Nonstationary Data Analysis
Julius S. Bendat, Allan G. Piersol · Wiley series in probability and statistics · 2010
Theoretical ideas, error formulas, and processing techniques do not generally apply when the data are nonstationary. Special considerations are required in these cases. Such considerations are discussed in this chapter. The measurement of nonstationary probability density functions can be a formidable task. The function computed by time-averaging data with a nonstationary mean square value will tend to exaggerate the probability density of low- and high-amplitude values at the expense of intermediate values. Mean values of nonstationary random processes and nonstationary mean square values can be estimated by using a special-purpose instrument or a computer. Two main steps are involved in the measurement. The first step is to obtain and store each record and the next step is to perform an ensemble average. The chapter also discusses the correlation structure, spectral structure of nonstationary data, and input/output relations for nonstationary data. Controlled Vocabulary Terms Correlation; ensemble average; mean square; mean values; nonstationarity; probability density function