Design and optimization of haze prediction model based on particle swarm optimization algorithm and graphics...

Area of Science:

• Environmental Science
• Computer Science
• Data Science

Background:

• Fine Particulate Matter (PM2.5) pollution is a major global environmental crisis impacting health and ecosystems.
• Accurate PM2.5 forecasting is essential but challenged by data volume and model accuracy limitations.
• Existing deep learning and Support Vector Regression (SVR) models have drawbacks in optimization, training time, and data handling.

Purpose of the Study:

• To develop an optimized and efficient model for predicting PM2.5 levels.
• To overcome the limitations of traditional and deep learning models in haze prediction.
• To enhance the speed, stability, and reliability of air quality forecasting.

Main Methods:

• Developed CUDA-based code to optimize the Support Vector Regression (SVR) algorithm.
• Combined SVR with intelligent algorithms: Genetic Algorithms (GA), Sparrow Search Algorithm (SSA), and Particle Swarm Optimization (PSO).
• Implemented a heterogeneous parallel computing approach using Central Processing Unit-Graphics Processing Unit (CPU-GPU).

Main Results:

• The combined intelligent algorithms with CPU-GPU parallel computing significantly improved SVR efficiency.
• The Particle Swarm Optimization-Central Processing Unit-Graphics Processing Unit-Support Vector Regression (PSO-CPU-GPU-SVR) model demonstrated substantial speed improvements (6.21-35.34 times faster than PSO-SVR).
• The PSO-CPU-GPU-SVR model achieved high accuracy, enhanced stability, and reliability in haze prediction.

Conclusions:

• The PSO-CPU-GPU-SVR model represents a breakthrough in efficient and accurate PM2.5 prediction.
• This approach offers significant advantages over existing methods for real-time air quality monitoring.
• The findings provide valuable insights for environmental decision-making and public health protection.