基于自监督学习的公路隧道照明节能算法  

Energy-saving Algorithm for Highway Tunnel LightingBased on Self-supervised Learning

作　　者：罗志伟 丘岳培 LUO Zhiwei;QIU Yuepei(Guangxi Fuhe Expressway Co.,Ltd,Nanning 530022,China;Guangxi Zhonghe Expressway Co.,Ltd,Nanning 530022,China)

机构地区：[1]广西富贺高速公路有限公司,广西南宁530022 [2]广西钟贺高速公路有限公司,广西南宁530022

出　　处：《常州工学院学报》2024年第4期38-43,共6页Journal of Changzhou Institute of Technology

摘　　要：针对公路隧道照明耗电大、费用高和行车安全受照明影响的问题,提出一种基于深度学习的公路隧道照明矩阵优化算法。该算法有效利用公路隧道里的现有设备,将ResNet和YOLO的骨干模型进行组合,结合行车安全标识和自监督学习技术,对损失函数端进行软约束,在确保行车安全和降低照明能耗的前提下,自动输出隧道光照参数矩阵,并利用该矩阵对隧道内的照明设备进行动态调节。实验表明,提出的算法在确保隧道照明能够保障行车安全的同时,降低了隧道照明能耗。A deep learning based optimization algorithm for highway tunnel lighting matrix is proposed to address the issues of high power consumption,high cost,and the impact of lighting on driving safety in highway tunnels.The algorithm effectively utilizes the existing equipment in the highway tunnel,combines the backbone models of ResNet and YOLO,and imposes a soft constraint on the loss function end on the basis of integrating the driving safety signs and self-supervised learning technology.While ensuring driving safety and reducing lighting energy consumption,it automatically outputs a tunnel lighting parameter matrix and dynamically adjusts the lighting equipment in the tunnel using this matrix.The experiment shows that the proposed algorithm reduces the energy consumption of tunnel lighting while ensuring that the tunnel lighting can guarantee driving safety.

关 键 词：公路隧道 安全节能 深度学习 机器视觉 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] U453.7[自动化与计算机技术—计算机科学与技术]