摘要

This paper puts forward a new modeling principle named partial grey model (PGM for short), based on the partial differential equation and grey model. Then, the diagonal transformation partial grey model (dtPGM) is proposed based on the principle of partial grey modeling, which is to transform matrix series through diagonal transformation to reduce calculation and optimize the result. An illustrative example is adopted for comparing the performance between dtPGM and the GM(1,1) rolling model. Results show that dtPGM model has higher accuracy than RGM and GM model when modeling for the long data sequence.