Vehicle Intelligent Classification Based on Big Multimodal Data Analysis and Sparrow Search Optimization.

Area of Science:

• Intelligent Transport Systems
• Machine Learning
• Radar Technology

Background:

• Vehicle intelligent classification is crucial for Intelligent Transport Systems (ITS).
• Dynamic traffic environments pose challenges to accurate vehicle classification.
• Existing methods struggle with accuracy in complex traffic scenarios.

Purpose of the Study:

• To propose a novel vehicle classification method using big multimodal data analysis.
• To enhance classification accuracy and efficiency in dynamic traffic conditions.
• To introduce a sparrow search algorithm extreme learning machine (SSA-ELM) for vehicle type identification.

Main Methods:

• Collected road vehicle data using a new pulse coherent radar (PCR).
• Extracted vehicle length, chassis outline, and height features for sample data.
• Employed extreme learning machine (ELM) for feature learning and classification.
• Optimized ELM initial weights and thresholds using the sparrow search algorithm (SSA).

Main Results:

• The proposed SSA-ELM method demonstrated notable advantages in classification accuracy.
• The SSA-ELM exhibited superior convergence speed compared to benchmark methods.
• Successfully classified vehicle types including cars, sport-utility vehicles, and buses.

Conclusions:

• The SSA-ELM approach offers a significant advancement in vehicle classification.
• This method provides a robust solution for intelligent transport systems.
• The integration of PCR data with SSA-ELM enhances the reliability of vehicle identification.