INTELLIGENT MULTI-AGENT QUALITY CONTROL SYSTEM FOR NATURAL GAS METERING BASED ON THE DUAL-CONTROL METHOD AND MACHINE LEARNING

Abstract

This paper addresses the problem of enhancing the reliability of commercial natural gas metering by developing an intelligent measurement quality control system for metering stations operating under conditions of significant uncertainty. This uncertainty is driven by the drift of sensor metrological characteristics, temperature and pressure fluctuations, random disturbances, and potential unauthorized tampering. An intelligent quality control system for energy resource metering is proposed, based on a multi-agent information and measurement architecture and a dual-control metering quality method. This method combines the physical verification of gas flow parameter consistency with the analysis of time series reconstructed via machine learning algorithms. The core of the developed system is an adapted dual-control method that implements parallel oversight of both the process (parameter consistency) and the outcome (balance relations) of the metering. To detect anomalies, leaks, and unauthorized extraction, machine learning algorithms (Isolation Forest, DBSCAN, and autoencoders) are applied. An integral metering quality indicator is proposed, providing a comprehensive, real-time assessment of measurement information reliability.Experimental validation of the proposed models was conducted using hybrid datasets that incorporate sensor drift, noise, and artificially introduced anomalies. The use of the Remaining Metrological Resource (RMR) indicator is proposed as a calculated time interval during which the integral natural gas metering quality indicator will remain within the established metrological tolerance before reaching a critical level of accumulated error. Тhe research results confirm the viability of hybrid approaches that combine physical and mathematical models of the gas metering process with machine learning methods. This facilitates a transition from traditional periodic metrology control to continuous intelligent monitoring of commercial energy resource metering stations, including those for natural gas.