Simulation-Integrated Multi-AI Agent Energy Platform (Demo)

HILS + Multi-Agent + Hybrid AI + MCP Runtime based Energy Management Visualization Demo

🇰🇷 프로젝트 개요 — HILS(Hardware-in-the-Loop Simulation) 기반 가상 EMS 환경에서 6개의 AI 에이전트(Forecaster, Optimizer, Scheduler, FaultDetection, HILSCoach, Orchestrator)가 공유 컨텍스트(MCP) 위에서 협업하며 PV/ESS/EV 등 분산전원(DER)을 최적 운영하는 과정을 시각화하는 에너지 관리 데모 플랫폼입니다. 실제 EMS 없이도 Streamlit 대시보드에서 에너지 최적화 AI의 동작 원리를 직관적으로 확인할 수 있습니다.

📸 Screenshots

Main Dashboard

Real-time HILS energy simulation with 6 AI agents status monitoring

Hybrid AI Architecture

ML + Rule-based + LLM three-tier architecture visualization

Agent Collaboration Timeline

Multi-agent decision flow and collaboration patterns

Agent Execution Distribution

Activity distribution across different AI agents

Hybrid AI Insights

Real-time insights from ML Optimizer, Rule-based Forecaster, and LLM Analyst

MCP Context Flow

Context server data flow showing shared state across agents

---

🎯 Project Purpose

This project is a demonstration platform showcasing advanced energy management technologies:

• HILS (Hardware-in-the-Loop Simulation) based closed-loop simulator
• ML + LLM Hybrid AI agent architecture
• Multi-Agent Collaboration framework
• MCP (Model Context Protocol) based Context server structure
• DER (PV/ESS/EV) management and optimization workflow visualization
• Real-time visualization with Streamlit UI

Core Concept: A demo UI that visually demonstrates how energy optimization AI works without requiring an actual EMS system

🛠️ Tech Stack

Category	Tech	Purpose
Language		전체 시스템 구현
Web UI		실시간 대시보드
Visualization		에너지 차트/그래프
Optimization		ESS 충방전 MILP 최적화
Simulation	SimPy 4.1+ · NumPy 1.26+ · Pandas 2.2+	HILS 시뮬레이션 & 데이터 처리
Testing	pytest 7.4+	단위 테스트 (17개, 100% 통과)
License	Custom License	개인/교육용 무료, 상업용 라이선스 필요

📁 Project Structure

smartEMS-MultiAgent-Demo/
├── app.py                          # Main Streamlit application entry point
├── requirements.txt                # Python dependencies
├── README.md                       # Project documentation
│
├── src/                            # Source code directory
│   ├── __init__.py
│   ├── agents/                     # AI agent modules
│   │   ├── __init__.py
│   │   └── agents.py               # All AI agents (Forecaster, Optimizer, etc.)
│   ├── core/                       # Core system components
│   │   ├── __init__.py
│   │   ├── simulator.py            # HILS-based simulator
│   │   └── mcp_core.py             # MCP Runtime (Context + Orchestration)
│   └── ui/                         # UI components
│       ├── __init__.py
│       ├── ui_components.py        # Reusable UI components (charts, cards, logs)
│       ├── hybrid_ai_ui.py         # Hybrid AI + Collaboration visualization
│       └── utils.py                # Utility functions (logging, state management)
│
└── tests/                          # Test directory
    ├── __init__.py
    └── test_system.py              # Unit tests (17 tests, 100% pass rate)

🔄 System Workflow

[HILS Simulator] 
    ↓ Calculate and generate ESS/PV/Load values
[MCP Runtime]
    ↓ Update context (time, ess_level, pv_output, load, price, etc.)
[Multi-AI Agents]
    ├─ Forecaster: Load/weather prediction
    ├─ Optimizer: MILP optimization
    ├─ Scheduler: Charge/discharge policy determination
    ├─ Fault Detection: Anomaly detection
    ├─ Orchestrator: Agent invocation order management
    └─ HILS Coach: HILS learning/tuning parameter calculation
[UI / Visualization]
    └─ Real-time energy graphs + Agent status dashboard + Log stream

🏗️ Core Technology Components

1. HILS-based Closed-loop Structure: Real-time feedback loop for control algorithm verification
2. ML + LLM Hybrid AI Agents: Combination of rule-based + optimization + LLM-style analysis
3. Multi-Agent Collaboration: 6 agents collaborating through shared context
4. Context Server-based MCP Structure: Platform for sharing state/history between agents
5. DER Control Flow: Integrated optimization of PV/ESS/EV
6. Extensible Tool Gateway Structure: Support for Modbus/OPC-UA integration (mocked in demo)

🚀 Quick Start

Installation

git clone https://github.com/lhg96/smartEMS-MultiAgent-Demo.git
cd smartEMS-MultiAgent-Demo

# (선택) 가상 환경 생성
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate

pip install -r requirements.txt

Execution

streamlit run app.py

Browser will automatically open at http://localhost:8501

Testing

python -m pytest tests/test_system.py -v

All tests passing: 17/17 ✅

🤖 Agent Architecture

6 Core Agents

Agent	Role	Technology
ForecasterAgent	Load/weather prediction	Stochastic modeling
OptimizerAgent	ESS charge/discharge optimization	MILP (PuLP) / Heuristic fallback
SchedulerAgent	Charge/discharge scheduling	Rule-based + SoC/ToU
FaultDetectionAgent	PV/ESS anomaly detection	Threshold-based detection
HILSCoachAgent	Learning parameter adjustment	Adaptive parameter tuning
OrchestratorAgent	Agent invocation management	Orchestration pattern

Hybrid AI Structure

• ML-based: MILP optimization (Optimizer)
• Rule-based: Scheduling, anomaly detection
• LLM-like: MockLLMAgent (strategic analysis)

🎨 UI Structure

┌─────────────────────────────────────────────────┐
│  Sidebar: Simulation Control                    │
│  - Scenario selection                           │
│  - Step length / speed adjustment               │
│  - Start / Stop buttons                         │
└─────────────────────────────────────────────────┘

┌────────────────────────┬────────────────────────┐
│  Left: HILS Simulation │  Right Top: Agent Status│
│  - ESS SOC graph       │  - 6 agent cards        │
│  - PV Output           │  - Status icons         │
│  - Load                │                         │
│                        ├────────────────────────┤
│                        │  Right Bottom: Live Logs│
│                        │  - Step-by-step metrics │
│                        │  - CSV download         │
└────────────────────────┴────────────────────────┘

┌─────────────────────────────────────────────────┐
│  Hybrid AI + Multi-Agent Collaboration Demo     │
│  ├─ Tab 1: Hybrid AI Architecture               │
│  │   - ML / Rule-based / LLM 3-tier visualization│
│  │   - Agent Decision Comparison Table          │
│  │   - Hybrid AI Insights Panel                 │
│  ├─ Tab 2: Agent Collaboration                  │
│  │   - Multi-Agent Decision Timeline Graph      │
│  │   - Agent Activity Distribution Pie Chart    │
│  └─ Tab 3: MCP Context Flow                     │
│      - MCP Context Updates (recent 5 steps)     │
│      - Context Server role explanation          │
└─────────────────────────────────────────────────┘

📊 Key Features

Simulation Control

• 3 Scenarios: Baseline / High Price Late Peak / Volatile Market
• Customization: Step length (5-60), speed (0.2-2.0s), price flip ratio
• Reproducibility: Fixed seed option

Real-time Monitoring

• Energy Charts: ESS SOC, PV, Load (Plotly spline curves)
• Agent Status: Color coding (✅🟢🟡⚠️❌)
• Live Logs: Scrollable + CSV download
• Metrics: Real-time ESS/PV/Load/Price display

Hybrid AI Demo (New)

• AI Architecture Visualization: ML/Rule-based/LLM 3-tier structure
• Agent Decision Comparison: AI type and recent decisions table for each agent
• Collaboration Timeline: Multi-agent decision flow graph
• Activity Distribution: Agent call count pie chart
• MCP Context Flow: Context server data flow visualization
• Hybrid AI Insights: Recent judgments from ML/Rule/LLM layers

🔧 Code Quality

Documentation

• ✅ All modules/functions have docstrings
• ✅ Type hints
• ✅ Detailed comments

Error Handling

• ✅ Try-except blocks
• ✅ Logging system
• ✅ Graceful degradation

Testing

• ✅ 17 unit tests
• ✅ 100% pass rate
• ✅ Integration tests

📚 Module Roles

src/core/simulator.py

• ESS charge level simulation
• PV output random modeling
• Load modeling
• HILS feedback loop support

src/core/mcp_core.py

• Shared context storage for entire system
• Agent invocation and coordination (orchestrate())
• MCP-like structure for state sharing between agents

src/agents/agents.py

• Implementation of 6 agents
• Hybrid AI structure (ML + Rule + LLM-like)
• Context-based collaboration

src/ui/ui_components.py

• Reusable UI components
• Charts, status cards, logs, metrics

src/ui/hybrid_ai_ui.py

• Hybrid AI architecture visualization
• Multi-agent collaboration timeline
• MCP context flow display

src/ui/utils.py

• Log formatting
• Session state management
• Helper functions

🐛 Troubleshooting

PuLP Optimization Error

pip install pulp

Streamlit Port Conflict

streamlit run app.py --server.port 8502

Test Failures

pip install -r requirements.txt --force-reinstall
python -m pytest tests/test_system.py -v

📝 Development Guide

Adding a New Agent

1. Add new Agent class to src/agents/agents.py
2. Inherit from BaseAgent and implement run() method
3. Add to agent dictionary in app.py
4. Add tests to tests/test_system.py

Adding New UI Components

1. Create component function in src/ui/ui_components.py or src/ui/hybrid_ai_ui.py
2. Import and use in app.py
3. Follow existing naming conventions (render_* functions)

Project Structure Guidelines

• src/agents/: All AI agent implementations
• src/core/: Core business logic (simulator, MCP)
• src/ui/: All UI-related components and utilities
• tests/: All test files with test_ prefix
• Keep app.py as the main entry point, import from src/

UI 수정 시

• 좌측 그래프: ui_components.py → render_energy_chart()
• 에이전트 카드: ui_components.py → render_agent_status_cards()
• 로그: ui_components.py → render_live_logs()

🎓 학습 순서

1. README.md 읽기 → 전체 개념 파악
2. simulator.py 보기 → HILS 시뮬레이션 로직
3. src/agents/agents.py - Understand each agent's role
4. src/core/mcp_core.py - Check MCP orchestration
5. app.py - Run and explore UI functionality

---

📄 License

Custom License - Free for Personal Use, Commercial License Required

This software is free to use for personal, educational, and non-commercial purposes. Commercial use requires a separate license agreement.

• ✅ Free: Personal use, education, research
• ❌ Requires License: Commercial use, production deployment, integration into commercial products

For commercial licensing inquiries, please contact: hyun.lim@okkorea.net

---

📞 Contact & Services

Development Consulting & Outsourcing Available

We provide professional consulting and development services for IoT, AI, and embedded systems projects.

👨‍💼 Project Manager Contact

• Email: hyun.lim@okkorea.net
• Homepage: https://www.okkorea.net
• LinkedIn: https://www.linkedin.com/in/aionlabs/

🛠️ Technical Expertise / 기술 전문 분야

• IoT System Design and Development / IoT 시스템 설계 및 개발
• Embedded Software Development / 임베디드 소프트웨어 개발 (Arduino, ESP32)
• AI Service Development / AI 서비스 개발 (LLM, MCP Agent)
• Cloud Service Architecture / 클라우드 서비스 구축 (Google Cloud Platform)
• Hardware Prototyping / 하드웨어 프로토타이핑

💼 Services / 서비스

• Technical Consulting / 기술 컨설팅

  • IoT project planning and design consultation / IoT 프로젝트 기획 및 설계 자문
  • System architecture design / 시스템 아키텍처 설계

• Development Outsourcing / 개발 외주

  • Full-stack development from firmware to cloud / 펌웨어부터 클라우드까지 Full-stack 개발
  • Proof of Concept (PoC) development / 개념 검증 개발
  • Production-ready system development / 상용 시스템 개발