Multi-Sensor AQI Fusion: Beyond PM2.5-Only Monitoring
A single PM2.5 sensor gives an incomplete picture — a room with low PM2.5 may still have dangerously high CO2 from occupancy or elevated TVOC from new furniture off-gassing. Multi-sensor fusion combines multiple signal sources into a single, complete Indoor Air Quality (IAQ) Score.
Recommended Sensor Stack
SDS011 (PM2.5/PM10, ±15%, 600 THB), BME680 (temperature, humidity, TVOC, 350 THB), SCD40 NDIR CO2 sensor (±50 ppm, 1,200 THB), and optional SGP41 (NOₓ index, 800 THB) — all connecting to an ESP32 or Raspberry Pi via I2C/UART.
Kalman Filter for PM2.5 Noise Reduction
SDS011 readings are noisy in Bangkok’s high-humidity environment. A simple Kalman Filter with configurable process variance (1e-4) and measurement variance (0.5) reduces noise from ±8 μg/m³ to ±2 μg/m³ at the 95th percentile — a significant improvement for threshold-based alerting.
IAQ Composite Score
Each parameter is normalized to a 0–100 sub-score (PM2.5: 75 μg/m³ = 100pts; CO2: 2,000 ppm = 100pts; TVOC: 500 ppb = 100pts) then combined with weights: PM2.5 (40%), CO2 (30%), TVOC (15%), temperature (10%), humidity (5%). This yields a 0–500 composite IAQ score usable in Home Assistant template sensors.
LINE OA Alert Automation
When IAQ > 100: immediate LINE notification + automatic air purifier activation. When IAQ > 150: Critical Alert (iOS Critical Notification sound) + closure of outdoor air intake damper. Scores below 50 generate no notifications, preventing alert fatigue.
Real-World Benefit
In a Bangkok townhouse pilot, the system detected a TVOC spike to 320 ppb from fresh paint in an adjacent unit — invisible to the PM2.5-only sensor — and automatically closed the ERV intake and notified residents 45 minutes before they noticed any smell.