10 Prediction of Residual Life of Oil and Gas Pipeline Corrosion …
113
Fig. 10.5 Deep learning network prediction model prediction result
10.6 Conclusion
The artificial intelligence detection method based on deep learning is an important
aspect of the future development of pipeline detection [13]. The article first introduces
the basic ideas of deep learning and the system structure, including the basic model
of the network, the network formation method and model structure, and the training
method. The feasibility of applying the pipeline corrosion detection in deep learning
system is discussed and further applied. Combined with the test, the corrosion life
prediction method of gas pipeline based on deep confidence network proposed in
this paper can effectively predict the corrosion rate of the pipeline with a minimum
error probability of 1.0. Combined with the corrosion time and the design of the
wall thickness, the remaining life of the pipeline is finally obtained. At present, there
are many kinds of techniques for pipeline corrosion detection. Relevant scholars
have studied and improved them, but they all have certain limitations. According to
the corrosion characteristics and types of pipelines, appropriate detection methods
should be selected to accurately predict pipeline life. Prevent accidents caused by
corrosion and leakage of gas pipelines.
References
1. Chindaprasirt, Mechanical properties, microstructure and drying shrinkage of hybrid fly ashbasalt fiber geopolymer paste. Construct Build. Mater 186(62–70) (2018)
2. A.P. Bhardwaj, T.C. Guedes Soares, M. S. Azad, W. Punurai, P. Asavadorndeja, Reliability
assessment of thick high strength pipelines with corrosion defects. Int. J. Pressure Vessels
Piping 177(103982)
3. V. Niesen, M. Gould, Detecting pipeline leaks. Mech. Eng. 139(11), 34–39 (2017)
113
Fig. 10.5 Deep learning network prediction model prediction result
10.6 Conclusion
The artificial intelligence detection method based on deep learning is an important
aspect of the future development of pipeline detection [13]. The article first introduces
the basic ideas of deep learning and the system structure, including the basic model
of the network, the network formation method and model structure, and the training
method. The feasibility of applying the pipeline corrosion detection in deep learning
system is discussed and further applied. Combined with the test, the corrosion life
prediction method of gas pipeline based on deep confidence network proposed in
this paper can effectively predict the corrosion rate of the pipeline with a minimum
error probability of 1.0. Combined with the corrosion time and the design of the
wall thickness, the remaining life of the pipeline is finally obtained. At present, there
are many kinds of techniques for pipeline corrosion detection. Relevant scholars
have studied and improved them, but they all have certain limitations. According to
the corrosion characteristics and types of pipelines, appropriate detection methods
should be selected to accurately predict pipeline life. Prevent accidents caused by
corrosion and leakage of gas pipelines.
References
1. Chindaprasirt, Mechanical properties, microstructure and drying shrinkage of hybrid fly ashbasalt fiber geopolymer paste. Construct Build. Mater 186(62–70) (2018)
2. A.P. Bhardwaj, T.C. Guedes Soares, M. S. Azad, W. Punurai, P. Asavadorndeja, Reliability
assessment of thick high strength pipelines with corrosion defects. Int. J. Pressure Vessels
Piping 177(103982)
3. V. Niesen, M. Gould, Detecting pipeline leaks. Mech. Eng. 139(11), 34–39 (2017)