EV Charging Schedule Control Using Two-Phase Objective Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing smart charging systems for electric vehicles face challenges in efficiently generating charging schedules that balance multiple conflicting objectives, such as minimizing energy costs, reducing peak load, and maximizing vehicle charging, while also adapting to changing preferences and system complexities.
Innovation Solution
The proposed method involves a two-phase approach: the first phase uses lexicographic optimization to generate charging schedules based on a hierarchy of objectives, and the second phase employs an objective function with weighted objectives, where the weights are determined offline using logged data from the first phase.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If lexicographic optimization is used to generate charging schedules with hierarchical objectives, then the ability to dynamically adapt to changing preferences is improved, but the computation time increases and becomes too slow for regular use
Solution Approach 1:
The patent pre-computes and stores Pareto optimal solutions during an offline phase before the system is put into operation. These pre-computed solutions are stored in a database and can be quickly retrieved during regular charging operations, eliminating the need for time-consuming real-time lexicographic optimization while maintaining the ability to handle different objective hierarchies.
Solution Approach 2:
The patent implements a two-phase approach where the system dynamically switches between offline pre-computation mode and online query mode. During offline phase, comprehensive Pareto solutions are pre-computed; during online phase, the system dynamically selects appropriate pre-computed solutions based on current charging demands and objective hierarchies, achieving both speed and adaptability.
2Productivity
If multiple conflicting objectives are combined using weighted sum approach, then the computation speed is improved, but the ability to represent hierarchical preferences and dynamically adapt to changing priorities is lost
Solution Approach 1:
The patent introduces a database of pre-computed Pareto optimal solutions as an intermediary between the weighted sum optimization method and the hierarchical objectives requirement. The database stores solutions generated with different weight combinations, allowing the system to quickly retrieve appropriate solutions without performing real-time weighted sum optimization, thus maintaining both speed and hierarchical preference representation.
Solution Approach 2:
The patent performs comprehensive weighted sum optimization and stores results in advance during an offline phase. This pre-computation covers various weight combinations and objective hierarchies, enabling fast retrieval during online operation without sacrificing the ability to represent different hierarchical preferences.
3Measurement precision
If lexicographic optimization is applied to each charging schedule in real-time, then the charging schedules accurately reflect current hierarchical objectives, but the system becomes too complex and slow for practical deployment with growing numbers of charging stations and vehicles
Solution Approach 1:
The patent segments the charging schedule generation process into two independent phases: offline pre-computation phase where comprehensive Pareto solutions are generated and stored, and online query phase where appropriate solutions are selected from pre-computed results. This segmentation reduces real-time computational complexity while maintaining accuracy.
Solution Approach 2:
The patent performs the computationally intensive lexicographic optimization in advance during an offline phase, storing results in a database. During regular operation, the system simply queries this database for appropriate pre-computed solutions, dramatically reducing real-time complexity while maintaining solution accuracy.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention regards a system and computer implemented method for charging electric vehicle. In a first phase, first charging schedules are generated by performing a lexicographic optimization based on a hierarchy of the charging, wherein charging objectives to be considered in generating the first charging schedule for one or more charging stations of the electric vehicle charging system are obtained, and a hierarchy of the charging objectives is determined. Charging of the electric vehicles is controlled according to the first charging schedule. In a second phase, second charging schedules are generated by performing an optimization based on an objective function with combined weighted objectives. The weights are learned in an offline optimization based on logged data from the first phase. In this second phase, charging of the electric vehicles is controlled according to the second charging schedule [Fig. 1]