Multi-task engine knock detection and intensity estimation based on temporal convolutional networks
Authors:
- 21 July 2026
- journal article
- Published by IOP Publishing in Measurement Science and Technology
- Vol. 37 (29) , 296103
Abstract
To address the challenge of reliable engine knock detection under diverse operating conditions, this paper proposes a multi-task framework based on a multi-head temporal convolutional network (TCN) for simultaneous detection and intensity estimation. In-cylinder pressure signals are transformed into a multi-channel representation through Savitzky–Golay detrending and engine-speed-adaptive angle-domain bandpass filtering to suppress low-frequency combustion trends and enhance oscillatory knock-related components. The proposed framework employs a multi-head TCN to regress knock intensity, quantified by the root mean square of the bandpass-filtered residual within a spark-relative or adaptive sliding window, and to classify knock occurrence, enabling a comprehensive evaluation of combustion stability. An operating-condition-adaptive multi-feature labeling strategy with robust normalization and Otsu-guided thresholding is adopted to ensure consistent supervision across varying engine states. To alleviate the adverse effects of class imbalance between knock and non-knock cycles, weighted sampling and class-balanced loss functions are incorporated to improve training stability and generalization. Experimental evaluations across multiple engine speeds and loads demonstrate that the proposed method surpasses conventional threshold-based indices and recent deep learning baselines, achieving superior accuracy, robustness, and computational efficiency suitable for real-time combustion monitoring and control-oriented applications. To ensure the reliability of the reported results, all metrics were averaged over three independent runs to mitigate the impact of network initialization and sampling randomness.Keywords
Funding Information
- Jilin Provincial Scientific and Technological Development Program (20240301005ZD)
This publication has 32 references indexed in Scilit:
- Knock feedback control strategies: A reviewAnnual Reviews in Control, 2025
- Effect of exhaust gas recirculation on knock of HCNG fueled spark ignition engine and prediction of knock intensity by improved particle swarm optimization-back propagation neural networkApplied Thermal Engineering, 2025
- Experimental investigation of knock control criterion considering power output loss for a PFI SI methanol marine engineEnergy, 2024
- Knock probability determination employing convolutional neural network and IGTD algorithmEnergy, 2023
- Knock Onset Determination with 1D CNN Using Random Search Hyperparameter Optimization and Data Augmentation in SI EngineInternational Journal of Automotive Technology, 2023
- Prediction of knock intensity and validation in an optical SI engineCombustion and Flame, 2023
- Knock recognition of knock sensor signal based on wavelet transform and variational mode decomposition algorithmEnergy Conversion and Management, 2023
- Impact of ammonia addition on knock resistance and combustion performance in a gasoline engine with high compression ratioEnergy, 2022
- A Novel Multiple Feature-Based Engine Knock Detection System using Sparse Bayesian Extreme Learning MachineCognitive Computation, 2022
- A phenomenological model of knock intensity in spark-ignition enginesEnergy Conversion and Management, 2017