In prognostics and health management (PHM) of nuclear, it is an important premise to determine the parameters and parts to be monitored in order to accurately predict and evaluate the health status of complex systems. However, the existing parameter selection mainly depends on engineering experience, there is no theoretical support and no method to follow. In this paper, a PHM parameter identification method based on FMMEA and functional performance analysis is proposed. Firstly, the component performance parameters are determined, and the key components are determined through FMMEA. Combined with functional performance analysis, the component parameters are transferred upward according to the level to obtain the product performance parameters, form the three-level parameter transfer relationship between component, sub circuit and product, and complete the parameter identification. This method gives the theoretical basis and data support for parameter selection, forms a complete identification process, and can reduce unnecessary data collection and improve the efficiency of health management in engineering application.