Bearing fault is an important issue which can lead to the failure of rotating machinery systems especially in wind turbines. There are some limitations on traditional fault diagnosis algorithms. Many of these algorithm process a single signal from each sensor to diagnose faults which is completely ignores interference between components. To address these limitations and make improved use of processing multi-dimensional signal inputs, our SiamFDnet based on deep learning method is proposed. In this study, we focus on wind turbine bearing fault diagnosis using multiple signal inputs. Our model is designed based on the Siamese network, which learns through whether the input pairs are similar or different. The siamese network, one of the few shot learning techniques also contributes to the improvement of class imbalance problem, a chronic problem in the field of fault diagnosis. Experimental results using Doosan Wind Turbine bearing dataset shows that multiple signals can be viewed and diagnosed comprehensively. At the conclusion, the effectiveness of SiamFDnet was verified by the Case Western Reserve University (CWRU) bearing fault diagnosis benchmark dataset.