Predictive Maintenance with an Arduino-based LoRa solution
summary
Claim: an Arduino + LoRaWAN setup detects vibration anomalies and predicts failure end to end. Hardware uses Makerban 1310 as host and Nikla Sense ME as sensor, sampling at 100 Hz; pipeline uses Edge Impulse spectral features (0–30 Hz) and trains a classifier that reaches 93% accuracy. Model compiles to an Arduino sketch, runs on the sensor, sends a 1/0 state over I2C to the host, the host uplinks via The Things Network to an IoT Cloud dashboard, and the demo of a finger-tapped fan succeeds.
AI-written summary · about
chapters (29)
- 0:00Introduction
- 0:20About Arduino
- 0:44The Arduino Platform
- 1:08What is Maintenance
- 2:47Predictive Maintenance
- 4:13Implementation
- 4:41Hardware
- 6:23Machine Learning
- 8:49IoT Cloud
- 10:23Inspect the data
- 11:21Demo
- 11:55Two approaches
- 12:51Demo goal
- 13:26Hardware setup
- 14:07Demonstration
- 14:42Training process
- 15:58Machine learning sketch
- 16:33Detection process
- 17:28Machine learning process
- 18:05Gather raw data
- 19:57Process data
- 21:55Extract features
- 23:14Train the model
- 24:43Training Sketch
- 26:30Data Forwarder
- 28:36Spectral Features
- 29:44Visualization
- 34:13Connecting to the cloud
- 36:26Dashboards
related talks
05mPredictive maintenance with sensor "BOB"
06mPredictive Maintenance
30mEnable predictive maintenance for your industrial equipment
25mPredicting the Internet of Things with Tensorflow.js
19mEnterprise AI-Powered Predictive Maintenance Platform
40mPredictive Maintenance
39mFrom a Weekend Hack, to a Future-Proof IoT Device
39mFrom a Weekend Hack, to a Future-Proof IoT Device