The embodiment of the invention provides a population information data fusion method and system. The first node in the system broadcasts the complete PCD meeting the fusion standard in the blockchainsystem; and the second node immediately starts training after receiving the broadcast, broadcasts a result after the training is completed, generates blocks and adds the blocks into the block main chain, other nodes which do not complete the training performs voting, and the block main chain selects the corresponding blocks to add according to a voting result and performs full-node broadcasting. If the second node votes correctly, a block value function is directly obtained, and if the second node votes incorrectly, a value function of the second node is obtained, thereby realizing model updating of reinforcement learning. In the embodiment of the invention, the blockchain technology, the reinforcement learning technology and the data fusion are grafted together, the learning training effect is improved, the data fusion efficiency is improved, and the workload of people is reduced.