基于卡尔曼滤波算法的新疆棉花产量预测

Prediction of cotton output in Xinjiang province based on Calman filter algorithm

棉花是重要的纺织原料,科学预测棉花的产量是制订棉花生产和纺织工业计划的基础。运用卡尔曼滤波算法对新疆棉花产量进行预测,并取得较好的效果。卡尔曼滤波算法的平均预测误差仅为0.869 79%,比自回归滑动平均(ARMA)模型的预测误差(3.390 34%)减小了74.345 20%。经预测,2018年新疆棉花产量为444.477 8万t。

Cotton is an important textile raw material, and the scientific prediction of cotton output is the basis for formulating cotton production and textile industry plans. The Caiman filtering algorithm was used to predict the cotton output of Xinjiang province, and good results had been achieved. The average predict error is only 0.869 79%, which is 74.345 20% less than the predict （3.390 34%）of the auto-regressive moving average（ARMA） model. It was predicted that cotton production in Xinjiang province will be 4.444 778 million ton in 2018.