The synthesis of methacrylic acid from biomass-derived itaconic acid is a green route,for it can get rid of the dependence on fossil resource.In order to solve the problems on this route such as use of a preciousmetal...The synthesis of methacrylic acid from biomass-derived itaconic acid is a green route,for it can get rid of the dependence on fossil resource.In order to solve the problems on this route such as use of a preciousmetal catalyst and a corrosive homogeneous alkali,we prepared a series of hydroxyapatite catalysts by an ionic liquid-assisted hydrothermal method and evaluated their catalytic performance.The results showed that the ionic liquid[Bmim]BF_(4) can affect the crystal growth of hydroxyapatite,provide fluoride ion for fluorination of hydroxyapatite,and adjust the surface acidity and basicity,morphology,textural properties,crystallinity,and composition of hydroxyapatite.The[Bmim]BF4 dosage and hydrothermal temperature can affect the fluoride ion concentration in the hydrothermal system,thus changing the degree of fluoridation of hydroxyapatite.High fluoride-ion concentration can lead to the formation of CaF_(2) and thus significantly decrease the catalytic performance of hydroxyapatite.The hydrothermal time mainly affects the growth of hydroxyapatite crystals on the c axis,leading to different catalytic performance.The suitable conditions for the preparation of this fluoridized hydroxyapatite are as follows:a mass ratio of[Bmim]BF4 to calcium salt=0.2:1,a hydrothermal time of 12 h,and a hydrothermal temperature of 130℃.A maximal methacrylic acid yield of 54.7%was obtained using the fluoridized hydroxyapatite under relatively mild reaction conditions(250℃ and 2 MPa of N_(2))in the absence of a precious-metal catalyst and a corrosive homogeneous alkali.展开更多
The unperturbed dimension and temperature character of poly(dibenzyl itaconate)s (PDBzI) are studied by a revised rotational isomeric state (RIS) method. The improved formulas of the mean-square radius of gyration, de...The unperturbed dimension and temperature character of poly(dibenzyl itaconate)s (PDBzI) are studied by a revised rotational isomeric state (RIS) method. The improved formulas of the mean-square radius of gyration, deduced by the pseudo-stereochemical equilibrium approach, may be used to investigate the configurational-conformational properties of atactic polymers with large side groups [poly(itaconates) for instance]. The calculated results showed that poly(itaconates) have larger dimension of the molecule than other vinyl polymers. Comparison of the dimension between considering and without considering side groups showed that the effect of large side groups on the unperturbed dimension for short-chain polymers is more obvious than that of long-chain polymers and, if the dimension of side groups increases, the effect also increases. The dimension differences of PDBzI between short-chain and long-chain polymers are investigated by the relation of characteristic ratios and temperature coefficients with temperature. Moreover, the dependence between the temperature coefficients and the tacticity of chains shows that the temperature characters of the isotactic, syndiotactic and atactic PDBzI chains have remarkable difference.展开更多
[目的]本文旨在解决在自然环境下不同成熟度苹果目标检测精度较低的问题。[方法]提出了一种改进的YOLOv5s模型SODSTR-YOLOv5s(YOLOv5s with small detection layer and omni-dimensional dynamic convolution and swin transformer bloc...[目的]本文旨在解决在自然环境下不同成熟度苹果目标检测精度较低的问题。[方法]提出了一种改进的YOLOv5s模型SODSTR-YOLOv5s(YOLOv5s with small detection layer and omni-dimensional dynamic convolution and swin transformer block),用于不同成熟度苹果检测。首先改进YOLOv5s的多尺度目标检测层,在Prediction中构建检测160×160特征图的检测头,提高小尺寸的不同成熟度苹果的检测精度;其次在Backbone结构中融合Swin Transformer Block,加强同级成熟度的苹果纹理特征融合,弱化纹理特征分布差异带来的消极影响,提高模型泛化能力;最后将Neck结构的Conv模块替换为动态卷积模块ODConv,细化局部特征映射,实现局部苹果细粒度特征的充分提取。基于不同成熟度苹果数据集进行试验,验证改进模型的性能。[结果]改进模型SODSTR-YOLOv5s检测的精确率、召回率、平均精度均值分别为89.1%、95.5%、93.6%,高、中、低成熟度苹果平均精度均值分别为94.1%、93.1%、93.7%,平均检测时间为16 ms,参数量为7.34 M。相比于YOLOv5s模型,改进模型SODSTR-YOLOv5s精确率、召回率、平均精度均值分别提高了3.8%、5.0%、2.9%,参数量和平均检测时间分别增加了0.32 M和5 ms。[结论]改进模型SODSTR-YOLOv5s提升了在自然环境下对不同成熟度苹果的检测能力,能较好地满足实际采摘苹果的检测要求。展开更多
针对煤矿高噪声、低照度、运动模糊与大批量煤矸混杂等复杂工况环境因素导致煤矸识别存在误检、漏检以及检测精度低的问题,提出一种基于CFS-YOLO算法的煤矸智能识别模型。采用ConvNeXt V2(Convolutional Neural Network with NeXt Units...针对煤矿高噪声、低照度、运动模糊与大批量煤矸混杂等复杂工况环境因素导致煤矸识别存在误检、漏检以及检测精度低的问题,提出一种基于CFS-YOLO算法的煤矸智能识别模型。采用ConvNeXt V2(Convolutional Neural Network with NeXt Units Version 2)特征提取模块替换主干网络末端的2个C3(Cross Stage Partial Bottle Neck Mudule)模块,通过将掩码自动编码器(Masked Autoencoders,MAE)和全局响应归一化(Global Response Normalization,GRN)层添加到ConvNeXt架构中,有效缓解特征崩溃问题以及保持特征在网络传递过程中的多样性;采用Focal-EIOU(Focal and Efficient Intersection Over Union)损失函数替换原CIOU(Computer Intersection Over Union)损失函数,通过其Focal-Loss机制和调整样本权重的方式优化边界框回归任务中的样本不平衡问题,提高模型的收敛速度和定位精度;添加无参注意力机制(Simple Attention Mechanism,SimAM)于主干网络每个C3模块的后端,凭借其注意力权重自适应调整策略,提升模型对尺度变化较大或低分辨率煤矸目标关键特征的提取能力。通过消融试验和对比试验验证所提CFS-YOLO模型的有效性与优越性。试验结果表明:CFS-YOLO模型对于煤矸在煤矿高噪声、低照度、运动模糊与大批量煤矸混杂等复杂环境下的检测效果均得到有效提高,模型的平均精度均值达到90.2%,相较于原YOLOv5s模型的平均精度均值提高了3.7%,平均检测速度达到90.09 FPS,可充分满足煤矸实时检测的需求。同时与YOLOv5s、YOLOv7-tiny与YOLOv8n等6种YOLO系列算法相比,CFS-YOLO模型对煤矿复杂环境的适应性最强且综合检测性能最佳,可为煤矸的智能高效分选提供技术支持。展开更多
基金supported by National Natural Science Foundation of China(21978066)Basic Research Program of Hebei Province for Natural Science Foundation and Key Basic Research Project(18964308D)the Key Program of Natural Science Foundation of Hebei Province(B2020202048).
文摘The synthesis of methacrylic acid from biomass-derived itaconic acid is a green route,for it can get rid of the dependence on fossil resource.In order to solve the problems on this route such as use of a preciousmetal catalyst and a corrosive homogeneous alkali,we prepared a series of hydroxyapatite catalysts by an ionic liquid-assisted hydrothermal method and evaluated their catalytic performance.The results showed that the ionic liquid[Bmim]BF_(4) can affect the crystal growth of hydroxyapatite,provide fluoride ion for fluorination of hydroxyapatite,and adjust the surface acidity and basicity,morphology,textural properties,crystallinity,and composition of hydroxyapatite.The[Bmim]BF4 dosage and hydrothermal temperature can affect the fluoride ion concentration in the hydrothermal system,thus changing the degree of fluoridation of hydroxyapatite.High fluoride-ion concentration can lead to the formation of CaF_(2) and thus significantly decrease the catalytic performance of hydroxyapatite.The hydrothermal time mainly affects the growth of hydroxyapatite crystals on the c axis,leading to different catalytic performance.The suitable conditions for the preparation of this fluoridized hydroxyapatite are as follows:a mass ratio of[Bmim]BF4 to calcium salt=0.2:1,a hydrothermal time of 12 h,and a hydrothermal temperature of 130℃.A maximal methacrylic acid yield of 54.7%was obtained using the fluoridized hydroxyapatite under relatively mild reaction conditions(250℃ and 2 MPa of N_(2))in the absence of a precious-metal catalyst and a corrosive homogeneous alkali.
基金Project supported by the National Basic Research Program (973) of China (No. 10574109)the Zhejiang Provincial Science and Technology Department (No. 2005C24008)the Natural Science Foundation of Zhejiang Province (No. Y604064), China
文摘The unperturbed dimension and temperature character of poly(dibenzyl itaconate)s (PDBzI) are studied by a revised rotational isomeric state (RIS) method. The improved formulas of the mean-square radius of gyration, deduced by the pseudo-stereochemical equilibrium approach, may be used to investigate the configurational-conformational properties of atactic polymers with large side groups [poly(itaconates) for instance]. The calculated results showed that poly(itaconates) have larger dimension of the molecule than other vinyl polymers. Comparison of the dimension between considering and without considering side groups showed that the effect of large side groups on the unperturbed dimension for short-chain polymers is more obvious than that of long-chain polymers and, if the dimension of side groups increases, the effect also increases. The dimension differences of PDBzI between short-chain and long-chain polymers are investigated by the relation of characteristic ratios and temperature coefficients with temperature. Moreover, the dependence between the temperature coefficients and the tacticity of chains shows that the temperature characters of the isotactic, syndiotactic and atactic PDBzI chains have remarkable difference.
文摘[目的]本文旨在解决在自然环境下不同成熟度苹果目标检测精度较低的问题。[方法]提出了一种改进的YOLOv5s模型SODSTR-YOLOv5s(YOLOv5s with small detection layer and omni-dimensional dynamic convolution and swin transformer block),用于不同成熟度苹果检测。首先改进YOLOv5s的多尺度目标检测层,在Prediction中构建检测160×160特征图的检测头,提高小尺寸的不同成熟度苹果的检测精度;其次在Backbone结构中融合Swin Transformer Block,加强同级成熟度的苹果纹理特征融合,弱化纹理特征分布差异带来的消极影响,提高模型泛化能力;最后将Neck结构的Conv模块替换为动态卷积模块ODConv,细化局部特征映射,实现局部苹果细粒度特征的充分提取。基于不同成熟度苹果数据集进行试验,验证改进模型的性能。[结果]改进模型SODSTR-YOLOv5s检测的精确率、召回率、平均精度均值分别为89.1%、95.5%、93.6%,高、中、低成熟度苹果平均精度均值分别为94.1%、93.1%、93.7%,平均检测时间为16 ms,参数量为7.34 M。相比于YOLOv5s模型,改进模型SODSTR-YOLOv5s精确率、召回率、平均精度均值分别提高了3.8%、5.0%、2.9%,参数量和平均检测时间分别增加了0.32 M和5 ms。[结论]改进模型SODSTR-YOLOv5s提升了在自然环境下对不同成熟度苹果的检测能力,能较好地满足实际采摘苹果的检测要求。
文摘针对煤矿高噪声、低照度、运动模糊与大批量煤矸混杂等复杂工况环境因素导致煤矸识别存在误检、漏检以及检测精度低的问题,提出一种基于CFS-YOLO算法的煤矸智能识别模型。采用ConvNeXt V2(Convolutional Neural Network with NeXt Units Version 2)特征提取模块替换主干网络末端的2个C3(Cross Stage Partial Bottle Neck Mudule)模块,通过将掩码自动编码器(Masked Autoencoders,MAE)和全局响应归一化(Global Response Normalization,GRN)层添加到ConvNeXt架构中,有效缓解特征崩溃问题以及保持特征在网络传递过程中的多样性;采用Focal-EIOU(Focal and Efficient Intersection Over Union)损失函数替换原CIOU(Computer Intersection Over Union)损失函数,通过其Focal-Loss机制和调整样本权重的方式优化边界框回归任务中的样本不平衡问题,提高模型的收敛速度和定位精度;添加无参注意力机制(Simple Attention Mechanism,SimAM)于主干网络每个C3模块的后端,凭借其注意力权重自适应调整策略,提升模型对尺度变化较大或低分辨率煤矸目标关键特征的提取能力。通过消融试验和对比试验验证所提CFS-YOLO模型的有效性与优越性。试验结果表明:CFS-YOLO模型对于煤矸在煤矿高噪声、低照度、运动模糊与大批量煤矸混杂等复杂环境下的检测效果均得到有效提高,模型的平均精度均值达到90.2%,相较于原YOLOv5s模型的平均精度均值提高了3.7%,平均检测速度达到90.09 FPS,可充分满足煤矸实时检测的需求。同时与YOLOv5s、YOLOv7-tiny与YOLOv8n等6种YOLO系列算法相比,CFS-YOLO模型对煤矿复杂环境的适应性最强且综合检测性能最佳,可为煤矸的智能高效分选提供技术支持。
