Model-based Knock Prediction and its Stochastic Feedforward Compensation

This article studies the correlation between incylinder mixture temperature at intake valve closing and the engine knock, along with cycle-to-cycle knock variability based on a knock predictive model developed earlier. Based on the correlated stochastic relationship, a feedforward knock limit control strategy is developed to reduce the knock cycle-to-cycle variability and maintain the knock mean-intensity within a desired up bound while keeping spark timing close to engine MBT (maximum brake torque) timing as close as possible. The proposed feedforward control strategy is evaluated based on the knock predictive model in terms of knock mean-intensity and standard deviation and demonstrated its capability of reducing the knock cycle-to-cycle variability under the knock intensity constraint.