Abstract: The calibration of the capacitance diaphragm gauge is a key step to ensure that its accuracy conforms to the standards. Both the noise interference introduced by the voltage signal of the standard vacuum gauge and the non-linear fitting error introduced by the differential capacitance signal of the vacuum gauge under test can affect the accuracy of calibration. In this paper, voltage signal is processed by Kalman filter, and based on the non-linear trend of the differential capacitance signal, quartic polynomial method was selected for function fitting. The experimental results show that this method can effectively reduce errors introduced during the calibration process. Mean squared error was reduced by 79.6%. The maximum absolute error was reduced by 48.8%.