A power grid parameter deviation identification method based on real-time measurement comprises the steps of judging whether big errors exist in the power grid parameters according to power grid real-time measuring data and state estimation calculation solutions; selecting suspicious target parameters, confirming the state of real-time measurement associated with the suspicious target parameters to be good, and data to be reliable according to the principle of generatrix balance; building the sub-objective function on associated real-time measurement which is good in state and reliable in data; setting a parameter deviation range, allowing the parameters to change in the range, operating state estimation, and calculating the sub-objective function of the associated real-time measurement until the sub-objective function is smallest; further selecting other real-time section data and even history section data, conducting the parameter deviation identification again, taking the average value of multiple times of parameter deviation identification as the final error modification value of the parameters if the error value of multiple times of parameter identification is stable. The method can be used for realization of on-line monitoring and on-line rectification of the grid parameter deviation and has positive significance on power grid analysis and power grid safety production.