光纤近场测量中自动聚焦系统的研究

在用近场法测量光纤几何尺寸时,关键在于调节光学系统,获得优质的光纤近场图象.而要使这样的测量系统实现自动化,关键在于寻找图象是否清晰的适当判据,继而构成自动调节与控制的闭环反馈系统.本文介绍了利用信息熵判别函数,可以较好地判断图象质量,从而实现了自动调焦和自动测量.     

傅里叶变换红外光谱结合化学计量学鉴别预包装纯菠萝汁中外源糖

为了快速、准确鉴别预包装纯菠萝汁是否掺有外源性糖,采用傅里叶变换衰减全反射中红外光谱技术采集900～1 500 cm-1范围内不同批次的预包装纯菠萝汁样品和掺入甜菜糖浆、大米糖浆、木薯糖浆的菠萝汁掺假样品的中红外光谱共计234例,以线性判别分析和支持向量机分析为掺假鉴别模型的建模方法,比较了全波段谱图与通过主成分载荷系数分析选取特征波长图谱的两种掺假鉴别模型。研究表明,全波段图谱的线性判别分析和支持向量机分类模型对验证集的判断正确率均高于88%;选取8个特征波长之后,线性判别分析模型验证集判断正确率提高至96.15%,支持向量机模型验证集判断正确率提高至94.87%,且模型输入变量由312个减少到8个。利用傅里叶变换衰减全反射中红外光谱技术结合化学计量学方法选取特征波长后建立的模型可以较好的应用于预包装纯菠萝汁外源糖的鉴别。     

LIF和CNN的矿井突水水源类型判别

矿井进入深部开采过程中,突水威胁分别来自顶板老空水和底板高压岩溶水。煤矿突水水源类型的在线识别能够预警煤矿水害,是矿井水害防治关键环节,对煤矿安全生产具有积极意义。代表离子法作为传统的煤矿突水水源类型识别方法,需要深入现场采集水样,密封处理后在实验室检测水样中7种典型的无机离子浓度,计算得到突水评价因子。这种存在检测周期过长、样品易被污染以及预警响应滞后、无法在线判别等不利因素。针对代表离子法方法的不足,提出了一种基于激光诱导荧光（LIF）和卷积神经网络(CNN)的矿井突水水源判别模型。首先,针对淮南矿业集团新集二矿的4种水体,2016年6月-2017年6月期间分批次取得161组水源样本,其中采空区积水46条,砂岩水59条,太灰水42条和奥灰水14条。用LIFS-405激光诱导荧光系统发射的405 nm激光检测水样,水体受激后得到突水水样的荧光光谱。主成分分析得到前10个主成分累计贡献率不足85%,4种水样无法有效直接辨识。针对该问题和水样荧光光谱中的随机高频波动干扰,采用一阶滞后滤波方法抑制波动频率较高的周期性干扰;针对线判别分析对数据更新率的要求,采用递推平均方法;在此基础上,提出了一种改进的递推平均一阶滞后平滑滤波方法,并对滤波处理后的荧光光谱进行自相关计算,得到二维自相关荧光光谱特征图。实验表明,采用改进后的滤波法处理方法,计算得到的4种测试水样的二维荧光光谱图较好的滤除了噪声干扰,并表现了出了明显的差异性。针对二维自相关荧光光谱特征图,构建了基于卷积神经网络（CNN）的突水水源类型判别模型,用于判别突水水源类型。该方法采用深度学习的模型框架,直接对二维自相关荧光光谱特征图进行识别,有效避免了PCA降维的片面性。理论分析和实验结果表明：该模型对水源类型的准确识别率达到了98%,是一种有效的矿井突水水源类型判别方法,为在线矿井突水水源类型判别方法提供了新的思路。     

WHITE ASPARAGUS HARVEST DATE DISCRIMINATION USING NIRS TECHNOLOGY

Asparagus is still an important resource for mid-size and small farms. It has been traditionally believed that factors such as the asparagus harvesting date have an influence on its quality. This research sought to identify the harvesting dates of different fruits by using Near Infrared Spectroscopy (NIRS) technology as quality indicators and the best zone a long of the asparagus to acquire the spectrum. All the asparagus tested came from the same manufacturer but had been canned on three different dates, giving a total of nine lots. There were one hundred asparagus per lot and the experiment data were taken from three different parts (tip, middle, and base) of each spear. Reflectance spectrum in the near infrared between 800–1700 nm was determined. Differences NIRS among the asparagus harvested on different dates were found. NIRS technology was able to classify the asparagus correctly in terms of harvest dates (71% well classified). The base of the asparagus turned out to be the best part to use in order to establish the harvest date.     

A novel k-word relative measure for sequence comparison

In order to extract phylogenetic information from DNA sequences, the new normalized k-word average relative distance is proposed in this paper. The proposed measure was tested by discriminate analysis and phylogenetic analysis. The phylogenetic trees based on the Manhattan distance measure are reconstructed with k ranging from 1 to 12. At the same time, a new method is suggested to reduce the matrix dimension, can greatly lessen the amount of calculation and operation time. The experimental assessment demonstrated that our measure was efficient. What's more, comparing with other methods’ results shows that our method is feasible and powerful for phylogenetic analysis.