VISUAL STATISTICAL-ANALYSIS OF 3-DIMENSIONAL STRUCTURAL DYNAMICS DATA

A scanning laser Doppler vibrometer provides high-spatial-density vibration measurements of a structure under test. These in-the-field laser-based measurements are the basis for deriving spatial. dynamics models. Statistical analysis of the experimental structural dynamics data is essential to determine the quality of the time signal data measurements. Statistically qualifying the time signal data at each point on the structure is particularly important to the process of deriving spatial dynamics models from the data. Analysis of numeric statistics of these large vibration data sets is tedious and time consuming. A new technique for visual statistical analysis using graphical computer modeling is introduced to aid the interpretation of the signal processing statistics measured at numerous spatial locations over a structure. Estimates of the error variance, maximum sample cross-correlation coefficients, and F test p-values for regression are computed for time signal data acquired at each measurement location on a structure. Colour and graphical glyphs added to the computer model quickly draw attention to areas on a structure where the time signal data measurements have poor statistical values. The visual statistical model can be a guide during data acquisition to indicate the quality of the acquired data as well as regions where additional measurements may be necessary.