The embodiment of the invention provides a multi-mode biological feature fusion method and device, and relates to the field of artificial intelligence. According to the embodiment of the invention, the identity of the to-be-identified user can be identified by using various biological characteristics. The method comprises the following steps: acquiring at least two biological characteristics of ato-be-identified user; R; respectively extracting characteristic parameters of biological characteristics contained in the at least two biological characteristics; P; performing normalization processing on the characteristic parameters of the biological characteristics contained in the at least two biological characteristics; carrying out weighted fusion on the characteristic parameters of the normalized biological characteristics to generate fusion characteristic parameters; and according to the fusion feature parameters, identifying the identity of the to-be-identified user by using a convolutional neural network. The method is applied to biological recognition.