Applications of AI and ML Techniques to Predict Backbreak and Flyrock Distance Resulting from Blasting

Abstract

Flyrock due to blasting has always remained risk and adverse environmental impact due to the past history of accidents with serious bodily injuries, fatalities, and damage to the properties. Backbreak is likely one of the causes for future flyrock during blasting. Hence, prediction of flyrock and backbreak is crucial. Various factors causing flyrock have been identified such as geology, rock mass properties, drill and blast design, impact of previous blast, failure to identify uncontrollable factors, personal and task factors, environmental factors, blast management practices, and lack technological tools. Input parameters based on blast design, rock mass properties, and explosives related factors such as powder factor, maximum charge per delay, charge per meter play crucial role in prediction of flyrock and backbreak. Empirical equations were initially developed based on blast design parameters for prediction of flyrock and backbreak. Statistical models as well as empirical equations do not have required accuracy for prediction of flyrock and backbreak. Various artificial intelligence (AI) techniques for prediction of flyrock and backbreak developed during last decade were reviewed. Artificial neural network, fuzzy interface system, and support vector machine were found common and useful. In addition, hybrid AI techniques showed better accuracy in prediction of flyrock and backbreak. Practical applications of AI techniques and need for future research are also discussed.