Research and /or Engineering Questions/Objective:
The detection accuracy and quality of a brake noise data logger depends on a large variety of test setups (e.g. microphone mounting and quality), signal processing settings, and is affected by different test philosophies (cabin or wheelhouse sound measurements). This can lead to different results for identical vehicle/brake set-ups, depending on the specific data logger, detection method and philosophy used. Although it seems essential, no procedure to assess the detection accuracy and quality of a mobile data logger exists so far.
Methodology: Within the German brake noise expert group (EKB), a procedure has been created and verified that enables a detection quality evaluation for brake noise data loggers. In principal, a simulation of brake events by a defined artificial signal catalogue is applied to the logger system. It consists of real world noise data (.wav format) and the corresponding operational parameters. Together with the introduced concepts of sensitivity (number of detected noise events vs. number of noisy events) and specificity (number of events correctly rated as "quiet" vs. number of quiet events), the logger's capability to correctly classify different events can be assessed.
Results: With respect to reliability and repeatability, the procedure was evaluated on four different logger systems, including individual test philosophies, algorithms and detection configurations. Showing highly reproducible outcomes for sensitivity and specificity the procedure proved itself as applicable. A data logger comparison based on this approach revealed differences in frequency resolution, number of detected noise events and peak level detection for different systems. In terms of an EKB specification, manufactures and users of brake noise data logger are now able to determine the noise detection capability of their system configuration and test method. In addition to that, the specification offers a reproducible basis to tune individual set-ups, in order to achieve a higher rating for sensitivity and/or specificity.
Limitations of this study:
The evaluation is so far restricted to tonal noise events and does not assess the instrumented vehicle set-up, as no real sensors are included.
What does the paper offer that is new in the field in comparison to other works of the author:
This certification enables a detailed assessment of the detection quality of a data logger system. Furthermore it can be used to compare different data logger systems, test philosophies (wheel house and cabin microphone) and system configurations. However, the certification does not claim to define which system or philosophy is superior but to align them to achieve comparable outcomes.
The work presented here enables manufactures and users of brake noise data loggers to determine the noise detection capability of a system configuration. This procedure will guaranty users of logger systems to have a defined minimum of accuracy and precision in objective vehicle noise assessments, making such assessments more reliable and meaningful.
Dirk Drechsler - Ford-Werke GmbH Kai Bode - Audi AG Karl Häsler - Daimler AG Axel Stache - TRW Automotive GmbH Claus Thomas - Porsche AG