Using multi-sensor data fusion for mechanical fault diagnosis can obtain higher accuracy， but traditional shallow fusion models have a weaker ability of nonlinear mapping and feature representation for complex data sets. Therefore， a multi-sensor data fusion method based on deep convolutional neural network （DCNN） for mechanical fault diagnosis is proposed. Firstly， multi-sensor vibration signals are extracted separately， and the feature vectors are obtained as one-dimensional feature surfaces to construct a set of multi-sensor feature surfaces， which is used as input of DCNN. Then， adaptive hierarchical fusion and extraction of multichannel features are realized with the deep network structure. Last， diagnosis results are output by softmax regression classifier. The experiment results show that this method has a higher ability of fault classification and identification， and a good robustness and adaptability under different noisy environments. This study provides a theoretical base for solving the multi-sensor information fusion problem of complex mechanical fault diagnosis.