Study On Blasting Fragmentation Prediction Of Rock-fill Dam Based On Random Forest Regression Algorithm

In the process of blasting quarry of dam materials,fragmentation control is one of the most important means to ensure the quality of dam construction.Blasting fragmentation not only affects the efficiency of excavation and loading machinery,but also affects the comprehensive cost of blasting construction,but also affects the comprehensive cost of blasting construction.At present,several shortcomings are left unsolved such as less consideration about factors of rock mass structure,low accuracy of model prediction and poor generalization ability of model in the study of blasting fragmentation prediction,which make it difficult to control the fragmentation of rockfill dam quickly and accurately.Aimming at the shortcomings of the current fragmentation prediction model,the problem of the structural property of rock mass face before blasting is studied,the structural property analysis model of rock mass face based on the transfer learning method is put forward,the fragmentation prediction model based on the random forest regression algorithm and the blasting management and fragmentation prediction analysis system are established,and the system is applied and verified through engineering examples.The acheievements in this study are as follows:(1)Aiming at the problem of rock blasting,rock blasting mechanism,rock blasting parameters and rock structure property analysis methods are introduced;The rock mass face property before blasting construction based on deep learning method is analyzed,and the transfer learning pre-training model and the rock mass face images are used to build the new calculating model in order to provide a scientific evaluating standard and effective reference for the selection of rock information parameters in blasting parameters.(2)Through the analysis of the traditional prediction model of blasting fragmentation,the shortcomings of the traditional model are summarized;Based on the knowledge of artificial intelligence and other emerging fields,the prediction problems of rock-fill dam blasting fragmentation by using random forest regression algorithm are analyzed,modeled and evaluated;Through the performance comparison of several commonly used fragmentation prediction models,the consistency,feasibility and superiority of the random forest regression model are verified.(3)Using the programming mode of C# language and MATLAB dynamic link library files,aiming at the fragmentation control problem in the blasting quarrying construction of rock-fill dam material,the database structure and system function modules are designed,the blasting management and block prediction analysis system is formed,and the informationalized,intelligentized analysis and management of blasting block prediction are realized.(4)In view of the construction process of rock-fill dam material blasting,through in-site sieve test and system test,the blasting construction of rock-fill dam material is simulated and analyzed,the in-site blasting construction fragmentation is predicted accurately,and the accuracy of the results meets the engineering requirements.