Wavelet-based and data-adaptive methods for time delay estimation in acoustic leak detection

Leakages in water distribution networks are a common issue encountered in the water industry. The acoustic cross-correlation method is generally employed for detecting and locating leaks in water pipes. One important factor that determines the effectiveness of this method is the accuracy of the time delay estimate which is usually obtained from the cross-correlation function (CCF) of two signals measured on the pipe. However, in some practical situations, accurate time delay estimate cannot be obtained using existing correlation-based time delay estimation (TDE) methods without first filtering the signals prior to calculating the CCF. Incorrect choice of filter cut-off frequencies limits the effectiveness of these methods. To deal with this issue, this paper proposes a scheme based on multi-resolution decomposition for accurately estimating time delays between leak signals. The proposed scheme first decomposes the signals at different scales using a shift-invariant wavelet transform or data-adaptive decomposition and then determines the time delay from the decomposed signals. Simulation and experimental results demonstrate the higher effectiveness of the proposed method compared to the commonly used basic and generalised cross-correlation TDE methods.