Statistical algorithm for the adaptation of detection thresholds

Abstract

Many event detection mechanisms in spark ignition automotive engines are based on the comparison of the engine signals to the detection threshold values. Different signal qualities for new and aged engines necessitate the development of an adaptation algorithm for the detection thresholds. An adaptation algorithm for the detection threshold is developed with the confidence interval method used as the basis for adaptation. The detection threshold is presented using a confidence interval with a given significance level. The adaptation is performed for an aged engine such that the significance level remains constant regardless of engine age and changing detection threshold values. This, in turn, guarantees the same event detection performance for new and aged engines/sensors. Adaptation of the engine knock detection threshold is given as an example.