Accelerating aircraft IFEC diagnostics with agentic AI on AWS

Panasonic Avionics Corporation offers in-flight entertainment and connectivity systems for a vast fleet serving numerous airlines and passengers worldwide. Diagnosing system issues across these diverse configurations is challenging due to the need for deep institutional knowledge, which can take hours to complete. To address this, Panasonic Avionics Corporation collaborated with AWS and the AWS Generative AI Innovation Center to create an AI-powered diagnostic system.

This system utilizes Amazon Bedrock, Amazon SageMaker, and AWS Glue to streamline the diagnosis process while maintaining accuracy. The challenge lies in translating raw operational data into actionable diagnostics for a fleet with various configurations and services. By leveraging AWS services, Panasonic Avionics Corporation can now process and analyze data more efficiently, reducing Mean Time to Detect (MTTD) and Mean Time to Resolve (MTTR) for system issues.

The solution consists of a multi-agent workflow system with three layers: a Trend Analyzer, parallel Diagnostic Agents, and a Summarizer. The Trend Analyzer identifies anomalies using performance indicators and detects patterns invisible in individual deployments. Parallel Diagnostic Agents perform specific diagnostic functions, such as correlation analysis and log pattern matching.

Finally, the Summarizer generates coherent diagnostic reports with root cause analysis and recommended actions, allowing engineers to focus on strategic improvements rather than manual investigations.