Objective and Subjective Classification of Creep Groan Noise

EB2020-STP-067
Full Paper + Presentation

Poster Pitch Video

Presentation Video

Abstract

Stick-Slip effects occur on the contact surface between the brake pads and the brake disc. They excite vibrations on the brake and axle system, which are transferred to large surfaces and generate audible noise. These stick-slip related Creep Groan phenomena induce vibrations on the entire brake system and influence structural vibrations of the vehicle. Therefore, feedback effects are created. A feedback effect at a mechanical system with more than three degrees of freedom produces instability and chaotic behaviour. Solving the problem of the Creep Groan occurrence on a complex brake system, by using analytical and/or numerical approach, is rather difficult. However, if sufficient data is available, the phenomenological approach can provide useable results. This is especially the case for a Creep Groan Noise, where additional mechanisms of noise generation from vibrating surfaces must be included in the analysis.

Creep Groan Noise is unpleasant and it can be considered as a defect, especially for drivers of luxury vehicles. Intensity and annoyance of Creep Groan noise depends on many factors. Therefore a subjective analysis of Creep Groan Noise is frequently included in vehicle testing. Unfortunately subjective tests reveal inherent uncertainties of subjective evaluation. A model for evaluation of Creep Groan noise annoyance is necessary to reduce the uncertainty and to improve the evaluation of Creep Groan. Experimental tests with over 1000 measurements were used for this purpose. Many different features from measured data were extracted and analysed. Selected features were used to classify Creep Groan into annoyance classes by Self-Classification Maps and a K-NN algorithm, reaching scores from 1 (very bad) to 10 (excellent). Results of the Self-Classification algorithm are correlated with objective evaluations. It was found that results provided by the Self-Classification algorithm are more dynamic than results provided by the experts’ subjective rating. Self-Classification algorithm has always attributed high intensity creep groan with scores below 3 while drivers seldom gave a score below 5. On the other hand, Self-Classification Algorithm has always attributed runs with no Creep Groan with score 10, while drivers rarely gave a score above 9.

EuroBrake Sponsors