In order to improve the discriminating capacity of the RFID tags for agricultural products traceability data, a new handling engine ORFID-CEP for complex events of agricultural products RFID Tag based on ontology, is proposed by introducing the event ontology model into the field of the agricultural products quality safety administration. In the new engine, the ontology model of RFID event, semantic space and the rules of the event ontology are defined. In the RFID practical application instances, the number of the fringe real-time events is so large that those methods used in current systems can not handle them in time. Thus, many significant events are lost. Aiming to overcome the problem, the transform system of work flow for the complex event ontology and the optimized strategies for event exploration, event operation and restrictive conditions of event appearance are established. Experimental results show that the new engine can mine more complex events with semantic information than the conventional Esper-CEP engine. Additionally, the new engine is steadier than the Esper-CEP engine. With increasing number of the events, the growth rate of the new engine for mining complex events raises more quickly than Esper-CEP, which indicates that the engine has the good ability of information process.