Leakage of backdoor privacy has become a major challenge with rapid development of industry Internet of Things (ⅡoT), causing serious threat to security and stability of industrial control system and internet of things. In this paper, some basic attributes are defined based on data feature of backdoor privacy in ⅡoT, upper semantics are extracted based on security threat in static and dynamic data flow, static and dynamic leakage degrees are generated based on multi-attribute decision-making, and finally security level and threshold are computed with grey correlation analysis. As a result, perception for leakage scenarios of backdoor privacy can be accomplished in static binary structure and dynamic data flow. Twenty seven types of backdoor privacy are chosen for testing in target environment to compute and analyze basic definitions, upper semantics and judgment semantics, and successful perception for leakage scenarios is performed via comparison between security level and threshold. In addition, effectiveness of this work is validated through comparison with other models and prototypes.