Abstract: To improve the accuracy and suppression level of network storm recognition by intelligent substation networking equipment, this paper proposes a message feature-based anti network storm method. Based on the characteristic values of various types of messages and the threshold for the number of identical messages, FPGA and driver simultaneously collect and classify the messages, accurately identify and suppress storm messages, and retain normal messages. The experimental results show that the anti network storm method based on message features can effectively identify storm messages and reduce CPU load. Compared with the operating conditions without introducing algorithms, the average CPU occupancy of the system decreased by 95.44%, ensuring the stability of the operation of intelligent substation networking equipment.