Plateau underground engineering geological safety risk assessment based on Bootstrap-SVM-BPNN

Underground engineering has become an extremely complex, highly uncertain and dynamic system problem due to the influence of various complex natural geological conditions and many unknown factors. Based on Bootstrap-SVM-BPNN optimization model, this paper predicts the underground engineering situation of plateau and analyzes the geological safety risk. The principle of Bootstrap algorithm and SVM algorithm is introduced, and the possibility of combining them is realized. The pseudo-sample set was built based on Bootstrap method, and the SVM model was used to train and predict it, and the mean and variance of the predicted value were calculated. The random error data set is constructed, and MEA-BPNN is used to train and predict the deviation of random error, so as to realize the hybrid algorithm of Bootstrap-SVM-BPNN geological safety prediction. Comparing and analyzing various models, the mixed learning model in this paper has the best prediction effect, R2 is up to 98%, RMSE value is 7.2. Finally, the Bootstrap-SVM-BPNN model was applied to practical engineering, and it was found that micro-filled and semi-filled types were still the most common positive samples in the real positive samples. The number of positive samples correctly predicted by interference type increased by 6, and AP increased by 56.15%. The accuracy of the micro-filling model was 41.41%, and the recall rate was 89.13%. The accuracy of the semi-filling type is 57.45%, and the recall rate is 96.43%. The use of the Bootstrap-SVM-BPNN optimization model for geological exploration can truly predict various disasters and reduce the negative impact of risk factors, thereby overall lowering the probability level of disaster occurrence.