A real-time online aircraft AI neural network system comprises a convolution fixed point sliding IP core, a pooling compression quantification IP core and a full connection compression fusion IP core.i + 1 layers are arranged in total, and each convolution fixed-point sliding window IP core and the pooling compression quantization core are the same in structure. A sensor signal 1 is main inertialnavigation data with the highest priority for the aircraft, and the sensor signal 1 is independently input into a unit network layer and needs to control input of 1 to i + 1 layers during second convolution. The aircraft heterogeneous sensor data is used as the input of the system; and the identification result is used as system output. a convolution sliding window IP core can rapidly achieve extraction of data features through a sliding window excluding redundant data; the pooling compression quantification IP core can improve the system execution efficiency by using a compression and quantization technology; the full-connection compression fusion IP core is compressed and fused after being deleted and quantized, and the output meets the high-reliability and low-power-consumption intelligent integrated processing requirement for a large amount of heterogeneous input data in the real-time flight process of the aircraft.