Predicting peak particle velocity in pre-splitting of gas-producing devices using improved particle swarm...

Area of Science:

• Geotechnical Engineering
• Computational Intelligence
• Civil Engineering

Background:

• Blasting construction necessitates precise vibration velocity prediction to safeguard infrastructure.
• Existing models often lack the accuracy required for effective vibration control.

Purpose of the Study:

• To develop a highly accurate prediction tool for peak particle velocity (PPV) in blasting construction.
• To introduce an Improved Particle Swarm Optimization (IPSO) Backpropagation (BP) Neural Network Model (IPSO-BP) incorporating frequency effects.

Main Methods:

• The IPSO-BP model was developed and trained using blasting construction data.
• Performance was evaluated against traditional BP neural networks, empirical formulas, and other advanced algorithms like GA-APSO-BP, GWO-SVR, MFO-BP, and RUN-XGBoost.
• Comparative analysis included training set performance, testability, and overall prediction accuracy.

Main Results:

• The IPSO-BP model demonstrated superior prediction performance compared to all benchmark models.
• The model showed high accuracy and reliability across training and testing datasets.
• Frequency (f) was successfully integrated, enhancing predictive capabilities.

Conclusions:

• The IPSO-BP model offers a reliable and accurate solution for predicting PPVs in blasting construction.
• This advancement can significantly improve the safety protocols for tunnels and surrounding structures.
• The study validates the effectiveness of integrating IPSO with BP neural networks for complex engineering predictions.