The present work discusses the application of multivariate statistical methods for the analysis of NVH data. Unlike conventional statistical methods which generally consider single-value, or univariate data, multivariate methods enable the user to examine multiple response variables and their interactions simultaneously. This characteristic is particularly useful in the examination of NVH data, where multiple measurements are typically used to assess NVH performance.
In this work, Principal Components Analysis (PCA) was used to examine the NVH data from a benchmarking study of hydraulic steering pumps. A total of twelve NVH measurements for each of 99 pump samples were taken. These measurements included steering pump orders and overall levels for vibration and sound pressure level at two microphone locations.
Application of the PCA method made it possible to examine the entire set of data at once. The results clearly showed two pump types as having significantly better NVH performance characteristics than the other pump types. This analysis also indicated which characteristics in particular were best for NVH.
As a result of using PCA, it was possible to obtain a more comprehensive picture of the relative NVH performance of the pumps in the study than more conventional “one-at-a-time” comparison methods. Lastly, because the complete set of NVH data was included in the analysis, use of this method increased confidence that the results accurately represented the relative performance of the pumps, and that some important attribute was not inadvertently missed through consideration of only a limited portion of the data.