In order to investigate the effects of time and temperature changes on blueberries after early decay, hyperspectral imaging technology combined with partial least squares and back-propagation neural network algorithms were used to carry out theoretical analysis and experimental validation, and partial least squares and back-propagation neural networks were used to obtain the time model and the temperature model of blueberry decay, and the modeling effects of these two algorithms were compared. The results show that with the increase of time, the blueberry decay will further deteriorate; along with the increase of temperature, the intensity of blueberry decay gradually increases, the effect of the model established based on the partial least squares method is more suitable for the detection of decayed blueberries, the coefficient of covariance and correlation coefficient of decayed blueberries are 0.131, 0.149, 0.932 and 0.921, respectively, and the error shows that the error is small and correlation tends to be consistent. The model established by partial least squares method can better show the effect of time and temperature on decayed blueberries, which provides a certain reference for the detection of micro-decay on the surface of blueberries.