Abstract
Optimizing Electric Vehicle (EV) charging infrastructure is a critical urban planning challenge characterized by high-dimensional uncertainty and periodic traffic flows. Traditional decision models often suffer from "temporal blindness," failing to recognize the cyclical nature of urban movement (e.g., the proximity of late-night and early-morning hours). This paper introduces the Complex T-Spherical Fuzzy C-Means (CTSF-CM) expert system, a novel unsupervised machine learning framework designed for spatio-temporal demand profiling. The system utilizes a complex phase term to mathematically represent temporal periodicity and a T-spherical fuzzy structure to filter environmental noise from meteorological sensors. Validated on 2,747 hourly instances from the I-94 Interstate Highway, the CTSF-CM system achieves a Peak Separation Accuracy of 87.21%, significantly outperforming standard K-Means and Intuitionistic Fuzzy models. Beyond numerical precision, the system provides geometrically interpretable decision boundaries that allow urban planners to distinguish between morning and evening service zones, ensuring grid stability and infrastructure efficiency in smart city environments.