Structural strength analysis of loess based on genetic algorithm and the prediction and optimization of coal mine gas emission

Loess is a special kind of broken granular material, and its structure is closely related to its physical and mechanical properties. From the macroscopic and microscopic study of loess to the multi-step comprehensive study of the development process of loess, in the study of the effect of medium on the structural strength characteristics of loess, many scholars have deeply discussed. At present, the main research direction of loess deformation and fracture is to clarify the nature and mechanism of macro-deformation of loess, especially from the change of microstructure. Therefore, it is of great scientific significance to study and explore the relationship between structure and soil shear strength, which can also provide guidance for specific engineering practice theory. In this paper, the deformation and microstructure characteristics of loess are analyzed according to the water content, and the Q3 loess in Y City is mainly studied. Based on genetic algorithm, the relationship between soil structure and deformation and failure characteristics is studied. However, JRBF network has many uncertain parameters, so it is difficult to determine the optimal value. Therefore, using RBF neural network to predict mine gas emission has certain limitations. In this paper, we use genetic algorithm to optimize and simplify the structure and parameters of RBF network and establish a GA-RBF prediction model to predict the amount of gas emission. The model keeps the best value of network parameters based on the principle of advantages and disadvantages and uses transformation and other operations to select escape algorithm. The simulation results show that the prediction accuracy of ca-rbf is better than that of traditional RBF, and the training speed is also greatly improved. According to the research of genetic algorithm, it is applied to loess structure and mine gas emission, which can effectively improve the loess structure and mine gas emission. © 2021, Saudi Society for Geosciences.