MUMBAI, India, Nov. 2 -- Intellectual Property India has published a patent application (202641126968 A) filed by Madhankumar C; Ms Sarika Suresh Sakhare - Sveris College Of Engineering; Ms Vahidhabanu Y - Anjalai Ammalmahalingam Engineering College; Ms Muthulakshmi G - Nehru Arts And Science College; Dr. Sushma T. C - Nitte Meenakshi Institute Of Technology Nmit, Nitte; Ms. P. Vijaya - Jai Shriram Engineering College; and Ms Sowmiya S - Kings Engineering College on October 24, 2026, for Ai-Driven Intelligent Mathematical Modeling Framework For Real-Time Scientific Computing, Predictive Analytics, And Complex Decision Optimization Using Deep Learning, Explainable Artificial Intelligence, And Digital Twin Analytics.
Inventors include Ms Sarika Suresh Sakhare - Sveris College Of Engineering; Ms Vahidhabanu Y - Anjalai Ammalmahalingam Engineering College; Ms Muthulakshmi G - Nehru Arts And Science College; Dr. Sushma T. C - Nitte Meenakshi Institute Of Technology Nmit, Nitte; Ms. P. Vijaya - Jai Shriram Engineering College; and Ms Sowmiya S - Kings Engineering College.
The application for the patent was published on October 30, 2026, under issue no. 44/2026.
Abstract: ABSTRACT AI-Driven Intelligent Mathematical Modeling Framework for Real-Time Scientific Computing, Predictive Analytics, and Complex Decision Optimization Using Deep Learning, Explainable Artificial Intelligence, and Digital Twin Analytics The present invention relates to an artificial intelligence-based mathematical modeling framework for real-time scientific computing, predictive analytics, and complex decision optimization. More particularly, the invention integrates mathematical modeling, numerical computation, deep learning, Explainable Artificial Intelligence (XAI), digital twin technology, multi-source data fusion, and optimization algorithms into a unified computational architecture configured to solve complex scientific and engineering problems, predict dynamic system behavior, and support intelligent decision-making under changing operational conditions. The proposed framework acquires heterogeneous data from scientific instruments, IoT sensors, experimental environments, simulation platforms, engineering systems, historical datasets, and other authorized data sources. A data-processing module performs data validation, noise reduction, normalization, feature extraction, and temporal synchronization before supplying the processed information to a mathematical modeling engine. The modeling engine incorporates differential equations, statistical models, numerical methods, optimization functions, probabilistic models, and data-driven mathematical representations to describe complex physical, scientific, and engineering processes. A deep learning intelligence module analyzes historical and real-time information to identify nonlinear relationships, estimate unknown model parameters, recognize complex patterns, predict future system states, and improve model performance. The framework supports hybrid modeling by combining physics-based mathematical equations with data-driven learning models, thereby enabling predictions that incorporate both established scientific relationships and observed system behavior. An adaptive parameter-estimation mechanism updates selected model parameters according to incoming observations, prediction errors, and changing operating conditions. The invention further incorporates a digital twin analytics module that maintains a dynamic virtual representation of a physical or computational system. The digital twin receives real time observations and updates its virtual state to reflect changes in the corresponding system. A simulation and optimization module evaluates alternative scenarios, performs numerical simulations, estimates potential outcomes, and identifies optimized solutions according to predefined objectives, constraints, uncertainty measures, and resource requirements. The framework may support applications involving scientific experimentation, engineering design, environmental modeling, energy systems, industrial processes, computational physics, resource allocation, and other complex decision environments. An Explainable Artificial Intelligence module generates human-readable explanations for predictions, estimated parameters, detected anomalies, and optimization recommendations by identifying influential variables, mathematical relationships, contributing features, sensitivity measures, confidence indicators, and relevant simulation results. The framework may further incorporate uncertainty estimation, sensitivity analysis, model validation, and error monitoring to assess the reliability of computational outputs. An interactive decision-support interface presents real-time predictions, model states, simulation results, optimization alternatives, and evidence-linked explanations to authorized users. Accordingly, the invention provides an adaptive, explainable, simulation-driven, and intelligent mathematical modeling framework that combines scientific computation with deep learning and digital twin analytics. The proposed framework improves the integration of physics-based and data-driven models, supports real-time prediction and parameter adaptation, enables systematic evaluation of alternative scenarios, and facilitates transparent, constraint aware decision optimization for complex scientific and engineering applications.
Disclaimer: Curated by HT Syndication.
