Flexible networked rural electrification using levelized interpolative genetic algorithm

Abstract

Networked rural electrification is an alternative approach to accelerate rural electrification. Using satellite photos and GIS tools, an electrical distribution network is used to connect villages and properly located generation facilities together to reduce electrification cost. To design the network, optimal paths connecting all node-pairs are identified, followed by finding a network topology that minimizes cost. Earlier work has illustrated that A* (A-star, an optimal path-finding algorithm) is inefficient for this application due to the complex topography in rural areas. The multiplier-accelerated A* (MAA*) algorithm overcomes key performance issues, but, like A*, produces only one path connecting each node-pair. Relying on one path increases project risk because adverse conditions, such as inaccurate GIS estimation, unexpected soil conditions, land-rights disputes, political issues, etc. can occur during implementation. In this paper, a hybrid path-finding method combining genetic algorithm and A* / MAA* algorithm is proposed. The proposed method provides a family of near-optimal paths instead of a single optimal path for routing. A family of paths allows a project implementer to quickly adapt to unexpected situations as new information becomes available, and flexibly change network topology before or during implementation with minimal impact on project cost.