A method of predicting a central fishing ground of flying squid family Ommastrephidae, includes three steps of setting spatial and temporal dimension, setting environmental factor, and establishing a central fishing ground prediction model. The spatial and temporal dimension includes three levels of spatial dimensions, and two levels of temporal dimensions of week and month. An SST is selected as a main environmental factor, and two environmental factors, i.e., SSH and Chl-a, are selected as a supplement. The environmental factors include four situations. According to the setting situations of the spatial and temporal dimension and the environmental factor, a set of sample schemes of 24 situations is established using permutation and combination method. An error backward propagation neural network model is established, wherein an input layer inputs data of the sample scheme set, and an output layer outputs a CPUE or a fishing ground grading index converted from the CPUE.
