Inversion of Rock Strength Parameters Using PSO-SVM Based on Monitoring Parameters with Drilling

The geotechnical characteristics of strata are crucial indicators for conducting subterranean engineering surveys and designs in the development of underground space engineering. This paper has developed a monitoring system while drilling and suggested a set of intelligent data processing schemes to address the current issues of low automation, inadequate utilization of borehole data, and challenges in identifying the lithology of the stratigraphic interface in the drilling process of engineering survey. On-site tests were conducted in Chongqing City and Guangdong Province, China, and a large amount of effective borehole data was obtained and extracted. The particle swarm optimization (PSO) algorithm was used to optimize the parameters of the kernel function based on the support vector machine (SVM) algorithm in order to deliver the best learning and prediction outcomes. The PSO-SVM algorithm was utilized to identify the key lithologies of the stratum and to predict the uniaxial compressive strength of the rock mass. The performance shows that the prediction results are in reasonable accordance with the on-site lithology catalog and laboratory mechanical test records, and PSO-SVM is excellent in identifying the critical lithology of the stratum and predicting the critical parameters of the rock mass.