| Title |
Leak Detection and Classification of Water Pipeline based on SVM using Leakage Noise Magnitude Spectrum |
| Authors |
최준규(Jungyu Choi) ; 임성빈(Sungbin Im) |
| DOI |
https://doi.org/10.5573/ieie.2023.60.2.6 |
| Keywords |
Water pipeline; Leak detection; FFT; Machine learning; SVM |
| Abstract |
The conventional water pipeline leak detection methods require the decision of experts at the leak sites. The water leak problems have resulted in variable costs and significant annual losses. This drives the demand for expert-based to sensor-based leak detection to reduce leak detection time and cost. Therefore, this paper proposes a model for classifying the class of data using leak detection data collected by leak detection sensors installed in the water pipeline. The proposed model classifies the data into five categories based on the SVM algorithm. The frequency spectrum density in the 10 Hz to 5,120 Hz band is a feature used in the SVM model. The data used for model training and evaluation is water and sewage data (water pipeline leak detection) provided by AI Hub. Training the fifty SVM model is conducted by randomly extracting the data. The performance of the final models is evaluated through the average of the 50 models, resulting in an accuracy of 0.8478, an F1 score of 0.8099, and an MCC of 0.7986. |