The invention disclose an HRRP (High Resolution Range Profiles) identification method based on a Gaussian cause-selecting Boltzmann machine, and puts forward a method for applying the Gaussian cause-selecting Boltzmann machine to the robust identification of the noise of the HRRP mainly by aiming at the deficiencies of an existing HRRP identification method. The method comprises the following implementation steps that: (1) preprocessing data; (2) constructing a Gaussian cause-selecting Boltzmann machine network; (3) initializing network parameters; (4) inputting a preprocessed HRRP training sample into the Gaussian cause-selecting Boltzmann machine; (5) carrying out iterative sampling on a hidden layer and a selection factor; (6) updating a visible layer; (7) updating the network parameters; and (8) training a linear SVM classifier, and outputting an identification result. By use of the method, dependency on HRRP data noise prior is lowered, meanwhile, the useful characteristics of original data can be effectively extracted, the method exhibits good noise robustness, and HRRP identification accuracy is improved.