The article considers the problem of determining the optimum amount of approximating the regression analysis of variance and covariance of the results of wavelet transforms transient vibration signals in the diagnosis of the state of vibration of industrial facilities through the use of cumulative statistics. The article shows that informative parameters in the analysis of vibration nodes of industrial facility for monitoring and diagnosing elected criterial F- statistics, obtained in the processing of the wavelet. The images with slice of scale and offset. Select the number of approximating line to achieve the minimum probability of error diagnostics is accompanied by the definition of a minimum number of violations of homogeneity of the wavelet coefficients in the regressions including favorites. The article proved the effectiveness of approaches involving the use of cumulative statistics.
vibrodiagnostics, wavelet-analysis, variance, F-statistics, error diagnostics