Combining exhaustive search and multi-objective evolutionary algorithm for service restoration in large-scale distribution systems

摘要

Service restoration problem in distribution systems emerges after the faulted areas have been identified and isolated. A solution is obtained by determining the minimal number of switching operations that results in a configuration with a minimal number of healthy out-of-service areas without violating the operational and radiality constraints. Recently a practical and efficient methodology was developed and demonstrated through tests performed on the real and large-scale distribution system of Londrina city (Brazil). This methodology, named MEAN-MH, combines a Multi-objective Evolutionary Algorithm with Node-Depth Encoding, Multiple criteria tables and an alarming Heuristic. As any methodology based on meta-heuristics, the MEAN-MH does not guarantee to find the optimal solution of the service restoration problem, even when the optimal solution requires operations only in normally open switches incident to the healthy out-of-service areas (named as Tier 1 normally open switches). This paper proposes an extension of MEAN-MH that incorporates an Exhaustive Search (ES) procedure as a previous stage before MEAN-MH. The proposed ES guarantees the generation and analysis of all possible radial configurations that restore the service to all healthy out-of-service areas requiring operations only in Tier 1 normally open switches. Therefore, when the optimal solution of the service restoration problem requires operations only in Tier I NO switches, the proposed methodology, named MEAN-MH+ES, guarantees the optimum. However, when the optimal solution requires operations also in other switches, the MEAN-MH+ES searches by a feasible solution minimizing both the number of switching operations and the number of healthy out-of-service areas. To demonstrate the effectiveness of the proposal, both MEAN-MH and MEAN-MH+ES are applied to two real and large-scale distribution systems of Brazil. Moreover, the results obtained by MEAN-MH+ES are compared with those found in another published work. (C) 2015 Elsevier B.V. All rights reserved.