A Fast-Acting Stochastic Approach to Knock Control

Abstract Stochastic methods arguably offer a more theoretically appropriate approach to knock control than conventional deterministic strategies but are typically much slower in their transient response. Recently, however, a new stochastic controller was developed which responds very rapidly to knock rates above the specified target and which maintains tight regulatory control about the knock limit. The response to overly retarded conditions, however, is still relatively slow. In this paper the algorithm is refined in order to speed up the rate of recovery from retarded conditions while maintaining tight control and fast retard action. The performance of the new algorithm is compared to a conventional knock controller. The comparison demonstrates that the stochastic controller presented in this work is able to operate at a mean spark timing that is more advanced with much less cyclic variation about this mean. The transient response to excess knocking events can be as fast, or faster, than the conventional controller though this depends to some extent on the particular realization of the random knock process. The transient response to retarded conditions is still not quite as fast as the conventional controller, but is at least four times faster than that previously achieved. Overall, the results of this work suggest that the stochastic controller will deliver increased torque and engine efficiency under knock limited conditions compared to a conventional deterministic controller under the same conditions.