基于麻雀搜索算法优化神经网络的生菜生理指标预测

Physiological Index Prediction of Lettuce Based on Neural Network Optimized by Sparrow Search Algorithm

生菜生理指标的精准预测对于植物工厂环境下数字化精准管理生菜生长具有重要意义。为了为植物工厂叶菜类作物生理指标的预测提供参考,以植物工厂的水培生菜为研究对象,采集5种营养液配方处理下水培生菜最长叶长、叶片数和株高的数据,以麻雀搜索算法优化的BP神经网络SSA-BP对生菜生理指标数据进行预测分析,并选取平均绝对误差、均方误差、均方根误差和平均绝对百分比误差作为精度指标对试验结果进行分析和评价。结果表明,SSA-BP神经网络对生菜最长叶长、叶片数、株高的预测平均绝对误差分别为9.21、0.563、8.34;均方误差分别为143.79、0.599、110.69;均方根误差分别为11.991、0.774、10.521;平均绝对百分比误差分别为15.639%、6.181%、13.318%,各项评价指标均优于传统BP神经网络,预测误差小于16%,但其预测误差提升不明显。利用SSA-BP神经网络模型可有效对生菜生理指标进行预测,该模型具有良好的预测准确性、泛化性。

The accurate prediction of lettuce physiological indexes is of great significance for the digital and accurate management of lettuce growth in plant factory environment.In order to provide a reference for the prediction of physiological indexes of leaf vegetable crops in plant factories,in this study,hydroponic lettuce from plant factories was taken as the research object,and the data of longest leaf length,leaf number,and plant height of hydroponic lettuce treated with 5 different nutrient solution formulations were collected.Physiological index data of lettuce were predicted by Sparrow Search Algorithm-Back Propagation neural network(SSA-BP).And mean absolute error(MAE),mean square error(Mean Square Error,MSE),root mean square error(RMSE),and mean absolute percentage error(MAPE)were used as precision indexes to analyze and evaluate the test results.The results showed that the mean absolute errors of SSA-BP neural network were 9.21,0.563,and 8.34 for predicting the longest leaf length,number of leaves,and plant height of lettuce.The mean square errors were 143.79,0.599,and 110.69.The root mean square errors were 11.991,0.774,and 10.521,respectively.The mean absolute percentage errors were 15.639%,6.181%,and 13.318%.All the evaluation indexes were better than those obtained by the traditional BP neural network.The prediction error in the experimental results was less than 16%,but the improvement of the prediction error was not obvious.Therefore,the SSA-BP neural network model could effectively predict the physiological indexes of lettuce,and the model had good prediction accuracy and generalization.