Scholarlink

<p>With the increasing importance of the coal mining industry in the global energy structure, the safety of underground coal mine operations has become an urgent international challenge. Especially, it is crucial to search underground targets quickly and accurately. Aiming at the challenge of balancing accuracy and efficiency in search tasks, this study proposes a multidimensional optimization method. First, multiple single features of search target and database target are extracted, and the features are binarized based on entropy to maximize efficiency with minimum loss of accuracy. Second, a similarity metric is calculated based on the binarized single features, and single feature weights are obtained using average peak correlation energy and correlation filtering consistency, which enable the system to adaptively select features to obtain the best results. Next, the similarity metrics of multiple single features are weighted to obtain a composite similarity metric, which is then ranked to return initial results and optimized to ensure that the best results are returned. Experimental results show that, compared with ShuffleNetV2, VGG16, ResNet18, DenseNet121, AlexNet, ConvNext, SqueezeNet, GoogleNet, EfficientNet, MnasNet, RegNet, MobileNet, the method proposed improves accuracy by 0.20%, 3.58%, 1.68%, 5.38%, 0.64%, 3.41%, 2.10%, 1.51%, 4.37%, 1.10%, 3.01%, 2.96%, and reduces efficiency by 0.1077&#xa0;s, 0.4764&#xa0;s, 0.0395&#xa0;s, 4.5547&#xa0;s, 0.4080&#xa0;s, 0.0872&#xa0;s, 0.0932&#xa0;s, 0.0925&#xa0;s, 0.1077&#xa0;s, 0.1457&#xa0;s, 0.0787&#xa0;s, and 0.1223&#xa0;s to be able to achieve better performance.</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

An adaptive multidimensional optimization method for goal search tasks in underground coal mines

  • 期刊论文

  • 发表日期：

    2025年4月16日

查看全文
  • Jiaojuan Wang,
  • Rui Shi,
  • Hongbin Dou,
  • Tongyang Dao,
  • Xuefeng Zhu,
  • Baoyu Wang

摘要

With the increasing importance of the coal mining industry in the global energy structure, the safety of underground coal mine operations has become an urgent international challenge. Especially, it is crucial to search underground targets quickly and accurately. Aiming at the challenge of balancing accuracy and efficiency in search tasks, this study proposes a multidimensional optimization method. First, multiple single features of search target and database target are extracted, and the features are binarized based on entropy to maximize efficiency with minimum loss of accuracy. Second, a similarity metric is calculated based on the binarized single features, and single feature weights are obtained using average peak correlation energy and correlation filtering consistency, which enable the system to adaptively select features to obtain the best results. Next, the similarity metrics of multiple single features are weighted to obtain a composite similarity metric, which is then ranked to return initial results and optimized to ensure that the best results are returned. Experimental results show that, compared with ShuffleNetV2, VGG16, ResNet18, DenseNet121, AlexNet, ConvNext, SqueezeNet, GoogleNet, EfficientNet, MnasNet, RegNet, MobileNet, the method proposed improves accuracy by 0.20%, 3.58%, 1.68%, 5.38%, 0.64%, 3.41%, 2.10%, 1.51%, 4.37%, 1.10%, 3.01%, 2.96%, and reduces efficiency by 0.1077 s, 0.4764 s, 0.0395 s, 4.5547 s, 0.4080 s, 0.0872 s, 0.0932 s, 0.0925 s, 0.1077 s, 0.1457 s, 0.0787 s, and 0.1223 s to be able to achieve better performance.