The support vector data description (SVDD) is a kind of single-value classification method, by which a single-value classifier can be built by using its normal state data samples even if the fault samples are lacking, thus revealing its normal operation. The orthogonal wavelet transform (OWT) has good performance for extracting the shock elements of a non-stable signal. We propose a new state evaluation method that uses the SVDD and the OWT and use the OWT to extract the peak-peak values of various detail signals, which are in turn used as input parame-ters of the classifier. We build the classification model of the classifier with the SVDD method to carry out the quan-titive evaluation of the state of the machine. We also use our method to do experimental analysis of the pitting faults on the inner ring of a rolling bearing and establish the quantitive indicators for evaluating its worsening perform-ance; the experimental results show that our method is effective.