Sistema robotizado autônomo para inspeção de tubulações
Nenhuma Miniatura disponível
Data
2017-03-13
Autores
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Federal de Goiás
Resumo
During one of the most critical water crises in Brazil, in 2015, a Folha de São Paulo Journal report showed that 37 % of all water treated in Brazil does not reach the final consumer, being the failures in pipes the main cause. This master’s work addresses this problem of waste and failure in piping, for both water and sewage. In this way the characteristics of these types of pipes as well as the most common types of faults are presented. Existing inspection piping methods are also presented. And in an attempt to help reduce this pipeline waste, a pipe failure detection method was developed to achieve a low cost solution so that it can be widely used.
The developed solution involved the development of a position control method using a reinforcement learning method, Q-Learning, which allows the robot to run through the pipe without tumbling, using only a gyroscope.
It was also developed, using infrared distance sensors and an accelerometer, a method of detecting failures in which it was possible to detect faults with only 2 mm of thickness, in all the times in which the robot passed through it. The developed method is able to overcome the instability and imprecision of the infrared sensor, allowing, during the tests, the detection of all the faults in which the robot crossed, having only 1 false failure accusation in 25 m of course at a speed 60 % greater than the maximum speed of a video inspection. The results were satisfactory, but more work on increasing the inspected surface area and redundancy of fault detection is still required.
Descrição
Citação
PINTO, D. S. S. Sistema robotizado autônomo para inspeção de tubulações. 2017. 93 f. Dissertação (Mestrado em Engenharia Elétrica e da Computação) - Universidade Federal de Goiás, Goiânia, 2017.