https://www.digikey.in/en/mylists/list/UVFM67HJ8A

Step 1: Collecting Components

I started by gathering all required components:

• Seeed XIAO ESP32S3 / ESP32 board
• MPU6050 vibration/accelerometer sensor
• Breadboard
• Jumper wires
• USB cable
• Laptop with Arduino IDE & Node.js

These were enough to build the full prototype.

Step 2: Wiring the Sensor

The MPU6050 was connected to the ESP32S3 using the I2C interface:

• VCC → 3.3V
• GND → GND
• SDA → GPIO 21
• SCL → GPIO 22

Step 3: Testing Sensor Readings

I uploaded a simple code in Arduino IDE to read raw accelerometer values from the MPU6050.
When I shook or tapped the sensor, I could see the values change on the Serial Monitor.
This confirmed that:

• Connections were correct
• Sensor was working
• Data was being received properly

Step 4: Collecting Vibration Data

Next, I collected vibration samples from a motor under different conditions:

• Normal running
• Loose mounting
• Misalignment
• Added load
• High vibration state

I saved these vibration readings as CSV files to use them for training the vibration detection model.

Step 5: Creating the Vibration Model

I processed the collected data and created a lightweight vibration classification model.
I mainly used:

• Time-domain vibration patterns
• Frequency features (FFT)

This model learns the difference between normal and abnormal vibration levels.

Once the model was ready, I exported it so that it could be deployed on the ESP32.

Step 6: Uploading Model to Microcontroller

I added the exported vibration detection model into my ESP32 code.
Now the microcontroller:

1. Continuously reads vibration data
2. Processes it
3. Compares it with the trained model
4. Detects abnormal vibration patterns

Whenever an abnormal vibration is detected, the ESP32 sends an alert message to the Node.js dashboard.

Step 7: Building the Real-Time Dashboard

I built a small Node.js server that receives alerts from the ESP32.
It shows:

• Alert message
• Timestamp
• Motor status
• Confidence level

This dashboard updates instantly whenever the ESP32 detects a vibration anomaly.

Step 8: Testing the Complete System

I mounted the sensor on a motor and performed various tests:

• Artificially loosening the screws
• Adding extra weight
• Changing speed
• Running the motor under stress

The system correctly detected abnormal vibration levels every time.

#include <Wire.h>
#include <MPU6050.h>

MPU6050 mpu;

void setup() {
  Serial.begin(115200);
  Wire.begin();               // SDA = 21, SCL = 22 (default for ESP32)
  mpu.begin();
  mpu.calcGyroOffsets();      // Auto-calibration
  delay(2000);

  Serial.println("accX,accY,accZ,gyroX,gyroY,gyroZ");
}

void loop() {
  mpu.update();

  Serial.print(mpu.getAccX()); Serial.print(",");
  Serial.print(mpu.getAccY()); Serial.print(",");
  Serial.print(mpu.getAccZ()); Serial.print(",");
  
  Serial.print(mpu.getGyroX()); Serial.print(",");
  Serial.print(mpu.getGyroY()); Serial.print(",");
  Serial.println(mpu.getGyroZ());

  delay(20);    // 50 Hz sampling (20 ms)
}

What this code does:

• Reads 6 DOF data (Accel + Gyro)
• Outputs CSV format (compatible with Edge Impulse Data Forwarder)
• Samples at 50 Hz (change delay() for faster/slower sampling)

Recommended Edge Impulse Impulse Settings (Perfect for Vibration/TinyML)

Sampling Parameters

Why?
1 second gives enough vibration cycles for FFT, 100 Hz satisfies Nyquist, and 200ms provides overlap for smooth inference.

Add ING a Processing Block

Select Spectral Features (FFT)  FOR  configure:

FFT Settings:

Additional Statistical Features:  RMS, Standard Deviation, Mean, Kurtosis. Skewness, Peak-to-Peak

These features improve anomaly detection significantly.

Learning Block

Choose one:

TinyML Classifier (Best for 2-class: Normal vs Abnormal)

K-means Anomaly Detection

Best for detecting unknown faults.

Model Settings

 Learning Rate: 0.005
 Epochs: 40–60
Batch Size: 32
Quantization: Int8 Optimized (for ESP32)

(3) Complete Node.js Backend + Frontend 

This creates a real-time alert dashboard using Express + Socket.io.

server.js  

const express = require("express");
const http = require("http");
const { Server } = require("socket.io");
const path = require("path");

const app = express();
const server = http.createServer(app);
const io = new Server(server);

app.use(express.json());

// Receive alert from ESP32
app.post("/alert", (req, res) => {
  console.log("Incoming Alert:", req.body);
  io.emit("vibration-alert", req.body);
  res.sendStatus(200);
});

// Serve dashboard
app.use(express.static(path.join(__dirname, "public")));

io.on("connection", () => {
  console.log("Dashboard Connected");
});

server.listen(3000, () => {
  console.log("Server running on http://localhost:3000");
});

Dashboard UI

Create a folder named public → Inside it create index.html:<!DOCTYPE html>
<html>
<head>
  <title>Vibration Monitoring Dashboard</title>
  <script src="/socket.io/socket.io.js"></script>
  <style>
    body { font-family: Arial; background:#f4f4f4; padding:20px; }
    .alert-box {
      background:#ffdddd; border-left:5px solid red; padding:15px;
      margin-bottom:10px; border-radius:8px;
    }
  </style>
</head>
<body>

<h2>Real-Time Vibration Alerts</h2>
<div id="alerts"></div>

<script>
  const socket = io();

  socket.on("vibration-alert", (data) => {
    const box = document.createElement("div");
    box.className = "alert-box";
    box.innerHTML = `
      <strong>⚠ Abnormal Vibration Detected!</strong><br>
      Device: ${data.device}<br>
      Confidence: ${data.confidence}<br>
      Time: ${new Date().toLocaleString()}
    `;
    document.getElementById("alerts").prepend(box);
  });
</script>

</body>
</html>

Run

Run these commands:npm init -y
npm install express socket.io
node server.js