Statistical Alignment
Mathematical operations for signal analysis compare two time-series datasets to determine their degree of similarity. Using cross correlation in non-destructive testing helps to identify the time delay between a transmitted ultrasonic pulse and its reflection. This technique measures the displacement of features within the signal.
It provides an effective way to locate flaws in moving webs.
Computational Execution
Digital systems multiply the received signal by a delayed version of the reference signal and sum the results. Applying cross correlation enables the detection of weak signals buried in the background noise of a paper machine. The peak value of the resulting function indicates the exact point where the two signals align.
Engineers use this data to calculate the speed of sound through the board. Accurate timing is essential for determining the thickness of the material. By comparing the input and output waveforms, the algorithm determines the phase shift caused by the paper structure.
This information helps to map the fibre orientation in the sheet.
Practical Boundary
Signal distortion from scattering at fibre boundaries reduces the accuracy of the peak detection. While cross correlation effectively handles random noise, periodic interference from mechanical vibration can create false matches. The sampling rate of the analogue to digital converter must be high enough to capture the necessary detail.
Success depends on the stability of the reference waveform.