Abstract: 【Objective】 To accurately predict the water demand of greenhouse watermelons. 【Methods】 A greenhouse watermelon transpiration prediction model was proposed by using a combination of watermelon stem flow rate and meteorological factors as feature variables as model inputs, and a Support Vector Machine (Support Vector Machine, SVM) was established and optimized by slime mold algorithm (slime mold algorithm, SMA). 【Results】 The experimental results showed that the combined use of meteorological factors and stem flow rate as inputs resulted in higher accuracy in predicting transpiration than using meteorological factors alone as model inputs, and the SVM prediction model optimized by SMA had the best prediction performance. 【Conclusion】 The R² and RMSE of the SMA-SVM transpiration prediction model considering stem flow rate in watermelon at three stages are 0.83, 0.87, 0.92, and 0.38, 0.31, and 0.15, respectively and the predicted values of the model are close to the actual values, and the predicted results are reliable.