在冻结法施工中,保证冻结壁稳定性至关重要,传统的现场检测方法因其间断性而无法提供实时监测,限制了对冻结壁潜在灾变的及时响应,采用冻土的深层原位精准探测是揭示冻结壁重大工程灾变机理及灾害预警的有效手段。基于卷积神经网络提出...在冻结法施工中,保证冻结壁稳定性至关重要,传统的现场检测方法因其间断性而无法提供实时监测,限制了对冻结壁潜在灾变的及时响应,采用冻土的深层原位精准探测是揭示冻结壁重大工程灾变机理及灾害预警的有效手段。基于卷积神经网络提出一种基于图像数据驱动的冻土强度智能识别方法,通过对93组试样的多角度图像捕获及随后的单轴抗压强度试验,标注试样图像与实际强度数据并结合图像数据增强技术,构建了深度学习模型训练所需的图像数据集;利用迁移学习深度残差网络34层(ResNet-34)模型,并对比其他不同模型的训练过程和测试结果,发现ResNet-34模型效果最佳,准确率为92.8%,且没有出现过拟合现象;应用深度学习模型对冻土强度的影响因素土质、温度和含水率进行识别,发现模型能有效识别出3个变量,证明了模型识别冻土强度的科学性和可靠性;此外研究了模型在不同干扰条件下的表现,模拟典型干扰场景并分析其对模型预测性能的影响,为后续改进数据增强策略和模型优化方向提供依据;引入Grad-CAM(Gradient-weighted Class Activation Mapping)可解释性分析方法揭示卷积神经网络在冻土强度识别任务中的内部特征提取过程,发现利用模型能够提取和分析冻土的关键图像特征,实现冻土强度的快速判识。研究成果为冻结壁状态的实时监控及早期灾害预警提供了新方法,可为冻结工程安全施工提供技术支撑。展开更多
Objective To comparatively observe the effect of electroacupuncture at digestive system-related lower he-sea points on the expressions of serum interleukin-1β（IL-1 β）, tumor necrosis factor-α（TNF-α） of colon t...Objective To comparatively observe the effect of electroacupuncture at digestive system-related lower he-sea points on the expressions of serum interleukin-1β（IL-1 β）, tumor necrosis factor-α（TNF-α） of colon tissues and high mobility group box 1 protein（HMGB 1） of ulcerative colitis（UC） model rats, and to explore whether there is relative specificity of electroacupuncture at Shàngjùxū（上巨虚 ST 37）, one of lower he-sea points of large intestine, in treatment of bowel diseases. Method A total of 60 SD rats were randomly divided into control group, model group, ST 37 group, Zúsānl？（足三里 ST 36） group, Xiàjùxū（下巨虚 ST 39） group and Yánglíngquán（阳陵泉 GB 34） group. There were ten rats in each group; five were males, and five were females. UC models were established by clysis with 2, 4, 6-trinitrobenzene sulfonic acid/alcohol solution. After modeling, treatment was conducted for ten days, specimens were collected, colonic ulcers and inflammation were inspected visually and scored. The content of serum IL-1β and the expressions of TNF-α and HMGB 1 in colon were detected through ELISA. Results 1 Compared with control group, the scores of colonic ulcers and inflammation, the content of serum IL-1β and the expressions of TNF-α（except ST 37 group） and HMGB 1 were all higher（P〈0.05, P〈0.01）; 2 compared with model group, the scores of colonic ulcers in ST 36 group and ST 37 group were lower obviously（P〈0.05, P〈0.01）; the expressions of IL-1β, TNF-α and HMGB 1 in the four treatment groups were lower obviously（P〈0.01）; 3 compared with ST 37 group, the expressions of IL-1β, TNF-α and HMGB 1 in other three treatment groups were higher obviously（P〈0.05, P〈0.01）; and the scores of colonic ulcers in ST 39 group and GB 34 group were higher obviously（P〈0.05）. Conclusion 1 The score of colonic ulcers can be reduced through electroacupuncture at ST 37, ST 36, ST 39 and GB 34, which can also reduce the content of serum IL-1β and the expressions of TNF-α and HMGB 1, and effectively inhibit inflammatory response of colon caused by UC; 2 the effect trend of the four acupoints in treatment of UC is： ST 37ST 36ST 39GB 34, and electroacupuncture at ST 37 has the best effect with relative specificity.展开更多
目的:(1)探讨番泻叶煎剂联合氢化可的松造成大鼠脾肾阳虚动物模型的最佳剂量。(2)结合现代研究,把中医证型与西医疾病结合起来,找到建立UC(脾肾阳虚型)模型的造模方法。方法:(1)采用不同浓度(8%、15%、20%、25%、30%)番泻叶煎剂2 m L/10...目的:(1)探讨番泻叶煎剂联合氢化可的松造成大鼠脾肾阳虚动物模型的最佳剂量。(2)结合现代研究,把中医证型与西医疾病结合起来,找到建立UC(脾肾阳虚型)模型的造模方法。方法:(1)采用不同浓度(8%、15%、20%、25%、30%)番泻叶煎剂2 m L/100 g灌胃和氢化可的松10 mg/kg颈部皮下注射15 d造脾肾阳虚动物模型。(2)用2,4二硝基氯苯-乙酸法造大鼠慢性溃疡性结肠炎模型。(3)2周后处死动物进行组织形态学评分及光镜评估。结果:实验各组动物均有不同程度出现大便稀溏、进食减少、体重减轻、肛温降低、弓背蜷缩、神态萎靡、反应迟钝、四肢不收、毛色枯槁、活动减少等脾肾阳虚的症状,但以15%番泻叶组造模效果最好。病理切片提示二硝基氯苯-乙酸法能够形成典型的慢性UC。结论:用15%番泻叶联合氢化可的松、二硝基氯苯(DNCB)-乙酸法建立脾肾阳虚型慢性溃疡性结肠炎的动物模型效果好,模型简单易行,重复性好,其症状与人UC(脾肾阳虚型)症状相似。展开更多
文摘在冻结法施工中,保证冻结壁稳定性至关重要,传统的现场检测方法因其间断性而无法提供实时监测,限制了对冻结壁潜在灾变的及时响应,采用冻土的深层原位精准探测是揭示冻结壁重大工程灾变机理及灾害预警的有效手段。基于卷积神经网络提出一种基于图像数据驱动的冻土强度智能识别方法,通过对93组试样的多角度图像捕获及随后的单轴抗压强度试验,标注试样图像与实际强度数据并结合图像数据增强技术,构建了深度学习模型训练所需的图像数据集;利用迁移学习深度残差网络34层(ResNet-34)模型,并对比其他不同模型的训练过程和测试结果,发现ResNet-34模型效果最佳,准确率为92.8%,且没有出现过拟合现象;应用深度学习模型对冻土强度的影响因素土质、温度和含水率进行识别,发现模型能有效识别出3个变量,证明了模型识别冻土强度的科学性和可靠性;此外研究了模型在不同干扰条件下的表现,模拟典型干扰场景并分析其对模型预测性能的影响,为后续改进数据增强策略和模型优化方向提供依据;引入Grad-CAM(Gradient-weighted Class Activation Mapping)可解释性分析方法揭示卷积神经网络在冻土强度识别任务中的内部特征提取过程,发现利用模型能够提取和分析冻土的关键图像特征,实现冻土强度的快速判识。研究成果为冻结壁状态的实时监控及早期灾害预警提供了新方法,可为冻结工程安全施工提供技术支撑。
基金Supported by National Natural Science Foundation of China:81173327/H2718
文摘Objective To comparatively observe the effect of electroacupuncture at digestive system-related lower he-sea points on the expressions of serum interleukin-1β（IL-1 β）, tumor necrosis factor-α（TNF-α） of colon tissues and high mobility group box 1 protein（HMGB 1） of ulcerative colitis（UC） model rats, and to explore whether there is relative specificity of electroacupuncture at Shàngjùxū（上巨虚 ST 37）, one of lower he-sea points of large intestine, in treatment of bowel diseases. Method A total of 60 SD rats were randomly divided into control group, model group, ST 37 group, Zúsānl？（足三里 ST 36） group, Xiàjùxū（下巨虚 ST 39） group and Yánglíngquán（阳陵泉 GB 34） group. There were ten rats in each group; five were males, and five were females. UC models were established by clysis with 2, 4, 6-trinitrobenzene sulfonic acid/alcohol solution. After modeling, treatment was conducted for ten days, specimens were collected, colonic ulcers and inflammation were inspected visually and scored. The content of serum IL-1β and the expressions of TNF-α and HMGB 1 in colon were detected through ELISA. Results 1 Compared with control group, the scores of colonic ulcers and inflammation, the content of serum IL-1β and the expressions of TNF-α（except ST 37 group） and HMGB 1 were all higher（P〈0.05, P〈0.01）; 2 compared with model group, the scores of colonic ulcers in ST 36 group and ST 37 group were lower obviously（P〈0.05, P〈0.01）; the expressions of IL-1β, TNF-α and HMGB 1 in the four treatment groups were lower obviously（P〈0.01）; 3 compared with ST 37 group, the expressions of IL-1β, TNF-α and HMGB 1 in other three treatment groups were higher obviously（P〈0.05, P〈0.01）; and the scores of colonic ulcers in ST 39 group and GB 34 group were higher obviously（P〈0.05）. Conclusion 1 The score of colonic ulcers can be reduced through electroacupuncture at ST 37, ST 36, ST 39 and GB 34, which can also reduce the content of serum IL-1β and the expressions of TNF-α and HMGB 1, and effectively inhibit inflammatory response of colon caused by UC; 2 the effect trend of the four acupoints in treatment of UC is： ST 37ST 36ST 39GB 34, and electroacupuncture at ST 37 has the best effect with relative specificity.
