Related Experiment Videos
AI-optimized BPNN model for port safety risk prediction and management
Fangxin Chen1, Jian Tan2, Le Cheng3
1China Communications Magazine Co., Ltd., Beijing, 100029, China.
Scientific Reports
|May 21, 2026
Summary
This study introduces an AI framework combining Bald Eagle Search (BES), Convolutional Neural Network (CNN), and Fuzzy Logic System (FLS) for enhanced port safety risk prediction. The model significantly improves accuracy and reduces false alarms for intelligent port management.
Area of Science:
- Artificial Intelligence
- Marine Safety
- Risk Management
Background:
- Traditional Backpropagation Neural Network (BPNN) models face limitations in data processing and optimization efficiency for port safety.
- Effective early warning and management of port safety risks are crucial for international maritime operations.
Purpose of the Study:
- To develop an intelligent prediction framework for port safety risks.
- To enhance the scientific accuracy of risk early warning and management efficiency in ports.
Main Methods:
- Integration of Convolutional Neural Network (CNN) for feature extraction.
- Application of the Bald Eagle Search (BES) algorithm for global parameter optimization of BPNN.
- Incorporation of Fuzzy Logic System (FLS) to manage uncertainty and fuzzy data.
Main Results:
- The ensemble model achieved high performance in port safety risk classification.
- Achieved a Root Mean Square Error (RMSE) of 0.012, accuracy of 98.5%, and Macro-F1 score of 0.982.
- Demonstrated low model inference latency of 12.5 milliseconds, suitable for real-time monitoring.
Conclusions:
- The proposed AI framework offers superior computational performance over traditional methods.
- Provides scientific decision support, reducing false alarms and improving emergency response.
- Presents a novel approach for AI application in port safety, driving intelligent and precise management transformation.
