Coal Geology & Exploration
Article Title
Deformation prediction of open-pit mine slope based on ABC-GRNN combined model
Abstract
Accurate prediction of slope deformation in open-pit mine is an important guarantee for effective disaster early warning of slope. The traditional slope deformation prediction method is unable to characterize and comprehensively analyze the effect of various factors on slope deformation. In view of this, the combined prediction model (ABC-GRNN) of an artificial bee colony algorithm (ABC) and the optimized generalized regression neural network (GRNN) was proposed for slope deformation in open-pit mine. In this prediction model, the following 5 factors affecting the slope deformation of open-pit mine are considered comprehensively: mining disturbance, rainfall, rainfall duration, temperature and humidity. Herein, the prediction results were compared and analyzed in combination with the measured deformation data and the artificial intelligence algorithms, such as genetic algorithm improved BP neural network (GA-BPNN) and support vector machine (SVM), based on Pingshuo Anjialing open-pit mine of China National Coal Group Corp.. The research results show that ABC algorithm could quickly help GRNN to optimize the appropriate transfer parameters and effectively predict the deformation. ABC-GRNN combined prediction model reduces the average absolute error from 292.9 mm, the average absolute percentage error of 0.691 3% and the root mean square error of 338.9 mm to 25 mm, 0.043 3% and 29.5 mm, respectively, which shows that the model has higher prediction accuracy. In addition, ABC-GRNN model converges faster than other models, and the minimum mean square error can be obtained through only 7 iterations. Compared with other prediction models, this model has higher prediction accuracy, stronger generalization ability and faster convergence speed, with higher practical value.