Industrial Vibration Monitor with Web Dashboard

Rotating machines rarely fail without warning — they warn in vibration first. Rising unbalance, developing misalignment and spalling bearings each leave a characteristic fingerprint in the vibration spectrum, but most small plants have no one reading spectra and learn about the fault when the machine stops. Professional analyzers are expensive and need trained analysts. This project puts the core of vibration analysis on student hardware: a MEMS accelerometer on the bearing housing, edge-computed FFT spectra, severity judged against the ISO 10816 standard, and a dashboard that marks the classic fault frequencies for the user. It teaches the signal chain from acceleration to diagnosis honestly, including what a low-cost MEMS sensor can and cannot resolve.

1. The MEMS accelerometer is stud-mounted (or magnet-mounted in the demo) on the drive-end bearing housing of the demo motor.
2. The ESP32 samples all three axes at 1.6 kHz in fixed blocks and applies a Hanning window to each block.
3. A 1024-point FFT converts each block to a velocity spectrum; the firmware extracts overall RMS velocity and the amplitudes at 1X, 2X and the configured BPFI.
4. The overall velocity is compared with the ISO 10816 zone limits for the machine group; zone entry raises the corresponding alarm level.
5. The DS18B20 bearing-temperature reading is sampled alongside and checked against its limit with a rising-trend watch.
6. Every 10 minutes a compact summary (overall velocity, zone, key amplitudes, temperature) publishes over MQTT; the dashboard renders the live spectrum, waveform and zone gauge.
7. The student records the healthy baseline at commissioning and documents the first deliberately-introduced fault (e.g. an unbalance mass) and its 1X signature for the report.

Tech stack:

• ESP32 (controller)
• ADXL345-class MEMS accelerometer
• DS18B20 temperature sensor
• Edge FFT signal processing
• MQTT telemetry
• Web dashboard (HTML, CSS, JavaScript)
• Arduino IDE (C/C++ firmware)
• 16x2 LCD for on-node readout

• [1.6 kHz triaxial sampling] An ADXL345-class MEMS accelerometer streams vibration data at 1.6 kS/s, fast enough to resolve spectra up to 500 Hz on a 50 Hz machine.
• [Edge FFT] A 1024-point FFT runs on the controller, so spectra are computed at the machine and only compact results travel over the network.
• [ISO 10816 severity gauge] Overall velocity in mm/s RMS is judged against ISO 10816 zones A (good) through D (danger) for the machine's power group.
• [Fault-frequency markers] The dashboard marks 1X (unbalance), 2X (misalignment) and the bearing's BPFI line, turning a raw spectrum into a readable diagnosis.
• [Live time waveform] A scrolling waveform view shows the raw vibration signal for classroom demonstration of periodic vs impulsive content.
• [Bearing temperature channel] A DS18B20 tracks bearing temperature alongside vibration, since heat plus vibration rise together in failing bearings.
• [Trend upload] Spectra summaries and overall values publish over MQTT every 10 minutes for long-term trend archives.
• [Buyer-run baseline] A commissioning procedure records the machine's healthy baseline spectrum, so every later alarm is judged against the student's own machine.

• Vibration sensor node: ESP32, MEMS accelerometer, temperature sensor, LCD, mounting hardware
• Demo motor rig with bearing-housing sensor mount
• Firmware with sampling, windowing, FFT and ISO zone logic
• Web dashboard source (spectrum, waveform, zone gauge, alarms)
• Baseline-recording and fault-introduction procedure for the report
• Bearing fault-frequency worksheet (BPFI from buyer-entered geometry)
• Project report PDF (vibration theory, FFT, ISO 10816, fault signatures)
• PPT presentation for final review
• Viva Q&A preparation document (sampling, aliasing, windowing, 1X/2X, bearing frequencies)