Speed prediction of a pipeline inspection gauge (PIG) based on differential pressure and acceleration with artificial neural networks
A device known as pipeline inspection gauge (PIG) runs through oil and gas pipelines performing various maintenance operations in the oil and gas industry. The PIG’s velocity, which plays a role in the efficiency of these operations, is usually obtained indirectly from odometers installed in it. A...
Spremljeno u:
Glavni autor: | |
---|---|
Daljnji autori: | |
Format: | doctoralThesis |
Jezik: | pt_BR |
Izdano: |
Universidade Federal do Rio Grande do Norte
|
Teme: | |
Online pristup: | https://repositorio.ufrn.br/handle/123456789/49553 |
Oznake: |
Dodaj oznaku
Bez oznaka, Budi prvi tko označuje ovaj zapis!
|
id |
ri-123456789-49553 |
---|---|
record_format |
dspace |
institution |
Repositório Institucional |
collection |
RI - UFRN |
language |
pt_BR |
topic |
Pipeline Inspection Gauge (PIG) Artificial neural networks Embedded systems Raspberry Pi CNPQ::ENGENHARIAS::ENGENHARIA ELETRICA |
spellingShingle |
Pipeline Inspection Gauge (PIG) Artificial neural networks Embedded systems Raspberry Pi CNPQ::ENGENHARIAS::ENGENHARIA ELETRICA Freitas, Victor Carvalho Galvão de Speed prediction of a pipeline inspection gauge (PIG) based on differential pressure and acceleration with artificial neural networks |
description |
A device known as pipeline inspection gauge (PIG) runs through oil and gas pipelines
performing various maintenance operations in the oil and gas industry. The PIG’s velocity, which plays a role in the efficiency of these operations, is usually obtained indirectly
from odometers installed in it. Although this is a relatively simple technique, the loss of
contact between the odometer wheel and the pipeline results in measurement errors. To
help reduce these errors, this work employed neural networks to estimate the speed of a
prototype PIG, using the pressure difference that acts on the device inside the pipeline
and its acceleration instead of using odometers; we built static networks (e.g., multilayer
perceptron) and recurrent networks (e.g., long short-term memory). We developed the
prototype PIG with an embedded system based on Raspberry Pi 3 to collect speed, acceleration, and pressure data for the model training. The Python library TensorFlow was used
to implement supervised neural networks. To train and evaluate the models, we used the
PIG testing pipeline available at the Petroleum Evaluation and Measurement Laboratory
of the Federal University of Rio Grande do Norte (LAMP/UFRN). Our results show that
the models were able to learn the relationship between the differential pressure, acceleration and speed of the PIG. The proposed approach can complement the odometer-based
systems, thereby increasing the reliability of speed measurement. |
author2 |
Salazar, Andrés Ortiz |
author_facet |
Salazar, Andrés Ortiz Freitas, Victor Carvalho Galvão de |
format |
doctoralThesis |
author |
Freitas, Victor Carvalho Galvão de |
author_sort |
Freitas, Victor Carvalho Galvão de |
title |
Speed prediction of a pipeline inspection gauge (PIG) based on differential pressure and acceleration with artificial neural networks |
title_short |
Speed prediction of a pipeline inspection gauge (PIG) based on differential pressure and acceleration with artificial neural networks |
title_full |
Speed prediction of a pipeline inspection gauge (PIG) based on differential pressure and acceleration with artificial neural networks |
title_fullStr |
Speed prediction of a pipeline inspection gauge (PIG) based on differential pressure and acceleration with artificial neural networks |
title_full_unstemmed |
Speed prediction of a pipeline inspection gauge (PIG) based on differential pressure and acceleration with artificial neural networks |
title_sort |
speed prediction of a pipeline inspection gauge (pig) based on differential pressure and acceleration with artificial neural networks |
publisher |
Universidade Federal do Rio Grande do Norte |
publishDate |
2022 |
url |
https://repositorio.ufrn.br/handle/123456789/49553 |
work_keys_str_mv |
AT freitasvictorcarvalhogalvaode speedpredictionofapipelineinspectiongaugepigbasedondifferentialpressureandaccelerationwithartificialneuralnetworks |
_version_ |
1773962666501472256 |
spelling |
ri-123456789-495532022-10-11T20:33:33Z Speed prediction of a pipeline inspection gauge (PIG) based on differential pressure and acceleration with artificial neural networks Freitas, Victor Carvalho Galvão de Salazar, Andrés Ortiz http://lattes.cnpq.br/7762013122037641 https://orcid.org/0000-0001-5650-3668 http://lattes.cnpq.br/7865065553087432 Doria Neto, Adrião Duarte https://orcid.org/0000-0002-5445-7327 http://lattes.cnpq.br/1987295209521433 Maitelli, André Laurindo Lima, Gustavo Fernandes de Villanueva, Juan Moisés Mauricio Pipeline Inspection Gauge (PIG) Artificial neural networks Embedded systems Raspberry Pi CNPQ::ENGENHARIAS::ENGENHARIA ELETRICA A device known as pipeline inspection gauge (PIG) runs through oil and gas pipelines performing various maintenance operations in the oil and gas industry. The PIG’s velocity, which plays a role in the efficiency of these operations, is usually obtained indirectly from odometers installed in it. Although this is a relatively simple technique, the loss of contact between the odometer wheel and the pipeline results in measurement errors. To help reduce these errors, this work employed neural networks to estimate the speed of a prototype PIG, using the pressure difference that acts on the device inside the pipeline and its acceleration instead of using odometers; we built static networks (e.g., multilayer perceptron) and recurrent networks (e.g., long short-term memory). We developed the prototype PIG with an embedded system based on Raspberry Pi 3 to collect speed, acceleration, and pressure data for the model training. The Python library TensorFlow was used to implement supervised neural networks. To train and evaluate the models, we used the PIG testing pipeline available at the Petroleum Evaluation and Measurement Laboratory of the Federal University of Rio Grande do Norte (LAMP/UFRN). Our results show that the models were able to learn the relationship between the differential pressure, acceleration and speed of the PIG. The proposed approach can complement the odometer-based systems, thereby increasing the reliability of speed measurement. Um dispositivo conhecido como pipeline inspection gauge (PIG) percorre oleodutos e gasodutos realizando diversas operações de manutenção na indústria de petróleo e gás. A velocidade do PIG, que desempenha um papel importante na eficiência dessas operações, geralmente é obtida indiretamente a partir dos hodômetros nele instalados. Embora esta seja uma técnica relativamente simples, a perda de contato entre a roda do hodômetro e a tubulação resulta em erros de medição. Para ajudar a reduzir esses erros, este trabalho empregou redes neurais para estimar a velocidade de um PIG protótipo, utilizando a diferença de pressão que atua no dispositivo dentro dos dutos e sua aceleração ao invés de utilizar hodômetros; construímos redes neurais estáticas (por exemplo, multilayer perceptron) e redes recorrentes (por exemplo, long short-term memory). Desenvolvemos o PIG protótipo com um sistema embarcado baseado em Raspberry Pi 3 para coletar dados de velocidade, aceleração e pressão para o treinamento do modelo. A biblioteca Python TensorFlow foi usada para implementar redes neurais supervisionadas. Para treinar e avaliar os modelos, utilizamos o duto de testes de PIGs disponível no Laboratório de Avaliação e Medição em Petróleo da Universidade Federal do Rio Grande do Norte (LAMP/UFRN). Nossos resultados mostram que os modelos foram capazes de aprender a relação entre a pressão diferencial, a aceleração e a velocidade do PIG. A abordagem proposta pode complementar os sistemas baseados em hodômetros, aumentando assim a confiabilidade da medição de velocidade. 2022-10-11T20:32:56Z 2022-10-11T20:32:56Z 2022-07-28 doctoralThesis FREITAS, Victor Carvalho Galvão de. Speed prediction of a pipeline inspection gauge (PIG) based on differential pressure and acceleration with artificial neural networks. Orientador: Andrés Ortiz Salazar. 2022. 87f. Tese (Doutorado em Engenharia Elétrica e de Computação) - Centro de Tecnologia, Universidade Federal do Rio Grande do Norte, Natal, 2022. https://repositorio.ufrn.br/handle/123456789/49553 pt_BR Acesso Aberto application/pdf Universidade Federal do Rio Grande do Norte Brasil UFRN PROGRAMA DE PÓS-GRADUAÇÃO EM ENGENHARIA ELÉTRICA E DE COMPUTAÇÃO |