基于SWT-ISSA-LSTM的地铁空气质量预测建模

Modeling of subway air quality prediction based on SWT-ISA-LSTM

为了提高地铁室内PM_(2.5)的预测精度,降低监测成本,提出了一种基于孤立森林算法(isolated forest,IF)、同步压缩小波变换算法(SWT)、改进麻雀搜索算法(improved sparrow search algorithm,ISSA)和长短期记忆网络(long short-term memory network,LSTM)的混合模型。首先,使用孤立森林算法检测并去除异常数据,在用SWT算法对原始PM_(2.5)数据进行去噪处理;其次,针对麻雀算法(SSA)易陷入局部最优、收敛速度慢的问题,利用正弦混沌、动态自适应惯性权重、高斯变异和反向学习策略改进麻雀算法,降低了SSA陷入局部最优解的概率,提高了麻雀算法的收敛速度和寻优能力;最后,利用ISSA对LSTM模型的参数进行寻优,构建ISSA-LSTM模型进行预测,得到最终的PM_(2.5)预测结果。实验结果表明,SWT-ISSA-LSTM模型在均方根误差比SWT-LSTM模型和SWT-SSA-LSTM模型分别降低了8.38和3.27μg/m^(3)。在拟合度方面,该模型比SWT-LSTM模型和SWT-SSA-LSTM模型分别高了10.6%和2.9%。

In order to improve the prediction accuracy of indoor PM_(2.5)and reduce the monitoring cost,a hybrid model based on isolated forest(IF),synchronous compression wavelet transform(SWT),improved sparrow search algorithm(ISSA)and long short memory network(LSTM)was proposed.Firstly,the isolated forest algorithm is used to detect and remove abnormal data,and the SWT algorithm is used to denoise the original PM_(2.5)data.Secondly,in order to solve the problem that sparrow algorithm(SSA)is easy to fall into local optimum and has a slow convergence speed,we use sine chaos,dynamic adaptive inertia weight,Gaussian mutation and reverse learning strategies to improve the sparrow algorithm,reduce the probability of SSA falling into local optimum,and improve the convergence speed and optimization ability of sparrow algorithm.Finally,ISSA was used to optimize the parameters of the LSTM model,and an ISSALSTM model was constructed for prediction,resulting in the final PM_(2.5)prediction result.The experimental results show that the SWT-ISSA-LSTM model has a 8.38 reduction in root mean square error compared to the SWT-LSTM model and the SWT-SSA-LSTM model,respectively 8.38 and 3.27μg/m^(3).In terms of fit,the model proposed in this article is 10.6%and 2.9%higher than the SWT-LSTM model and SWT-SSA-LSTM model,respectively.