Prediction model of water inrush risk level of coal seam floor based on KPCA-DBO-SVM

To predict the risk of water inrush from coal seam floor more effectively, a prediction model of water inrush risk level of coal seam floor based on KPCA-DBO-SVM is proposed. Firstly, the risk level of water inrush of coal seam floor is graded based on the influencing factors of water inrush from coal seam floor which are determined by data of water inrush accident and related literature, and probability of water inrush in working face. Secondly, Kernel Principal Component Analysis (KPCA) is used to reduce the dimension of high-dimensional features of the influencing factors, Then, results of feature extraction are input into the DBO-SVM model. Penalty parameters and kernel parameters of Support Vector Machine (SVM) are optimized by Dung Beetle Optimization algorithm (DBO). Next, these data is mapped to high-dimensional space by SVM to separate. In this way, water inrush risk level of coal seam floor is predicted. Finally, 94 groups of sample data are selected and divided into training set and test set. Prediction results of KPCA-DBO-SVM model are compared with these that DBO-SVM, PSO-SVM and PSO-BPNN models. The results show that the accuracy of KPCA-DBO-SVM model prediction is increased by 0.18, 0.12, 0.29 respectively; the macro precision is increased by 0.16, 0.11, 0.27 respectively; the macro recall rate is increased by 0.14, 0.10, 0.28 respectively; and the Macro-F₁ is increased by 0.15, 0.10, 0.28 respectively. The KPCA-DBO-SVM model is applied to three coal mine working face to verify the stability and universality of the model whose prediction results are consistent with the actual engineering situation. Therefore, KPCA-DBO-SVM model is suitable for the risk prediction of water inburst of coal seam floor.