A multi-source heterogeneous data fusion multi-core classification method comprises the following steps: taking distributed new energy access information, power grid production management informationand geographic meteorological information of a running power grid as samples; manually processing the samples as training samples; calculating distributed new energy access information, power grid production management information and geographic meteorological information, performing a plurality of multi-core learning tests on the public parameter Pb by the public parameter Pb to compare and calculate a difference public parameter Pbd, and obtaining a total public parameter sub-core number Pa=Pb*Pbd; carrying out kernel learning on the Pa kernels by using a training sample to construct a random kernel; putting the trained random kernel function into a test point to carry out fusion classification on the multi-source heterogeneous data; and performing lossless compression, encryption and storage on the classified data. And the integrated distributed new energy related data is provided for each department, the management level and quality of provincial and municipal companies on the distributed new energy are improved, and customized services with different requirements are provided for all social circles.