基于误差补偿的物流需求混沌预测模型  

Logistics demand chaotic prediction model by error compensation

作　　者：徐琴[1] 

机构地区：[1]武汉商学院商贸物流学院,湖北武汉430056

出　　处：《南京理工大学学报》2018年第1期126-132,共7页Journal of Nanjing University of Science and Technology

基　　金：湖北省高等学校省级教学研究项目(2012458)

摘　　要：针对当前单一物流需求预测模型误差大的缺陷,以获得更优的物流需求预测结果为目标,提出了基于误差补偿的物流需求预测模型。首先通过混沌理论对历史数据进行处理,挖掘物流数据中隐含的变化特点,建立物流需求预测的学习样本,然后采用最小二乘支持向量机对学习样本进行建模和预测,并采用自回归滑动平均模模型对最小二乘支持向量机的预测残差序列进行建模和预测,最后通过残差预测结果对物流需求预测结果进行误差补偿,并通过具体实例对模型性有效性进行测试。结果表明,该模型降低了物流需求的预测误差,明显改善了物流需求的预测效果,并且可以推荐其它预测领域。Research on logistics demand prediction can provide valuable reference information for scientific planning of logistics park. Aiming at high prediction error of single logistics prediction models and get better prediction results,a novel logistics demand prediction model based on error compensation is designed. Firstly,historical data of logistics demand prediction are dealt by chaos theory to data the change rule and establish a sample of logistics demand prediction;secondly least squares support vector machine is used to model and predict learning samples while an error prediction model is established by using auto regressive moving average to estimate the residual sequence of least squares support vector machine predicting results;at last,the logistics demand prediction result is error compensation,and the performance is tested. The results show that it can effectively reduce error of logistics demand prediction,obviously improve the effect of logistics demand prediction,and be used to other prediction fields.

关 键 词：物流需求量 误差补偿 混沌理论 学习样本 最小二乘支持向量机 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]