It is generally difficult to obtain the data of circuit breaker failures. Considering this situation, an experiment platform for circuit breaker failures is built to obtain the failure data. To overcome the deficiency of traditional monitoring, which is based on single feature and thus is hard to judge the statues of the circuit breaker, this paper proposes the method of multi-dimensional mapping. An algorithm based on mathematical morphology is proposed to filter current signals, and the feature values reflecting the waveform differences are extracted. For vibration signals morphological wavelet transform is used to extract their feature values, i.e. the energy of various frequency bands. Support vector machine is used for combining the feature values of current signals and vibration signals to form multi-dimensional mapping and to diagnose the status of circuit breakers. A great number of experiments are conducted and the results prove the effectiveness of the proposed method