MUMBAI, India, Sept. 14 -- Intellectual Property India has published a patent application (202641107786 A) filed by Pragati Engineering College on September 08, 2026, for Self-Learning Digital Twin Framework With Reinforcement Learning For Industrial Automation.
Inventors include Mr. M Sunil Raj; Mrs. P Gayathri; Mr. P. Ram Prasad; and Mr. A. Phani Bhaskar.
The application for the patent was published on September 11, 2026, under issue no. 37/2026.
Abstract: The invention proposes an Adaptive Digital Twin System using Explainable Artificial Intelligence for real-time process monitoring, prediction, optimization, and control in industrial environments. The system collects continuous operational data from sensors measuring parameters such as temperature, pressure, flow rate, vibration, speed, energy consumption, equipment condition, and product quality. An Industrial Internet of Things or edge gateway performs signal conditioning, noise filtering, time stamping, normalization, feature extraction, and communication with the processing platform. The processed information is supplied to an adaptive Digital Twin that maintains a continuously synchronized virtual representation of the physical process. The Digital Twin combines physics-based models, data-driven models, historical process information, and expert knowledge to improve modelling accuracy. It performs real-time data synchronization, state estimation, model calibration, simulation, what-if analysis, and prediction of future process behaviour. An Artificial Intelligence prediction engine detects anomalies, forecasts product quality, estimates equipment degradation, and evaluates operational risks. An optimization engine analyses alternative operating conditions and identifies suitable process parameters for improving production efficiency, energy utilization, quality, equipment reliability, and safety. An Explainable Artificial Intelligence module generates understandable reasons for every significant prediction and optimization recommendation. The explanation identifies influential process variables, feature contributions, confidence levels, expected outcomes, and possible risks associated with the proposed action. A decision-validation module verifies each recommendation against safety limits, equipment restrictions, process constraints, expert rules, and operator-defined policies. Depending on the selected operating mode, a validated action may be automatically transmitted to a process controller, forwarded to an operator for approval, or presented as decision-support information through a dashboard. A continuous feedback mechanism compares predicted results with the actual response of the physical process. The observed difference is used to recalibrate Digital Twin parameters, update expert rules, retrain Artificial Intelligence models, and compensate for equipment ageing, process drift, sensor variation, and changing operating conditions. Edge-based execution allows critical monitoring and control functions to continue during communication failures. The proposed invention thereby provides an adaptive, transparent, safe, and scalable framework for real-time industrial process optimization. It can be applied to manufacturing, chemical processing, energy systems, food production, textile industries, water management, mining, transportation, cold-chain facilities, and other process-oriented sectors.
Disclaimer: Curated by HT Syndication.
