A health management system based on knowledge is proposed. The straindata of large aircraft structure is sampled by adopting wireless sensor network. Monitoring and diagnosing the strain field of the structure, the health management is realized. Through the ontology modeling of the domain knowledge, the sharing and reuse of the knowledge can be achieved. Building the related reasoning rules and with the help of reasoner, domain knowledge involved in health management can be inferred. With this knowledge, the related functional agent can be organized into a multi agent system to achieve the health management task. A strategy of voting mechanism is proposed in data fusion to improve the veracity and reliability of diagnosis. A system of stain state health management system is shown to verify the system and its efficiency.