Slope Stability Evaluation Based on Selective Ensemble of Improved D-S Evidence Theory
Aiming at the difficulty of selecting slope stability prediction algorithm and the high risk of misjudgment of a single model,a slope stability evaluation method based on the selective ensemble of improved D-S evidence theory is estab-lished to provide a methodological basis for the preliminary evaluation of slope stability.Based on the main influencing factors of slope stability,a large-scale slope stability evaluation dataset was constructed using the limit equilibrium method.Introducing a base learner selection technique based on margin distance minimization to enhance the generalization ability of selective en-semble model.Propose an improved D-S evidence theory to fuse base learner information,reduce uncertainty and fuzziness in the decision-making process of selective ensemble model,and solve the problems of existing slope stability evaluation models that are prone to misjudgment and inconsistent results.The simulation experiment results show that the improved D-S evidence theory selective ensemble method can directly and objectively evaluate the slope stability state without complicated numerical modeling and calculation iteration process,and give the instability probability of slope from the perspective of information theo-ry.Compared with the traditional machine learning method,this method effectively improves the prediction accuracy of slope stability,reduces the uncertainty of the prediction results,and realizes the wide-scale slope stability evaluation with fast speed,high accuracy and good robustness.
slope stabilityD-S evidence theoryensemble learningselective ensembleinstability probability
万方数据
维普
