Grey system theory based parallel combination forecast method and its application

Due to the complexity and importance of crop yield forecasting, many works have been done during the past decades. In this paper, a reasonable division of influencing respects of crop yield per unit area is presented, and a novel intelligent parallel combination model consisting of a GM(1,1) model from the grey system theory and a intelligent evolution algorithm is referred to put into practice as well, which is a comprehensive reflection of both social production levels and environmental factors in the crop yield forecasting results, with respective individual models' merits. In the mentioned model, errors of each individual forecasting model (GM(1,1) models both for the social influence forecasting and for the environmental influence forecasting) are put into an intelligent evolution model to determine the own weights in the combination model. A experiment based on the observed historical data from Shulehe basin in Gansu province in Northwestern of China is carried out to verify the reasonability and validation of the combination model, which shows that the novel model will yield lower error level than that of original individual models with less risky in practice, and is relatively more intuitive and more feasible than some other conventional approaches applying agronomy analysis as well.