This thesis develops a modular system for the automated diagnosis of vessel connectivity issues in coastal environments, based on device telemetry. Building on an analysis of common root causes, we design dedicated subsystems for detection and state assessment. At its core is a decision model that combines a centralized KPI view, automated threshold- and pattern-based evaluation, persistent state snapshots, and similarity search to determine the most likely cause and a recommended solution. The system is integrated into the existing product, shortening diagnostic time, reducing the scope of manual checks, and laying the groundwork for proactive user alerts. Drawing on previously resolved cases, the system finds similar situations and, with the support of a large language model, produces clear explanations and recommends concrete remediation steps.
|
