卷烟烟气焦油量的ARMA预测模型研究  

ARMA Forecast Model of Tar of Cigarette Smoke

作　　者：陆鸣 李东亮[2] 许自成[2] 刘秀彩 许寒春 

机构地区：[1]福建中烟技术中心,福建厦门361022 [2]河南农业大学国家烟草栽培生理生化研究基地,河南郑州450002

出　　处：《安徽农业科学》2008年第14期5696-5698,共3页Journal of Anhui Agricultural Sciences

摘　　要：[目的]为计算机辅助配方研究提供理论依据。[方法]以同一品牌不同批次卷烟的烟气焦油量为研究对象,采用时间序列分析法,建立了卷烟烟气焦油量的预测模型,并进行了模型预测验证。[结果]ARMA(22,)模型的AIC、LF和FPE值在各模型中均最小,所以选择ARMA(2,2)为卷烟烟气焦油量的预测模型,即:(1-1.622 q-1+0.844 q-2)y(t)=(1-1.836 q-1+1.02 q-2)e(t)。根据对模型残差序列进行的白噪声检验判定,建立的ARMA(2,2)模型是显著有效模型,模型预测验证表明模型预测精度达99.51%,平均相对误差为0.49%,属于一级(优等)模型。时间序列一般用于短期预测,不能用于长期预测。[结论]该研究建立的卷烟烟气焦油量的ARMA(2,2)预测模型的预测精度高、误差小,可以用于卷烟烟气焦油量的短期预测。[Objective] The study aimed to provide a base for computer-aided formula design. [Method] With the tar of cigarettes from different batches of cigarettes with the same trademark as tested materials and by using time series sequence method, the forecast model of tar of cigarette smoke was built and validated. [Result] Among forecast models, the values ofAIC. LF and FPE were all least in ARMA（2,2）model. namely （ 1- 1.622 q^-1＋0.844 q^-2） y（ t ） = （1-1.836 q^-1 ＋ 1.02 q^-2） e （t）, so it was selected as forecast model for tar of cigarette smoke. The established ARMA （2,2）model was significantly effective model according to autocorrelation check of the residuals on the model residual sequenee oThe forecast precision of ARMA （2,2）model reached 99.51% and the average relative error was 0.49%. so the ARMA（ 2,2 ）model belonged to the first grade model （excellent model）. However, time series model was usually used for short-term forecast but not for long-term forecast, [Conclusion] The forecast model of tar of cigarettes smoke built in this study had a better precision, lower error and could beused for short-term forecast of the tar of cigarettes smoke.

关 键 词：卷烟 烟气焦油量 ARMA模型 时间序列分析 

分 类 号：S572[农业科学—烟草工业]