基于遗传神经网络的番茄光合速率模型英汉.docVIP

基于遗传神经网络的番茄光合速率模型 胡瑾，何东健※，张海辉 王智永 王小春 （西北农林科技大学机械与电子工程学院，陕西杨凌 712100） 摘 要光合速率模型是开展光环境细粒度调控是理论基础，已成为是决定光环境调控成效关键，因此，如何基于智能算法构建高效精准的光合速率模型已成研究关键。由于BP神经网络算法初始权值为任意随机数，在建立回归预测模型容易陷入局部平坦区，现有型(基于神经网络光合速率模的)存在收敛速度慢、训练时间长等的问题。针对上述问题，本文提出了一种基于遗传算法的启发式神经网络黄瓜的光合速率模型，通过遗传算法优化初始权值，有效提高模型的性能。模型建立多因子嵌套试验获得825组黄瓜幼苗光合速率测试数据的基础上，通过BP网络结构构建和数据预处理，基于遗传算法进行网络权值和阈值优化，以及基于LM训练法的网络训练,建立基于遗传神经网络的黄瓜光合速率模型。在此基础上，进一步对比分析本文建立的遗传神经网络的光合速率预测模型与未优化的神经网络预测模型的训练性能和模型精度。试验结果表明，遗传神经网络的光合速率预测模型其训练效果与模型精度均优于神经网络预测模型，模型预测值与实测值间的相关系数为0.987，光合速率绝对误差小于±0.5 关键词（小五、黑体）：农业；工程；编辑；科技；论文（小五、楷体。） 中图分类号： 文献标志码： 文章编号：（小五、黑体） Tomato Photosynthetic Rate Based on Genetic Neural Network Abstract The photosynthetic rate model is the theoretical basis to carry out the light environment fine-grained control, which have become the key to determine the effectiveness of light environment control. Therefore, how to build efficient precision for photosynthetic rate model based on intelligent algorithm has become critical on the study . Due to the initial weights of BP neural network algorithm for arbitrary random number, the establishment of regressive prediction model is easily trapped in partial flat area. The existing type of photosynthetic rate model based on neural network exists the problem, such as slow convergence speed, long training time and so on. In view of the above problem, this paper presents a photosynthetic rate model of heuristic neural network for cucumber based on genetic algorithm. The performance of model can be effectively improved by genetic algorithm optimizing initial weights. The model establishes the multi-factor nesting experiment to obtain 825 group of cucumber seedling photosynthesis rate test data in the foundation. By BP network structure construction and data preprocessing, based on the genetic algorithm to optimize network weights and threshold, as well as based on LM training method of network training, the photosynthetic rate model of heuristic neural network for