SVM Methods For Respose Analysis Of Masonary Walls And Statically-presureed Pipe Piles

The Support Vector Machine(SVM), originated in1995, is applied in manyfields. It has special superiority in solving small-sample, non-linear andhigh-dimensional recognition problems. And can be applied to other machinelearning problems, such as function fitting. At present, the application research ofSVM is mainly concentrated in expand application, but expand application is usuallya simple input and output problem, which does not involve the research ofApplication Method and the further analysis or explanation to the calculated results.To solve the above problems, this thesis is based on the reason why few peoplechoose to accept and popularize SVM though it has better caculated result in manyproblems and whether can we use the fact that the caculated results of SVM is betterto serve scientific analysis. Combining with the test data of the masonry wall andthe static pressure pile experiment, this thesis develops the SVM AdjustmentPrediction Method (adjust the SVM model before modeling) and the SVM WeightCaculated Method (a new calculation method of index weight).In chapter2, uses SVM method derictly to predict the cracking mode ofmasonary wall panels, the failure load of masonary wall panels and the ultimatebearing capacity of static pressure pile, and also analyses the calculated results. Thefirst purpose of this chapter is to give an introduction about traditional applicationmethods of SVM, the second purpose is to establish the foudation for proposing theSVM Adjustment Prediction Method.Because SVM method predicting cracking load of masonary wall panels isunsatisfactory, Chapter3proposes the SVM Adjustment Prediction Method in orderto predict the cracking load of masonary wall panels, and applies this method topredict the ultimate bearing capacity of static pressure pile.In order to solve the problem proposed by the innovative rule of SVMAdjustment Prediction Method in chapter3, Chapter4proposes the SVM WeightCaculated Method. And proves the effectiveness of this mathematical method andcompares it with several other objective weighting methods. And this chapter alsoanalyses weights of the factors which have effect on the ultimate bearing capacity ofstatic pressure pile.The SVM in the SVM Adjustment Prediction Method and the SVM WeightCaculated Method only plays a calculation tool’s role. These two methods can notonly be applied in all works of life, but also choose various calculation tools, such asartificial neural networks, for various problems. That is, the two methods both haveflexibility and robustness.