| In recent years,with the popularity of e-commerce,a new economic form under the background of "internet plus",express delivery industry,as the core carrier of communication between upstream and downstream of the industrial chain,connecting merchants and consumers,has gradually become an important part of residents’ lives and economic and social development.According to the latest data from the State Post Bureau,the business volume of China’s express delivery industry has been increasing year by year since 2010.In 2020,the business volume of China’s express delivery exceeded 83.3 billion pieces,and the annual increment of express delivery business volume(19.86 billion pieces)was close to that of 2015.Facing the intensified competition of market homogenization and the high management and operation costs of express delivery enterprises,this study focuses on the prediction method of daily average express delivery business volume based on deep learning algorithm,in order to provide technical support for the daily dynamic resource allocation and personnel deployment of enterprises,promote the development and transformation of enterprises towards digitalization and intelligence,and realize the flexibility,diversification and precision of management mode.This study mainly has the following three research contents and conclusions:(1)A provincial express business forecasting model based on GRU deep learning algorithm is proposed.Based on pre-processing,the experiment is carried out by using Y company’s provincial express business data set,and compared with other classical algorithms,and the effectiveness of the model is discussed in three situations: ordinary days,holidays and e-commerce festivals.Experimental results show that the algorithm in this chapter can accurately and quickly predict the daily average express business volume.The prediction accuracy of the algorithm in this study is as high as 97.50%.At the same time,this method has good anti-interference performance.In special periods such as holidays and e-commerce activities,slight fluctuations or sharp increase in data volume will not have a great impact on the final prediction results of express delivery business volume.(2)Using GRA method to select the relevant influencing factors of express delivery business volume forecast,and fusing them with the data set,then using PCA to reduce the dimension of the data,and selecting the main principal components to replace the multi-features of the original data for further research.In this paper,a prediction model of municipal express traffic based on PCA-LSTM algorithm is proposed.Compared with the model based on LSTM,GRU and Cat Boost algorithm,the prediction accuracy of this algorithm is as high as 96.24%.The performance of the model under normal days,holidays,e-commerce festivals and multivariate conditions is discussed.It is found that the model has good anti-interference performance in special period,but the multivariate data fusion has little influence on the model.(3)Using the user interface toolbox in MATLAB R2020 b,the express traffic forecasting software based on deep learning algorithm is developed and designed.The software mainly includes the provincial express business forecasting algorithm based on GRU algorithm and the municipal express business forecasting algorithm based on PCA-LSTM algorithm,and is tested by using the data of Y Company in Shaanxi Province and other cities.The experimental results show that the software can complete the integration of this research algorithm and realize the prediction of the express delivery business volume of the research target.Its advantage lies in its simple operation and visual processing process and results. |
