Prognostics and health management (PHM) technology is composed of remaining useful life (RUL) prediction and health management. Health management is the ultimate goal of applying PHM technology. For multivariate degradation equipment, in this paper, we develop a joint decision for spare part ordering and equipment replacement based on prognostic information. First, according to the RUL probability distribution of the multivariate degradation equipment predicted by deep learning, a multi-objective joint decision model considering both cost and availability is established. Then, to determine the optimal spare part ordering and equipment replacement time, the trade-off between cost and availability is formulated through a constructed decision boundary. Finally, a commercial modular aviation propulsion system dataset is used to verify the effectiveness of our proposed method.