💻⚡️ Edge AI Supercapacitor & Capacitor Analyzer

🛠️ Hardware Setup Configuration

Asynchronous profiling connection circuit featuring the ESP32 development board and direct-drive logic-level switching line.

---

📊 Real-Time Dashboard

Live telemetry preview capturing analytical calculations, ideal curves, and component health scaling modules.

An automated, hardware-software integrated testing system designed to characterize the health and dynamic behaviors of capacitors and supercapacitors. By performing an active discharge test controlled via an ESP32 microcontroller, the system captures high-resolution real-time voltage and current curves. The accompanying dashboard utilizes TensorFlow.js to compute structural deviations from ideal models, providing a precise State of Health (SoH) estimation directly at the edge.

---

Key Features ⚙️

• Active Discharge Testing: Programmatic control of an N-Channel Logic-Level MOSFET (IRLZ44N) to execute fast, clean discharge cycles.
• High-Resolution Real-Time Telemetry: Real-time acquisition of voltage, current, and power consumption metrics using an INA219 sensor via an optimized 400 kHz $I^2C$ bus speed (Wire.setClock(400000)).
• Mathematical Curve Integration: Capacitance is computed locally on the microcontroller through numerical trapezoidal integration ($Q = \int I , dt$) between defined $80% \rightarrow 40%$ voltage thresholds.
• TensorFlow.js Edge AI Diagnostics: Computes the Mean Squared Error (MSE) between the actual physics-based discharge curve and an ideal calculated RC model, instantly uncovering internal degradation factors like elevated ESR or dielectric leakage.
• Dynamic Visual Dashboard: Built-in Web Serial API connection interface, real-time numeric readouts, interactive multi-dataset graphs via Chart.js, and automated health scoring animations.

---

Theoretical Framework

Unlike standard low-capacity capacitors that exhibit near-perfect exponential decay curves described by $V(t) = V_0 \cdot e^{-\frac{t}{\tau}}$, real-world supercapacitors operate as complex electrochemical systems. They are better modeled as distributed networks where capacitance values can shift dynamically as a function of the voltage domain: $C = f(V)$.

1. Internal Resistance (ESR) Computation

The hardware computes initial Equivalent Series Resistance drops right as load connection stabilizes:

$$ESR = \frac{V_{\text{gol}} - V_{\text{sarcina}}}{I_{\text{init}}}$$

2. Numerical Integration for Capacitance

Because constant resistance cannot be assumed under active load profiling, charge ($Q$) is evaluated using trapezoidal integration slices:

$$C = \frac{Q}{\Delta V} = \frac{\int_{t_{\text{start}}}^{t_{\text{stop}}} I_{\text{dis}}(t) dt}{V(t_{\text{start}}) - V(t_{\text{stop}})}$$

3. State of Health (SoH) Optimization

TensorFlow.js processes the recorded timeline tensors to isolate component age and structural fatigue via a custom loss analysis:

$$MSE = \frac{1}{N}\sum_{k=1}^{N}(V_{\text{ideal}}[k]-V_{\text{actual}}[k])^2$$

---

Hardware System Architecture

The hardware pipeline utilizes an efficient minimalist component layout optimized for breadboard structural integrity:

• MCU: ESP32 Development Board (handles microsecond task timing loops, asynchronous UART telemetry, and gate switching).
• Sensor: INA219 Zero-Drift Digital Current/Power Monitor (monitors up to $32\text{V}$ and $2\text{A}$ domains natively via $I^2C$).
• Switching Element: IRLZ44 Logic-Level N-Channel MOSFET (allows direct drive logic from ESP32 pins without a dedicated driver).
• Power/Bus Configuration: Dual-line $I^2C$ connection tracking alongside structural active load configurations across target test capacitor nodes.

---

📂 Project Structure

├── analizator.cpp      # ESP32 Arduino Core firmware (sampling engine & telemetry loops)
├── index.html          # HTML5/CSS3 Dashboard featuring Chart.js & TensorFlow.js engines
└── README.md           # Documentation

Technical Setup & Deployment

1. Firmware Flash

1. Open analizator.cpp in the Arduino IDE or PlatformIO.
2. Ensure you have installed the Adafruit INA219 library dependencies.
3. Connect your ESP32 to your workstation via USB.
4. Verify the target pinouts (GATE_PIN 25 and native $I^2C$ mapping).
5. Flash the sketch to your board at a standard $115200\text{ bps}$ upload speed.

2. Dashboard Deployment

1. Open the index.html file natively within any modern web browser supporting the Web Serial API (e.g., Google Chrome or Microsoft Edge).
2. Click the 🔌 Conectează ESP32 button in the dashboard header.
3. Select the matching operational COM port from your system's device prompts.
4. Make sure your target supercapacitor is charged, and trigger a cycle using 🚀 PORNEȘTE TEST DESCĂRCARE.

---

Aknowledgements

This work was developed for SCSS (Sesiunea de Comunicări Științifice Studențești) 2026, incorporating AI‑based tools to support the prototyping process.