翻译 鉴于此本文立足于实测数据搭建用于识别换道决策和预测换道轨迹的深度学习架构充分挖掘实测数据的潜在特征解决现有研究识别精度低、预判时间短的问题以更加准确描述实际换道行为具体研究内容如下：

Given this, this article is based on actual test data and builds a deep learning architecture for identifying lane-changing decisions and predicting lane-changing trajectories, fully mining the potential features of the actual test data and solving the existing problems of low recognition accuracy and short prediction time in previous research, in order to more accurately describe actual lane-changing behavior. The specific research content is as follows: