Abstract: [Objective] Corn silage is a high-quality roughage.With the precise nutrition needs of dairy farming,the regular nutrition index test results can be quickly and environmentally friendly,providing timely and effective data support for the formulation of feed formula,which is conducive to improving the production performance and cattle health of Northeast dairy farms. [Method] In this paper, Near Infrared Spectrum（NIRS）analysis technology was used to establish prediction models of corn silage dry matter（DM） and starch in different sample states. Meanwhile,NIRS analyzers from two different manufacturers were compared to select the one more suitable for Northeast dairy farms. [Result] In this study,138 corn silage samples were selected as the calibration set and 23 were selected as the validation set. The results showed that：（1）After establishing the prediction model of fresh corn silage dry matter（DM） and starch,compared with the results of the NIRS analyzers used by manufacturers A and B for practical detection,and the results of manufacturer B were better than those of manufacturer A,and fresh corn starch prediction model established by manufacturer A should be further optimized.（2）The NIRS analyzer of manufacturers A and B was used to predict the starch of corn silage samples after drying and crushing,which was better than that of fresh corn starch model.（3）Further data collection and optimization are needed for fresh corn silage prediction model and dried and crushed corn silage prediction model. [Conclusion] The prediction model of maize silage dry matter（DM）and starch established in this study can be used for the actual prediction and production practice of maize silage, and the near infrared spectrum analyzer of manufacturer B is more suitable for the use of dairy farms in northeast China.

Key words: Northeast dairy farm, corn silage, NIRS analyzer, wet chemistry, application example