REAL-TIME LEAK DETECTION FROM SIGNAL FILTERING VIA INDEPENDENT COMPONENT ANALYSIS

Código

XXIV-SBRH0646

Autores

Thacio Carvalho Pereira, Bruno Melo Brentan, GUSTAVO MEIRELLES LIMA, Daniel Bezerra Barros

Tema

SE04.B - Análise de sistemas de distribuição de água e segurança de redes de abastecimento

Resumo

In the field of water supply, large amounts of water are lost through leaks in pipes, which causes great environmental impacts and high costs for companies linked to the basic sanitation sector. Indirect leak detection systems with sensors allocated along the distribution networks, present benefits in optimizing the allocation of sensors in the network and reducing costs linked to the implementation of equipment. Computational approaches based on mathematical and statistical models have been widely used in the processing of hydraulic data, such as flow and demand, since it is possible to compare the behavior of the application of parameters, with the expected behavior for the system. This work aims to know the real-time behavior of known leaks, through signal processing by the method of Blind Source Separation (BSS), implementing a calibrated algorithm for a water supply network. From the analysis in different sample spaces, leak patterns and a sensitivity of the simulation for two leaks obtained can be observed.