Linear Dynamic Controllers Evolved by Genetic Regulatory Network Based Artificial Cells

In this paper a linear representation of a synthetic genetic regulatory network (GRN) model is derived and it is used for evolving linear dynamic controllers for nonlinear systems. A case study is considered in which running the genetic algorithm on the elements of the system matrix of a linear controller is unable to evolve and reach the control ends, while running the genetic algorithm on the genes of an artificial cell with linear regulatory networks evolves and a linear controller is achieved. This justifies the computational burden imposed on computations due to GRN dynamics as GRN representation increases the evolvability of the controller.