Parallel Enabled and Stability-Aware Self Optimizing Control With Globalized Constrained Nelder-Mead Optimization Algorithm

Parallel computation enhanced by Digital Twin can accelerate the execution of real time optimization algorithms for closed loop control systems. In this paper, a parallel implementation of the stability-aware Globalized Constrained Nelder-Mead Self Optimizing Control method is presented. It is based on the Analysis, Control, Parallel Execution approach, leveraging the use of Digital Twin instances for the simultaneous cost function evaluation during the most intensive function evaluation steps. Obtained results shows that the concurrent function evaluations using Digital Twins reduce the execution time and accelerate the convergence of the optimization algorithm.