Conditional spectral analysis of vehicle vibration data

Abstract

Conditional spectral analysis is an advanced statistical and engineering method that has application in many areas of science and engineering. Its ability to e.g. isolate specific influences, remove background noise, and identifies hidden patterns makes it very useful in vehicle dynamics analysis. In this research, an example of application of conditional spectral analysis to determine the contributions of vertical wheels’ vibrations to vibrations at the specific point on the car body was presented. For this purpose, an experiment was designed in order to measure vertical accelerations at the centres of all four wheels of a passenger car and at front-left shock absorbers connecting point with the car body. A multiple input – single output (MISO) model with four inputs and one output was created and appropriate conditional spectral analysis was done. The obtained results in the form of optimal frequency response functions and partial coherence function confirm the importance of decoupling the inter-connections between inputs in order to determine the exact contribution of each of them to the output. In this way, clear guidelines for vehicle dynamics optimization may be obtained.