Intellectual Property India Publishes Patent Application for An Explainable Artificial-Intelligence Framework For Real-Time Decision Support And Anomaly Detection In Autonomous Cyber-Physical Systems With Physics-Consistency Cross-Checking And Safe-State Fallback

MUMBAI, India, Aug. 12 -- Intellectual Property India has published a patent application (202641093051 A) filed by Prameela. R; Rashmi Purad; Dr. Sunetra Chatterjee; Roopa. A; Sunil Kunchoor Basavaraju; Ambica. N; Arunkumar. A; Bangalore Institute Of Technology; and Sir M Visvesvaraya Institute Of Technology on July 31, 2026, for An Explainable Artificial-Intelligence Framework For Real-Time Decision Support And Anomaly Detection In Autonomous Cyber-Physical Systems With Physics-Consistency Cross-Checking And Safe-State Fallback.

Inventors include Prameela. R; Rashmi Purad; Dr. Sunetra Chatterjee; Roopa. A; Sunil Kunchoor Basavaraju; Ambica. N; Arunkumar. A; Bangalore Institute Of Technology; and Sir M Visvesvaraya Institute Of Technology.

The application for the patent was published on August 07, 2026, under issue no. 32/2026.

Abstract: ABSTRACT OF THE INVENTION Title: AN EXPLAINABLE ARTIFICIAL-INTELLIGENCE FRAMEWORK FOR REAL-TIME DECISION SUPPORT AND ANOMALY DETECTION IN AUTONOMOUS CYBER-PHYSICAL SYSTEMS WITH PHYSICS-CONSISTENCY CROSS-CHECKING AND SAFE-STATE FALLBACK The invention relates to an explainable-artificial-intelligence framework for real-time decision support and anomaly detection in autonomous cyber- physical systems that overcomes the limitations of single-model, opaque and operator-blocking prior art. The framework comprises a physical-sensor telemetry collector (101), a control-command and actuator-state collector (102), a network and bus-event collector (103) and an environment and context collector (104); a real- time signal aggregator and feature extractor (105); a physics-based consistency model (106) derived from the CPS's own governing dynamics; an ensemble anomaly- detection subsystem comprising an autoencoder detector (107), an LSTM sequence detector (108) and an isolation-forest detector; an ensemble fusion and risk scorer (109); an XAI attribution engine (110) implementing SHAP, attention attribution and counterfactual explanations; a safety-state knowledge base and rules (111); a safety fallback controller (112); an operator decision-support dashboard (113); an immutable audit ledger (114); and a safe-state CPS actuation gateway (115). Safety- critical indications trigger an unconditional transition of the CPS to a pre-declared safe state prior to operator interaction; non-safety-critical indications are surfaced with SHAP, attention and counterfactual explanations to the operator; and every automated and human step is recorded in the immutable audit ledger. The framework is expressly a decision-support and safety-fallback layer operating within the constraints of the underlying CPS platform's own functional-safety architecture. Number of Claims: 5 Number of Drawings: 3

Disclaimer: Curated by HT Syndication.