voltgrid-ai

Autonomous Edge-AI renewable microgrid & IoT sentinel with sub-5ms fault mitigation.

Comment

💡 Inspiration

Modern electrical grids are facing an existential crisis. The rapid adoption of intermittent renewables (solar and wind), extreme climate-induced grid stress, and the sudden surge of massive inductive loads (such as multi-bay ultra-fast EV charging plazas) have made centralized power distribution dangerously fragile. Traditional utility grids rely on slow, manual SCADA interventions or crude regional blackouts that leave critical infrastructure—such as hospitals and trauma wards—at catastrophic risk.

We asked ourselves: What if localized community microgrids could think, self-heal, and dynamically balance themselves at the edge in single-digit milliseconds before a frequency collapse even ripples through the main grid?

That question gave birth to VoltGrid AI.

🌐 What It Does

VoltGrid AI is an enterprise-grade cyber-physical microgrid management system and IoT digital twin powered by real-time Edge-AI. It continuously monitors distributed energy resources (DERs)—including rooftop solar arrays, wind turbines, and utility-scale Battery Energy Storage Systems (BESS)—to deliver uninterrupted, clean power.

Key Capabilities:

  1. High-Frequency IoT Telemetry (10Hz): Monitors AC bus voltage, grid frequency (60.00 Hz), active/reactive power, power factor, and Total Harmonic Distortion (THD) via real-time WebSockets.
  2. Cyber-Physical Digital Twin: A high-performance canvas engine rendering dynamic, smooth electron particle flows between renewable generation sources, BESS storage, substations, and consumption sectors.
  3. Autonomous Edge-AI Dispatch Engine:
    • Sub-5ms Decision Loop: Detects power deficits and overcurrent hazards faster than traditional mechanical circuit breakers.
    • Tiered Priority Load Shedding: Automatically throttles or disconnects non-essential loads (e.g., EV charging plazas, robotic manufacturing lines) to guarantee 100% uninterrupted power to critical facilities like St. Jude Hospital (Priority 1).
    • Seamless Islanding & Synthetic Inertia: Detects upstream transmission blackouts and instantly transitions the microgrid into autonomous islanding mode, switching BESS inverters into Grid-Forming (GFM) mode.
  4. Resilience Fault Sandbox: Allows operators to inject real-world emergencies with a single click:
    • Substation Blackout & Islanding
    • Solar Cloud Collapse (sudden 85% generation drop)
    • Transformer Arc Flash & Harmonic Distortion
    • Hospital Critical Trauma Wing Surge
    • BESS Thermal Runaway Warning
  5. AI Root-Cause Incident Investigator: Automatically compiles IEEE 1547 and NERC CIP compliant diagnostic post-mortems with root-cause analysis and actionable engineering mitigation steps with JSON export.
  6. Physical ESP32 Embedded Firmware: Production C++ sketch reading real I2C current/voltage sensors (INA219), thermal sensors (DHT22), and actuating opto-isolated solid-state relays with hardware-level safety bypasses.

🛠️ How We Built It

VoltGrid AI is engineered as a clean, decoupled full-stack architecture across three distinct tiers:

  1. Frontend (Next.js & Material UI):

    • Built with Next.js App Router and Material UI (MUI) with a custom dark-mode command-center theme.
    • Real-time HTML5 Canvas particle animation rendering electrical current velocity and direction based on live telemetry.
    • Live oscilloscope waveform visualizers tracking voltage and frequency harmonic stability.
    • Resilient WebSocket client hook with auto-reconnection and REST API fallback.

  2. Backend (Node.js, Express & WebSockets):

    • Multi-tiered Node.js architecture (controllers/, services/, models/, config/).
    • High-throughput WebSocket server broadcasting 10Hz physics-based telemetry.
    • Edge-AI decision service executing autonomous tiered shedding rules and IEEE 1547 protection logic.
    • REST endpoints for manual overrides, fault injection, and automated incident report generation.

  3. Embedded Hardware & Firmware (ESP32 C++):

    • Syntactically complete Arduino/ESP32 sketch communicating via MQTT.
    • High-side sensing using Texas Instruments INA219 and Aosong DHT22.
    • Fast-acting hardware interrupt tripping solid-state relays in <10 microseconds upon thermal runaway.
    • Complete Bill of Materials (BOM) and pinout schematic documentation.

🧗 Challenges We Ran Into

  • Modeling Electrical Grid Physics: Translating real-world grid dynamics—such as the generator swing equation ($d\Delta f/dt \propto P_{gen} - P_{load}$), active power balance, and harmonic distortion (THD)—into a high-frequency mathematical simulation engine in Node.js.
  • Sub-10ms UI Rendering Performance: Rendering animated electron streams across interconnected nodes alongside high-speed oscilloscope charts without triggering React re-render lag. We solved this using a decoupled HTML5 Canvas animation loop with double-buffered telemetry state.
  • Fail-Safe Priority Locking: Ensuring that automated AI decisions could never, under any programmatic circumstance, disconnect the priority healthcare node (St. Jude Hospital), enforced through both software validation and hardware-level normally-closed relay architectures.

🏆 Accomplishments That We're Proud Of

  • True Full-Stack Architecture: Zero monolithic shortcuts—cleanly decoupled Next.js frontend, independent Node.js backend with live WebSockets, and production-grade ESP32 firmware.
  • Aesthetic Command Center: An interface that looks and feels like a state-of-the-art grid operations room at Tesla Energy or an ISO control room.
  • Sub-5ms Autonomous Response Time: Demonstrating how edge intelligence can protect microgrids from catastrophic brownouts and blackouts.

📚 What We Learned

  • Deepened our knowledge of IEEE 1547-2018 interconnection standards and anti-islanding regulations for distributed energy resources.
  • Advanced state synchronization techniques between high-frequency WebSocket streams and Material UI dashboards.
  • Cyber-physical safety designs: How software AI dispatchers interact with deterministic, hardware-level protective relays.

🚀 What's Next For VoltGrid AI

  • Federated Multi-Microgrid Coordination: Allowing adjacent neighborhood microgrids to trade energy peer-to-peer via local smart contracts.
  • Hardware-in-the-Loop (HIL) Testing: Integrating with physical solar microinverters using industrial SunSpec Modbus RTU / RS485.
  • Wholesale Day-Ahead Price Arbitrage: Integrating live wholesale electricity pricing APIs to dynamically charge BESS during negative pricing windows and discharge during peak tariff hours.

Built With

Share this project:

Updates

Submission history