Abstract: A bionic model of bat bio-sonar is established through computer simulation in this study, with a focus on investigating its orientation measurement performance. A model for simulating bat orientation perception is created via a binaural broadband interferometer, utilizing binaural interference spectra. The model incorporates a convolutional neural network (CNN) to achieve extraction and perception of orientation features. Additionally, the principle of multi-frequency point defuzzification is elaborated upon in the model. The simulation results reveal the proposed bionic model achieves a direction measurement accuracy of 0.74° and 0.116° in the rough search mode and precise search mode, respectively, at a signal-to-noise ratio of 25 dB. The model demonstrates high accuracy in orientation recognition and defuzzification capability. Compared to the technique that employs broad phase difference to determine target orientation information, the proposed model exhibits analogous direction finding efficiency but superior defuzzification capacity.