Improved support vector machine regression in multi-step-ahead prediction for rock displacement surrounding a tunnel

A dependable long-term prediction of rock displacement surrounding a tunnelis an eective way to predict rock displacement values in the future. a multi-step-aheadprediction model, which is based on a support vector machine (svm), is proposed forpredicting rock displacement surrounding a tunnel. to improve the performance of svm,parameter identication is used for svm. in addition, to treat the time-varying featuresof rock displacement surrounding a tunnel, a forgetting factor is introduced to adjustthe weights between new and old data. finally, data from the chijiangchong tunnelare selected to examine the performance of the prediction model. comparative resultspresented between svmff (svm with a forgetting factor) and an articial neural networkwith a forgetting factor (annff) show that svmff is generally better than annff.this indicates that a forgetting factor can eectively improve the performance of svm,especially for time-varying problems.