Abstract: The treatment of corn and rice straw after harvest has always been an urgent problem to be solved in agricultural production. Mechanized straw returning to the field has been an important means of straw treatment of crops after harvest and also an important measure to protect black soil resources. Combined with relevant literature, the key links was single prediction model selection based on cointegration test and optimal combination prediction model selection based on minimum error index was proposed. The quadratic function model, ARIMA model and H-W non-seasonal model were selected as the single forecasting model for the mechanization degree of straw returning to the field by using the method of cointegration test. The combined forecasting model was built according to the minimized sum of the absolute error value method, Shapley method and Theil inequality coefficient and IOWAO model method. The forecasting accuracy of the combined models was compared by SSE, MAE, MSE, MAPE and MSPE. It was proved that Theil inequality coefficient and IOWAO combined model was the better model to forecast the mechanization degree of crop straw returning to the field. The results show that the mechanization degree of straw returning to field in Heilongjiang Province will be steadily improved from 2022 to 2026. The average annual increase will be 4.52%, which will reach 74.19% in 2026, an increase of 22.62% over 2021. After 2022 the mechanization degree of straw return in Heilongjiang Province will enter a rapid development period. The combined prediction results provide theoretical basis for determining and implementing mechanized straw treatment measures and have practical significance for protecting and restoring the productive capacity of black soil resources.