Enhanced Engine Misfire Diagnosis through Artificial Neural Networks

Strategic Insights -

Researchers from Minia University have developed an advanced engine misfire diagnosis system using artificial neural networks (ANNs). The innovative approach combines vibration and acoustic emission (AE) signals to improve detection accuracy. By integrating multiple sensors and employing Bayesian Optimization techniques, the ANN model achieved an impressive 94.11% accuracy in engine misfire detection.

Key Takeaways:

  • The researchers extracted 20 time- and frequency-domain statistical features from vibration and AE signals, which were then used to develop a complete matrix for classifying the data.
  • The selection algorithms chose the most essential features for classifying the data, resulting in a significant improvement in accuracy.
  • The use of multiple sensors, including AE and vibration, increased accuracy by providing a more comprehensive picture of combustion events.
  • The ANN model was trained using Bayesian Optimization techniques, which helped to optimize hyperparameters and improve the accuracy of the model.
  • The research concluded that multi-signal methods in engine diagnostics are effective, providing a pathway for predictive maintenance of internal combustion vehicles.
  • The study highlighted the importance of integrating multiple sensors and employing advanced machine learning techniques to improve engine misfire detection.
  • Mohamed H. Abdelati, Automotive and Tractors Engineering Department, Faculty of Engineering, Minia University, was the lead researcher on the project.
  • Additional authors included M. Mourad, Al-Hussein Matar, and M. Rabie.

Statistics:

  • 20 time- and frequency-domain statistical features were extracted from vibration and AE signals.
  • The ANN model achieved an accuracy of 94.11% in engine misfire detection.
  • The use of multiple sensors increased accuracy by 30% compared to using a single signal.
  • The study utilized Bayesian Optimization techniques to optimize hyperparameters and improve the accuracy of the ANN model.
  • The research was published in the Journal of Engineering and Applied Science in 2025, with proceedings available online.

Sources:

  • Enhanced engine misfire diagnosis through integration of vibration and acoustic emission signals using artificial neural networks. Journal of Engineering and Applied Science, 2025,72(1):1-17.
  • https://doi-org.sdpl.idm.oclc.org/10.1186/s44147-025-00703-y