A tool for validating and benchmarking signal processing techniques applied to machine diagnosis

• Published in: Mechanical systems and signal processing 2020-05, Vol.139, p.106618, Article 106618
• Main Authors: Buzzoni, Marco, D’Elia, Gianluca, Cocconcelli, Marco
• Format: Article
• Language: English
• Subjects: Algorithms; Computer simulation; Construction standards; Gear fault; Gearboxes; Non-stationarity; Researchers; Rolling element bearing fault; Signal generation; Signal processing; Signatures; Simulated signal; Time invariant systems
• ISSN: 0888-3270; 1096-1216
• DOI: 10.1016/j.ymssp.2020.106618

This paper proposes a tool for the generation of synthetic mechanical signatures of faulted gears and bearings in non-stationary conditions. Synthetic signals are commonly used to provide early validation of vibration-based diagnostics techniques. In the absence of a standard tool, each researcher builds his/her own model keeping into account the characteristic features expected in a faulted condition. This approach implies two main consequences. The first one is that the complexity of simulated signals strictly depends on the researcher skills that could lead to excessively simple models that compromises the validation process. The second one is that there is not a widely accepted algorithm, or tool, for the generation of synthetic signals. This lack hinders simpler and consistent comparisons among signal processing techniques. Hence, the main goal of this paper is to supply to the scientific community a standard tool for the generation of synthetic mechanical signatures. The tool is provided under Creative Commons license and it allows to simulate different possible situations: from the basic case of a single bearing fault to the case of a multi-stage gearbox having either a localized fault and a distributed bearing fault. Among several features, the proposed tool could take also into account variable speed profiles of the input shaft, cyclostationary contributions and effects of linear time-invariant systems.