In the manufacturing process, environmental factors such as tasks, natural conditions, power levels, etc., restrict the state change of objects and their relationships. Intelligent manufacturing units need to adaptively understand and judge events and complex situations under different context constraints, and a adaptive situation identification method based on complex event processing is proposed naturally, to make real-time optimization decision reasonable. In view of the phenomenon that the influence of context constraints on event discrimination is neglected, a context-aware hierarchical event model is built, and new operators of events such as contemporaneity, context and collaboration are given, while manufacturing situation model and aggregation process are proposed. Aiming at the shortage of generating method on situation model in knowledge base, the mapping association between an object and environment data is established firstly, and such sensed information is transformed into context based events. Integrating the distance calculation of ordinal, nominal variable and adaptive entropy weight method, a improved mixed clustering method is put forward to deal with the diversity and relevance of complex event instance attributes, providing service support for real-time situation identification. Finally, 4 real data sets and 1 simulation data set of are employed for manufacturing process. Experiment results have verified the validity and adaption of the proposed model and method in large-scale problem, and expounded that context factors can significantly improve the accuracy of event judgment and situation recognition in complex manufacturing applications.