基于ELM算法的设施农业小气候环境因子预测模型

Prediction model of protected agriculture environment factors based on Extreme Learning Machine algorithm

设施环境调控技术是设施农业生产过程中的关键技术,设施农业小气候环境因子预测模型技术是设施环境调控技术的核心和基础。针对现有的设施农业小气候预测模型无法对设施环境因子进行快速而精确预测的问题,文中提出了一种基于极限学习机(ELM)的方法来建立设施农业环境因子预测模型。该模型以采集的设施内环境因子历史数据样本作为预测模型的输入,采用ELM的方法建立设施农业小气候环境因子预测模型。Matlab仿真实验结果表明,在预测模型训练速度和精度两个方面,文中ELM模型都优于传统的前馈BP神经网络算法和基于结构风险最小化准则的SVM算法。该模型学习速度快,模拟精度高,可以为设施农业环境因子调控提供决策支持。

Facilities environmental control technology is a crucial technique for protected agriculture production, while environment factor prediction model of protected agriculture mieroclimate is the core and basis of the control of protected agriculture environment. Aiming at the existed problem that protected agriculture environment control system could not quickly and accurately predict the temperature and humidity inside the greenhouse, a new method of Extreme Learning Machine was proposed to predict the environment factor of protected agriculture. To establish environment factor prediction models, historical data of indoor environment factor were chosen to be as input of neural networks. Matlab simulation results show that our models have an advantage in learning speed and accuracy compared with the traditional BP and SVM algorithm. The model with fast learning speed could have good generalization which could provide scientific basis for the regulation of environment factor.