Industrial IoT Predictive Maintenance
ESP32-based multi-sensor condition monitoring, signal filtering, and anomaly detection for pumps and motors.
Overview & Research Motivation
An applied Industry 4.0 condition monitoring system designed for industrial electric motors and pump systems. The platform utilizes ESP32 microcontrollers connected to DS18B20 digital temperature sensors, SW-420 piezoelectric vibration sensors, current transducers, voltage sensors, and optical speed encoders.
Raw telemetry is filtered at the edge using Butterworth bandpass and Kalman filters before extracting time-domain and frequency-domain features (RMS, peak-to-peak, kurtosis, crest factor, spectral power). The extracted features are transmitted via MQTT to an edge analytics dashboard for automated anomaly detection, health classification, and remaining-useful-life (RUL) estimation workflows.
The Core Systems Problem
- Unexpected motor failure in industrial plant environments causes catastrophic production downtime, yet commercial enterprise predictive maintenance systems are cost-prohibitive for small-scale manufacturing facilities.
My Specific Technical Contributions
- Designed and wired the multi-sensor hardware circuit on ESP32 microcontrollers with optical isolation.
- Implemented embedded C++ signal processing routines for digital Butterworth filtering and FFT feature extraction.
- Built MQTT telemetry publisher and real-time dashboard for plant operators with threshold-based alerting.
System Architecture & Verification Pipeline
Edge Sensing Hardware
ESP32 acquisition unit sampling vibration, current, voltage, and temperature at high sampling rates.
Digital Signal Processing
On-device Kalman and Butterworth filtering to eliminate electrical noise before computing kurtosis and RMS.
MQTT Edge-to-Cloud Broker
Lightweight pub-sub pipeline routing structured sensor payloads to condition monitoring servers.
Technical Stack & Tools
Connected Systems & Inquiries
Exploring how evolving cyber-physical state can remain independently verifiable under privacy and adversarial constraints.
A permissioned Cosmos SDK ledger for connecting supply chain state with MES and digital twin telemetry.
Encountered-type haptics system coordinating distributed mobile robotic units for realistic tactile feedback in VR.