Optimal routing algorithm for trips involving thousands of ev-charging stations using Kinetica-Graph

TL;DR

This paper presents a graph-based routing algorithm optimized for electric vehicle trips involving thousands of charging stations, utilizing combinatorial optimization, a fixed graph topology, and enhanced shortest path computations within Kinetica-Graph.

Contribution

It introduces a novel combinatorial optimization algorithm with a fixed topology and re-purposed Dijkstra solver for efficient EV routing among thousands of charging stations.

Findings

Successfully solved example trips demonstrating the algorithm's effectiveness.

Reduced computational costs through optimized shortest path calculations.

Implemented as a scalable, multi-threaded API within Kinetica-Graph.

Abstract

This paper discusses a graph based route solving algorithm to find the optimal path for an electric vehicle picking the best charging locations among thousands to minimize the total cumulative driving distance between the end points of the trip. To this end, we have devised a combinatorial optimization algorithm and a fixed storage graph topology construction for the graph road network of the continental USA. We have also re-purposed our existing Dijkstra solver to reduce the computational cost of many shortest path solves involved in the algorithm. An adaptive and light weight spatial search structure is also devised for finding a set of prospective stations at each charging location using uniform bins and double link associations. The entire algorithm is implemented as yet another multi-threaded at-scale graph solver within the suite of Kinetica-Graph analytics, exposed as a restful API endpoint and operable within SQL. Several example trips are solved and the results are demonstrated within the context.