摘要
蔬菜价格因天气、季节、政策等因素的影响,具有波动间歇性特点,利用一般预测模型对蔬菜价格的短期预测会产生误差。本文将MEEMD用于处理蔬菜价格时序信号的波动特性,并与KELM模型结合,提出了基于MEEMD-KELM的蔬菜价格短期预测方法。本文首先采用MEEMD将蔬菜价格时序信号进行分解,然后通过排列熵值判断异常分量并剔除,进而得到完备性较高的模态分量,将各模态分量作为新的信号分别代入KELM模型中进行训练和预测,最后再将分量预测结果合成为整体预测结果。本文以上海批发市场青菜日平均价格为例进行仿真实验,并与传统模态分解法对比,实验表明本文模型具有较高的精度和稳定性。
Due to the influence of weather, season, policy and other factors, vegetable prices have the characteristics of intermittent fluctuations. The short-term prediction of vegetable prices using general prediction model will produce errors. In this paper, MEEMD is used to deal with the fluctuation characteristics of vegetable price time series signal, and combined with KELM model, a short-term vegetable price forecasting method based on MEEMD-KELM is proposed. In this paper, the time series signal of vegetable price is decomposed by MEEMD, and then the abnormal components are judged and eliminated by permutation entropy, and the modal components with high completeness are obtained. As new signals, each modal component is substituted into KELM model for training and prediction, and finally the component prediction results are synthesized into the overall prediction results. In this paper, the average daily price of vegetables in Shanghai wholesale market is taken as an example to carry out the simulation experiment, and compared with the traditional mode decomposition method, the experiment shows that the model has higher accuracy and stability.
出处
《价格理论与实践》
北大核心
2020年第9期68-71,178,共5页
Price:Theory & Practice
基金
上海市决策咨询课题“三农专项,提升上海蔬菜绿色生产能力研究”(2020-N-5)
上海市哲学社会科学规划课题“土地产权安全对上海新型农业经营主体抵押贷款影响机制研究”(2017BGL020)
上海市软科学研究重点项目“上海都市农业众创空间发展模式及对策研究”(17692105600)。
