VISUAL STATISTICAL-ANALYSIS OF 3-DIMENSIONAL STRUCTURAL DYNAMICS DATA

A scanning laser Doppler vibrometer provides high-spatial-density vibr ation measurements of a structure under test. These in-the-field laser -based measurements are the basis for deriving spatial. dynamics model s. Statistical analysis of the experimental structural dynamics data i s essential to determine the quality of the time signal data measureme nts. Statistically qualifying the time signal data at each point on th e structure is particularly important to the process of deriving spati al dynamics models from the data. Analysis of numeric statistics of th ese large vibration data sets is tedious and time consuming. A new tec hnique for visual statistical analysis using graphical computer modeli ng is introduced to aid the interpretation of the signal processing st atistics measured at numerous spatial locations over a structure. Esti mates of the error variance, maximum sample cross-correlation coeffici ents, 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 poo r statistical values. The visual statistical model can be a guide duri ng data acquisition to indicate the quality of the acquired data as we ll as regions where additional measurements may be necessary.