Abstract Forecasting crop water requirement is an important basis for formulating a reasonable irrigation system. In view of the deficiencies of BP neural network, genetic algorithm(GA)has the characteristics of strong global search capability, and a prediction model of crop water requirement based on GA-BP neural network was established. Taking the experimental crops of the Lamu Farm in Guangzhou as the object of research, the results show that the mean square error and certainty coefficient of the test set of crop water requirement prediction model based on BP neural network are 0.037 and 0.648 respectively. The square error and the certainty coefficient of the rest set of crop water demand forecasting model based on the GA-BP neural network are 0.013 and 0.882 resiectively. The GA-BP crop water demand forecasting model has a convergence rate, certainty coefficient and performance better than the BP crop water demand forecasting model.
CHEN Wufen, ZHANG Qianhua, HUANG Zhengwu, et al. Prediction of pH value of hot springs water quality based on regression support vector machine [J]. People′s Pearl River, 2016,37(7): 94-97.(in Chinese)
XIA Zehao, WENG Shaojie, LUO Wei, et al. Research on crop water demand prediction model based on grey neural network [J]. Chinese agricultural machinery chemistry report,2015,36(2):219-223.(in Chinese)
SHANG Zhigen. Prediction of crop water requirements based on neural network ensemble[J].Software guide, 2018,17(1): 46-48.(in Chinese)
LI S, KANG S, LI F, et al. Evapotranspiration and crop coefficient of spring maize with plastic mulch using eddy covariance in northwest China[J]. Agricultural water management, 2008, 95(11):1214-1222.
YANG Z F, SUN T, CUI B S, et al. Environmental flow requirements for integrated water resources allocation in the Yellow River Basin, China[J]. Communications in nonlinear science & numerical simulation, 2009, 14(5):2469-2481.