Objective: Correlation between alkali treatment process and carrageenan gel quality and prediction model were studied. Method: Carrageenan was extracted from Eucheuma muricatum by alkali extraction method， and a reliable and measurable model was established to predict the gel properties and molecular characteristic structure of carrageenan under different alkali treatment process conditions using artificial neural network (ANN) model. Results: Artificial neural network model fit the total set with R2 = 0.991 better than the multivariate linear model (MLR)， with mean absolute percentage errors (MAPEgel strength = 1.71%， MAPEmolecule = 18.84%， and MAPEsulphate group = 1.91%， MAPE3，6-endoethergalactose = 3.68%) are lower than the MLR model. In addition， in actual production， when the key process parameters of alkali treatment (temperature， time， and concentration) were 65 ℃， 6 h， and 10%， the gel strength of carrageenan was measured to be 614.32 g/cm2， and the gel strength of carrageenan was predicted to be 617.13 g/cm2 by the constructed prediction model of the correlation between carrageenan processing alkali treatment and its gel quality. The results show that the predicted value of the model is close to that of the actual production. Conclusion: The prediction performance of ANN model is better than that of multivariate linear model， and ANN model is more accurate than MLR model in predicting carrageenan gel properties and molecular characteristic structure. This study provides a technical reference for the intelligent and precise manufacturing of carrageenan high-quality products， so as to expand the application of carrageenan in biomedicine and other fields.

carrageenan  /  gel strength  /  characteristic molecular structure  /  multiple linear model  /  artificial neural network model