With　the　rapid　development　of　low-Earth　orbit　(LEO)　satellite　networks,　this　area　is　becoming　increasingly　popular　due　to　its　economic　prospects,　high　bandwidth,　and　low　latency.　However,　the　dynamic　nature　of　satellites　leads　to　frequent　switching　of　inter-satellite　connections.　As　the　scale　of　satellite　networks　grows,　the　transmission　of　large　data　becomes　a　key　challenge.　In　response　to　these　challenges,　we　propose　an　integrated　solution　named　CAAST　(Cache　and　Anycast　for　Satellite　Transmission).　By　employing　caching　techniques　and　leveraging　the　characteristics　of　anycast　communication,　CAAST　stores　data　at　satellite　nodes　and　integrates　them　into　the　routing　system,　which　supports　the　direct　addressing　and　utilization　of　cached　data　within　the　satellite　network.　Additionally,　we　propose　a　dynamic　performance　evaluation　based　on　the　Incremental　Learning-based　Technique　for　Order　Preference　by　Similarity　to　Ideal　Solution　(IL-TOPSIS)　for　selecting　cache　locations　and　a　Hierarchical　Fallback　Strategy　(HFS)　to　address　cache　misses　within　CAAST.　Simulation　results　indicate　that　CAAST　performs　excellently　in　terms　of　data　transmission　rate　and　network　throughput.　It　significantly　reduces　transmission　overhead　and　improves　data　transmission　success　rates,　making　it　a　promising　solution　for　addressing　large　data　transmission　in　satellite　networks.