PCA-based leakage detection method for water supply systems considering complex Fourier components

The water leakage leads to deficient water supplies, roads caving in, leakage in buildings, and secondary disasters. In this study, we propose the PCA-based automatic water leakage detection method considering complex Fourier components. The water leakage sounds and pseudo sounds, such as gas flow sounds, water usage sounds etc., are collected by microphone put on the ground. The Fourier spectra are obtained through the short-time Fourier transform (STFT). Then the principal component analysis (PCA) is applied to complex Fourier components of each collected sound data. The contribution ratio and the kurtosis of the eigenvector of the first principal component show the good ability to distinguish the water leakage sounds from the pseudo sounds. Therefore, the feature vectors are created from these PCA parameters. Based on them, the Support Vector Machine (SVM) is built. The results show that the classification can reach a very high accuracy. At last, applicability of the proposed water leakage detection method is well demonstrated.