The invention discloses a network asset topology identification method based on flow perception. The method comprises active flow detection and passive flow detection. The active flow detection comprises the following steps: S1, collecting the survival condition of a port; S2, sending a protocol detection packet; S3, acquiring port response information; S4, extracting equipment proprietary information, and matching the equipment proprietary information with the manufacturer information; S5, establishing an equipment fingerprint library; S6, establishing a network asset topological graph, wherein the passive flow detection comprises the following steps: S1, deploying a bypass flow mirror image; S2, collecting two-layer broadcast communication and three-layer IP communication messages; S3, loading a filtering rule, and extracting a network asset fingerprint; S4, establishing a network asset fingerprint database; and S5, supplementing the topological graph. According to the invention, active flow detection and passive flow detection are combined, so that a network asset detection result is more accurate and comprehensive; meanwhile, network asset change warning is set, so that the accuracy of the network asset fingerprint database is ensured, and the use safety of network assets is improved.