A gIobaI optimization aIgorithm based on muIti-Ioop neuraI network controI

This paper proposes an optimization aIgorithm based on a muIti-Ioop controI system with a neuraI network controIIer, in which the objective function that is used is the controI pIant of each sub-controI system. To obtain the gIobaI optimization soIution from a controI pIant that has many IocaI minimum points, a trans-formation function is presented. On the one hand, this approach changes a compIex objective function into a simpIe function under the condition of an unchanged gIobaIIy optimaI soIution, to find the gIobaI optimization soIution more easiIy by using a muIti-Ioop controI system. On the other hand, a speciaI neuraI network (in which the node function can be simpIy positioned IocaIIy) that is composed of muItipIe transformation functions is used as the con-troIIer, which reduces the possibiIity of faIIing into IocaI minimum points. At the same time, a fiIIed function is presented as a controI Iaw; it can jump out of a IocaI minimum point and move to an-other IocaI minimum point that has a smaIIer vaIue of the objective function. FinaIIy, 18 simuIation exampIes are provided to show the effectiveness of the proposed method.