The　DeNOX　efficiency　of　the　Selective　Catalyst　Reduction　(SCR)　system　relates　closely　to　the　urea　injection,　and　the　starting　injection　time　of　urea　injection　depends　on　the　exhaust　gas　temperature.　For　the　diesel　vehicle　driving　on　real　roads,　the　exhaust　gas　temperatures　change　significantly.　At　high　temperatures,　the　SCR　system　actively　injects　urea;　whereas　at　low　temperatures,　no　urea　injection　happens　and　the　catalyst　reaction　relies　solely　on　the　stored　ammonia　in　the　catalytic　container.　Thus,　it　is　important　to　investigate　the　SCR　DeNOX　efficiency　under　condition　of　different　temperature　ranging　from　low　to　high　temperature.　In　this　paper,　a　heavyduty　diesel　vehicle　was　tested　under　stationary　idling　condition　with　first　rapid　acceleration,　then　high-speed　holding　running　and　finally　deceleration,　and　the　NOX　emission　performance　at　various　exhaust　temperatures　was　acquired　under　two　distinct　operating　conditions:　active　injection　of　ammonia　at　higher　temperatures　and　no　ammonia　injection　at　lower　temperatures.　Furthermore,　a　deep　learning　model　was　built　for　predicting　NOX　downstream　of　the　SCR　by　utilizing　exhaust　flow　rate,　average　exhaust　temperature,　and　NOX　emission　concentration　upstream　of　the　SCR　as　input　parameters,　and　this　model　was　verified　with　a　model　fit　of　0.98.　The　results　show　that　the　SCR　DeNOX　efficiency　achieves　99　%　with　active　urea　injection　during　high-temperature　stages,　and　30-40　%　during　low-temperature　stages,　relying　on　ammonia　storage.　During　the　ammonia　storage　stage,　the　SCR　DeNOX　efficiency　exhibits　high　sensitivity　to　temperature,　with　a　rate　of　decreases　of　1.3　%/degrees　C　per　unit　temperature.　In　the　active　ammonia　injection　stage,　the　SCR　DeNOX　efficiency　decreases　at　a　rate　of　0.5　%/degrees　C　per　unit　temperature.　Various　reaction　stages　exhibit　a　significant　relationship　between　SCR　DeNOX　efficiency　and　exhaust　temperature.