The invention discloses a GNSS spoofing interference detection method based on a convolutional neural network. The GNSS spoofing interference detection method comprises the following steps that (1) inthe signal acquisition phase, the number N<peak>, greater than an acquisition threshold, of correlation peaks in a two-dimensional matrix A generated during signal acquisition is detected, if N<peak>is no less than 2, it is considered that a spoofing signal exists, if N<peak> is less than 2,the step two continues to be conducted; and (2) a detection matrix As is obtained by intercepting data inthe +/-2 chip area on pseudo code phase axes of the correlation peaks of the two-dimensional matrix A, after data preprocessing, the convolutional neural network conducts detection training and classification, and finally a detection result is obtained. The detection method has the good detection effect and high applicability, and the detection period is in the signal acquisition stage, and complexity is moderate; and the problem difficult to detect when the pseudo phase difference delta T between the spoofing signal and a real signal is within two chips can be solved.