Abstract: The accuracy of pulmonary function detection depends on the measurement accuracy and calibration method of flow sensor.Conventional calibration approaches struggle to account for the sensor’s nonlinear behavior across both high- and low-flow ranges, often leading to cumulative clinical errors. This study introduces a high-precision segmented calibration method designed for pulmonary function testing. Firstly, the standard flow generator is used to obtain the calibration data points of each flow rate section within the range of the sensor (0−14 L/s). According to the nonlinear characteristics of the sensor input-output, the flow range is divided into four intervals, and the linear or cubic polynomial regression model is established in each interval.Then, a 3 L calibration cylinder was used to calibrate the low flow velocity of about 1L/s, and the calibration coefficient K was obtained by combining the regression model and the calibration algorithm. The K value was used to convert the flow, and then the global error was corrected by the piecewise regression model, and the key pulmonary function indicators were calculated after correction.Finally, short-term, long-term and inter-group performance verification experiments were carried out on pulmonary function instruments according to the standard of "YY/T 1804-2021 spirometer for measuring human forced expiratory volume at time", and the index error and volume linearity were used as the evaluation basis. The experimental results show that only 2 of the 26 mean volume error indexes FVC and FEV₁ exceed the limit, and the volume linearity of more than 15 indexes is less than 3%. The overall performance meets the acceptance criteria. This method significantly improves the calibration accuracy and stability of the flow sensor in pulmonary function detection, and provides a reliable technical support for the wide application of pulmonary function testing instruments in clinical practice.