METHOD OF RECORDING AND ANALYZING ACOUSTIC EMISSION SIGNALS IN A SYSTEM FOR DIAGNOSTIC MONITORING OF PRODUCTION FACILITIES Russian patent published in 2019 - IPC G01N29/14 

FIELD: acoustics.

SUBSTANCE: use: for recording and analyzing acoustic emission signals in a system of diagnostic monitoring (DMS) of production facilities. Summary of invention consists in the fact thatdetection and extraction of signals AE on the background of noise is carried out by mathematical processing of the recorded time series of noisy signals AE by means of their digital filtration in a sliding time window with a random noise component using three-stage digital filter – one-dimensional band-pass normalizing frequency filter, adaptive spectral-correlation filter with variable delay line, two-dimensional statistical frequency-time filter – in three stages. At the first stage, by means of band-pass filtering of time series of noisy signals AE, improvement of useful signal / noise ratio, normalization of random noise component and minimization of influence of pulsed broadband interferences of different nature is performed. At the second filtration step, the useful signal is detected by dividing the band-pass filtering result in the current time window into the noise and signal components using a single-channel adaptive spectral-correlation filter with an automatically selected value of the input delay and the controlled ratio of the useful signal / noise at the output. At the third stage, the useful signal is extracted by removing the residual noise component occurring in the error channel at the output of the adaptive filter from the detected signal component using a statistical filter, operation of which is based on automatic detection of a noise fragment containing only a stationary component stationary within the current time window, and calculation of the corresponding two-dimensional time-frequency spectrum. Further mathematical processing of the obtained digital data is carried out by removing zero noise readings outside the useful signal AE and, respectively, forming a length-compressed time series consisting of useful signal samples, and subsequent calculation of its envelope based on two-dimensional convolution with sliding time window. That is followed by calculating diagnostic features of presence of defects – sources of useful signal AE in the monitoring object, classification of detected sources and decision on current technical state of the monitoring object.

EFFECT: high reliability and efficiency of existing acoustic emission DMS engineering devices, buildings and structures.

5 cl, 1 tbl, 4 dwg