Recently, UAV formations, as a typical instance of aerospace multi-robot systems, have played an increasingly crucial role in military confrontation due to their extraordinary synergy. As the basis of performance optimization for UAV formations, performance assessment not only provides credible reference for the offline design of their parameters and structure, but also offers strong decision support for the online update of their configuration, thus becoming the key to maintaining their high-performance operation. In most cases, the program for iteratively assessing the performance of UAV formations is deployed and operated in ground stations. During periodic information transmission between the UAV formation and the ground station, delay is inevitable, leading to the asynchrony of temporal information received from different UAVs. Moreover, for each assessment, it is the observation information of UAV formations over a time period, rather than that at a time instant, that should be utilized to reduce the risk of poor decision-making. Therefore, the belief rule base with asynchronous sequential inputs is constructed with the help of cross correlation and discrete wavelet transformation, on which a strategy is further designed for the online performance assessment of UAV formations. Pertinent simulation is performed, showing the feasibility and validity of the proposal.