Plan First, Judge Later, Run Better: A DMAIC-Inspired Agentic System for Industrial Anomaly Detection

2026-06-03

AI Summary

This work addresses the limitations of existing large language model (LLM) agents in industrial anomaly detection, which often prioritize execution over planning and struggle with efficiently handling multimodal heterogeneous data. Inspired by the DMAIC (Define, Measure, Analyze, Improve, Control) quality management framework, the authors propose a novel multi-agent architecture that follows a “plan–evaluate–execute” paradigm. The approach first distills heterogeneous inputs into standardized operating procedures (SOPs), then leverages a pre-trained, execution-free evaluator to rank candidate strategies, thereby avoiding costly trial-and-error. This study is the first to integrate the structured DMAIC workflow into LLM-based agent systems and enables strategy assessment prior to execution. Evaluated on four multimodal industrial datasets, the method achieves an average anomaly detection performance improvement of 37.76% over current agent-based baselines, significantly enhancing both accuracy and cost efficiency.

Abstract

Large language model (LLM) agents have shown promise in automating complex data-analysis workflows, but their reliable deployment remains challenging in high-stakes industrial scenarios. Industrial anomaly detection (IAD) is essential for manufacturing quality, safety, and efficiency, yet existing LLM-based IAD agents mainly focus on execution while under-exploiting strategy formulation. Consequently, they struggle to handle heterogeneous modalities in a unified and cost-effective manner. Inspired by the DMAIC quality-management framework, we propose DMAIC-IAD (DMAIC-inspired Agentic Industrial Anomaly Detection), a "Plan First, Judge Later" multi-agent system that aligns LLM agents with structured industrial problem-solving. DMAIC-IAD distills heterogeneous references into standardized operating procedures (SOPs) before strategy generation, and introduces a pre-trained execution-free judge model to rank candidate strategies without costly runtime trials. Extensive experiments across four modalities show that DMAIC-IAD improves average detection performance over applicable agentic baselines by 37.76%.

Problem

• Industrial Anomaly Detection
• LLM Agents
• Heterogeneous Modalities
• Strategy Formulation
• Reliable Deployment

Innovation

• DMAIC
• LLM agents
• industrial anomaly detection
• strategy planning
• execution-free judging