Scholarlink

<p>The safety risk management of tunnel construction is very important, while the traditional risk identification method relies on experience and expert judgment, but it is difficult to deal with sudden safety hazards in the construction process in the face of complex and dynamic construction environment. To solve these problems, this paper proposes a safety risk identification model of tunnel construction based on BPNN (Back Propagation Neural Network). A detailed index system is constructed to ensure the timeliness and accuracy of data through real-time collection and update. Invite experts to score, get the evaluation index weight. The model uses error backpropagation algorithm to train large-scale data, which can effectively deal with complex nonlinear relations and improve the accuracy of risk prediction. The results show that the BPNN model is better than the traditional Analytic Hierarchy Process (AHP) and Bayesian network in terms of prediction accuracy and real-time, with an average error of 0.0186, and with the increase of training time, the accuracy of BPNN is significantly improved. In the process of extending the training time from 10 to 100&#xa0;s, the BPNN’s accuracy improved from 82.34 percent to 97.81 percent. This model can effectively reduce the potential accident rate and provide scientific decision support for tunnel construction safety management.</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

Application of BP neural network model algorithm in safety risk identification of tunnel construction

  • 期刊论文

  • 发表日期：

    2025年1月19日

查看全文
  • Qiming Liu

摘要

The safety risk management of tunnel construction is very important, while the traditional risk identification method relies on experience and expert judgment, but it is difficult to deal with sudden safety hazards in the construction process in the face of complex and dynamic construction environment. To solve these problems, this paper proposes a safety risk identification model of tunnel construction based on BPNN (Back Propagation Neural Network). A detailed index system is constructed to ensure the timeliness and accuracy of data through real-time collection and update. Invite experts to score, get the evaluation index weight. The model uses error backpropagation algorithm to train large-scale data, which can effectively deal with complex nonlinear relations and improve the accuracy of risk prediction. The results show that the BPNN model is better than the traditional Analytic Hierarchy Process (AHP) and Bayesian network in terms of prediction accuracy and real-time, with an average error of 0.0186, and with the increase of training time, the accuracy of BPNN is significantly improved. In the process of extending the training time from 10 to 100 s, the BPNN’s accuracy improved from 82.34 percent to 97.81 percent. This model can effectively reduce the potential accident rate and provide scientific decision support for tunnel construction safety management.