example, stations located in high-traffic areas or near major residential zones were allocated more chargers, while those in less dense areas received fewer. This granular, demand-responsive approach is a hallmark of the APSOGA method and a key reason for its superior performance.
The research also explored the sensitivity of the planning outcomes to the relative importance placed on operator costs versus user costs. By adjusting the weights assigned to these two objectives in the model, the researchers demonstrated how the optimal plan changes under different policy priorities. When the economic viability of the charging network is the primary concern (a higher weight on operator costs), the model recommends fewer, larger stations with higher utilization rates, even if it means slightly longer travel distances or wait times for users. This scenario might be appropriate in the early stages of EV adoption, when building a sparse but efficient network is more critical than maximizing convenience. Conversely, when user experience is prioritized (a higher weight on user costs), the model suggests building more stations, potentially with lower individual utilization, to ensure that charging is always nearby and readily available. This approach is better suited for mature EV markets where high user satisfaction is key to sustaining growth.
This sensitivity analysis provides valuable guidance for urban planners and policymakers. It underscores that there is no one-size-fits-all solution to charging infrastructure. The optimal strategy depends on the specific context, including the current level of EV penetration, the city’s development goals, and the available budget. The APSOGA framework, with its flexible weighting system, offers a powerful tool for tailoring plans to these unique circumstances.
The implications of this research extend beyond the immediate improvement of charging network efficiency. By providing a more rational and balanced approach to infrastructure planning, it contributes to the overall sustainability of the transportation sector. Efficiently utilized charging stations reduce the need for redundant construction, conserving materials and energy. Shorter travel distances to chargers mean less energy is wasted on “charging trips,” further reducing the carbon footprint of EV ownership. Moreover, a reliable and convenient charging experience is essential for overcoming consumer hesitation and accelerating the transition to electric mobility.
The work of Huang Ziqing, Lin Bing, Lu Yu, Liu Dui, and Wang Mingfen represents a significant step forward in the field of intelligent transportation systems. Their APSOGA algorithm demonstrates the power of combining different computational intelligence techniques to solve complex real-world problems. It moves the conversation from a simplistic focus on quantity—how many chargers can we build?—to a more sophisticated consideration of quality—how can we build a charging network that is truly optimal for all stakeholders?
Looking ahead, the researchers acknowledge that their model can be further refined. Future work will incorporate dynamic factors such as real-time traffic flow, the impact of conventional vehicles on EV travel patterns, and the integration of existing gas stations into the charging network. They also plan to consider different types of chargers (fast, slow) and the constraints imposed by the power grid to ensure system stability. As the EV landscape continues to evolve, so too will the tools needed to manage it.
In conclusion, the transition to electric vehicles is not just about changing the power source of our cars; it requires a complete rethinking of our transportation infrastructure. The research published by the team from Fujian Normal University provides a sophisticated, evidence-based methodology for building a charging network that is both economically sound and user-centric. As cities around the world grapple with the challenges of decarbonization, studies like this offer a blueprint for a smarter, more sustainable future.
Huang Ziqing, Lin Bing, Lu Yu, Liu Dui, Wang Mingfen, College of Physics and Energy, Fujian Normal University; Journal of Fujian Normal University (Natural Science Edition), DOI: 10.12046/j.issn.1000-5277.2024.02.003