Sistema para captação de vazão com uso de membrana piezoelétrica

Abstract

With the recent water crisis and the high rate of water losses in Brazil, an opportunity was seen to develop a work to contribute to the management of water losses. This project presents a system consisting of hardware, processing and 3D modeling with the objective of detecting the difference between flows by vibroacoustic methods. For this, an acquisition circuit was made that captures the vibration of a pipe through a piezoelectric membrane. The signal from this sensor is conditioned by an instrumentation amplifier, has its offset adjusted and is sent to the ESP32 microcontroller. The signal is then converted from analogue to digital and sent through the serial port, for subsequent pattern recognition analysis – with different learning models – on a computer using the Python language. To accommodate the system and to couple the piezoelectric sensor to the pipe, a casing and a support were developed, printed in 3D, using filament and resin techniques, respectively. The experimental method consisted of collecting 10 samples with 10 seconds duration for 4 different flows (no flow, low flow, medium flow and high flow) in a gas shower. The results showed that the best among the 3 chosen models was the SVC, derived from Support Vector Machine, with test accuracy greater than 95%.