Time-series　data　mining　plays　an　important　role　in　big　data　decision　making　because　it　can　reveal　the　development　pattern　of　things.　Similar　concatenation　of　temporal　data　is　a　fundamental　prerequisite　for　data　twinning.,　whose　core　objective　is　to　find　all　similar　temporal　data　pairs　according　to　a　given　similarity　metric.　Dynamic　temporal　regularization　(DTW)　has　been　widely　used　in　many　fields.,　such　as　target　detection.,　trend　prediction　and　fault　identification.,　as　the　best　data　alignment　method　on　temporal　data.　We　use　deep　learning　to　extract　vehicle　trajectories　and　perform　behavioral　pattern　learning.,　and　finally　use　an　improved　DTW　algorithm　to　pre-process　the　trajectory　data　and　solve　the　distance　function　to　achieve　matching　between　the　trajectories　of　the　event　sequence　to　be　measured　and　the　typical　trajectory　data　patterns.　By　comparing　the　indicators　with　the　unimproved　DTW　algorithm.,　the　research　results　show　that　this　traffic　condition　recognition　method　is　stable　and　reliable.,　and　can　maintain　high　matching　accuracy　with　significantly　reduced　computation.,　high　success　rate　and　good　real-time　performance.　©　2023　IEEE.