Dear @Daozhi.Tang,
Well, it looks like your modeling an idling turbine that goes unstable. At certain yaw errors in the idling condition, a known blade-edgewise instability exists, e.g., as discussed in the following forum topic: Designing for yaw errors using FAST. In the absence of more information, I would guess this is the issue you are running into.
FYI: You can set the fixed wind direction in either TurbSim (HFlowAng) or the InflowWind module of OpenFAST (PropagationDir), but the latter is preferred because the wind will propagate along the wind direction specified in InflowWind (PropagationDir) regardless of the setting of HFlowAng in TurbSim.
Best regards,