Aimed at the problem that pulse loss is a significant impact on mode recognition accuracy, an intelligent working mode recognition algorithm for airborne radar is proposed based on PDW reconstruction. Firstly, frequent DTOA items are extracted based on the first-order differential sequence of TOA. Secondly, the radar’s true PRI and the PRI variation pattern are derived from these frequent items. Thirdly, the computed number of pulses under conditions of different PRI are to complete PDW reconstruction. Finally, the reconstructed PDW is inputted into a pre-built temporal convolution network (TCN) for recognizing the onboard radar working mode. Such a method can exploit the temporal correlation within the PDW sequence to further capture the features of the PDW sequence. The results show that the proposed method achieves a mode recognition accuracy of over 95% when the pulse loss rate is less than 70%.