Application of Weighted Distance Discrimination Criterion to Optimization of Supporting Program for Deep Foundation Pit

Abstract

Based on the principle of Mahalanobis distance discriminant analysis, the discrimination of influential degree of the different index to optimization of supporting programs for deep foundation pit is taken into account, using the weights obtained by the analytic hierarchy process to modify the Mahalanobis distance, by the new calculation method of covariance matrix, the weighted distance discreminant analysis criterion is proposed. 6 factors influencing the optimization of supporting programs are used to be estimating indexes; and a weighted distance discriminant analysis model is established by learning from a great deal of deep foundation pit cases. Comparing the model with the distance discriminant analysis model, the conclusion is obtained that the index weights can significantly reduce the ratio of mis-discrimination, and it also shows the necessity and the effectiveness of the introduction of index weights. A set of data in some deep foundation pit engineering is used to test the discriminant ability of this model. The results show that this method is an efficient one in optimization of supporting program for deep foundation pits and can be used in actual engineering.