Power System Database Assistant

A versatile LangGraph-powered chatbot that connects to different database types (GraphDB, PostgreSQL, AWS RDS, AWS Neptune) for querying power system equipment data.

Features

• Multi-Database Support: Compatible with both graph databases (GraphDB, AWS Neptune) and relational databases (PostgreSQL, AWS RDS)
• Advanced LLM Integration:
  • Auto model selection based on query complexity
  • Manual model selection (Gemini, Llama3)
  • Optimized query generation
• Sophisticated Query Processing:
  • Automatic conversion from natural language to appropriate query language (SPARQL/SQL)
  • Query validation and optimization
  • Result caching for better performance
• Power System Domain Specialized:
  • Built-in knowledge of electrical grid equipment
  • Specialized schema for transformers, lines, substations, etc.
• Interactive Web Interface:
  • Simple and intuitive chat interface
  • Database and model selection options
  • Connection configuration capabilities

Technical Architecture

The application follows a multi-stage processing pipeline powered by LangGraph:

1. Model Selection: Chooses between Gemini and Llama3 models based on user preference or query complexity
2. Query Optimization: Converts natural language to the appropriate query language
3. Query Validation: Checks for syntax errors and semantic correctness
4. Query Execution: Executes the validated query against the selected database
5. Response Generation: Creates a formatted response from the query results

Database Schema

The system supports both relational and graph database schemas for power system equipment:

Relational Schema (PostgreSQL/AWS RDS)

• Equipment (id, name, description, type)
• Transformer (id, name, apparent_power, voltage_level, status)
• Terminal (id, name, connected)
• Substation (id, name, latitude, longitude)
• Line (id, name, length, voltage_level)
• Various relationship tables (EquipmentSubstation, TerminalConnection)

Graph Schema (GraphDB/AWS Neptune)

• Similar entities as vertices with relationships defined as edges
• Supports both SPARQL and Gremlin query languages for AWS Neptune

Setup and Installation

See the guide.txt file for detailed installation instructions.

Quick Start

1. Clone this repository
2. Install dependencies (see install_deps.txt for details)
3. Configure your .env file with API keys and database connection (see .env.example)
4. Run the application: python main.py or gunicorn --bind 0.0.0.0:5000 --reuse-port --reload main:app
5. Open http://localhost:5000 in your browser

Usage Examples

The system can answer a wide range of questions about power system equipment:

• "What are all the transformers with voltage level above 300kV?"
• "Show me the substations located in the northern region"
• "Which terminal connects transformer TX001 to line L1?"
• "List all equipment at the Main Substation"

API Keys and Secrets

This project requires:

• Google API key for Gemini model access
• GROQ API key for Llama3 model access (optional)
• Database credentials if connecting to real databases

Contributors

• [Your Name/Team]

License

[Your License Information]