Research on cold chain logistics demand forecast based on LSTM model

Abstract

Cold chain logistics demand is affected by multiple factors and exhibits significant nonlinear, dynamic and time series dependence characteristics. Traditional forecasting methods are difficult to accurately capture its long-term changing patterns, resulting in large deviations in forecast results. In order to improve the accuracy of cold chain logistics demand forecasting and optimize industry resource allocation, infrastructure construction and development planning, this article takes the total national cold chain logistics demand as the research object, selects 12 core indicators from macroeconomics, population consumption, industry development and other dimensions as influencing factors, collects relevant statistical data from 2005 to 2024, and builds a cold chain logistics demand forecasting model based on long short-term memory network (LSTM). The core parameters of the LSTM model are optimized based on time series characteristics, and the prediction accuracy of the model is verified through three indicators: root mean square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). Research has confirmed that the LSTM model has excellent applicability and accuracy in cold chain logistics demand forecasting, and its prediction results can effectively support the scientific and intelligent development of the cold chain logistics industry.