A hierarchical data fusion method for detection of the leak of gas pipelines based on wireless sensor network

To improve the accuracy and reliability of the leak monitoring of gas pipelines by using wireless sensor networks(WSN), this paper puts forward a hierarchical data fusion algorithm based on the combination of the wavelet support vector machine (SVM) method and the evidence theory. The algorithm is described below. In the signal level fusion, the noise elimination for primitive signals is conducted using the wavelet transform technology, and leak characteristic parameters are totally extracted as well. In the attribute fusion, a multi-classifier model based on SVM is constructed, and characteristic parameters as input vectors are sent to the multi-classifier for initial recognition. In the decision level fusion, the evidence combination is accomplished using the improved evidence combination methods at the sink node for final decision making. The experimental results show that the approach could improve the precision of the leak location detection and reduce the undetected rate as well as the false alarm rate.