Abstract: Wind turbine gearbox has high failure rate in its complex operation. In order to overcome the drawbacks of the conventional BP (Back Propagation) neural network which is easy to trap into local optimal, a new fault diagnosis model of turbine gearbox is presented based on BP neural network optimized by crisscross optimization (CSO)algorithm. Considering the instability and complexity of vibration signal of the wind turbine gearbox. The fault feature for gearbox is first extracted and a new error analysis model is established with assessment factors, CSO is then used to optimize the weights and bias of BP neural network, and finally the trained neural network is used to test the samples. The simulation and comparison with other methods show that the proposed method is effective and efficient in fault diagnosis of wind turbines.