Solving Algebraic Riccati Equation real time for Integrated Vehicle Dynamics Control

In this paper we present a comparison study of different computational methods to implement State Dependent Riccati Equation (SDRE) based control in real time for a vehicle dynamics control application. Vehicles are mechatronic systems with nonlinear dynamics. One of the promising nonlinear control methods to control vehicle dynamics is based on SDRE. In this method, an Algebraic Riccati Equation (ARE) is solved at each sample to generate the control signal. However solving ARE is computationally complex. In this work, Extended Kalman Filter (EKF) iterative, Schur, Eigenvector, and Hamiltonian methods to solve ARE real time are implemented and studied for their timing, accuracy, and feasibility. Three methods, Schur, Eigenvector, and Hamiltonian are found to have an average calculation time of 3.9, 2.5, and 1.6 milliseconds on a dSPACE real time processor. This timing is acceptable as the controller sampling time is 10 milliseconds. In addition to the least processing time, the Hamiltonian based approach yields the lowest quadratic cost for SDRE based Integrated Vehicle Dynamics Control (IVDC).