EV Route Planning With On-Route Energy Replenishment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing approaches to electric vehicle route planning for multi-service delivery and on-route energy replenishment struggle with scalability and computational efficiency, especially in large problem instances involving hundreds of users and multiple operational constraints.
Innovation Solution
A method and system utilizing an agent-based learning approach for electric vehicle route planning, which incorporates battery swapping and vehicle-to-grid energy supply to optimize routes for multi-service delivery and on-route energy replenishment, while minimizing energy usage and capital costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If existing approaches are used for electric vehicle route planning, then basic routing functionality is provided, but scalability and computational efficiency deteriorate in large problem instances involving hundreds of users and multiple operational constraints
Solution Approach 1:
The patent segments the complex routing problem into multiple independent components: route planning module, energy management module, and constraint satisfaction module. This segmentation allows each module to be optimized independently and processed in parallel, improving computational efficiency for large-scale problems with hundreds of users and multiple constraints.
2Ease of manufacture
If electric vehicles are used for last-mile delivery, then operational cost and emission goals are improved, but challenges arise in route optimization including minimizing energy usage and exploiting charging opportunities
Solution Approach 1:
The system performs preliminary energy assessment and charging station identification during the route planning phase. By pre-calculating energy requirements and identifying optimal charging opportunities before execution, the system minimizes actual energy consumption during delivery operations while maintaining cost-effectiveness.
Solution Approach 2:
The patent implements a feedback mechanism where real-time energy consumption data from electric vehicles is continuously monitored and fed back to the route optimization system. This feedback enables dynamic adjustment of routes and charging schedules to minimize energy usage while meeting delivery constraints.
3Adaptability or versatility
If multi-service deliveries are implemented, then service capability is enhanced, but capital costs and availability of charging infrastructure worsen
Solution Approach 1:
The patent designs the electric vehicle fleet and infrastructure to perform multiple functions: delivery operations, energy storage, and vehicle-to-grid power supply. This multi-functionality allows the same assets to generate revenue through multiple channels, offsetting the high capital costs of acquiring electric vehicles and charging infrastructure.
Solution Approach 2:
The system dynamically adjusts operational parameters such as vehicle routing, charging schedules, and discharge timing based on market conditions and grid demands. By optimizing these parameters, the system maximizes revenue from multi-service operations while minimizing the effective capital cost through improved asset utilization.
4Speed
If reinforcement learning is used for EV routing, then routing speed is improved (24 times faster than MILP and GA), but solution accuracy may be compromised (within 20% of optimal)
Solution Approach 1:
The patent segments the routing problem into critical and non-critical components, applying reinforcement learning to non-critical aspects where approximate solutions are acceptable, while using exact methods for critical constraints requiring high precision. This hybrid approach maintains computational speed while improving overall solution accuracy.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
This disclosure relates generally to method and system of electric vehicle route planning for multi-service delivery and on-route energy replenishment. Last mile delivery is a critical component of supply chains that impacts both customer experience and delivery cost. The method disclosed processes a received user request comprising a current location of the user, one or more required services, and a time window rendered between each of the required services, and generates a graph for the user request. Further, a trained learning agent generates a route map indicating a plurality of waypoint locations for the electric vehicle to visit each node in accordance with minimized trip cost of fleet and time duration where each node has a state action pair for the electric vehicle. The learning agent learns continuously during the interaction with the delivery environment and obtains feedback for every associated action.