MUMBAI, India, July 13 -- Intellectual Property India has published a patent application (202641083349 A) filed by Vignan'S Nirula Institute Of Technology And Science For Women on July 07, 2026, for Adaptive Multi-Agent Federated Edge Intelligence Framework For Autonomous Industrial Iot Anomaly Detection And Predictive Maintenance.
Inventors include V. Pavani; P. Sandhya Krishna; K V S S Rama Krishna; K. Aruna Kumari; K. Triveni; Bhavya Sri Reddy; P Pranathi; J Keerthi; B Lavanya; and D Rathnamani.
The application for the patent was published on July 10, 2026, under issue no. 28/2026.
Abstract: The current invention is described as an Adaptive Multi-Agent Federated Edge Intelligence Framework for Autonomous IIoT Anomaly Detection and Predictive Maintenance in order to overcome the drawbacks associated with the traditional centralized industrial monitoring system and static federated learning. The present framework incorporates the integration of distributed edge computing, adaptive federated learning, autonomous multi-agent coordination, intelligent client prioritization, communication-aware optimization, predictive maintenance, and industrial cybersecurity in one single framework for the purpose of industrial intelligence in real-time. The data from industrial sensors are processed locally at distributed edge nodes without transmitting any sensitive operational information to the centralized cloud servers in order to maintain organizational privacy. Autonomous software agents monitor the process of resource allocation, communication scheduling, collaborative learning, trust evaluation, cybersecurity monitoring, and knowledge evolution continuously in order to make intelligent adaptation to dynamic industrial environments. The hybrid deep anomaly detection engine examines time-based, contextual, and structural estimate equipment degradation, the amount of life left for equipment, maintenance priorities, and the operational behavior in an industrial environment, whereas the threshold adaptation process dynamically adjusts the threshold for anomaly detection based on changes in operating conditions. The identified anomalies can be used to identify risks and perform predictive maintenance. The continuous process of knowledge evolution ensures gradual model improvement without having to train the model again, ensuring that the model remains adaptable despite the aging of equipment, production variations, and cybersecurity threats. The results of the experiments show a better performance in terms of anomaly detection accuracy, lower communication costs, scalability, privacy protection, and predictive maintenance capabilities than traditional systems.
Disclaimer: Curated by HT Syndication.
