Techniques for prediction of the statistical variation of multiple-input-multiple-output system response

Analytical statistical techniques are investigated for analysis of the variation of radiated noise from complex built-up structural-acoustical systems. The method of moments and Monte Carlo simulation techniques are identified as the most useful and feasible methods for application to noise control design, The methods are demonstrated for two available automobile samples, Frequency response function data were measured for 98 Isuzu Rodeo sport utility vehicles and for 57 Isuzu pickup trucks, The automobile studies illustrate application of the methods and interpretation of the results of measurements. In these studies, the mean, the standard deviation, the cumulative density function, and probabilities are computed for single-input-single-output systems, The predicted results are compared to measured results. In addition, the extension of the methods to multiple-input-multiple-output systems is discussed and the bootstrap technique for identifying the sensitivity of the prediction to the effects of finite sample size is illustrated. (C) 1997 Institute of Noise Control Engineering.