Abstract Through defect analysis on traditional GM(1,1) and the mechanism description of improved base on the weight of PGM(1,1), we consider that if the same parameters are taken in GM(1,1) when constructing background values, then the prediction error of the model cannot be sufficiently reduced. Different parameters are applied at different times to improve the GM(1,1) background value sequence formula. This kind of background value construction method and grey element N are applied to the GM(1,1) to build a new albino equation. On the basis of the establishment of the new albino equation, the modified initial value through the Runge-Kutta method is applied to calculate the accumulated value of the simulation sequence. To resolve the introduction of many parameters, the particle swarm optimization algorithm is used to find optimal parameters which satisfy the relative error, so the PSD-GM model based on the particle swarm optimization algorithm and the weighted grey combination is established. The application of an engineering example shows that fitting precision of the new model is high, the predictive effect is good, and the predictive accuracy of the new model is improved significantly compared with the other two models.