Fast urbanization and the widespread use of location-aware electronics have generated large amounts of trajectory data from city transportation networks. Because of changing network topologies, complex spatial relationships, and movement patterns that happen at certain times, it is still hard to efficiently mine trajectory patterns and accurately guess where the next location will be. The present models for predicting mobility often miss the graph-based geographical linkages and temporal periodicity that are part of real-world transportation systems. This paper proposes the Graph-Enhanced Periodic Mobility Learner (GE-PML) as a remedy to these limitations; it can extract trajectory patterns and intelligently predict the next destination. The proposed (GE-PML) combines a graph representation of the city's transportation network with a framework for learning about mobility regularly. The first stage is to draw the paths on a transport network. The nodes on the graph represent places, and the edges show how the places are connected depending on mobility and topology. A graph neural network is used to understand how places are related to each other in space and how they interact with each other. At the same time, a periodic temporal encoder shows trip patterns that happen on a regular basis, like every day or week. To make valid trajectory embeddings for predicting the next location, these temporal and spatial representations are combined in a way that changes based on attention. When evaluated on real-world datasets of urban mobility, experiments reveal that GE-PML always exceeds the best baselines in terms of F1-score, recall, precision, and accuracy. The suggested model efficiently includes both structural and temporal mobility elements, resulting in an increase in F1-score of roughly 10-15%, a decrease in prediction error of about 14%, and an improvement in prediction accuracy of around 12-18%. Finally, GE-PML is great for ITS, TM, and LS applications because it improves trajectory pattern mining and next-location prediction by using both graph structure and periodic mobility behavior together. This results in a 16% total performance boost. This method makes it much easier to capture difficult spatial transitions and long-term periodic behaviors, especially in transport networks that are dense and varied. For urban transportation systems, GE-PML is the best way to find out where to go next by looking at patterns in trajectories. Its ability to mimic graph-enhanced spatial linkages and periodic temporal dynamics at the same time can be very useful for intelligent transportation applications, including route planning, traffic management, and personalized mobility services.