# Nonlinear correction of thermocouple based on neural network

......
1 Introduction Thermocouples are widely used in the field of temperature measurement because of their simple structure, easy manufacturing and wide temperature range. However, the thermocouple nonlinear correction problem (also called linearization) seriously affects the temperature measurement accuracy. Both the international and domestic calculation standards give a thermoelectric potential-temperature relationship table, that is, a thermocouple index table. The conversion relationship can be based on the look-up table method, but this method is very inconvenient in the application process. A better method can use the neural network technology to establish the corresponding mathematical model and improve the linearity of the thermocouple. The neural network has powerful memory capacity, high-speed parallel computing capability and nonlinear transformation characteristics, which can be re-learned at any time and can be used to effectively correct the nonlinearity of the system. 2 Thermocouple Nonlinearity Thermocouples have a wide variety of types, specifications, and structures. There are almost all serious nonlinear problems, and the output signal has a nonlinear relationship with the measured temperature. This brings errors to the measurement results. In this paper, the nonlinear correction of nickel-chromium-nickel silicon thermocouple (K type) is performed by neural network technology. The structure diagram of the K-type thermocouple is shown in Figure 1. Figure 1 Thermocouple structure diagram The measured temperature t is measured by a thermocouple, and the corresponding thermoelectric potential E(t, 0) is output. K type heat(View original)

Previous£ºPhilosophical Thinking on Bridge Aesthetics in China £ü
Next£ºMeasuring RF signals with the ADL5902 TruPwr detector