The invention provides a neural network-based wind power farm wind turbine suspension electricity quantity assessment method. The method comprises the steps that a BP neural network model is built; the BP neural network model is trained by adopting training sample data on the basis of initialized basic parameters, obtained in the step 2, of the BP neural network model to obtain a trained BP neural network model; and the sum of the overall farm inside wind turbine suspension electricity quantity of a wind farm (j) at the t moment and the overall farm outside wind turbine suspension electricity quantity of the wind farm (j) at the t moment is solved to obtain the overall farm wind turbine suspension electricity quantity value of the wind farm (j) at the t moment. The method has the advantages that the wind power farm wind turbine suspension electricity quantity can be easily, quickly and precisely assessed, and therefore a power network dispatching center can better regulate the peak load and the load flow.