基于VMD和LSTM方法的北京市PM_(2.5)短期预测

Short-Term Prediction of PM_(2.5) in Beijing Based on VMD-LSTM Method

雾霾问题是与社会发展息息相关的热点问题,为了进行PM_(2.5)浓度预测,为有效防治雾霾提供依据,本文提出了改进的VMD(变分模态分解)和LSTM(长短时记忆)神经网络相结合的PM_(2.5)预测模型VMD-LSTM。首先利用阈值法确定VMD方法的分解数目,将历史数据分解成不同序列,然后对每个序列进行预测,最后将每个序列的预测结果求和得到最终的预测结果。将VMD-LSTM模型应用到北京市PM_(2.5)序列的短期预测中,并利用7种评价指标将其与ARIMA(整合移动平均自回归)、RFR(随机森林回归)、LS-SVR(最小二乘支持向量回归)、LSTM等9种模型进行比较。结果表明,在其中的5个误差评价指标中,VMD-LSTM模型表现最优,仅有1个误差指标评价位列第二,在协议指数评价中,VMD-LSTM模型最接近于1,精度最高。其中:VMD-LSTM模型的均方误差为41.10,均方根误差为6.42,平均绝对误差为5.79,协议指数为0.97;而RFR、VMD-LS-SVR、ARIMA和LSTM等9种模型的均方误差范围为60.72~1 058.07,均方根误差范围为7.79~32.53,平均绝对误差范围为7.45~26.14,协议指数为0.39~0.95。相比于其他模型,本文提出的VMD-LSTM模型精度最高。

Haze is a hot issue closely related to social development.In order to predict PM_(2.5) concentration and provide basis for its effective prevention and control,the PM_(2.5) prediction model VMD-LSTM is proposed based on the combination of the improved VMD(variational modal decomposition) and LSTM(long and short-term memory) neural network.Firstly,the threshold method is used to determine the decomposition number of VMD method,then the historical data is decomposed into different sequences,further each sequence is predicted,and the final prediction result is obtained by summing the prediction results of each sequence.The VMD-LSTM model is applied to the short-term prediction of PM_(2.5) series in Beijing,and its result is compared with nine models such as ARIMA(autoregressive integrated moving average),RFR(random forest regression),LS-SVR(least squares support vector regression),LSTM and so on,by using seven evaluation indexes.The comparison results show that among the five error evaluation indexes,the VMD-LSTM model performs best,with only one error index ranking second.In the protocol index evaluation,the VMD-LSTM model is closest to 1 and has the highest accuracy;The mean square error of VMD-LSTM model is 41.10,the root mean square error is 6.42,the mean absolute error is 5.79,and the protocol index is 0.97.The mean square error range of RFR,VMD-LS-SVR,ARIMA,SVR,and LSTM models is from 60.72 to 1 058.07,the root mean square error range is from 7.79 to 32.53,the mean absolute error range is from 7.45 to 26.14,and the protocol index range is from 0.39 to 0.95.The VMD-LSTM model proposed in this paper has the highest accuracy.