Method and device for predicting content of components in sample

The invention belongs to the field of analysis and detection, and particularly relates to a method for predicting the content of components in a sample, which comprises the following steps: accurately predicting the temperature of the sample according to the infrared spectrum of the sample, and automatically distributing or establishing a classification model and a segmented quantitative model with crossed concentration segments in real time by taking the predicted temperature as a label; and predicting the component content in the sample through the model. The invention also relates to a device for predicting the component content in the sample. According to the method, the influence of temperature fluctuation on model analysis precision is reduced, the influence of prediction accuracy degradation caused by a wide correction concentration range on a global quantitative model is avoided, the model generalization ability is enhanced, and the accuracy and precision of a prediction result are improved. 本发明属于分析检测