Cloud Model of Slope Stability Evaluation Based on PSO‐SVR

Recently, the cloud model method has made some progress in slope stability evaluation. However, the subjective weighting method of indicator weight value is influenced by subjective factors. Also, the importance of decision‐makers to different indicators cannot be reflected in the objective weighting method. Therefore, the method of using PSO‐SVR (particle swarm optimization–support vector regression) to determine the weight value of each indicator was proposed. The slope safety factors of different combinations of evaluation indicators were simulated by using GeoStudio software. And 963 combinations were obtained. PSO‐SVR was used to predict the cloud model′s weight. The R ‐squared of the test sample set of each evaluation indicator was greater than 0.8. It was indicated that the PSO‐SVR model for each evaluation indicator had good generalization ability. The cloud model was used to evaluate the nine sections of the Life Park′s slope in Nanjing County, Zhangzhou City, Fujian Province, China. In comparison to the engineering investigation report′s analysis results, it was discovered that the cloud model method based on PSO‐SVR was accurate to 77.8%, which was a good result. Therefore, it could be concluded that the cloud model of slope stability evaluation based on PSO‐SVR is reliable.