Industrial RAG System for Machine Maintenance Analysis

Overview

This project implements a Retrieval-Augmented Generation (RAG) system designed to analyze structured industrial data such as machine maintenance logs, incident reports, production reports, and SOPs (Standard Operating Procedures).

Unlike generic chatbot-style RAG demos, this system focuses on data correctness, auditability, and retrieval reliability. The design prioritizes retrieval validation before generation and avoids unnecessary complexity.

---

Key Features

• Structured Data Ingestion from JSONL files
• Metadata-Preserving Vector Storage using ChromaDB
• Retrieval-First Design (retrieval verified independently)
• Strict Hallucination Control via prompt constraints
• Deterministic & Auditable Outputs
• Terminal-Based Workflow (no notebooks required)

---

Dataset Structure

The dataset is split into four JSONL files, each representing a distinct document type:

• logs.jsonl — Machine maintenance logs
• incident.jsonl — Incident and anomaly descriptions
• report.jsonl — Daily production reports
• sop.jsonl — Standard Operating Procedures

Each record follows this structure:

{
  "id": "unique-id",
  "document_type": "maintenance_log | incident | production_report | sop",
  "text": "RAG-optimized natural language content",
  "metadata": {
    "machine_id": "optional",
    "date": "optional",
    "severity": "optional",
    "system": "optional"
  }
}

---

System Architecture

JSONL Files
   ↓
Custom Loader (preserves metadata)
   ↓
SentenceTransformer Embeddings
   ↓
Chroma Vector Database
   ↓
Similarity Search (Retrieval)
   ↓
Prompt Construction
   ↓
Local LLM (Generation)

Retrieval can be executed independently without invoking the language model.

---

Technologies Used

• Python 3.10+
• LangChain (document abstraction & prompt templates)
• ChromaDB (vector storage)
• SentenceTransformers (embeddings)
• Hugging Face Transformers (local LLM)
• PyTorch

---

Usage

1. Ingest Data

python db.py

This loads all JSONL files, embeds documents, and persists vectors in the chroma/ directory.

---

2. Retrieval-Only Testing

python test.py

This script performs similarity search and saves results to a text file for inspection.

---

3. Retrieval + Generation

python main.py

The system retrieves relevant documents and generates an answer strictly grounded in retrieved context.

If no answer is present in the data, the system responds with:

No root cause found in available documents.

---

Design Decisions

Why RAG instead of Fine-Tuning?

• Data is frequently updated
• Requires traceability and explainability
• Lower cost and easier iteration

Why No Reranking?

• Small, clean dataset
• Strong embeddings
• Metadata-driven retrieval sufficient

Why No Chain-of-Thought (yet)?

• Avoids masking retrieval errors
• Keeps outputs concise and auditable