Abstract In this paper, a closed-loop knock limit control strategy is proposed for spark-ignition (SI) engines to tune the spark advance (SA) timing to borderline where knock probability is less… Click to show full abstract
Abstract In this paper, a closed-loop knock limit control strategy is proposed for spark-ignition (SI) engines to tune the spark advance (SA) timing to borderline where knock probability is less than a prespecified value and achieve the tradeoff between knock and fuel efficiency. An analysis of experimental data is implemented firstly to provide evidence that logarithm of knock intensity can be regarded as a cyclically uncorrelated process which obeys normal distribution at every cycle. Thus, the knock control problem should be considered under a stochastic framework. The effect of SA on logarithm of knock intensity probability distribution is discussed and, based on this, the statistical learning-based knock controller is presented. The new controller achieves an improved regulatory response compared to a likelihood-based strategy in the experimental validation.
               
Click one of the above tabs to view related content.