A Memetic Algorithm to Schedule Planned Grid Maintenance

The combination of local search operators with a genetic algorithm has provided very good results in certain scheduling problems, particularly in timetabling and other maintenance scheduling problems. In fact, in previous work, the local search operators alone have been seen to produce good quality results for the thermal generator maintenance scheduling problem. However, even better results have been obtained (for this problem) by adopting a genetic algorithm/local search hybrid approach. The resulting algorithm from such an approach has been termed a Memetic Algorithm. This paper investigates the use of such an algorithm for the scheduling of transmission line maintenance for a known benchmark problem that has been previously addressed in the literature using a combination of a genetic algorithm and greedy optimisers. The problem discussed here is concerned with the scheduling of essential maintenance for an electricity transmission network where every transmission line must be maintained once within a specified time period. The objective is both to avoid situations where sections of the network are disconnected, and to minimise the overloading of lines which are in service. We look particularly at the “four–node problem” presented and we present and discuss in some detail the proposed memetic algorithm and compare it against a variety of algorithms, including the local search operators on their own, and also a range of algorithms which apply the local search operators to randomly generated solutions, thus demonstrating the utility of the memetic approach. A comparison is made with the results from previous work, and it will be shown that, for this problem, the proposed memetic algorithm can produce good quality solutions in a very short space of time.