基于可见/近红外光谱技术的板栗产地识别

采用可见/近红外光谱分析技术对河北和安徽两个产地的板栗进行检测分级,获得板栗样品在600~1 100 nm波长区间的可见/近红外光谱,采用偏最小二乘判别分析（PLSDA）进行建模,比较不同预处理方法和波长范围对PLSDA模型精度的影响。结果显示,不同预处理方法对模型性能影响较大,一阶导数预处理在全波长范围所建PLSDA模型性能最优,校正集和验证集的决定系数分别为0.884和0.863,均方根误差分别为0.170和0.191。不同波长范围也会影响模型的预测和识别性能,在波长区间为750~1 000 nm的光谱所建立的PLSDA模型性能要高于全波长光谱所建立的模型性能,其中经过Savitzky Golay平滑预处理后,模型性能的提高最为明显,且其与原始光谱在校正集和验证集的敏感性和特异性均达到最优,即识别率均可达到100%。因此,可见/近红外光谱分析技术能够对板栗产地进行鉴别。     

鲜枣品种和可溶性固形物含量近红外光谱检

采用近红外光谱分析技术无损鉴别鲜枣品种和测定其可溶性固形物含量.对3个不同品种的鲜枣进行光谱分析,各获取30个样本数据.采用平滑法和多元散射校正方法对样本数据进行预处理,再用主成分分析法对光谱数据进行聚类分析并获得各主成分数据.将样本随机分成75个建模样本和15个预测样本,将建模样本的主成分数据作为BP神经网络的输入变量,鲜枣品种和可溶性固形物含量作为输出变量,建立3层人工神经网络鉴别模型,并用该模型对15个预测样本进行预测.结果表明,在阈值设定为±0.17的情况下该模型对预测集样本品种鉴别准确率达到100%,可溶性固形物含量预测值与实测值相对偏差小于10%.     

碧螺春茶叶的真伪鉴别技术——基于漫反射式高光谱成像技术

提出了基于漫反射式高光谱成像技术结合模式识别的碧螺春茶真伪鉴别方法。试验以漫反射式高光谱成像系统采集碧螺春茶样图像数据;通过主成分分析优选特征图像,并利用灰度共生矩阵从中各提取4个纹理特征参量;最后,利用支持向量机的模式识别方法构建茶叶真伪鉴别模型。训练和预测集的判别率分别达到100%和96.25%。研究表明,利用漫反射式高光谱成像技术结合支持向量机模式识别方法鉴别碧螺春茶品质真伪是可行的。     

洋梨硬度的便携式可见/近红外漫透射检测技术

采用便携式可见/近红外检测仪快速检测阿巴特、康佛伦斯和五九香梨硬度.采集洋梨漫透射光谱(500～1 010 nm),经二阶导数和卷积平滑处理后,分别建立偏最小二乘法和多元线性回归模型.相关系数法和遗传算法用于选择偏最小二乘法建模变量,预测均方差分别为7.780 N和8.080 N,相对预测误差分别为26.24%和29.71%.多元线性回归模型使用7个变量,预测均方差和相对预测误差分别为7.740 N和26.10%.结果表明:建立多品种洋梨硬度全局模型是可行的,便携式仪器可用于洋梨硬度现场检测.     

洋梨硬度的便携式可见/近红外漫透射检测技术

采用便携式可见/近红外检测仪快速检测阿巴特、康佛伦斯和五九香梨硬度。采集洋梨漫透射光谱(500～1010nm)，经二阶导数和卷积平滑处理后，分别建立偏最小二乘法和多元线性回归模型。相关系数法和遗传算法用于选择偏最小二乘法建模变量，预测均方差分别为7.780N和8.080N，相对预测误差分别为26.24%和29.71%。多元线性回归模型使用7个变量，预测均方差和相对预测误差分别为7.740N和26.10%。结果表明：建立多品种洋梨硬度全局模型是可行的，便携式仪器可用于洋梨硬度现场检测。     

基于可见/近红外光谱与SIMCA和PLS-DA的脐橙品种识别

脐橙的优良品种选育是通过芽变选种获得，为了选择一些具有特殊性状的脐橙芽变进行培植，需要对脐橙的品种进行鉴别。该研究应用可见/近红外光谱分析方法结合软独立模式分类（SIMCA）和偏最小二乘判别分析（PLS-DA）模式识别方法对赣南脐橙的品种进行识别。研究结果表明，采用原始近红外光谱结合SIMCA方法，实现了纽贺尔、奈弗宁娜、华脐以及朋娜4种脐橙的100%的识别；应用近红外光谱结合PLS-DA方法对校正样本建立判别模型，其校正及验证结果与实际分类变量的相关系数均大于0.970，交叉验证均方根误差（RMSECV）和预测均方根误差（RMSEP）都小于0.100，利用模型对验证集中纽贺尔、奈弗宁娜、华脐以及朋娜4种脐橙的识别率均为100%。为脐橙优良品种的选育提供快速鉴别分析方法。     

基于太赫兹衰减全反射技术的花生品种快速鉴别

利用太赫兹衰减全反射技术,结合距离匹配算法,对不同品种花生进行快速分类鉴别。实验随机采集花育36号、鲁花1号和鲁花9号共3个花生品种、总计60个花生样本在0.3~3.6 THz的太赫兹衰减全反射光谱,随后对其吸收系数进行一阶导数及归一化预处理,并建立花生品种的距离匹配快速鉴别模型。距离匹配模型结果显示:3个类别的总体识别准确率可达93.3%,仅有1个样本预测错误。研究结果表明,利用太赫兹衰减全反射技术实现花生品种快速鉴别具有一定的可行性,具有操作简单且快速高效等特点。     

基于低秩自动编码器及高光谱图像的茶叶品种鉴别

提出一种基于低秩自动编码器及高光谱图像技术的茶叶品种鉴别方法。应用高光谱成像系统采集5个品种的茶叶样本高光谱图像数据,利用ENVI软件确定高光谱图像的感兴趣区域(ROI),并提取茶叶样本在ROI的平均光谱作为该样本的原始光谱数据。由于高光谱信息量大、冗余性强且存在噪声,运用自动编码器和低秩矩阵恢复结合的低秩自动编码器(LR-SAE)对原始光谱数据进行降维,在自动编码器降维基础上加入去噪处理,提取鲁棒判别特征。在此基础上应用支持向量机(SVM)和Softmax分类算法对降维后的茶叶样本高光谱数据分类。通过5折交叉试验验证,LR-SAE-SVM模型的预测集准确率达到99.37%,SAE-SVM模型的预测集准确率为98.82%;LR-SAE-Softmax模型的预测集准确率达99.04%,SAE-Softmax模型的预测集准确率为97.99%。研究结果表明,相较于未进行去噪处理的传统自动编码器,LR-SAE降维之后的分类建模效果有所提升,将其应用于茶叶品种鉴别是可行、高效的。     

基于可见/近红外高光谱的八角茴香与莽草无损鉴别

基于高光谱成像技术提出了一种八角茴香与其伪品莽草的快速鉴别方法。实验采集400~1000nm范围的高光谱数据,依据样本和背景像素点的光谱特征差异,选择850nm和450nm下的图像并进行差运算,结合阈值法去除背景信息,利用线性拉伸去除样本高度引入的阴影噪声像素点,再结合二值图像区域标记法从样本高光谱数据中自动提取其平均光谱数据;利用平均光谱数据,采用连续投影算法（Successive projections algorithm, SPA）选取了4个最优波长：533、617、665、807nm;基于最优波长下的光谱数据,建立了偏最小二乘判别（Partial least square discrimination analysis,PLSDA）模型,模型对鉴别八角和莽草的总体准确率为98.4%;利用所建多光谱模型对外部验证集数据进行预测,总体分类准确率为97.9%。利用常规图像处理技术同时对外部验证集数据进行处理,并对两种技术方法进行了比较,结果表明,依托高光谱成像技术建立的八角和莽草辨识的多光谱分析方法简单、高效,易于实现动态在线便携式检测。     

近红外玉米品种鉴别系统预处理和波长选择方

以7个品种玉米籽粒的鉴别系统为研究对象,对比研究了6种预处理方法和波长选择对模型鉴别能力的影响.结果表明,在被比较的6种预处理方法中,一阶导数方法能够使模型有更好的鉴别性能.使用一阶导数预处理和全光谱区的模型平均正确识别率和正确拒识率最高,分别为98.6%和98%,有5个品种的模型的正确识别率和正确拒识率都达到了100%.波长选择对一阶导数模型没有明显作用,但能使标准正态变量变换和矢量归一化模型鉴别准确度得到较大提高.     

Phosphorus losses from an artificially drained rural lowland catchment in North-Eastern Germany

In a small, extensively artificially drained lowland catchment (15.5 km²) in Mecklenburg-Vorpommern (North-Eastern Germany), the dynamics and the extent of total phosphorus (TP) and total reactive phosphorus (TRP) losses as well as the discharge were monitored at different scales for three winter seasons of 6 months each. Ranging from 0.036 to 0.044 mg TP l⁻¹ and from 0.030 to 0.037 mg TRP l⁻¹, average phosphorus concentrations in the discharge of a collector drain, a ditch draining arable land and a small brook were low. Elevated concentrations occurred during intensive snowmelt events. Probably due to the re-mobilisation of phosphorus under anaerobic conditions, concentrations (0.137 mg TP l⁻¹ and 0.076 mg TRP l⁻¹) in a ditch draining grassland on degraded peat were significantly higher than at the other sites characterised by mineral soils. Generally, phosphorus concentrations increased with discharge at all sites except for the grassland, although not during each single discharge event. Surprisingly, a dependency on the fertilisation practices could not be found. The phosphorus losses per winter season were low, with a maximum of 270 g TP ha⁻¹ and 211 g TRP ha⁻¹. Using a two-component mixing model based on baseflow separation and parameter optimisation, it was estimated that around 53, 60 and 56% of the TP losses from the collector drain, from the ditch and from the brook as well as 53, 68 and 45% of the TRP loads were exported via a fast flow component. This component accounted for 18-23% of the total discharge. At all measurement stations, there were large differences between the partitioning patterns of the single discharge events. Our study has not only shown the event-based behaviour of the P losses and the possible occurrence of high P concentrations due to preferential flow, but also that the highest potential of eutrophication in this lowland landscape originates from drained, degraded, and intensively used peatlands.     

Agricultural land management strategies to reduce phosphorus loads in the Gippsland Lakes, Australia

A target to reduce phosphorus flows into the Gippsland Lakes in south-eastern Australia by 40% in order to improve water quality has previously been established by stakeholders. This target, like many others worldwide, has been set mostly on the basis of environmental concerns, with limited consideration of issues such as technical feasibility, socio-economic constraints, political factors and associated costs and benefits. An integrated analysis at the catchment scale is undertaken to assess the agricultural land management changes required to achieve this target, and to evaluate the cost-effectiveness of these changes. Based on assumptions used, it appears technically feasible to achieve a 40% reduction in P load entering the Lakes. However, the least-costly way of doing so would require around A$1 billion over 25 years, a dramatic increase in the current levels of funding provided for management. Results of a sensitivity analysis indicate that there is little or no chance of investment in a 40% reduction being cost-effective. On the other hand, a 20% P reduction could be achieved at much lower cost: around $80 million over 25 years and requiring more modest land-management changes. Reliance on voluntary adoption of ‘Current Recommended Practices’ is unlikely to deliver changes in management practices at the scale required to have sufficient environmental impacts. Enforcement of existing regulations for the dairy industry would be amongst the most cost-effective management strategies. The major implications of this work for agriculturally induced diffuse-source pollution include the need for feedback between goal setting and program costs, and consideration of factors such as the levels of landholder adoption of new practices that are required and the feasibility of achieving those adoption levels. Costs, landholder adoption of new practices and socio-political risks appear neglected in the formulation of many water quality programs. The study provides a demonstration of an approach to integrated multidisciplinary research addressing complex environmental problems with multiple decision makers, multiple stakeholders, and high uncertainty. On the evidence of experience in this study, the approach deserves consideration in other contexts.     

Assessment of irrigation and environmental quality at the hydrological basin level: II. Salt and nitrate loads in irrigation return flows

Irrigation return flows may induce salt and nitrate pollution of receiving water bodies. The objectives of this study were to perform a salt and nitrogen mass balance at the hydrological basin level and to quantify the salt and nitrate loads exported in the drainage waters of three basins located in a 15,500 ha irrigation district of the Ebro River Basin (Spain). The main salt and nitrogen inputs and outputs were measured or estimated in these basins along the 2001 hydrological year. Groundwater inflows in the three basins and groundwater outflow in one basin were significant components of the measured mass balances. Thus, the off-site impact ascribed solely to irrigation in these basins was estimated in the soil drainage water. Salt concentrations in soil drainage were low (TDS of around 400–700 mg/l, depending on basins) due to the low TDS of irrigation water and the low presence of salts in the geologic materials, and were inversely related to the drainage fractions (DF = 37–57%). However, due to these high DF, salt loads in soil drainage were relatively high (between 3.4 and 4.7 Mg/ha), although moderate compared to other areas with more saline geological materials. Nitrate concentrations and nitrogen loads in soil drainage were highest (77 mg NO₃⁻/l and 195 kg N/ha) in basin III, heavily fertilized (357 kg N/ha), with the highest percentage of corn and with shallow, low water retention flood-irrigated soils. In contrast, the lowest nitrate concentrations and nitrogen loads (21 mg NO₃⁻/l and 23 kg N/ha) were found in basin II, fertilized with 203 kg N/ha and preponderant in deep, alluvial valley soils, crops with low N requirements (alfalfa and pasture), the highest non-cropped area (26% of total) and with fertigation practices in the sprinkler-irrigated fields (36% of the irrigated area). Thus, 56% of the N applied by fertilization was lost in soil drainage in basin III, as compared to only 16% in basin II. In summary, a low irrigation efficiency coupled to an inadequate management of nitrogen fertilization are responsible for the low-salt, high-nitrate concentrations in soil and surface drainage outflows from the studied basins. In consequence, higher irrigation efficiencies, optimized nitrogen fertilization and the reuse for irrigation of the low-salt, high-nitrate drainage waters are key management strategies for a better control of the off-site pollution from the studied irrigation district.     

基于近红外光谱技术的叶面药液浓度检测

提出了一种应用近红外光谱技术快速检测叶面药液浓度的方法。采用漫反射测量方式获取了叶面药液的近红外光谱。选用标准偏差归一化、三点滑动平均滤波和一阶导数为最优组合预处理。通过7种波段方案的对比,得出最优波段为350～1 900 nm。采用偏最小二乘法建立了叶面药液质量浓度与光谱反射率的定量分析模型。其预测集相关系数为0.994,预测均方根误差为0.039。结果表明,利用近红外光谱技术检测叶面药液浓度具有实际指导意义。     

农村剩余劳动力就业问题的探讨

论述了我国农村剩余劳动力就业的意义，并针对我国农村剩余劳动力的现状，提出多渠道促进农村剩余劳动力就业的途径和办法。     

车用发动机润滑油失效规律与更换周期研究

研究在正常使用条件下１０Ｗ／３０－ＱＣ级润滑油用于某型车用发动机时所表现的失效规律,确认期服从两参数Ｗｅｉｂｕｌｌ分布,并给出分布的尺度参数η和形状参数ｍ之最佳线性无偏估计（ＢＬＵＥ）同时,还为定量分析,计算车用发动机润滑油的使用可靠性和合理确定其更换周期提供了一套可供借鉴的方法,并给出了具体的计算实例。     

草莓的采收与贮藏

草莓果味酸甜爽口，营养价值高，为人们所喜爱的应市鲜果之一。由于草莓易受损伤和微生物侵染，因此其采收、贮藏过程极为重要。介绍草莓采收过程中的注意事项以及11种贮藏保鲜技术的实施步骤和操作方法，推介9种加工方式，为草莓产业化提供技术支持。     

乡（镇）农业技术推广体系存在的问题及对策

乡（镇）农技推广体系是农业发展的关键。介绍朝阳市农业技术推广体系的现状,针对目前体制及运行机制、服务手段等方面存在的问题,提出适合朝阳市乡（镇）农技推广体系改革与发展的对策,以期对全省农技推广体系改革和发展起到一些借鉴作用。     

水稻“3414”肥料试验研究

以海城地区水稻主栽品种辽星一号为研究对象,开展水稻＂3414＂氮磷钾肥料试验,研究水稻在不同氮磷钾水平配比时的产量效应、最佳施肥量及最佳氮磷钾配比。结果表明,本地区最佳经济产量为725.5kg/667m2,氮磷钾最佳施肥量分别为N16.1kg/667m2、P2O58.5kg/667m2、K2O7.3kg/667m2,氮磷钾的最佳施肥配比为1：0.53：0.45。     

Quantifying the sustainability of agriculture

The rural sustainability index is a scientifically based tool to quantify the performance of agriculture. The sustainability of crop production is quantified from three perspectives; people, planet and profit. Within each perspective, one condition was selected that must be met to warrant agriculture. These are: No hazardous work should be used within the crop production chain; agricultural crops should not be grown on land allocated to nature by national law or regulations and, when a GM-crop is present or is introduced in a region, it should not harm development opportunities of other farmers. If these excluding conditions are met, the sustainability of agriculture is assessed through five performance indicators on school attendance, water use and consumption, fertilizer use, pesticide use, and farm income. For each of the five indicators, critical values and target values have been given that limit the transition range between non-sustainable and sustainable production. The five indicators are combined into a sustainability index. The index aims at improving the socio-economic position of farmers while protecting the environment.