The invention discloses an automatic data diagnosis and identification method based on a desulfurization system of a coal-fired power plant. According to the method, automatic identification is carried out on data, collected on the site, of the desulfurization system through the statistical method based on the process principle; automatic identification of data immutability, data ultralimit, data fluctuation abnormality, parameter coupling abnormality, the flue gas quantity, the limestone consumption and the gypsum yield is achieved, data collection accuracy is improved, and result evaluation fairness is guaranteed.