Tracking Action Potentials of Nonlinear Excitable Cells Using Model Predictive Control

摘要

We present explicit and online Model Predictive Controllers (MPCs) for an excitable cell simulator based on the non-linear FitzHugh-Nagumo model. Despite the plant's nonlinearity, we are able to formulate the model predictive control problem as an instance of quadratic programming, using a PieceWise Affine (PWA) abstraction of the plant. The speed-versus-accuracy tradeoff for the explicit and online versions is analyzed on various reference trajectories. Our MPC-based approach, enabled by the PWA abstraction, presents a framework for designing automated in silico biomedical control strategies for excitable cells, such as cardiac myocytes and neurons.