Enhancing electric vehicle charging infrastructure: A framework for efficient charging point management

• Created a real-time dataset containing 1100 charging points information with parameters such as IMEI number, voltage, current, power, energy, frequency, time stamp, status of the charging point and firmware version. • Analysis of various parameters of the real-time dataset including charging station information and charging point status; the EV charging points are categorized into clusters. • The road traffic pattern is analysed by leveraging GPS modules installed in the EVs. The approximate time for EVs to reach the charging stations is estimated. • By exploiting the existing scheduler data, the availability of the charging point to accommodate dynamic demand charging requests and provide uninterrupted service is forecasted. • Combining these insights, charging point operators can anticipate the EV charging patterns and customer behavior, enabling them to optimize resource allocation and prevent underutilization of charging points. The rise of electric vehicles (EVs) in the transportation sector aids in curbing global greenhouse gas emissions yet efficiently integrating them into the existing infrastructure presents challenges in guaranteeing the real-time availability of charging points and the dynamic nature of electric mobility. This paper presents a novel dynamic demand scheduling framework that utilizes predictive analytics to address the issue of emergency charging requests; situations where an EV urgently require to reach a charging point due to critically low battery levels. The framework is integrated with advanced dynamic demand scheduling algorithm (ADDSA), which utilizes real-time charging data collected from Trivandrum, Kerala state, India. Using the comprehensive dataset, the framework identifies delayed EVs and considers the charging point status (active, idle or faulty) and charging point pricing to optimize the charging station allocation. By employing the K-Means clustering algorithm, the ADDSA categorizes charging points based on their performance and availability. To evaluate the effectiveness of these clusters, we utilize internal metrics such as the Silhouette score, Calinski-Harabasz (CH) index, and Davies-Bouldin (DB) index. Our findings demonstrate that K-Means outperforms other clustering algorithms, including DBSCAN, K-Medoids, Agglomerative clustering, and Gaussian mixture models (GMM), with a CH score of 1200, a Silhouette score of 0.45, and a DB score of 0.74. In the final stage of ADDSA, groups of available charging points along with their pricing information is generated, facilitating informed decision-making for EV users. With the rapid growth of the EV population, our unique dynamic demand scheduling framework, featuring real-time constraints, offers a promising solution for efficiently addressing the emergency charging needs of EVs.