Optimization of smooth blasting parameters for mountain tunnel construction with specified control indices based on a GA and ISVR coupling algorithm

• Published in: Tunnelling and underground space technology Vol. 70; pp. 363 - 374
• Main Authors: Liu, Kaiyun, Liu, Baoguo
• Format: Journal Article
• Language: English
• Published: Oxford Elsevier Ltd 01.11.2017; Elsevier BV
• Subjects: Blasting (explosive); Compressive strength; Control; Decoupling; Field test; Genetic algorithms; Geological mapping; Highway construction; Machine learning; Mountain tunnel; Mountain tunnels; Optimization; Optimization with specified control indices; Overburden; Quality; Regression models; Smooth blasting; Support vector machines; Support vector regression; Training; Tunnel construction; Tunnels; Underground construction; Support vector regression; Optimization with specified control indices; Field test; Mountain tunnel; Smooth blasting
• ISSN: 0886-7798; 1878-4364
• DOI: 10.1016/j.tust.2017.09.007

•An ISVR model which mapping the relation of smooth blasting parameters was proposed.•The optimal parameters of the ISVR model can be searched by coupled genetic algorithm.•A control optimization method for smooth blasting parameters was proposed.•The control optimization method was verified in practical tunnel engineering.•The method can be spread in similar engineering to provide smooth blasting parameters. The smooth blasting method has been widely used in the construction of mountain tunnels to decrease the volume of overbreak or underbreak and maintain the tunnel outline in the design shape. However, due to the shortcomings of existing optimization theories and the complexity of rock masses, optimizing the smooth blasting parameters in arbitrary geological conditions with specified control indices is challenging. Eighteen on-site smooth blasting experiments were conducted during the construction of the long Foling highway tunnel in China. These experimental data were used as the training samples for machine learning. By training these samples, an improved support vector regression (ISVR) model was proposed to map the relation between the inputs, comprising the geological conditions (the basic quality [BQ] grade of the rock mass, saturated uniaxial compression strength of rock, and overburden depth) and control indices and the outputs of the smooth blasting parameters, including the spacing of perimeter holes and relief holes, minimum burden and linear charge concentration of perimeter holes. A genetic algorithm (GA) was coupled with an ISVR algorithm to automatically search the optimal parameters of the ISVR model during the training process. Using the ISVR model, the optimization of smooth blasting parameters can be obtained based on certain geological conditions of surrounding rock and specified control indices, including the crown settlement, thickness of the blasting damage zone (BDZ) in which the travelling velocity of ultrasonic waves is reduced significantly due to explosive vibration, volume of overbreak or underbreak, and radial decoupling ratio. According to the application results of the Foling tunnel, the ISVR model was shown to be superior since it can outperform certain existing models. As geological conditions and control indices are comprehensively considered, the proposed ISVR model of smooth blasting parameters is expected to be more feasible and reliable and is thus recommended for use in similar tunnel projects.