Research on thermistor signal conditioning circuit based on genetic algorithm-BP neural network

It is an important method to detect ocean turbulence by using rapid temperature gradient change, and the accuracy of temperature measurement is an important factor affecting the quality of turbulence measurement. However, the current turbulence temperature sensor is a FP07 thermistor with negative temperature coefficient, which is a nonlinear element and affects the quality of turbulence observation. Faced with this problem, this paper proposes a low-cost two-stage linearization scheme based on thermistor temperature sensing. The first stage is a signal analog conditioning circuit based on operational amplifier, which is used to output the voltage signal and perform the first-stage linear amplification. In the second stage, the two-stage nonlinear compensation of BP neural network based on genetic algorithm optimization is used to make up for the shortcomings of the nonlinear output of the thermistor furtherly. The verification shows that the linearizer realized by the method in this paper has good linearity and accuracy in the temperature range of-2 °C -32 °C, up to 0.005 °C accuracy range. Compared with the standard BP network and the existing linearization method, the results are obviously better, which proves the feasibility and superiority of the proposed method in the linearization of the thermistor.