Smart Manufacturing IoT-Cloud Monitoring Dataset for Predictive Maintenance

Dataset Overview

This comprehensive dataset contains real-time IoT sensor data collected from industrial machines in a smart manufacturing facility. Designed for Industry 4.0 applications, it enables research in predictive maintenance, anomaly detection, and cloud-based monitoring systems for optimizing manufacturing operations and preventing unexpected equipment failures.

Key Features

• Real-time sensor data from multiple industrial machines
• Multi-modal measurements: temperature, vibration, pressure, and acoustic sensors
• Machine operational status and performance indicators
• Cloud integration metrics for IoT-to-cloud data flow analysis
• Labeled data for normal operation and various fault conditions
• High-frequency sampling for detailed temporal analysis
• Production line context including machine IDs and process stages
• Suitable for both supervised and unsupervised learning approaches

Data Structure

The dataset is structured with the following key components:

• Machine Identifiers: Unique IDs for different equipment and production lines
• Sensor Readings: Temperature (°C), Vibration (mm/s RMS), Pressure (bar), Acoustic signals (dB)
• Operational Metrics: Machine speed (RPM), Load percentage, Power consumption (kW)
• Temporal Information: Timestamps with millisecond precision
• Maintenance Labels: Normal, Warning, Fault, Critical status indicators
• Fault Types: Bearing wear, misalignment, imbalance, overheating
• Cloud Metrics: Data transmission latency, upload frequency, connectivity status
• Production Context: Shift information, product type, batch numbers

Data Collection Method

Data was collected from IoT sensors installed on industrial machinery in an active manufacturing facility. Wireless sensor nodes transmit real-time measurements to edge gateways, which aggregate and forward data to cloud platforms for storage and analysis. The dataset includes both normal operating conditions collected during regular production and fault scenarios introduced during controlled maintenance experiments.

Research Applications

• Predictive maintenance model development for reducing downtime
• Real-time anomaly detection in manufacturing processes
• Cloud-based IoT architecture evaluation and optimization
• Digital twin development for virtual factory simulation
• Condition-based monitoring system design
• Fault diagnosis and classification in rotating machinery
• Production efficiency optimization through sensor analytics
• Edge-to-cloud computing tradeoff analysis for IIoT

Machine Learning Use Cases

• Multi-class classification for fault type identification
• Binary classification for fault detection (normal vs. abnormal)
• Time series forecasting for remaining useful life (RUL) prediction
• Unsupervised anomaly detection using autoencoders and isolation forests
• Deep learning (CNN, LSTM, Attention mechanisms) for sequential patterns
• Sensor fusion for improved fault diagnosis accuracy
• Reinforcement learning for adaptive maintenance scheduling
• Transfer learning across different machine types and factories
• Real-time streaming analytics for immediate fault alerts