In order to achieve rapid identification of Wolfberry from different geographical origin, an electronic tongue identification method based on Hilbert-Huang transform (HHT)-Linear Discriminant Analysis (LDA) was proposed. Taking the four geographical origins (Ningxia, Xinjiang, Gansu and Qinghai) of wolfberry as experimental materials, the voltammetry electronic tongue was used to collect the “fingerprint” information of different geographical origins, and then the Ensemble empirical modal decomposition (EEMD) was used to carry out the original signal of the electronic tongue. The scale decomposition obtained a set of intrinsic mode functions (IMF), and finally its singular spectral entropy and Hilbert marginal spectrum were collected as feature vectors. On this basis, LDA was used to establish a nonlinear combination prediction model for the production area. The experimental results showed that HHT-LDA was better than the algorithm of Feature Point Extraction (FPE), Principal Component Analysis (PCA) and Discrete Wavelet Transform (DWT). The overall classification accuracy and kappa coefficient of Wolfberry from unknown origin reached 98% and 0.973, respectively, indicating that the model had a good identification performance.