MUMBAI, India, Sept. 28 -- Intellectual Property India has published a patent application (202641112488 A) filed by Madhankumar C; Dr. B. Nagarajan - Madanapalle Institute Of Technology & Science Deemed To Be University; Ms Vinodhini Chidhambaram - Mvj College Of Engineering; Mrs. Adlin Stephi S - Kings Engineering College; Mrs. S. Geetha - V. S. B. Engineering College; Dr. M. Maheswaran - Nehru Institute Of Engineering And Technology; Mr K. C. Palanisamy - Erode Sengunthar Engineering College; Mrs. V. K. Poornima - Jct College Of Engineering And Technology; and M. S. Vinu - Jct College Of Engineering And Technology on September 19, 2026, for Federated Explainable Ai Engine For Privacy-Preserving Real-Time Smart Infrastructure Management.
Inventors include Dr. B. Nagarajan - Madanapalle Institute Of Technology & Science; Ms Vinodhini Chidhambaram - Mvj College Of Engineering; Mrs. Adlin Stephi S - Kings Engineering College; Mrs. S. Geetha - V. S. B. Engineering College; Dr. M. Maheswaran - Nehru Institute Of Engineering And Technology; Mr K. C. Palanisamy - Erode Sengunthar Engineering College; Mrs. V. K. Poornima - Jct College Of Engineering And Technology; and M. S. Vinu - Jct College Of Engineering And Technology.
The application for the patent was published on September 25, 2026, under issue no. 39/2026.
Abstract: ABSTRACT Federated Explainable AI Engine for Privacy-Preserving Real-Time Smart Infrastructure Management The present invention relates to an intelligent artificial intelligence-based infrastructure management system, and more particularly to a Federated Explainable Artificial Intelligence (XAI) Engine configured for privacy-preserving, real-time monitoring, prediction, anomaly detection, and decision support for smart infrastructure. The proposed invention integrates Internet of Things (IoT) sensor networks, edge computing, Federated Learning, artificial intelligence, explainable AI, secure communication, infrastructure digital representations, real time analytics, and an authorized management interface within a unified architecture. The system is configured to process distributed infrastructure information while reducing the requirement to transfer sensitive raw operational data from individual infrastructure sites to a centralized learning environment. The system acquires real-time information from heterogeneous sensors deployed across smart infrastructure assets, including buildings, transportation systems, energy facilities, water systems, communication infrastructure, industrial installations, and other critical or public infrastructure. The sensor information may comprise vibration, temperature, pressure, structural strain, electrical parameters, energy consumption, equipment status, environmental conditions, occupancy information, traffic conditions, and other infrastructure-related measurements. Edge computing nodes perform localized data preprocessing, filtering, synchronization, feature extraction, anomaly detection, and sensor-health assessment before transmitting selected information or privacy-preserving model updates. The Federated Learning engine enables multiple infrastructure sites, organizations, departments, or edge nodes to collaboratively train artificial intelligence models while retaining corresponding raw datasets within their local environments. Local models are trained using authorized infrastructure data, and selected model parameters, gradients, or other privacy preserving updates are transmitted to a secure aggregation mechanism. The aggregation mechanism combines the updates to generate a shared AI model that can be distributed to participating nodes according to authorization and model-governance policies. Secure aggregation, encryption, access control, differential privacy, model-update validation, and other privacy-preserving mechanisms may be incorporated to improve protection of sensitive infrastructure information. The AI analytics engine utilizes the shared or locally deployed models to perform real-time infrastructure-condition assessment, anomaly detection, predictive maintenance, equipment health estimation, energy-consumption forecasting, failure prediction, demand prediction, operational-performance analysis, and risk assessment. The system can correlate information from multiple infrastructure assets and sensor sources to identify abnormal conditions and predict potential degradation. A real-time monitoring mechanism continuously evaluates sensor measurements, infrastructure conditions, network status, and analytical outputs to generate prioritized alerts and decision-support information. A principal feature of the invention is the Explainable AI engine, which generates human readable explanations associated with predictions, classifications, anomaly alerts, and recommendations. The explanations may identify contributing sensor observations, relevant features, historical patterns, infrastructure dependencies, detected deviations, confidence indicators, and other factors influencing an AI-generated output. The system may further provide evidence-linked information through an authorized management interface to enable infrastructure operators and decision-makers to understand the basis of an analytical result. The platform may additionally maintain a knowledge representation of infrastructure assets, components, dependencies, operational conditions, maintenance information, and historical events. Such information can be utilized to improve contextual analysis and explainability. The system may dynamically identify relationships between infrastructure components and assess the potential propagation of abnormal conditions across interconnected assets. The proposed invention further incorporates secure communication and access-control mechanisms for protecting infrastructure information and federated model updates. Role-based access, device authentication, encryption, audit logging, data-integrity verification, and secure model-management mechanisms may be employed throughout the architecture. Accordingly, the invention provides a privacy-preserving, distributed, explainable, and real-time AI framework for smart infrastructure management, enabling collaborative intelligence across geographically distributed infrastructure environments without requiring centralized collection of corresponding raw operational datasets. The system improves infrastructure visibility, predictive maintenance capability, anomaly identification, energy management intelligence, operational resilience, and transparency of AI-based decision support while maintaining appropriate human oversight for consequential infrastructure management decisions.
Disclaimer: Curated by HT Syndication.
