Study on Optimization of Hole Location and Blasting Parameters of Carbon Dioxide Gas Explosion Assisted By Machine Learning

As a non-traditional blasting method, carbon dioxide (CO2) gas blasting technology breaks rocks or soil through high-pressure gas generated by rapid gasification of liquid CO2. By collecting geological data, blasting parameters and blasting effect data, an efficient machine learning model is trained to predict and optimize the position of gas explosion holes and blasting parameters. Random forest regression model (RFR) is selected as the main machine learning model because of its excellent performance in dealing with complex geological data and optimization of blasting parameters. Through correlation analysis, the characteristic variables that significantly affect the blasting effect are screened out, and the data are processed by Z-score standardization method. The experiment was carried out in Wandong Mountain open pit mine, which verified the application effect of CO2 gas blasting hole technology, especially its safety and rock crushing efficiency. Data processing and analysis show that the machine learning model can accurately predict the vibration under different blasting parameters, which provides a scientific basis for parameter optimization. The optimized blasting parameters can control the vibration more effectively and keep good rock crushing effect. The research results show that CO2 gas explosion technology has obvious advantages in vibration control, especially suitable for blasting near buildings or sensitive areas, which provides practical experience and data support for the application of CO2 gas explosion technology in open pit mines.