The embodiment of the invention provides a network intelligent platform malicious data detection method and device based on mining. The method mainly comprises the following steps: storing a plurality of integrated feature vectors Table and a data packet type corresponding to each Table into a prior data bank, extracting the field value of each field of a data packet to be tested, and calculating a distance value between the data packet to be tested and the center of each stored Table according to the field value of each field of the co-ordinated data packet to be tested; and according to the distance value between the data packet to be tested and the center of each Table and the action radius of each Table stored in the prior data bank and the corresponding data packet type, judging whether the data packet to be tested is a malicious data packet or not. The embodiment of the invention can be used on network intelligent platforms, such as a mobile internet gateway and the like, to carry out effective malicious data detection to the data packet, the property and privacy safety of a user can be guaranteed, and the competiveness of an operator can be enhanced.