Here a general method is proposed for type identification of unknown thermocouple. The thermo e.m.f. vs. temperature characteristic data of different known types and unknown type of thermocouple sensors are complied to form feature matrix. The Principle component analysis method is applied to this feature matrix to reduce the dimension of feature variables of each type. The outcome of this PCA is PCA score data of dimensions (n×2), where n being the total of types. From the score plot and minimum distance classification method the similarity of the unknown type to any of the known class types is determined.