摘要
根据植物电信号非平稳、非线性等特点,以盆栽植物君子兰和观音莲为研究材料,对采集的两种植物电信号进行自回归模型(AR模型)分析,并采用自适应技术对模型参数进行优化。首先采用Akaike信息检验准则估计植物电信号的模型阶次,然后基于最小二乘参数估计法建立植物电信号模型并拟合,最后利用自适应AR参数算法对两种植物电信号进行预测。结果表明:基于最小二乘法的AR模型对植物电信号进行短期预测是可行的。为了解决预测中误差随预测点数增加而增大的问题,提出了基于Kalman滤波算法的AR模型参数估计法,使得预报精度以及收敛速度得以优化。进一步预报的数据可作为温室或塑料大棚控制系统中重要的输入参数,该方法为植物生物信息学提供了新的研究基础。
According to nonstationary and nonlinear characteristics of electrical signals in plants,two kinds of potted plants：Clivia miniata and Alocasia macrorrhiza were chosen as research materials and an autoregressive model（AR model）analysis was done for electrical signals acquired from two kinds of plants by the adaptive technology to optimize the model parameters.Firstly the model orders in terms of Akaike information criterion were estimated,and then the models of signals were made by using estimating the parameters based on least square in two plants and fitting.Finally the prediction of electrical signals in two plants by adaptive AR parameters models was discussed.The research result showed that short-term forecast for electrical signals in plants based on least square is feasible.In order to solve the question that the error will increase with prediction points increasing,AR model parameter based on Kalman filter algorithm was given to optimize the prediction precision and convergence speed.Furthermore,forecasting data can be used as the important input parameters in the automatic control system of greenhouse and plastic trellis,and this way will provide a new research basis for plant bioinformatics.
出处
《南京农业大学学报》
CAS
CSCD
北大核心
2012年第6期142-147,共6页
Journal of Nanjing Agricultural University
基金
江苏省农机局科研基金资助项目(GXZ06007)
江苏省农机三项工程项目(NJ2010-02)
关键词
植物电信号
自回归模型
自适应
拟合
预测
electrical signals in plants
autoregressive model（AR model）
adaptive
fitting
prediction
