Estimation of future power consumption level in smart grid: Application of fuzzy logic and genetic algorithm on big data platform

Summary In an environment where the importance of new and renewable energies is growing, the balance between energy production and consumption cannot be achieved easily. Although it is possible for the one with surplus power to supply it to the one experiencing a shortage, power demand can hardly be expected to exceed supply or both to stay at equal levels at all times. In such case, a network limited to an individual or a small group can be regarded as a small single node so that the larger network consisting of these nodes can be represented with a graph based on the topology of dispersed nodes. This situation is similar to the prisoner's dilemma wherein the most ideal situation for the nodes is to collaborate with each other; in a situation wherein betrayal takes place, however, the Nash equilibrium can hardly be expected. Such situation between the nodes is almost the same condition repeatedly laid down to the prisoners who consistently and competitively pursue maximum profit and can be considered a game. Thus, this study attempted to devise a method of gaining maximum profit and predicting future power demands by using a genetic algorithm based on the game theory‐based fuzzy logic that seeks maximum profit by making the best choice. A scheme that can avoid a possible “prisoner's dilemma” situation in a new and renewable energy transaction environment was devised based on the game theory. For such scheme, the fuzzy theory was adopted to reflect the power demand, supply, and values; by developing a greedy algorithm, the optimal values were reflected under each given environment to set a foundation on which the situation wherein the cooperative nodes could be placed at a greater disadvantage than the uncooperative nodes can be avoided with a tit‐for‐tat algorithm wherein the genetic algorithm was reflected as well.