[Aim] This study was conducted to improve the efficiency of the pest control used for crystal pears, and to build a prediction model to determine the effecctiveness of treatments against diseases and pests. [Method] The grey model (GM) was used to model environmental factors relevant for crystal pears to obtain a pest forecast. The time response formula and parameter estimation were derived through the differential equation, and a grey model (OIVGM) for optimizing the initial value was established. The OIVGM was combined with the BP neural network prediction model (BP), and this grey BP neural network prediction combination model (OIVGM-BP) was used to optimize the initial value. [Result] In this paper, the stability of the model is measured by the unit root test. After the first-order difference processing of OIVGM-BP, the T statistic ( -5.487654)is less than the 5% critical value (-2.878073). The data series is stable indicate that OIVGM-BP can predict stably. This paper measures the adaptability of OIVGM-BP by using the white noise test method. The P value of the residual of OIVGM-BP starts from the second order and is greater than 0.05, indicating that the adaptability of OIVGM-BP is good, and each order has passed the white noise test. The experimental results show that the average prediction accuracy of LRM, GM, TSM, BP and OIVGM-BP for six diseases and insect pests of pear rust, powdery mildew, rot, pear yellow aphid, pear binary aphid and pear wood lice are 70.81%, 70.09%, 69.74%, 65.64% and 83.01% respectively, the prediction accuracy of OIVGM-BP is better than the other four classical prediction models. [Conclusion] OIVGM-BP can effectively predict diseases and insect/pest infestations in crystal pears and guide agricultural production.