基于季节调整的GRU-BiLSTM旅游客流量预测方法

打开文本图片集

中图分类号：TP391.9 文献标志码：A 文章编号：1673-5072（2025）03-0318-07

Abstract：Addressing issues such as weak non-linear processing capability and low prediction accuracy for strong seasonal data while using single forecasting models in short-term tourist flow，a combined prediction model based on Seasonally Adjusted Gated Recurrent Unit（GRU）and Bidirectional Long Short-Term Memory（BiLSTM）was developed for short-term tourist flow forecasting.Initially，seasonal indices were used to adjust the data to eliminate seasonal impacts.Subsequently，GRU was utilized for its fewer parameters and high training eficiency in non-linear modeling.Finally，BiLSTM was applied in a bidirectional manner to capture the characteristics of time series data more comprehensively.Taking tourists and weather data from Jiuzhaigou scenic area as experimental subjects，the results indicate that the GRU-BiLSTM combined model outperforms the single models by 23% to 44% on four evaluation indicators.Additionally，the seasonally adjusted GRU-BiLSTM combined model shows improvement of 13% to 39% in accuracy on the same indicators compared to that before adjustment.

Keywords： passenger flow prediction;seasonal adjustment; deep learning;gated recurrent unit; bidirectional long short-term memory network

近年来，旅游业蓬勃发展，逐渐成为经济增长和社会发展的重要推动力[1]。（剩余8505字）