Scholarlink

<p>In the recent past there is an increasing demand to monitor and regulate the traffic flow and implement an efficient traffic management system. This can be achieved by implementing dynamic and proactive traffic control systems. Traffic flow forecasting is used as an essential tool for developing a traffic control system in intelligent transportation systems (ITS). In this paper, traffic flow forecasting is presented using LSTM. LSTM is used to forecast traffic flow patterns and model temporal dependencies, enabling more accurate predictions of transport system behavior. Bio-inspired algorithms are commonly employed to optimize the model’s parameters to improve the performance. In the present work, optimizing the performance of LSTM parameters is initiated with the Cuckoo Search Evolutionary Algorithm. The cuckoo search optimization is used in finding the optimum weights and bias values for the LSTM network. Finally, based on the error, the weights are dynamically adjusted in the LSTM network. The results of this forecast analysis propagate that the Cuckoo Search has improved the performance of LSTM with the R2 of 0.98.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Day-Ahead Traffic Flow Forecast Using LSTM and Cuckoo Search Optimization

  • 期刊论文

  • 发表日期：

    2025年8月14日

查看全文
  • V. Rajalakshmi,
  • P. Sharon Femi,
  • A. Kala

摘要

In the recent past there is an increasing demand to monitor and regulate the traffic flow and implement an efficient traffic management system. This can be achieved by implementing dynamic and proactive traffic control systems. Traffic flow forecasting is used as an essential tool for developing a traffic control system in intelligent transportation systems (ITS). In this paper, traffic flow forecasting is presented using LSTM. LSTM is used to forecast traffic flow patterns and model temporal dependencies, enabling more accurate predictions of transport system behavior. Bio-inspired algorithms are commonly employed to optimize the model’s parameters to improve the performance. In the present work, optimizing the performance of LSTM parameters is initiated with the Cuckoo Search Evolutionary Algorithm. The cuckoo search optimization is used in finding the optimum weights and bias values for the LSTM network. Finally, based on the error, the weights are dynamically adjusted in the LSTM network. The results of this forecast analysis propagate that the Cuckoo Search has improved the performance of LSTM with the R2 of 0.98.