Experimental comparison of gcc-based time delay estimation methods for ultrasonic leak detection in compressed air system

Air leakage is one of the most significant energy waste factors in compressed air systems which account for about 10% of total industrial energy consumption. It is estimated that about 10%~40% of the compressed air is wasted through leakage in most plants. A new ultrasonic leak detection method based on time delay estimation (TDE) is proposed to locate the compressed air leak for preventing energy waste in pneumatic systems. The accuracy of detection is highly dependent on the performance of the TDE method. Performances of six typical TDE methods based on generalized cross correlation (GCC) are compared, and these methods are the basic cross correlation (BCC), the Roth impulse response, the phase transform (PHAT), the smoothed coherence transform (SCOT), the WEINER processor, and the Hannan-Thomson (HT) processor. The experimental results show that: Firstly, the accuracy and precision of time delay estimation increases with the observation interval for all these methods. Secondly, the success rates of Roth, PHAT, SCOT and HT are much higher than that of BCC and WEINER, among which the HT processor performs best with a highest success rates closely followed by the PHAT processor. Thirdly, the HT processor which is a maximum likelihood estimator gives the minimum standard deviation of the time delay estimate; however, the standard deviations of all these GCC methods are very small. The HT processor outperforms other GCC methods in terms of success rate and standard deviation. Consequently, it is preferable to apply the HT processor for this particular purpose.