With　the　rapid　development　of　the　Internet　of　Things　(IoT)　and　big　data　technology,　traditional　logistics　management　has　gradually　exposed　problems　such　as　information　silos,　lagging　response　and　inefficient　resource　allocation,　and　it　is　urgent　to　introduce　intelligent　and　real-time　monitoring　and　decision-making　systems　to　improve　the　overall　operation　efficiency.　In　this　paper,　a　logistics　intelligent　monitoring　system　based　on　big　data　of　the　Internet　of　Things　is　proposed,　and　the　overall　design　of　the　system　is　based　on　the　three-layer　fusion　architecture　of　＇edge-cloud-intelligent　decision-making＇:　in　the　perception　layer,　high-frequency　dynamic　data　collection　of　logistics　nodes　is　realized　through　the　deployment　of　heterogeneous　IoT　devices　including　RFID,　GPS,　and　environmental　sensors;　At　the　edge　layer,　a　lightweight　deep　learning　model　MobileNet　is　introduced　for　initial　anomaly　detection　and　event　annotation　to　reduce　cloud　load　and　latency.　In　the　cloud　computing　layer,　a　distributed　storage　and　processing　framework　based　on　Apache　Spark　is　used　to　construct　a　real-time　event　stream　processing　mechanism,　and　on　this　basis,　a　Time　Series　Convolutional　Network　(TCN)　fusion　model　is　superimposed　to　realize　the　deep　feature　modeling　and　multi-step　trend　prediction　of　node　anomalies　in　the　logistics　chain.　At　the　decision-making　level,　a　reinforcement　learning-driven　dynamic　resource　scheduling　mechanism　was　designed,　which　was　based　on　Deep　Deterministic　Policy　Gradient　(DDPG)　to　intelligently　optimize　vehicle　scheduling,　path　planning　and　warehouse　management　strategies　according　to　environmental　changes.　The　comprehensive　experiments　on　the　logistics　dataset　and　the　actual　operation　data　show　that　the　proposed　system　improves　the　accuracy　of　anomaly　detection　by　24.7%　compared　with　the　traditional　method.　©　2025　IEEE.