Abstract:
The MATLAB-implemented K-means clustering algorithm is applied to a real-time public Electric Vehicle (EV) charging dataset in this study. The objective is to identify distinct patterns by locating centroids and creating four unique clusters. The analysis focuses on the distribution of data points and the identification of peak hours for each pattern, which are intricately linked to these centroids. The primary targets are to segregate the dataset into four distinct usage patterns and to ascertain the number of data points within each cluster. The insights gleaned are crucial for optimizing the placement and scheduling of charging stations. Such strategic planning facilitates the convenience for electric vehicle operators and contributes to the sustainable expansion of the EV charging infrastructure
Aim:
Identify the electric vehicle (EV) charging patterns like “Convenient,” “Gradual,” “Anxious,” and “Urgent” based on real-time charging behavior using a k-means clustering algorithm.
Objectives:
- Gather details regarding EV charging session from various charging stations, then use the principle of Pareto to filter the data and points out the most crucial factors.
- Depending on users’ charging behaviors, determine the ideal number of clusters with the goal of learning more about all the factors impacting users’ charging choices.
- Implement k-means clustering in conjunction using Minkowski embedded distance to figure out the proportion of consistent users in every cluster.
- Examine the density of peak charging hours for every individual cluster.
Fig: EV daily charging dataset
Fig: Elbow curve