CAJ | 학술논문

光照度对植物的生长发育有着不可忽视的重要作用，它的强弱直接影响到植物的光合作用、物质代谢、结构形态和农作物产量等。通过重点研究大规模样本点的BP神经网络空间插值方法，进而提高计算光照分布的准确性和有效性。利用分组训练、组合优化的改进BP神经网络的权值和阈值，训练BP神经网络预测模型求得最优解。实验结果表明该模型的相对误差较小，对光照度预测的可信度较高，可以用于预测光照分布。
광조도대식물적생장발육유착불가홀시적중요작용，타적강약직접영향도식물적광합작용、물질대사、결구형태화농작물산량등。통과중점연구대규모양본점적BP신경망락공간삽치방법，진이제고계산광조분포적준학성화유효성。이용분조훈련、조합우화적개진BP신경망락적권치화역치，훈련BP신경망락예측모형구득최우해。실험결과표명해모형적상대오차교소，대광조도예측적가신도교고，가이용우예측광조분포。
Light inte nsity on plant growth and development has an important role that directly affects the strength of plant photosynthesis, metab-olism, morphology and crop yields. Focuses on the large-scale sample points through BP neural network spatial interpolation methods, thus improving the accuracy of calculation of light distribution and effectiveness. The use of packet training, combinatorial optimization of improved BP neural network weights and thresholds, the training BP neural network prediction model optimal solutions. Experimental re-sults show that the model is small relative error on the credibility of the higher light intensity projections can be used to predict the light distribution.