Height of Hydraulic Fracture Zone Based on PSO_LSSVM Model

• This study develops a particle swarm optimization algorithm based on adaptive inertia weight and a least squares support vector machine model to achieve height prediction of water conducting fracture zones. Predicting the height of water conducting fracture zones can prevent mine water damage and increase the upper limit of mine mining. To achieve its effective prediction, this study selects four main control parameters for prediction, including the proportion coefficient of hard rock in the overlying strata, the inclination distance of the working face, and the thickness and depth of mining. Grey correlation analysis is performed on the collected sample data. At the same time, this study develops a particle swarm optimization algorithm based on adaptive inertia weight and a least squares support vector machine model to achieve height prediction of water conducting fracture zones. The results showed that in the training set, the RMSE, MAE, and MAPE of the research model were only 1.45m, 1.04m, and 2.17%, respectively, which were significantly lower than other comparison models. In practical applications, the absolute and relative errors between the predicted results of the research model and the actual measured values were 4.41m and -7.05%, which were significantly better than other prediction methods. This indicates that the designed fusion model has high prediction accuracy and can meet the application requirements of practical engineering. The research model provides important technical support for the safety production of mines and the protection of the surface ecological environment.