Mathematically Improved XGBoost Algorithm for Truck Hoisting Detection in Container Unloading.

Area of Science:

• Engineering
• Computer Science
• Security Systems

Background:

• Conventional truck hoisting detection methods in ports face challenges with high costs, weather sensitivity, and low accuracy.
• There is a need for improved, non-intrusive solutions for effective port security and container handling.

Purpose of the Study:

• To propose and evaluate a novel, non-intrusive approach for truck hoisting detection.
• To enhance the accuracy and reduce the cost of detecting abnormal container hoists in port environments.

Main Methods:

• Utilized Hall sensors to collect electrical signals (voltage and current) from truck hoisting operations.
• Developed a mathematical model to process and augment physical information from electrical signals.
• Employed an extreme gradient boosting (XGBoost) model, trained on processed data, for abnormal hoist identification.

Main Results:

• The proposed approach demonstrated excellent performance in experimental trials at multiple stations.
• Achieved a false positive rate not exceeding 0.7% and zero false negatives.
• The XGBoost model showed improved performance in identifying abnormal hoists.

Conclusions:

• The non-intrusive detection method effectively reduces costs and improves accuracy in container hoisting detection.
• This approach offers a viable solution for enhancing port security and operational efficiency.