Kernelized technique for outliers detection to monitoring water pipeline based on WSNs

Currently, the technology for sensing and control has become the potential for significant advances not only in science and business but equally important on a range of industrial applications. In addition to reducing costs and increasing efficiencies for monitoring systems, Wireless Sensor Networking (WSN) is expected to bring consumers a new generation of conveniences. However, there are issues when treating extremely interrelated, composite, and noisy databases with a large dimension. For that purpose, outliers detection techniques (ODT) are used for an effective monitoring system to ensure the safety of a transport process. Therefore, in this paper, a novel model of outliers detection and classification has been developed to complement the existing hardware redundancy and limit checking techniques. To overcome these problems, we present a Combined Kernelized Outliers Detection Technique (CKODT) based WSN for damages detection in water pipeline. Initially, training pressure measurement is collected from the sensors which are implemented outside the pipe. Next, these data are fed into the data reduction algorithm as known as Kernel Fisher Discriminant Analysis (KFDA) to create discriminant vectors. Then, these vectors were utilized as inputs for the One Class Support Vector Machine (OCSVM) method to classify the feature sets which were extracted using the proposed technique. The main objective of this work was to combine the advantages of these tools to enhance the performance of the monitoring water pipeline system. The accuracy of our Combined Kernelized Outliers Detection Technique for classification was analyzed and compared with variety of techniques. The experimental results showed the improvements of the proposed framework compared to other techniques in the context of damage detection in the monitoring water pipeline process. (C) 2019 Elsevier B.V. All rights reserved.