The invention discloses a surgical navigation system of mandibular angle osteotomy, which comprises the following steps: S1, establishing an osteotomy surface prediction model learning version based on a multitask convolutional neural network and according to the relevant data of past mandibular angle osteotomy patients, carrying out training by using the relevant data of a newly enrolled group ofmandibular angle osteotomy patients to obtain a stable osteotomy surface prediction model, and then superimposing data sets in a danger area to obtain an osteotomy surface prediction model stabilization version; S2, inputting the relevant information of mandibular angle osteotomy patients into an osteotomy surface prediction model, and predicting the change range of a post-operation face 3D effect; S3, drawing a projected three-dimensional image, which can be seen through and fit with a surgical field real time image, on the screen of a lens according to the maximum osteotomy amount and pre-operation CT; and S4, carrying out function superimposition on an osteotomy surface prediction model continuously, testing and improving continuously, and perfecting the surgical navigation system. Thesystem predicts a post-operation effect through establishing models, real-time rendering is performed during operation, and the operation accuracy is improved.