Evolutionary single objetive algorithms
Ormazabal, as a company specialised in the electrical sector, needs to have more knowledge of its electrical networks, so that it can have more applications in them. Its main advantages would be to achieve synchronised systems, automatic fault detection, inventory improvement, to be able to estimate the optimal connection of new users and theft detection by means of an automatic system. Taking into account that the cost of the equipment and its updating is high, the idea is to mathematically obtain a solution that can be executed periodically and that provides information on these networks. Because of this, Ormazabal seeks to find the topology (feeder and phase) for each consumer. In order to do this, there are different aspects to take into account. Firstly, the objective function must be representative, and it must have a logic behind it. On the other hand, the scalability of the problem must be taken into account, as it must be applicable to more scenarios. Finally, the correlation between the value of the objective function and the precision of the number of counters must be correctly assigned. In this way, the objective is to have a solution in which each Smart Meter has its line and phase optimally associated. For this, the metric to be used to evaluate the solution will be measured in the number of well associated lines and the amount of energy that is correctly classified.
Evolutionary algorithms have been used to solve this feeder-mapping problem. These offer ease of selection of the best possible individual given a specific problem. Among the types of optimisation problems, this one falls into the category of single-objective combinatorial optimisation, since the search space is numerable and the problem to be solved is a single-objective problem. The algorithm that has offered the best results has been Coevolution.
Cooperative coevolution is related with different agents collaborating to solve problems. In artificial intelligence, t he way this collaboration is articulated is by means of sharing information Cooperation promotes two complementary advantages
- It helps maintaining diversity in the evolutionary system.
- It enables the rapid dissemination of breakthrough solutions