Overview
Developed an intelligent energy monitoring system for power equipment units. Combined IoT hardware, real-time data analytics, and machine learning to optimize equipment usage patterns and identify energy waste.
Problem
Research labs run expensive equipment 24/7 without visibility into energy consumption. Challenges:
- No per-equipment energy tracking
- Equipment left running unnecessarily
- Peak demand charges during high-usage hours
- No data to inform procurement decisions
Hardware Solution
Energy Monitoring Devices
- Raspberry Pi with current transformers (CT sensors)
- 30 monitoring nodes across lab equipment
- WiFi connectivity for data transmission
- Local LCD displays showing real-time consumption
Sensor Configuration
- Split-core CT sensors for non-invasive installation
- Voltage monitoring for power quality
- Power factor measurement
- 1-second sampling rate for accuracy
Software Implementation
Data Collection
- Python scripts reading sensor data
- MQTT broker for real-time messaging
- Time-series database (InfluxDB)
- Redundant storage on edge devices
Machine Learning
- Usage pattern recognition
- Anomaly detection for equipment malfunctions
- Predictive models for optimal scheduling
- Seasonal trend analysis
User Interface
- Real-time dashboard (React + D3.js)
- Mobile app for equipment owners
- Email alerts for unusual consumption
- Weekly/monthly energy reports
Key Features
Real-Time Monitoring
- Equipment-level energy consumption
- Cost calculations based on time-of-use rates
- Power quality metrics
- Comparative analysis across similar equipment
Predictive Analytics
- Optimal scheduling recommendations
- Peak demand forecasting
- Equipment lifecycle cost projections
- Idle time detection and alerts
Reporting & Insights
- Automated monthly reports
- Cost allocation by research group
- Carbon footprint calculations
- Benchmark against industry standards
Business Impact
- $5K+ annual savings identified through optimization
- 25% reduction in peak demand charges
- 15% overall energy consumption decrease
- Identified 3 malfunctioning units consuming excess power
- Improved scheduling of high-power equipment
- Data-driven procurement for energy-efficient replacements
Environmental Impact
- 75 tons CO2 reduction annually
- Contributed to lab's sustainability goals
- Informed green building certification application
Technical Details
- Hardware: Raspberry Pi 3B+, SCT-013 current transformers
- Communication: MQTT, WiFi
- Database: InfluxDB, PostgreSQL
- Backend: Python, Flask
- ML: scikit-learn, Prophet for forecasting
- Frontend: React, D3.js, Chart.js
- Deployment: Docker containers, systemd services