The invention provides a data correction method for improving measurement precision. The data correction method comprises the steps of a, acquiring to-be-corrected continuous indirect measurement data, and establishing a segmented nonlinear correction model I and a segmented nonlinear correction model II; b, setting a hysteresis interval q-p, wherein critical thresholds satisfy a relation of q<p, and on the condition that the value of the measured continuous indirect data is lower than the critical threshold p, performing correction by means of the segmented nonlinear correction model I, and on the condition that the value of the measured continuous indirect data increase to higher than the critical threshold p, performing correction by means of the segmented nonlinear correction model II, and on the condition that the value of the measured continuous indirect data gradually reduces to lower than the critical threshold q, performing correction by means of the segmented nonlinear correction model I. According to the data correction method, data are corrected by means of a segmented nonlinear mode. Compared with a traditional segmented linear correction, the data correction method is advantageous in that less correction points in segmented nonlinear correction are realized; and furthermore after the hysteresis interval is introduced, high-frequency data jump caused by high-frequency correction model switching at critical points can be prevented.