Abstract
The information was reviewed to orderly arrange theoretical provisions and to devise practical recommendations for the diagnostic examination of a system for the anti-corrosion protection of metal oil and gas pipelines. A set of informative parameters for modeling functional relations and determining polarization potential in the system underground metal structure – cathodic protection was formed. It was proposed to apply the algorithm of prediction of corrosive current, the approach of non-linear programming, as well as the neural network, including the corresponding methods of learning, for a pipeline section, taking into account the polarization potential on the outer surface. The test set to evaluate the effectiveness of a neural network was formed. The above-mentioned information is essential for the improvement of the equipment of distant control of metal structures of oil and gas enterprises, that is, the procedures for correct measuring and evaluating direct and alternating voltages, as well as polarization potential in a pipeline. The methods and algorithms of neural networks, which are used to control the anticorrosive protection of pipelines, were explored. The study of the effectiveness of artificial neural networks, specifically, a two-layer network of direct propagation with the function of prediction of the resource of metal pipes. Using the polarization potential, we detected the capability of neural networks to perform inaccessible for conventional mathematics operations of processing, comparison, classification of images, capability of self-learning and self-organization relative to pipelines. The qualimetric quality criterion for a pipeline section, taking into consideration the optimal range of polarization potential was improved. We developed the method for prediction of the polarization potential of a cathodic protection plant and transitional specific resistance of the insulating coating on the surface of an metal structure using a neural network. Taking into consideration the results of analysis of polarization potential and transitional specific resistance, the methodology of formation of information support for procedures of degradation of anticorrosive dielectric coating and metal on the outer surface of an metal structure, as well as for predicting its resource, was designed
Highlights
Analysis and monitoring of technical condition of underground metal pipelines (UMP) at oil and gas enterprises is important because the damage and destruction of the elements of structures in the process of operation can lead to dangerous and/or catastrophic consequences
The application of neural networks [10] taking into consideration the criteria of the type [9] makes it possible to develop the methodology for the analysis of functional connections for the system “UMP – plant of cathodic protection (PCP)”, enhance the quality of PCP, but does not give grounds to predict the resource of metal structures
Transitional resistance of the insulating coating was determined and the appropriate test set for the evaluation of the effectiveness of a neural network was formed
Summary
Analysis and monitoring of technical condition of underground metal pipelines (UMP) at oil and gas enterprises is important because the damage and destruction of the elements of structures in the process of operation can lead to dangerous and/or catastrophic consequences. Applied physics principle of using artificial neural networks (ANN) involves continuous and automatic control of defects and damages caused by adverse conditions during the operation of pipelines and metal structures. It is appropriate to analyze this type of complex systems with a large amount of damage (defects) using the ANN and take into consideration the elements of artificial intelligence of a certain level [1] in order to avoid errors in predicting operating conditions. It is appropriate to consider the “UMP – PCP” system as a complex integrated system taking into account the set of energy and kinetic parameters It is obvious that the relevance of the research is caused by the fact that during development of algorithms of diagnosing UMP and designing improvements for the PCP based on the ANN, it is advisable to perform a selection of the ANN type
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
