The invention discloses an icing early warning determination method based on multi-source data fusion. The invention relates to the technical field of icing early warning, and solves the problem that in the prior art, the possibility of icing in each area cannot be predicted and the accuracy of the early warning is reduced. Through The icing prediction unit analyzes the environmental information to predict whether each area is icing, divides the area where the circuit to be detected is located into several sub-areas, obtains the environmental information, and obtains the icing prediction coefficient of the sub-area through the formula Xi, if the icing prediction coefficient Xi of the sub-region ≥ the threshold value of the icing prediction coefficient of the sub-region, the corresponding sub-region is marked as an icing sub-region, and the wind direction of the cold air is obtained, and then according to the time sequence of the cold air arriving at the sub-region , to sort each ice-covered sub-area, to predict the ice-covered area in each area, and to judge the time sequence of the ice-covered areas, so as to improve the accuracy of early warning.