Increasing knock detection sensitivity by combining knock sensor signal with a control oriented combustion model. (1st April 2022)

• Record Type: Journal Article
• Authors: Pla, Benjamín; Bares, Pau; Jimenez, Irina; Guardiola, Carlos
• Is Part Of: Mechanical systems and signal processing. Volume 168(2022)
• Publication Date: 2022-04-01
• Subjects: Knock -- Knock sensor -- Resonance -- Combustion model; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2021.108665
• Languages: English
• ISSNs: 0888-3270

Abstract

Abstract: Knock phenomenon reduces the thermal efficiency and restricts performance improvement in spark-ignited engines. Reliable and rapid knock recognition is crucial for the engine knock control. Among the wide set of knock detection techniques, those based on in-cylinder pressure sensors provide the most precise recognition. However, pressure sensors are still affected by challenges such as durability and cost. For on-board applications, knock is usually detected by vibration signal, but the accuracy is limited due to natural vibration and external noises. In this paper, a novel knock recognition method based on knock sensor signal is proposed. The method consists of the comparison of a resonance index obtained through the knock sensor signal and a combustion model capable of estimating the fraction of mass burned, and thus being able to estimate if the amplitude in the knock sensor signal is produced by the auto-ignition of certain amount of fuel. The proposed method was compared with a fixed threshold for knock sensor resonance intensity, the improvements were quantified by using as reference a high sensitive knock recognition method based on cylinder pressure. Results show that the proposed method is able to improve the accuracy in over a 10 % of knock detection than using one set threshold over the entire cycle. Highlights: A novel knock recognition method based on knock sensor signal is proposed. A combustion model is used to develop a threshold as a function of combustion evolution. Knock recognition is based on the comparison of a resonance index and the threshold. The accuracy is improve in a 10% by comparing with a fixed threshold classification.