Predicting time-delays under real-time scheduling for linear model predictive control

Digital implementations of model predictive controllers as tasks on an embedded computer introduce time-delays in feedback control loops. Time-delays may cause performance degradation or even instability, but are difficult to predict when multiple tasks are executing concurrently on the same processor under real-time scheduling. In this paper, we propose an analytical timing model for real-time scheduling. Based on this model, we are able to perform an online prediction of time-delays caused by real-time scheduling. Then, a model predictive controller is designed on the basis of a process model that contains the predicted time delays. Example designs are demonstrated to show the effectiveness of the proposed method.