Prediction of the seismic capacity of RC structural walls by non-parametric multidimensional regression

The use of a non-parametric multidimensional regression is suggested for the prediction of the seismic capacity of reinforced concrete structural walls. This capacity is expressed in terms of the shear strength, drift, ductility and failure mode of the walls. For the application of the method a data base is needed which contains data on the results of tests performed on the structural elements under consideration. The data base used in the study was compiled from the available literature, and includes data from laboratory tests carried out on 262 structural walls. The method was validated by comparing the predicted and test results. A comparison with some available closed-form empirical predictions has been made. The results show that the accuracy of the predictions obtained by the proposed method is, on average, higher than that provided by existing formulae. The proposed method has several advantages when compared to closed-form empirical expressions based on standard linear regression: (1) The relationship between the input and output variables is not selected a priori by a prediction law. (2) It takes into account the currently available data base, which can be updated when new information becomes available. (3) An arbitrary number of input and output variables can be taken into account using the same data base. (4) The method can be applied directly to similar problems in structural and earthquake engineering, wherever a suitable data base is available.