Improving the diagnostics of underground pipelines at oil­and­gas enterprises based on determining hydrogen exponent (PH) of the soil media applying neural networks

Abstract

A set of key parameters and information flows has been formed to simulate stages of probing the outside surface of underground metal pipelines (UMP) taking into account pH of the soil contacting with the pipe metal. Specimens of 17G1S steel placed in acid, alkaline and neutral media were examined using a polarization potential meter in a complex with a contactless current meter. Principles of application of neural networks (NN) in processing experimental results were formulated. A database has been developed. It meets the initial conditions for controlling the soil pH at the boundary with the metal under real conditions. Elements of the optimization approach for assessing pH of a coated UMP in the soil medium were proposed. The approach is based on the multiplicative qualimetric criterion of quality for the UMP section taking into account two groups of coefficients. The first group of coefficients refers to the internal coefficients and characterizes the metal pipeline and the second group refers to the external medium (i.e., soil electrolyte). Elements of the optimization approach for assessing pH of the coated pipeline in the soil medium were proposed. An NN was presented for the pipeline-coating system, which: 1) is capable of solving the problem of cluster analysis and image classification; 2) makes it possible to process data without their prior spectral transformation operating with discrete counts of information signals. The proposed NN type allows it to dynamically expand its own knowledge base of possible types of defects in controlled objects (pipelines) in the process of operation. With the help of the NN, soil pH was assessed for an UMP of 17G1S steel for three situations. The above information is important for improving the methods for controlling oil-and-gas enterprise UMPs, in particular, the methods for a correct assessment of anode current density in metal defects taking into account nonlinear character of informative parameters.