The invention provides an active bad data recognition and correction method and belongs to the field of power system energy management and monitoring. The active bad data comprises the following steps that firstly, according to real-time network topology structure conditions, a network partitioning rule is generated or a corresponding network partitioning rule is called online from a database, and then a primary network is divided into a series of sub-networks; secondly, according to telemetering configuration of nodes and branches and collected active information, active data initialization is conducted on the primary network and each sub-network, and the nodes and the braches are filled with the active information; thirdly, in a cloud environment, distributed data checking is conducted on each sub-network, and suspicious data are corrected through redundant information; fourthly, the primary network integrates the correction results of all the sub-networks and conducts overall recognition on the correction results of all the sub-networks, and remaining suspicious data are corrected. By the adoption of the active bad data recognition and correction method, the defect that residual contamination and residual submerging occur or a calculation amount is large in an existing state estimation process is overcome.