Application Of The Improved Support Vector Machine In Slope Stability Evaluation And Parameter Inversion

Safety problems of slope have been an important aspect of engineering geology,and it achieves significant theoretical value and practical meaning. It’s one of the keyissues to analyse the stability of slope correctly and gain the slope parameters both athome and abroad.Recently, many computing methods of artificial Intelligence are introduced intothe engineering geology and geotechnical field, which provid some new ideas forstability analysis of slope and back analysis of parameters.Support vector machine(SVM) has special advantage on solving limited samples and complex nonlinearproblems, and it has been received extensive attention in slope engineering as aresearch focus in the intelligent method. Support vector machine improved based on acuckoo search algorithm was adopted in stability analysis of slope and back analysisof parameters. The main works in this paper are as follows.(1)The cuckoo search algorithm has global searching ability and does not easy tofall into a local optimal solution, and this method has relatively strong adaptability forits less undetermined parameters and simpleness. The applicable ability was verifiedtaking back analysis of parameters in one-dimensional groundwater system.(2)Focus on the selection of kernel function parameter and error warning factor,Support Vector Machine was improved based on the cuckoo search algorithm and thealgorithm was improved using the global optimization abilities of cuckoo searchalgorithm.(3)Support vector classification algorithm was used in preliminary stabilityanalysis of test samples, and the improved Support Vector Machine was used in safetycoefficient calculations of test samples. The result shows that the method isreasonable and reliable.(4)Taking a simple slope of dam for example, sample datas with40water headsand permeability coefficients were created with uniform design in SEEP of GeoStudio.The permeability coefficients were back analysed based on the improved SupportVector Machine. Comparing with back analysis value and theoretical value, and the (5)result shows that it’s possible to back analyse the permeability coefficients ofslope of dam with the improved support vector machine.