This page has only limited features, please log in for full access.
Small leaks in water distribution networks have been a major problem both economically and environmentally, as they go undetected for years. We model the signature of small leaks as a unique Directed Acyclic Graph, called the Lean Graph, to find the best places for k sensors for detecting and locating small leaks. We use the sensors to develop dictionaries that map each leak signature to its location. We quantify leaks by matching out-of-normal flows detected by sensors against records in the selected dictionaries. The most similar records of the dictionaries are used to quantify the leaks. Finally, we investigate how much our approach can tolerate corrupted data due to sensor failures by introducing a subspace voting based quantification method. We tested our method on water distribution networks of literature and simulate small leaks ranging from [0.1, 1.0] liter per second. Our experimental results prove that our sensor placement strategy can effectively place k sensors to quantify single and multiple small leaks and can tolerate corrupted data up to some range while maintaining the performance of leak quantification. These outcomes indicate that our approach could be applied in real water distribution networks to minimize the loss caused by small leaks.
Ary Mazharuddin Shiddiqi; Rachel Cardell-Oliver; Amitava Datta. An Advanced Sensor Placement Strategy for Small Leaks Quantification Using Lean Graphs. Water 2020, 12, 3439 .
AMA StyleAry Mazharuddin Shiddiqi, Rachel Cardell-Oliver, Amitava Datta. An Advanced Sensor Placement Strategy for Small Leaks Quantification Using Lean Graphs. Water. 2020; 12 (12):3439.
Chicago/Turabian StyleAry Mazharuddin Shiddiqi; Rachel Cardell-Oliver; Amitava Datta. 2020. "An Advanced Sensor Placement Strategy for Small Leaks Quantification Using Lean Graphs." Water 12, no. 12: 3439.
Leaks in water pipeline networks have cost billions of dollars each year. Robust leak quantification (to detect and to localize) methods are needed to minimize the lost. We quantify leaks by classifying their locations using machine learning algorithms, namely Support Vector Machine and C4.5. The algorithms are chosen due to their high performance in classification. We simulate leaks at different positions at different sizes and use the data to train the algorithms. We tune the algorithm by optimizing the algorithms' parameters in the training process. Then, we tested the algorithms' models against real observation data. We also experimented with noisy data, due to sensor inaccuracies, that often happen in real situations. Lastly, we compared the two algorithms to investigate how accurate and robust they localize leaks with noisy data. We found that C4.5 is more robust against noisy data than SVM.
Ary Mazharuddin Shiddiqi. INCREASING THE ROBUSTNESS OF CLASSIFICATION ALGORITHMS TO QUANTIFY LEAKS THROUGH OPTIMIZATION. JUTI: Jurnal Ilmiah Teknologi Informasi 2020, 18, 1-8 .
AMA StyleAry Mazharuddin Shiddiqi. INCREASING THE ROBUSTNESS OF CLASSIFICATION ALGORITHMS TO QUANTIFY LEAKS THROUGH OPTIMIZATION. JUTI: Jurnal Ilmiah Teknologi Informasi. 2020; 18 (1):1-8.
Chicago/Turabian StyleAry Mazharuddin Shiddiqi. 2020. "INCREASING THE ROBUSTNESS OF CLASSIFICATION ALGORITHMS TO QUANTIFY LEAKS THROUGH OPTIMIZATION." JUTI: Jurnal Ilmiah Teknologi Informasi 18, no. 1: 1-8.