Wood defect classification based on image analysis and support vector machines

This paper addresses the issue of automatic wood defect classification. A tree-structure support vector machine (SVM) is proposed to classify four types of wood knots by using images captured from lumber boards. Simple and effective features are proposed and extracted by partitioning the knot images into three distinct areas, followed by utilizing a novel order statistic filter to yield an average pseudo color feature in each area. Excellent results have been obtained for the proposed SVM classifier that is trained by 800 wood knot images. Performance evaluation has shown that the proposed SVM classifier resulted in an average classification rate of 96.5% and false alarm rate of 2.25% over 400 test knot images. Future work will include more extensive tests on large data set and the extension of knot types.     

Nondestructive evaluation of veneer quality using acoustic wave measurements

Two nondestructive evaluation methods, impact-induced stress wave techniques and ultrasonics, were investigated to detect lathe checks and knots in veneer, which were identified as key veneer quality properties for some engineered applications. Measurements included wave velocity and attenuation in the directions parallel and perpendicular to the grain. The results showed that both techniques were sensitive to lathe checks when using wave propagation perpendicular to grain. For wave transmission parallel to grain, signals showed some sensitivity to knots. There was no significant difference in wave velocity measurements between stress wave and ultrasonic techniques. Regression models based on stress wave velocities in these two orthogonal directions were developed to estimate the veneer quality index giving a coefficient of determination ranging between 0.39 and 0.50. Received: 20 December 1998     

Predicting field performance using seedling quality assessment

Seedling quality is defined as fitness for purpose. Attributes for assessing seedling quality are grouped in material attributes that can be rapidly assessed by any number of direct or indirect methods, and performance attributes that are assessed by subjecting whole seedlings to certain environmental regimes and evaluating their growth response. Material attributes comprising morphology, bud dormancy, water relations and nutrition and performance attributes comprising frost hardiness, vigour and root growth potential are discussed with regard to relevance for predicting field performance. Measurement techniques and relevance in assessing seedling quality are presented for different physiological attributes: electrolyte leakage, enzymatic activity, water potential, water conductance, mineral nutrition, food reserves, mitotic index, gas exchange, chlorophyll content, plant growth regulators, stress-induced volatile emission, and nuclear magnetic resonance. Methods with future potential in improving the prediction of field performance are identified and discussed. The methods presented in this respect are chlorophyll fluorescence, infrared thermography, machine vision systems, integrated approaches and ecophysiological evaluations.     

我国南方道路除冰机振动滚筒的研究

2008年春节期间,我国南方许多省份发生了百年一遇的冰雪灾害,许多高速公路和普通公路上冻结了一层几厘米至十多厘米厚的由冻雨形成的密实冰层,致使数万辆客车和货运汽车滞留在公路上长达一个星期,给国家和人民生活带来了巨大的损失,迫切需要研制一种适应南方的这种特殊冰层的道路除冰机。目前,已提出在我国目前大量应用的L50装载机上安装一个带除冰振动滚筒的设计构想,但还没有具体的结构设计。针对这种振动除冰滚筒、除冰羊角、振动液压系统以及振动轴和偏心块式振动机构,进行了比较详细的理论分析设计计算。     

基于BP神经网络的泉州湾红树株高预测模型

在分析测量数据的基础上,提取红树的平均基径、基径数、平均胸径、胸径数等特征参数,建立了预测红树株高的人工神经网络模型。采用Levenberg-Marquardt优化算法改进了BP神经网络算法;采用训练好的BP神经网络模型对距堤坝25,50,75 m 3个采样点的株高进行预测,预测值和实测值的均方根误差分别为0.000 6,0.002 2,0.004 1,相关系数分别为0.99,0.95,0.94。结果表明利用BP神经网络对红树株高进行预测是可行的。     

基于灰色理论的西丰县森林资源预测研究

根据西丰县2005年森林资源二类调查和(2006-2010年)森林资源年度变更统计数据,采用灰色GM(1.1)预测方法,对有林地面积、活立木蓄积和森林覆盖率建立预测模型。经精度分析,该模型拟合度较高。通过该模型预测了2020年和2025年有林地面积、活立木蓄积和森林覆盖率,为未来林业政策的制定提供数据支持。     

Development of LVL frame structures using glued metal plate joints I: bond quality and joint performance of LVL-metal joints using epoxy resins

Bond quality and joint performance between laminated veneer lumber (LVL) and metal plates were investigated. Commercially fabricated LVL made of Douglas fir veneer and bonded with phenol-formaldehyde resin as well as three types of epoxy adhesives were used. Various surface preparations and treatments were applied to ordinary steel, stainless steel, and aluminum plates to remove the weak boundary layer that is incompatible with the resin and form a stable adherend layer that is chemically and mechanically compatible with the resin. Small specimens were tested in shear to search the most suitable metal surface for bonding with LVL. Generally, shear strength obtained for the specimens bonded with aluminum plates was lower than those bonded with ordinary steel plates. Among them chemically treated (ChT) and roughened (R) surfaces have demonstrated superior performance. To investigate strength performance and bond quality, LVL beams jointed with metal plates were tested while bending. The best results were obtained for specimens bonded with zinccoated metal plates, though good results were obtained also for ChT and R plates. However, the fracture proved to be fragile when no drift pins were used, even for high-performance surface treatments. The usage of drift pins was necessary to add toughness and avoid the brittle status of the fracture.     

Predicting field performance of five irrigated tree species using seedling quality assessment in Burkina Faso,West Africa

Five exotic tree species (Acacia angustissima (Mil.) Kuntze, A. mangium Wild, Gliricidia sepium (Jacq.) Alp., Leucaena hybrid (L×L), and Leucaena leucocephala (Lam.) de Wit) were investigated to determine whether parameters of nursery seedling stock quality could be used to predict their field performance in a plantation irrigated with treated waste-water to produce fodder and wood. Plants were grown in the nursery in two contrasting rooting substrates (ordinary nursery soil and sand), predicted to have different effects on resource allocation. Three categories of morphological indicators were measured, i.e., plant dimensions (height, diameter, root length), plant weights (shoot, root and whole plant weights) and indices (sturdiness quotient ‘SQ’, shoot:root dry weight ratio ‘SRR’ and Dickson’s quality index ‘DQI’). In the nursery, all species performed better in the ordinary nursery soil for all growth parameters except root length. Thus ordinary nursery substrate appeared superior to sand in terms of plant quality. However, a follow up at plantation phase revealed that only some morphological attributes or ratios were suitable to predict field performance for the five tested species in irrigated plantation. In addition, the effect of the substrate observed at the nursery stage had disappeared 12 months after out planting due to the availability of water and nutrients provided by the treated waste water used for the irrigation. The results showed that root collar diameter and DQI appeared to be the most appropriate indicators to predict the outplanting performance of the five tested species in a short-rotation irrigated plantation in semi-arid Burkina Faso. The former measure is simpler and non-destructive.     

光皮树果实专用型螺旋冷榨机的压榨性能研究

开展了能源植物光皮树果实压榨制油工艺研究,对比了自制硬果核油料专用小型螺旋冷榨机（XLK-2010型）与德国进口的螺旋冷榨机（KOMETCA-59-G型）的压榨效果。重点考察了在出饼孔分别为5mm和6mm的条件下,不同榨机榨轴转速对光皮树果实的生产效率、饼粕残油率和出饼温度的影响。结果表明：在相同条件下,自制的小型螺旋冷榨机与进口设备相比,压榨效率相当,但饼粕残油率和出饼温度略低,设备制造成本大大降低。该研究初步建立了光皮树果实压榨制油工艺参数,为大型专用型冷榨机的研制提供了理论支撑。     

基于声发射技术模拟蛀木害虫羽化孔洞缺陷检测

[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A...     

基于神经网络的沥青路面基层裂缝应力强度因子预测模型

沥青路面裂缝尖端应力强度因子是判断裂缝扩展的重要指标,然而应力强度因子是路面材料及结构参数的复杂函数,针对其计算的复杂性,本文根据神经网络理论,考虑了荷载、裂缝长度、基层模量及厚度、面层模量及厚度和底基层模量等主要因素,采用三层BP网络近似计算沥青路面基层裂缝的应力强度因子,建立了沥青路面基层裂缝应力强度因子的预测模型。通过大量的计算与验证,训练好的BP网络精度高、速度快、泛化能力强,且易于实现,可以应用于工程设计与验算。     

基于模糊综合评价法的东江湖水质评价

以湖南省郴州市的东江湖为研究对象,运用模糊综合评判模型对3个监测断面的24个监测因子进行综合评判。结果表明:东江湖水质总体状况较好,符合功能区水环境质量标准,其中1月、4月、7月东江湖水质对I类水的隶属度最高,10月对Ⅱ类水的隶属度最高。水质污染因子中最重要的污染因子为铅,与东江湖流域内存在金属尾矿区有关。评价结果为污染控制提供了科学依据。     

Estimating the moisture excluding effectiveness of surface coatings on Southern Yellow pine using acoustic emission technology

In this research, acoustic emission technology was used to estimate the moisture excluding effectiveness of coatings on wood. Moisture interaction with wood causes damage, and coatings such as paints and varnishes can effectively limit this interaction. Over time, the moisture excluding effectiveness of wood is reduced by the actions of weathering and the resulting wood-water interaction leads to problems due to the dimensional instability of the wood. Five coating types, at two spreading levels, were applied to samples of southern yellow pine. The samples were exposed to three weathering conditions to simulate the effects of exposure then immersed in water while the moisture interaction was monitored using acoustic emission monitoring equipment. Cumulative counts as well as changes in size and moisture content were measured or calculated. The results show that acoustic emission technology is an effective and sensitive estimator of the moisture excluding effectiveness of coatings on wood. The acoustic emission method is sensitive to local changes and coating fractures whereas the more traditional methods are sensitive to gross changes in sample characteristics. Of five coatings tested, the most effective one was found to be an oil-based paint, and the worst performance was by a water-based, clear poly-urethane. Received: 10 September 1998     

地质雷达在依勃公路路面质量检测中的研究与应用

本文介绍了地质雷达在公路路面强度和厚度检测的研究。     

运用烟雾机防治落叶松球蚜效果分析

湖北省建始县近年落叶松球蚜危害日趋严重，且有扩大趋势。利用6HYB-25A型烟雾机防治落叶松球蚜，进行机动喷烟，经过防治效果检查，防治56h后的校正死亡率达89.3%；防治成本平均为29.55元/hm^2。     

基于主成分分析的茶油品质综合评价研究

为了建立一套适合茶油品质评价的方法,并为茶油应用范围的拓展、茶油品牌的建设及油茶栽培种植的立地选择提供参考依据,对9个立地的茶油的14个品质指标进行了测定,采用主成分分析法建立了茶油综合品质评价模型,并基于主成分得分及综合品质得分进行了优良度排序。结果表明:1)主成分分析结果显示,前4个主成分的累积贡献率为85.065%,决定第1主成分的主要是茶油的感官品质、氧化稳定性及脂肪酸不饱和度;决定第2主成分的主要是出油率及茶油的功能性成分含量;决定第3主成分的主要是茶油的理化品质;决定第4主成分的主要是茶油色泽;2)9个立地茶油的综合品质得分排序为:西坡中坡>南坡下坡>西坡下坡>西坡上坡>南坡中坡>东坡下坡>南坡上坡>东坡中坡>东坡上坡,各主成分得分排序与综合品质得分排序间存在差异;3)茶油品质整体表现为西坡优于南坡优于东坡,下坡优于中坡优于上坡。基于上述结果说明立地是影响茶油品质的重要因素,坡向及坡位因子对茶油的综合品质均有一定影响。     

基于近红外光谱及 BP 神经网络分析法磁预测森林土壤有机碳含量

为快速测定森林土壤的有机碳含量,从取自小兴安岭带岭林业局东方红林场的120个土壤样品中采集350～2500 nm的土壤近红外光谱数据,对光谱做一定的预处理后,运用主成分分析法压缩提取前8个主成分,结合BP神经网络非线性方法建立土壤有机碳含量的预测模型并进行验证。结果表明,验证集的相关系数为0.78002,均方根误差为0.5002,预测集的相关系数为0.84941,均方根误差为0.4538。应用近红外光谱技术及BP神经网络非线性方法建模可以有效地预测土壤的有机碳含量,为野外大面积快速测定森林土壤碳含量提供了技术依据。     

基于扩散指数模型的四川生态旅游业发展景气预测

运用等权一致扩散指数模型,分析了四川生态旅游发展的市场前景和景气循环周期。结果表明:(1)四川省生态旅游业虽然经历了2008年的短暂低谷期,但从总体上看处于持续增长的态势;(2)2007~2012年四川省生态旅游业经历了"景气-不景气-景气-不景气-景气"的循环过程。短期内,四川生态旅游业会持续当前的景气状态。     

基于GM(1,1)的天然次生林空间结构预测

【目的】系统地分析已有天然次生林林分空间结构数据,通过灰色模型预测天然次生林林分空间结构各指标未来的发展趋势,对天然林经营具有十分重要的意义。【方法】以湖南大围山自然保护区典型次生林为研究对象,依据结构化森林经营理论,选取了混交度、竞争指数、角尺度、空间密度指数、开阔比数、大小比数、林分综合均质性指数作为天然次生林林分空间结构合理性评价与预测的量化指标,构建了基于GM(1,1)的天然次生林林分空间结构灰色预测模型。模型将2008年林分空间结构各指标的平均值作为初始值,并在研究区设置了面积为20 m×20 m的5个研究样地(M1,M2,M3,M4,M5),利用保护区2008-2018年典型样地林分空间信息,预测了研究区调研样地2019年-2021年林分空间结构各指标未来的变化趋势。【结果】利用精度检验机制对该模型的精确度进行了有效性检验,检验结果表明,所有指标预测合格概率P合=71.43%,良好的概率P良好=22.86%,优的概率P优=5.71%,表明该预测模型符合精度检验要求。【结论】样地未来3 a各指标整体变化尺度不大,林分空间结构基本稳定。从各指标在2008-2018年时空上的变化规律来看,各样地林分平均竞争指数、平均大小比数及平均空间密度指数是影响林分均质性指数的关键指标。