Sensor-Based Ozone Monitoring and Forecasting in a Synchrotron Radiation Laboratory Using Autoregressive Integrated...

Area of Science:

• Environmental Science
• Analytical Chemistry
• Data Science

Background:

• Ozone monitoring is crucial for safety and stable conditions in laboratory environments, especially during high-energy radiation operations.
• Enclosed facilities risk ozone accumulation, necessitating reliable monitoring systems.
• The National Synchrotron Radiation Research Center (NSRRC) requires precise ozone level tracking.

Purpose of the Study:

• To investigate short-term ozone concentration prediction using sensor data.
• To evaluate the performance of different forecasting models for ozone levels.
• To establish a framework for real-time environmental monitoring and safety management in laboratories.

Main Methods:

• Collected ozone concentration data using a UV absorption-based ozone analyzer at NSRRC.
• Analyzed sensor data as a time-series dataset under controlled conditions.
• Compared three forecasting models: Autoregressive Integrated Moving Average (ARIMA), Long Short-Term Memory (LSTM), and linear regression.

Main Results:

• The ARIMA model demonstrated superior predictive performance compared to LSTM and linear regression for the small-sample dataset.
• ARIMA achieved high R-squared values (89.5% at 5cm, 86.3% at 10cm, 81.1% at 15cm) in the Right direction.
• ARIMA also showed stable predictive performance in the Up direction, indicating its reliability.

Conclusions:

• Classical time-series models, like ARIMA, are effective for analyzing sensor data in environments with limited data.
• Integrating sensing devices with predictive analytics offers a promising approach for real-time environmental monitoring.
• The proposed framework supports enhanced safety and operational stability in laboratory settings.