Abstract: Abstract: Physiological and biochemical indices play a significant role in the evaluation of wheat storage quality. The changes in the storage environment and time will not only cause the deterioration of wheat quality but also cause significant changes of wheat physiological and biochemical indices. However, the information expressed is quite different, which brings great inconvenience to the analysis of wheat storage quality. To solve this problem, a multi-index analysis model of wheat storage quality was proposed in this paper. The falling number, germination rate, peroxidase, fatty acid, conductivity, reducing sugar were selected. Seven physiological and biochemical indices of malondialdehyde were selected as the key factors in this paper. First, the KMO (kaiser-meyer-olkin) and Bartlett's sphericity method were used to test the wheat index, and it was found that the KMO measure value was 0.807 > 0.7, SIG value was less than 0.001, which indicated that the selected index and the measured data were suitable for factor analysis. Then the correlation of wheat index was calculated by the European similarity coefficient and PCA(principal components analysis). The results showed that the distance between the falling number and fatty acid, reducing sugar value, malondialdehyde and conductivity was small, which indicated that the effect of these five indices on the quality of wheat was the same. Meanwhile, these five indices are very far away from germination rate and peroxidase, indicating that they are different in expressing the quality of wheat. In addition, the distance between the germination rate and peroxidase is very close. It shows that the two indices reflect the same quality in some degree. The sensitivity of fatty acid is the highest (0.186), which indicates that it has the greatest influence on the evaluation result, so this index can be used as the key index to evaluate wheat quality, and the sensitivity of peroxidase (0.160) is the least, and it has the least influence on the evaluation result. After a comprehensive analysis, the peroxidase was eliminated and the other six indices were retained for further calculation. Then, a systematic classification method based on nearest element and square Euclidean distance is used to cluster the wheat data. The classification results show that the wheat sample is composed of three kinds of data with obvious classification characteristics. The primary classification of wheat was obtained by analyzing the distribution of fatty acid data, and the discriminant function was obtained by using the Fisher discriminant method to train wheat data. According to the discriminant function, the center value of excellent wheat was -5.699. The center value of medium wheat was 1.316 and the center value of poor wheat was 3.945. By comparing the distance between the value of unknown wheat under this function and the center value of these three kinds of wheat, the unknown wheat classification can be identified. If a batch of wheat has the smallest distance to one center value of these three kinds of wheat, then it would be identified as this classification. The final test analysis shows: The result of the classification of wheat storage quality by the discriminant model in this paper is up to 88.9% in accordance with the classification of an actual reference standard. The analysis model in this paper is basically correct, which can not only provide technical support for the construction of quality evaluation system of stored wheat, but also guide the analysis and discrimination of other grain crops to a certain extent.