Dielectric testing of spark plugs using neural networks

Production testing methods for spark plugs have changed very little over the years. This paper forms part of an on-going series of publications from the Author about new spark plug testing techniques. The paper specifically addresses noted troublesome faults within the fabric of the spark plug insulator: chips, punctures and cracks. Many of these faults are notoriously difficult to detect and reproduce. This paper describes a novel method of spark plug dielectric testing, offering potential for detection and elementary diagnosis of faults. High voltage pulse waveforms are applied to the test sample and the resulting waveforms are classified by a neural network. The experimental work has produced promising results, indicating that neural networks offer potential for the future of spark plug testing.