首页 | 本学科首页   官方微博 | 高级检索  
相似文献
 共查询到19条相似文献,搜索用时 125 毫秒
1.
基于小波神经网络的木材内部缺陷类型识别的研究   总被引:7,自引:0,他引:7  
齐巍  王立海 《林业科学》2006,42(8):63-68
利用榆木标准试件,在实验室内用超声波检测仪器对试件进行缺陷分类检测,检测信号作为原始信息.各类试件的原始信号用小波包分解,计算缺陷试件与完好试件在小波包第5层各结点的信号能量变化值.试验发现:木材缺陷引起能量的变化值主要由木材缺陷的大小或严重程度来决定,亦即木材的缺陷程度越严重,能量的变化幅度就越大;对小波包5层分解后各信号结点的能量变化值进行分析,发现在32个结点中,(5,0)结点在各类缺陷试件中能量值变化最大;使用经小波压缩后的信号作为神经网络的输入,形成应用频带能量变化值和应用(5,0)结点小波包系数的2个不同输入特征的人工神经网络.对比分析2个网络识别木材缺陷类型的能力,(5,0)结点小波包系数作为特征训练得到的神经网络检测精度更高.  相似文献   

2.
【目的】针对木材蛀干害虫羽化孔洞缺陷检测问题,通过对声发射信号的时频分析,研究木材蛀干害虫羽化孔洞缺陷的AE信号特征。【方法】首先,对无孔洞和3种不同尺寸的钻孔缺陷的木材试件,参照ASTM-E976标准采用铅芯折断方式产生AE源,通过采样频率为500 kHz的2通道木材声发射信号采集系统获取原始AE信号。然后,对原始AE信号进行降噪滤波,再对滤波后的信号进行小波分解并重构AE波形,分析孔径对AE信号频率分布的影响。最后,采用信号相关性分析法和时差定位法,计算AE信号的传播速率,进而分析羽化孔径对传播速率的影响。【结果】AE信号通过无孔洞、5 mm孔洞、8 mm孔洞、15 mm孔洞缺陷,AE信号的主频率分别为38、43、51、117 kHz,AE信号传播的平均速率分别为1 380.7、1 067.3、848.6、437.1 m/s。【结论】AE信号在不同孔径缺陷的木材试件传播时,随着孔洞缺陷的增大,AE信号的幅值发生明显衰减,AE信号的主频率增大,AE信号传播的平均速率减小。为了验证试验结果和结论,找了具有天然蛀干害虫羽化孔洞缺陷的木材试件,孔洞直径约为20 mm,使用相同的方法采集和处理AE信号,AE信号通过天然蛀干害虫羽化孔洞缺陷的主频率为121 kHz,AE信号传播的平均速率为324 m/s。  相似文献   

3.
由于小波基缺乏平移不变性,传统小波及小波包去噪算法可能使信号急剧变化部分产生人为振荡现象.提出了基于平移不变的小波包去噪方法,对所分析的信号进行循环平移,利用软或硬阀值对该信号的小波包系数进行压缩,重构信号,再进行相反的循环平移,通过多次的平移—消噪—平移,平均所获得的结果,从而消除小波包基的平移依赖性.对比普通小波包去噪,该方法能有效地消除人为振荡现象,使去噪后的信号更光滑,更逼近真实信号.  相似文献   

4.
不同含水率云南松声发射信号特征   总被引:1,自引:0,他引:1  
李杨  许飞云 《林业科学》2019,(6):96-102
【目的】研究不同含水率条件下云南松试件声发射信号的传播规律,探讨含水率对声发射信号波形的响应,为云南松声发射源定位提供依据,为木材内部缺陷无损检测提供基础数据。【方法】以云南松为试验材料,采用NI高速数据采集设备和LabVIEW软件构建云南松试件声发射信号采集平台,利用铅芯折断模拟声发射源,对绝干、气干、生材和饱水4种含水率状态下的试件表面进行声发射信号采集,通过时差法计算信号在4种含水率状态下的平均传播速率,并运用小波分析对声发射信号波形进行分解和重构,根据软阈值量化方法消除各通道系数和经阈值量化后的各高频层系数,去除非主能量信号以便从噪声中析取微弱的声发射信号。【结果】试验中传感器均以接收表面波信号为主;随着含水率增加,云南松试件表面声发射信号波形和平均声速均大幅度衰减,绝干状态下声发射信号时域波形幅值达到±5.2 V,平均声速可达4 208.77 m·s^-1 ,而饱水状态下信号时域波形幅值仅为±0.6 V,平均声速降至1 414.07 m·s^-1 ,气干和生材状态下信号时域波形幅值和平均声速分别为±4、±2 V和 3 331.79 、 2 328.73 m·s^-1 ,且各含水率状态之间平均声速差在876.98~1 003.06 m·s^-1 范围内;小波变换能有效将“淹没”在噪声中的声发射信号析取出来,4种含水率状态下试件声发射信号频域波形范围在40 ~150 kHz之间,且气干状态下波形峰值出现在110 kHz左右,其余3种均在50 kHz左右达到峰值。【结论】含水率增加可显著改变云南松声发射信号和传播特征,其信号波形和平均声速均与含水率降低呈正比;小波变换在信号降噪处理方面具有明显优势,不仅可去除信号中的大量噪音,而且不会破坏有用信号,保证信号完整性,能更大程度降低对不同含水率云南松试件声发射信号的分析误差,为云南松表面声发射源定位研究及内部无损检测给予试验数据支持;作为一种木材声发射信号采集与分析平台,研究结果可为不同含水率云南松受压变形破坏全过程的声发射特征分析提供必要的基础理论依据。  相似文献   

5.
针对木结构中常见的榫卯连接与螺栓连接节点,利用压电传感技术对其进行健康工况和损伤工况的数值模拟研究,其中榫卯连接考虑榫头的平动、转动、缺损工况,螺栓连接考虑预紧力松动工况;对压电片所接收到的信号进行以小波包分解,并以各频段信号能量的总和作为能量指标,对不同损伤工况进行定量评估,最后以螺栓连接为例进行实验验证。数值计算和实验结果均表明,压电片所接收到的信号能量随损伤程度的不断增加而减小,利用压电传感技术可以有效识别不同类型连接节点处不同程度的损伤,该结论可为工程实际提供参考。  相似文献   

6.
基于振动模态分析的落叶松节子定位的无损检测   总被引:2,自引:0,他引:2  
木材作为结构材使用时,节子的存在影响木材的强度,降低了木材的利用价值.为了节约木材,提高木材的使用率,人们一直在研究木材缺陷无损检测的方法.Beall(2000)首次使用表面传感器研究人造板的力学性能和内部缺陷,收到良好效果.日本学者小玉泰羲(2000)用连续小波变换分析处理利用敲击木材所获得的声信号,对气干木材的节子缺陷进行检测,研究发现,当木材含有节子缺陷时,声信号的共振频率降低、声信号共振的持续时间缩短.  相似文献   

7.
榫卯松动是木结构产生损伤的主要原因,为准确评估木构件的榫卯松动程度,采用压电陶瓷和数字图像相关法对木构件拔榫距离进行监测试验。首先基于压电陶瓷主动感应法和小波包能量法得到时域信号图,分析拔榫距离与电压值的关系。然后利用数字图像法实时检测拔榫过程中榫卯表面应变场信息变化。结果表明:拔榫距离越大,传感器接收能量和电压值越少;数字图像法可实时监测构件表面应变场信息变化,并根据应变演化规律检测木构件拔榫距离。试验结果可为木构件榫卯松动程度的检测研究与实际工况测量提供依据。  相似文献   

8.
小波分析在刀具磨损状态检测中的应用   总被引:1,自引:1,他引:0  
提出一种基于小波分析的声发射刀具磨损状态的判别方法,将占有声发射信号主能量的频段区间提取出来进行信号重构作为识别刀具磨损状态的特征向量。实验证明,该方法具有较高的准确性。  相似文献   

9.
木材损伤断裂过程的声发射特性与Felicity效应   总被引:5,自引:0,他引:5  
以2种针叶材、2种阔叶材为试材,研究无缺陷试件和含横纹裂纹试件在弯曲破坏过程中材料内部微结构演化的声发射特性,并利用声发射特征参数对几种损伤类型进行辨识.结果表明:1)无缺陷试件在加载初期声发射事件发展较为缓慢,且出现的主要是一些低振幅的AE信号,而大量高振幅AE信号出现在峰值载荷附近及韧性断裂阶段;2)利用声发射监测含裂纹试件在三点弯曲载荷下的损伤并断裂全过程,可以明显地识别裂纹尖端启裂和扩展的不同阶段;3)声发射信号的特征与损伤模式有关,胞壁断裂对应的AE特征为高幅值、高能量及长持续时间,而胞壁界面损伤与层裂损伤和细胞屈服与压溃损伤对应的AE特征为低幅度、低能量及短持续时间;4)木试件在低载荷水平下呈现Kaiser效应,而在高载荷水平下呈现Felicity效应,应用Felicity比能够较好地反映木结构的损伤程度.  相似文献   

10.
基于MEA-BP神经网络对木材内部缺陷诊断的研究   总被引:1,自引:0,他引:1  
为了提高木材内部缺陷的自动识别率,采用电阻层析成像(ERT)的方法获取电导率波动信号,通过小波包变换对采集的数据进行3层小波包分析,对八维特征向量进行提取,利用思维进化算法(MEA)优化权值和阈值,孔洞、节子、腐朽试样各45组数据,进行BP神经网络训练,每种缺陷20组作为测试集,识别木材内部缺陷。结果表明:MEA-BP神经网络对木材孔洞、节子和腐朽的识别率分别为96.92%、95.38%和92.31%,该模型解决了复杂组合的优化问题,提高了搜索效率,并且达到最佳的预测效果。  相似文献   

11.
基于ANSYS的圆锯片模态分析和振动分析   总被引:1,自引:0,他引:1  
利用ANSYS有限元软件,通过对圆锯片的模态分析,研究其低频范围的横向振动问题.通过软件分析得出圆锯片相应模态下的固有频率和振型.并利用动态信号测试分析系统对圆锯片相应模态的固有频率进行检测,得出圆锯片的固有频率与节圆数和节径数的变化规律,圆锯片边缘的位移矢量和与节圆数、节径数的关系,并指出在不了解圆锯片弹性模量、泊松比和密度等参数的条件下利用动态信号测试分析系统对圆锯片的模态进行检测具有一定的可行性.  相似文献   

12.
Acoustic emission (AE) monitoring was used to trace directly the fracture intensity in cylinders of lime wood subjected to variations in temperature and relative humidity (RH) in their environment. High-frequency components produced by mechanical fracturing were extracted from the raw AE signals using the wavelet transforms. The accumulated energy of these components depended on the magnitude and rate of the RH variations. The AE activity correlated well with predictions of the numerical modeling carried out as the first part of the present investigations. In particular, the AE activity became negligible below the allowable magnitude for the rapid RH variation predicted by the simulation, or when the time interval allowed for the RH variation was long enough. Furthermore, AE proved capable of tracing the progressive evolution of damage at the microlevel, which preceded failure of wood discernible from the macroscopic perspective.  相似文献   

13.
The orientation of wood fibers and the resulting surface roughness of radiata pine are studied during a peripheral milling with acoustic emission control. Studies were undertaken in machining process for radiata pine, where a surface quality problem appears when cutting is not well controlled. It is known that acoustic emission signals are a good monitoring element of the surface roughness with different cutting conditions. Previous studies have shown that changing cutting conditions affects acoustic emission signals; these variations are related to surface roughness and chip formation. The purpose of this study is to evaluate and quantify the surface quality and the consumption of cutting energy in radiata pine with differing grain angles, and to determine the relationships of grain angle with acoustic emission signals in order to better optimize cutting conditions.  相似文献   

14.
The primary bottleneck to extracting wood defects during ultrasonic testing is the accuracy of identifying the wood defects. The wavelet energy moment was used to extract defect features of artificial wood holes drilled into 120 elm samples that differed in the number of holes to verify the validity of the method. Wavelet energy moment can reflect the distribution of energy along the time axis and the amount of energy in each frequency band, which can effectively extract the energy distribution characteristics of signals in each frequency band; therefore, wavelet energy moment can replace the wavelet frequency band energy and constitute wood defect feature vectors. A principal component analysis was used to normalize and reduce the dimension of the feature vectors. A total of 16 principal component features were then obtained, which can effectively extract the defect features of the different number of holes in the elm samples.  相似文献   

15.
Browsing by large herbivores on planted and naturally regenerated conifer seedlings (Picea abies and Pinus sylvestris) was recorded in 104 clear‐cuts in east‐central Sweden during 1990 and in 47 clear‐cuts in 1991. The number of seedlings browsed and the browsing patterns were analysed in relation to seedling type. Browsing frequencies were also compared between forest stands with different site productivities and subjected to different management practices. The variation in the number of seedlings browsed in 1990 was explained mainly by seedling category. Among both planted and naturally regenerated seedlings, pine was browsed more than spruce. Two‐year‐old containerized seedlings of spruce was browsed more than 4‐yr‐old bare‐rooted spruce. In 1991, browsing was more equal among species and seedling types. Number of seedlings with their leader browsed and the amount of biomass left after browsing differed significantly between seedling types. Seedlings that had been browsed in 1990 experienced significantly higher browsing frequencies in 1991 when compared with unbrowsed seedlings. The effects of stand characteristics were not found to be significant.  相似文献   

16.
The genetic structure in Danish indigenous and introduced provenances of beech (Fagus sylvatica L.) is described. Ten stands are analysed based on variation revealed by seven enzyme systems. The present study reveals significant allele frequency differences between populations with a small absolute amount of genetic differentiation, with no pattern related to geographic distance. The genetic differentiation between indigenous populations measured by Wright's F ST‐statistic does not deviate significantly from zero, and only 2.9% of the total genetic variation in the foreign populations was due to interpopulation differences, The allele frequencies in relatively small and isolated populations do not differ from the frequencies found in larger areas with beech forest. The low genetic differentiation of isozyme level between indigenous and foreign populations was opposed by significant differences in foliation performance measured on three pairs of neighbouring stands. Foliation performance in adjacent stands of indigenous and introduced beech indicates that quantitative characteristics must be considered separately from neutral variation at single loci.  相似文献   

17.
料慈竹种群结构光能利用率的初析   总被引:2,自引:0,他引:2  
按照生态学原理,分析料慈竹(Bambusa distegia)种群结构中的不同年龄叶的生长、不同年龄和不同胸径叶面积变化规律、叶面积指数与产量的关系、料慈竹种群结构的产量与光能利用率。  相似文献   

18.
Forest trees, populations, races, species, and taxonomic groups above the species level display rich variation in biochemical markers. The variation stems from inherited modifications that trace back in time, through converging ancestries, towards common progenitors. Past movements of continents, mountain building events, and climate changes isolated forest populations one from another and provided critical challenges to the lineages that survived to the present day. A wealth of molecular variants in forest trees characterize these widely-distributed, large, long-lived, outbreeding, organisms. Forest trees have an abundance of rare variants and over one third of all the alleles (different forms of one gene) occur only rarely ( < 2% frequency) in a few trees of a species sample. Those rare alleles may either represent new variation or persistent forms of genes that have low adaptive value under present conditions. From another perspective, however, the largest share of genetic variation in forest trees is due to the presence of multiple alleles found at intermediate frequencies for only a small percentage of all the genes, and those alleles are commonly widespread throughout species areas. These common alleles may mark genes that track historical events in lineages or mark genes with adaptive significance in present populations. Evidence from enzyme studies supports the conclusion that highly comparable functional genes are common to different forest taxa. Future research will be toward understanding the phenotypic expression of particular genes and revealing the relative importance of genetic variants to adaptation and growth.  相似文献   

19.
Experimental determination of the disposition of Rhagoletis cerasi L. (Dipt., Trypetidae) to fly An apparatus was described which serves for registering the flight frequencies of insects. It consists of a cage with insects hanging under an electronic balance, which registers the flight as reduction of weight. These electric signals of the balance are transformed by an electronic connexion into countable pulse, which are printed.  相似文献   

设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号