Smart Charging Strategy Boosts EV Infrastructure Health
As electric vehicles (EVs) continue to gain traction worldwide, the backbone of their adoption—charging infrastructure—is facing increasing scrutiny. While much attention has been placed on vehicle performance and battery technology, the long-term reliability and health of charging stations themselves have often taken a backseat. A groundbreaking study published in the Chinese Journal of Automotive Engineering presents a novel approach to not only extend the lifespan of charging equipment but also enhance the overall efficiency of charging networks. Led by Lu Yang, Liu Niexuan, and Huang Hongye from State Grid Electric Power Research Institute, NARI Group Corporation, the research introduces a health-based scheduling strategy that dynamically allocates EVs to charging points based on real-time equipment conditions.
The shift toward electrified transportation is no longer a distant vision—it is unfolding rapidly across global markets. Governments are setting aggressive targets for internal combustion engine phase-outs, automakers are rolling out new EV models at an unprecedented pace, and consumers are increasingly embracing the technology. However, behind the sleek exteriors and zero-emission promises lies a critical infrastructure challenge: how to ensure that the charging network can sustain long-term operational demands without excessive maintenance costs or premature failures.
Traditional charging systems operate on a first-come, first-served basis, where vehicles are assigned to the nearest available charger. While this method appears efficient in the short term, it often leads to uneven wear and tear across charging units. Some chargers may be overused, subjected to repeated high-power cycles and prolonged operation, while others remain underutilized. Over time, this imbalance accelerates degradation, increases failure rates, and ultimately raises operational costs for charging station operators.
Recognizing this gap, the team from NARI Group developed a comprehensive framework that redefines how EVs are routed to charging stations—not just based on availability, but on the health status of each individual charging unit. Their strategy, titled “Charging Scheduling Strategy for Electric Vehicles Based on Charging Pile Health,” moves beyond reactive maintenance models and introduces a proactive, data-driven optimization system.
At the core of the innovation is a multi-dimensional health assessment model that evaluates charging piles across six key modules: power, control, communication, charging gun, critical components, and environmental interactions. Within these categories, the researchers identified 26 distinct performance and safety indicators. These include electrical parameters such as output voltage and current accuracy, voltage and current regulation precision, current balancing across parallel modules, power factor, and overall system efficiency. On the safety side, the model incorporates thermal conditions—such as control module temperature, communication module heat levels, and charging gun temperature—as well as battery pack thermal extremes and the physical aging of components, particularly the charging gun’s service life.
Each of these indicators contributes to an overall health score, which is normalized to a scale from 0 to 1, where 0 represents complete failure and 1 indicates optimal condition. The model uses entropy-weighted methods to objectively assign importance to each metric, avoiding subjective bias in the evaluation process. This ensures that factors with greater variability and impact on longevity—such as frequent component failures or thermal stress—are given higher weighting in the final health calculation.
One of the most significant contributions of this research is its integration of equipment health into the charging dispatch decision-making process. Instead of treating all chargers as functionally identical, the system treats them as assets with varying degrees of wear, much like a fleet manager would assess the condition of vehicles in a delivery network. By doing so, the algorithm can prioritize healthier units for longer or higher-power charging sessions, while directing shorter or lower-demand charges to units that may already be showing signs of degradation.
To operationalize this concept, the researchers converted charging load into time-based variables, allowing for a unified framework where both duration and equipment condition could be optimized simultaneously. This transformation simplifies the scheduling problem and enables charging operators to make decisions based on two primary inputs: how long a vehicle needs to charge and the current health status of available charging points.
The optimization model was designed with two primary objectives. First, it seeks to maximize the cumulative health of all charging units over time, ensuring that no single charger is overburdened. Second, it aims to minimize the variance in health levels across the station, promoting balanced usage and preventing the emergence of weak links in the infrastructure. These dual goals align with the economic interests of operators, who benefit from longer equipment lifespans and reduced downtime, while also supporting sustainability by lowering the need for frequent replacements and associated resource consumption.
To solve this multi-objective optimization problem, the team employed the NSGA-II algorithm—a well-established evolutionary computation method known for its effectiveness in handling complex, non-linear systems with competing objectives. Unlike traditional optimization techniques that may converge to suboptimal solutions, NSGA-II generates a set of Pareto-optimal solutions, allowing operators to choose the best trade-off between system-wide health and individual unit preservation.
The real-world validation of the strategy was conducted using data from 10 charging stations in a region of Jiangsu Province, China. The dataset included a full year of operational parameters and charging records, providing a robust basis for simulation. Three distinct scenarios were compared: a baseline with no optimization (first-come, first-served), the proposed health-based strategy optimized using a particle swarm algorithm, and the same strategy optimized using NSGA-II.
The results were striking. When the NSGA-II-optimized health-based scheduling was implemented, the annual health of the entire charging station improved by 18.54% compared to the unoptimized scenario. On an individual level, each charging unit saw an average health improvement of 1.85%. In contrast, the particle swarm-optimized version achieved a more modest 2.76% improvement in station health and 0.27% at the unit level, highlighting the superior performance of the NSGA-II approach in navigating the complex trade-offs inherent in the system.
Equally important was the impact on health variance across the station. In the unoptimized case, the monthly health variance of the charging station grew over time, peaking at 0.0020—indicating increasing disparity in equipment condition. This trend suggests that without intelligent scheduling, some chargers degrade faster than others, leading to an unstable and inefficient network. Under the NSGA-II-optimized strategy, however, the maximum monthly variance was reduced to just 0.0015, and the overall trend was one of stabilization. This uniformity in health levels means that maintenance can be planned more predictably, spare parts inventory can be optimized, and unexpected failures can be minimized.
Further analysis revealed that the optimized strategy led to a more balanced distribution of service cycles and energy delivery across the chargers. In the unoptimized scenario, usage patterns were relatively uniform, but this masked underlying inefficiencies—some chargers were handling more high-power sessions than others, accelerating wear. The health-based model adjusted assignments dynamically, ensuring that no single unit was consistently exposed to the most stressful charging profiles.
From an operational standpoint, the implications are profound. Charging station operators often face a difficult balancing act between maximizing throughput and minimizing maintenance costs. Overuse of equipment leads to higher failure rates and more frequent repairs, while underuse results in idle capacity and lost revenue. The health-based scheduling model offers a way to achieve both goals: by extending the operational life of charging units, it reduces long-term capital expenditure, while intelligent load distribution ensures that capacity is used efficiently.
Moreover, the strategy supports broader energy system goals. As EV adoption grows, the demand for electricity will rise, placing additional strain on the grid. Smart charging systems that consider equipment health can be integrated with grid-responsive features, such as load shifting and peak shaving, to provide even greater value. For instance, during periods of high grid stress, the system could prioritize charging on healthier units that are more capable of handling variable power inputs, while preserving weaker units for off-peak use.
The research also opens the door to predictive maintenance models. By continuously monitoring health indicators, operators can identify early signs of degradation—such as rising temperature in the control module or increased voltage fluctuation—and schedule preventive maintenance before a failure occurs. This shift from reactive to predictive maintenance not only improves reliability but also enhances customer satisfaction, as fewer charging sessions are interrupted by outages.
Another advantage of the model is its scalability. While the current study focused on a single charging station with 10 units, the underlying principles can be applied to larger networks, including highway fast-charging corridors and urban charging hubs. With the addition of real-time communication between vehicles and infrastructure, the system could even support dynamic rerouting of EVs to nearby stations based on health and availability, creating a truly intelligent charging ecosystem.
The authors acknowledge that real-world implementation will require integration with existing charging management platforms and standardization of health data formats. However, the foundational work presented in this study provides a clear roadmap for how such systems can be designed and optimized. The use of open, transparent algorithms like NSGA-II also ensures that the approach can be replicated and adapted by other researchers and operators.
Looking ahead, the team suggests that future work could explore the integration of grid-side objectives, such as demand response and renewable energy utilization, into the scheduling model. For example, during periods of high solar generation, the system could prioritize charging on units that are both healthy and located near distributed energy resources, maximizing the use of clean power while minimizing grid congestion.
In conclusion, the research by Lu Yang, Liu Niexuan, and Huang Hongye represents a significant step forward in the evolution of EV charging infrastructure. By shifting the focus from simple availability to holistic equipment health, the proposed strategy offers a more sustainable, efficient, and cost-effective approach to managing charging networks. As the world moves toward a fully electrified transportation future, innovations like this will be essential in ensuring that the infrastructure can keep pace with demand—without sacrificing reliability or longevity.
The study demonstrates that smart charging is not just about when and how fast a vehicle charges, but also about how the charging process impacts the very equipment that enables it. In doing so, it sets a new standard for what it means to operate a truly intelligent and resilient EV charging network.
Lu Yang, Liu Niexuan, Huang Hongye, State Grid Electric Power Research Institute, NARI Group Corporation, Chinese Journal of Automotive Engineering, DOI: 10.3969/j.issn.2095‒1469.2024.06.09