Some progress has been made on how to identify important nodes in temporal hyper-networks. Eight centrality methods for measuring the importance of nodes in this type of network were defined, focusing on the different topological properties and time characteristics of the network, and comprehensively considering the importance of nodes in this type of network from multiple perspectives. At the same time, a SI spreading model in temporal hyper-networks was constructed. Based on this model, an evaluation method for randomly removing nodes was put forward to measure the effectiveness of the defined centrality methods. The results show that the betweenness centrality method which is based on the fastest arrival path is a good indicator in identifying important nodes in temporal hyper-networks. Furthermore, degree and hyper-degree centrality methods which consider time resolution can optimize the conventional degree and hyper-degree centrality methods by finding the optimal time resolution of the network. This addresses the limitation of traditional methods in effectively considering network temporal information and performs comparably to the betweenness centrality method in multiple real networks.