The thermocouple is widely used in industries for precise temperature measurement. The primary constraint is its nonlinearity. Sensor linearization techniques based on multilinear model approach are popular due to the simplicity and transparency of local linear models (LMs). However, the decomposition of the nonlinear range into suitable LMs is essential for the correct representation of its nonlinear characteristic. This article presents a novel and systematic data-driven approach using an included angle method to determine optimal LMs. Later, the Takagi-Sugeno (T-S) fuzzy interpolation technique is effectively used to combine the LMs. The efficacy of the proposed method is demonstrated in a simulation environment using J-type thermocouple for a wide range of 0 °C-760 °C and real-time linearization of K-type thermocouple over a scale of 0 °C-320 °C using Arduino platform. The results show that the proposed technique reveals satisfactory performance indices compared with other popular methods. © 1963-2012 IEEE.