Auxiliary Power Unit (APU) is an essential component of civil aircraft. In order to meet demands for aircraft APU maintenance, failure rate of aircraft APU should be predicted. Since aircraft APU is a repairable system, its performance degradation can be considered as a Generalized Renewal Process (GRP). Therefore, aircraft APU failure rate prediction can be implemented by utilizing Weibull-based GRP. However, Aircraft APU maintenance effect which is required in Weibull-based GRP is difficult to quantify. In order to solve this problem, an improved Weibull-based Generalized Renewal Process (IWGRP) model is proposed in this work. To be specific, utilizing some key test parameters after repairs, aircraft APU maintenance effect is estimated. Based on aircraft APU maintenance effect, conditional failure rate and hazard rate of aircraft APU can be obtained. Then, conditional failure rate and hazard rate are utilized to predict aircraft APU failure rate. Finally, the performance of IWGRP is validated by real data from China Southern Airlines (CSA). Experiment results demonstrate that IWGRP is suitable for aircraft APU failure rate prediction.