Knock Detection and Estimation Based on Heat Release Strategies

Engine knock has been studied extensively over the years. Its undesired effects on drivability, its potential to damage an engine, and its impact on limiting the compression ratio are the main reasons why it remains a current topic of research. This paper focuses on exploiting the connection between auto-ignition and knock. A new method based on the frequency analysis of the heat release traces is proposed to detect and estimate auto-ignition/knock robustly. Filtering the heat release signal with the appropriate bandwidth is crucial to avoid misdetection. The filter settings used in this paper are found using spectral analysis of the heat release signal. By using the proposed method, it is possible to detect auto-ignition/knock even under the presence of undesired sensor resonance effects and noise from mechanical and electrical sources. The proposed algorithm is evaluated with different engines running at wide open throttle; and, its results are compared to a commonly used knock evaluation parameter. This new method of knock detection is shown to be robust and more sensitive for different types of engines and speeds.