The invention discloses a mixed auxiliary variable separation and dimension reduction method based on independent subspace false neighboring point discrimination. The mixed auxiliary variable separation and dimension reduction method based on the independent subspace false neighboring point discrimination is characterized by including the following steps: 1, determining n original auxiliary variables probably related a primary variable, and acquiring value data of the n original auxiliary variables and the primary variable to form a sample set; 2, respectively calculating weighing values of the n original auxiliary variables through the independent subspace false neighboring point discrimination; 3, forming an original auxiliary variable sequence; 4, utilizing a least square regression method to build a model, and determining the best auxiliary variable according to a minimum mean square error (MSE); 5, obtaining separated independent signal soft measurement model. The mixed auxiliary variable separation and dimension reduction method based on the independent subspace false neighboring point discrimination can find a variable set containing mixed auxiliary variables on the basis of the optical modeling effect to perform separation, achieves dimension reduction, simplifies auxiliary variable information, simultaneously reduces model complexity, and improves soft measurement effectiveness.