Prediction model of K2CsSb photocathode reflectivity based on LSTM北大核心CSCD

针对目前K₂CsSb光阴极制备过程中无法预判光阴极生长状态的问题,提出一种基于长短期记忆（LSTM）循环神经网络的K₂CsSb光阴极反射率预测模型。一维原始反射率数据集经过清洗、筛选、序列化等预处理手段后重构为二维数据输入模型。为充分利用反射率数据在时序上高度相关的特性,采用双层LSTM网络提取特征,预测结果通过全连接层输出,以均方误差（MSE）作为模型预测效果的评判标准。实验结果表明,该模型的网络结构合理且在不同数据集下的表现良好,预测准确率可达99.21%。该模型可运用在K₂CsSb光阴极的制作过程中,通过反射率预测值反馈调节工艺参数以趋近目标走势,对提高光阴极性能具有促进作用。     

光电容积脉搏波的睡眠呼吸暂停综合征筛查方法

睡眠呼吸暂停综合征（SAS）素有“睡眠杀手”之称。由于其诊断金标准多导睡眠监测仪（PSG）的限制,诊断率一直偏低。由于呼吸暂停发生时会引发心率节奏的变化,因此利用心电图（ECG）通过心率变异性（HRV）分析可以实现SAS的自动筛查。但是,ECG-SAS方法所用电极穿戴繁琐、材质致敏性较高,影响睡眠安适度。鉴于脉率变异性（PRV）分析与HRV分析高度相关,并且光电容积脉搏波（PPG）信号相对ECG信号获取方式更加简单,不仅电极不易致敏,而且更易于穿戴,对睡眠干扰小。由此,提出利用同步采集的PPG信号和ECG信号,应用相同的建模方法,比较二者的疾病识别能力。应用反向传播（BP）神经网络,分别建立PPG-SAS与ECG-SAS自动筛查模型,并采用十折交叉验证法及受试者工作特征（ROC）曲线对模型进行对比与评估。实验数据来源于MIT-BIH Polysomnographic Database,共8 248个样本,其中正常样本6 227例。首先采用三层BP神经网络,默认参数下建立PPG-SAS与ECG-SAS模型,使用十折交叉验证法及ROC曲线进行模型分类准确性的对比;然后依次改变影响分类性能的隐层节点数、训练函数以及传递函数,建立多个PPG-SAS与ECG-SAS模型,从中选取各自的最优模型再进行对比。通过比较识别率、预测率以及ROC曲线面积,采用默认参数的PPG-SAS模型优于ECG-SAS模型。通过比较平均分类准确率,隐层节点数为50、训练函数为一步正割算法、隐含层传递函数为双曲正切S型函数时,PPG-SAS模型得到的最高识别率与预测率分别为80.30%和80.13%;隐层节点数为50、训练函数为一步正割算法、隐含层传递函数为径向基时,ECG-SAS模型的最高识别率与预测率分别为77.60%和77.67%。以上实验结果均表明PPG信号的SAS分类能力较ECG信号更具优越性,由此证明了PPG信号筛查SAS的可行性及可靠性,为临床SAS病症的早期发现及诊断率提升奠定理论基础。     

Intelligent Online Monitoring of Rolling Bearing: Diagnosis and Prognosis

This paper suggests a new method to predict the Remaining Useful Life (RUL) of rolling bearings based on Long Short Term Memory (LSTM), in order to obtain the degradation condition of the rolling bearings and realize the predictive maintenance. The approach is divided into three parts: the first part is the clustering to detect the damage state by the density-based spatial clustering of applications with noise. The second one is the health indicator construction which could give a better reflection of the bearing degradation tendency and is selected as the input for the prediction model. In the third part of the RUL prediction, the LSTM approach is employed to improve the accuracy of the prediction. The rationale of this work is to combine the two methods—the density-based spatial clustering of applications with noise and LSTM—to identify the abnormal state in rolling bearings, then estimate the RUL. The suggested method is confirmed by experimental data of bearing life cycle, and the RUL prediction results of the model LSTM are compared with the nonlinear au-regressive model with exogenous input model. In addition, the constructed health indicator is compared with the spectral kurtosis feature. The results demonstrated that the suggested method is more appropriate than the nonlinear au-regressive model with exogenous input model for the prediction of bearing RUL.     

Using Convolutional Neural Network and a Single Heartbeat for ECG Biometric Recognition

The electrocardiogram (ECG) signal has become a popular biometric modality due to characteristics that make it suitable for developing reliable authentication systems. However, the long segment of signal required for recognition is still one of the limitations of existing ECG biometric recognition methods and affects its acceptability as a biometric modality. This paper investigates how a short segment of an ECG signal can be effectively used for biometric recognition, using deep-learning techniques. A small convolutional neural network (CNN) is designed to achieve better generalization capability by entropy enhancement of a short segment of a heartbeat signal. Additionally, it investigates how various blind and feature-dependent segments with different lengths affect the performance of the recognition system. Experiments were carried out on two databases for performance evaluation that included single and multisession records. In addition, a comparison was made between the performance of the proposed classifier and four well-known CNN models: GoogLeNet, ResNet, MobileNet and EfficientNet. Using a time–frequency domain representation of a short segment of an ECG signal around the R-peak, the proposed model achieved an accuracy of 99.90% for PTB, 98.20% for the ECG-ID mixed-session, and 94.18% for ECG-ID multisession datasets. Using the preprinted ResNet, we obtained 97.28% accuracy for 0.5-second segments around the R-peaks for ECG-ID multisession datasets, outperforming existing methods. It was found that the time–frequency domain representation of a short segment of an ECG signal can be feasible for biometric recognition by achieving better accuracy and acceptability of this modality.     

基于长短时记忆神经网络的能谱核素识别方法

针对新兴的能谱核素识别方法在混合放射性核素的噪声环境中存在识别速度慢、准确率较低等问题,提出了基于长短时记忆神经网络(LSTM)的能谱核素识别方法。实验使用溴化镧(LaBr3)晶体探测器,分别对环境中60Co、137Cs放射性源分组测量得到能谱数据集,首先使用数据平滑方法和归一化方法进行数据预处理,然后将能谱数据按时间序列分组以获得可用的输入序列数组,最后训练LSTM模型得到预测结果。通过基于BP神经网络和卷积神经网络(CNN)的两个能谱识别模型进行对比,得到在测试集中平均识别率分别为83.45%和86.21%,而LSTM能谱识别模型平均识别率为93.04%,实验结果表明,该能谱模型在核素识别效果中表现较好,可用于快速的能谱核素识别设备上。     

LIF结合LSTM神经网络的矿井水源识别

矿井水害对煤矿安全生产存在巨大威胁,所以快速识别矿井突水水源,对煤矿水灾预警及灾后救援工作开展都有重大意义。激光诱导荧光（LIF）技术具有快速、高效、灵敏度高等特点,克服了传统水化学方法识别时间长的缺点。循环神经网络（RNN）在解决长序列训练过程中产生的梯度消失、梯度爆炸等问题上存在明显不足,而特殊变体RNN即长短期记忆（LSTM）神经网络很好地弥补了RNN的短板及缺陷。提出了将LIF技术与LSTM算法相结合,应用在矿井突水水源快速识别中。实验样本采自淮南矿区,以砂岩水和老空水为原始样本,并将砂岩水和老空水按照不同比例混合配置成5种混合水样,共7种待测水样进行实验。首先采用最大最小值归一化（MinMaxScaler）、平滑滤波（SG）以及标准正态变量变换（SNV）三种预处理方法对原始光谱数据进行预处理,减少原始光谱数据存在的噪声和干扰信息。之后为防止数据量过大,维度过高,将包括原始光谱数据在内的四组数据再进行LDA降维至3维。最后分别搭建LSTM识别模型,从测试集预测准确率、训练集准确率变化趋势以及训练集损失函数变化趋势三个方面进行比较,选择最优模型。其中SG+LDA+LSTM和Original+LDA+LSTM在测试集预测准确率上都能达到100%,MinMaxScaler+LDA+LSTM测试集预测准确率在98.57%,SNV+LDA+LSTM准确率最低,只有87.14%;在训练集准确率变化趋势表现上,SG+LDA+LSTM能够保持良好的学习,很快达到100%,Original+LDA+LSTM和MinMaxScaler+LDA+LSTM也能达到100%的准确率,但在前几次训练过程中会有准确率下降的情况出现,SNV+LDA+LSTM训练集准确率在训练次数内并未达到100%;SG+LDA+LSTM损失函数变化趋势也具有很好的收敛性和稳定性,Original+LDA+LSTM,MinMaxScaler+LDA+LSTM以及SNV+LDA+LSTM在损失函数变化趋势上表现并不出色。结果表明,4组模型中,SG+LDA+LSTM模型是最适合应用于矿井突水识别,该方法补充了矿井突水水源识别工作的内容,为矿井突水识别提供了新的思路。     

Physics-informed recurrent neural networks for linear and nonlinear flame dynamics

This paper demonstrates the ability of recurrent neural networks (RNNs) to predict the linear and the nonlinear response of a premixed laminar flame to incoming velocity perturbations. We develop data-driven models, which require the velocity and heat release rate fluctuations as input data. Both time series are obtained from Direct Numerical Simulations (DNS) of a laminar flame. The length of the signals, and, hence, the cost of the simulation, is comparable to those used in the linear framework of System Identification. A more robust type of RNNs, namely long short term memory (LSTM), is employed to reduce the dependency on large datasets. The LSTM framework is modeled as a time series regression problem and four models are trained with decreasing data set lengths. All purely data-driven models accurately predict the unsteady time series of the heat release rate and, hence, the Flame Transfer Functions (FTFs). We further improve the model accuracy by incorporating a physical constraint, namely the low-frequency limit for perfectly-premixed flames, into the LSTM model. This step reduces the required data length compared to the purely data-driven approach. The proposed model, called PI-LSTM, is able to reproduce the linear and the nonlinear FTFs for amplitudes up to 50% of the laminar flame based on one numerical simulation, where the length of the time series is 100 ms.     

Taming the Chaos in Neural Network Time Series Predictions

Machine learning methods, such as Long Short-Term Memory (LSTM) neural networks can predict real-life time series data. Here, we present a new approach to predict time series data combining interpolation techniques, randomly parameterized LSTM neural networks and measures of signal complexity, which we will refer to as complexity measures throughout this research. First, we interpolate the time series data under study. Next, we predict the time series data using an ensemble of randomly parameterized LSTM neural networks. Finally, we filter the ensemble prediction based on the original data complexity to improve the predictability, i.e., we keep only predictions with a complexity close to that of the training data. We test the proposed approach on five different univariate time series data. We use linear and fractal interpolation to increase the amount of data. We tested five different complexity measures for the ensemble filters for time series data, i.e., the Hurst exponent, Shannon’s entropy, Fisher’s information, SVD entropy, and the spectrum of Lyapunov exponents. Our results show that the interpolated predictions consistently outperformed the non-interpolated ones. The best ensemble predictions always beat a baseline prediction based on a neural network with only a single hidden LSTM, gated recurrent unit (GRU) or simple recurrent neural network (RNN) layer. The complexity filters can reduce the error of a random ensemble prediction by a factor of 10. Further, because we use randomly parameterized neural networks, no hyperparameter tuning is required. We prove this method useful for real-time time series prediction because the optimization of hyperparameters, which is usually very costly and time-intensive, can be circumvented with the presented approach.     

拉曼光谱结合LSTM长短期记忆网络的樱桃产地鉴别研究

现在樱桃市场上存在着大量以次充好的不良现象,严重损害了名牌樱桃的品牌经济效益,所以亟需一种能对不同产地樱桃实现快速无损鉴别的技术。拉曼光谱溯源技术作为光谱溯源技术的一种,由于具有快速、高效、无污染、无损分析等优点,逐渐得到相关研究者的重视。长短期记忆（LSTM）网络是一种具有记忆性的反馈神经网络,它是循环神经网络的一种变体。LSTM网络克服了循环神经网络中梯度消失的缺点,适合处理序列敏感的问题和任务,目前被广泛应用在语音识别、图像识别和手写识别等领域,但LSTM网络在产地溯源方面的应用还有待研究。基于此,提出了一种LSTM网络与拉曼光谱技术结合的能对不同产地樱桃实现快速无损鉴别的技术。将来自美国、山东和四川的369个樱桃作为研究样本,用拉曼光谱仪在785 nm激光下获得了不同产地樱桃的光谱数据。并且以每条经过基线校正后的拉曼光谱数据作为网络输入数据,基于LSTM网络构建了能对不同产地樱桃实现快速鉴别的判别模型,并且以样本判别准确率A、样本精确率P、样本召回率R和样本F值作为评价指标,探究了不同预处理方法对LSTM网络判别模型性能的影响。结果表明：当样本训练集和测试集的比例为85∶38时,直接采用原始拉曼光谱数据的LSTM网络模型的产地鉴别能力不高,鉴别准确率为79.87%。但当使用预处理过后的拉曼光谱数据,模型的鉴别准确率维持在92%以上。并且光谱经过SG+MSC预处理后模型的鉴别准确度最好,鉴别准确率达99.12%。同时在采用SG+MSC预处理的方法下,LSTM网络鉴别模型的精确率、召回率、F值均较高,表明了所提出的LSTM网络模型有较好的性能可实现对不同产地樱桃的鉴别,为樱桃的产地溯源提供了一种新的思路。     

基于PCA-BP神经网络的光谱消光法颗粒粒径反演算法研究

光谱消光法广泛应用于颗粒粒径测量领域,在利用光谱消光法对颗粒粒径进行反演的过程中,由于颗粒的消光系数存在理论复杂、计算繁琐、收敛速度慢以及求解不稳定等问题,很大程度上影响了整个反演过程的快速性和准确性。且在众多波长的消光数据中,存在较多重复冗余的信息,也很大程度上增加了反演算法的时间。针对光谱消光法粒径反演算法计算繁琐、反演效率低的问题,提出了基于主成分分析（PCA）和BP神经网络的光谱消光颗粒粒径分析方法。基于Mie散射理论对不同粒径、不同波长下的光谱消光值进行了仿真计算,通过对光谱消光数据集的主成分分析及各个波长综合载荷系数的计算,实现了最优特征波长的选取,利用降维后的光谱消光数据训练了PCA-BP神经网络模型,并利用该网络模型计算了粒径颗粒分布。通过仿真计算,比较了PCA-BP神经网络模型与传统的BP神经网络模型的预测精度,并分析了波长数目对两种神经网络模型预测结果的影响。针对训练得到的PCA-BP神经网络模型开展光谱消光法粒径参数反演算法的验证实验,搭建了光谱消光法颗粒粒径参数测量实验系统,测量了粒径范围在0.5～9.7 μm内的6种不同粒径参数的聚苯乙烯标准颗粒。仿真和实验结果表明：基于主成分分析方法可确定各个波长向量之间的相关性,利用综合载荷系数选取最优特征波长对应的消光值对整体的光谱数据具有较好的代表性,可实现光谱数据的降维。相比传统的BP神经网络模型,基于PCA-BP神经网络模型的颗粒粒径分布的分析方法预测精度更高,对于较分散颗粒系的分布参数的预测有更加明显的优势。而且,被选取的波长数较少时,PCA-BP神经网络模型依然有较高的预测精度。利用训练好的PCA-BP神经网络模型对颗粒粒径参数进行实验验证,预测结果可瞬时输出,颗粒粒径分布误差在5%以内,验证了该算法的可行性。     

基于混合神经网络和注意力机制的混沌时间序列预测

为提高混沌时间序列的预测精度,提出一种基于混合神经网络和注意力机制的预测模型(Att-CNNLSTM),首先对混沌时间序列进行相空间重构和数据归一化,然后利用卷积神经网络(CNN)对时间序列的重构相空间进行空间特征提取,再将CNN提取的特征和原时间序列组合,用长短期记忆网络(LSTM)根据空间特征提取时间特征,最后通过注意力机制捕获时间序列的关键时空特征,给出最终预测结果.将该模型对Logistic,Lorenz和太阳黑子混沌时间序列进行预测实验,并与未引入注意力机制的CNN-LSTM模型、单一的CNN和LSTM网络模型、以及传统的机器学习算法最小二乘支持向量机(LSSVM)的预测性能进行比较.实验结果显示本文提出的预测模型预测误差低于其他模型,预测精度更高.     

混沌时间序列的局域高阶Volterra滤波器多步预测模型

依据相空间邻近轨道演化相似性特点建立训练模式,提出了基于自适应高阶非线性Volterra滤波器（HONFIR）的混沌时间序列多步预测模型（MSP-HONFIR）;通过定义距离相似度、趋势相似度来衡量轨道演化相似度,提出了混沌吸引子邻近轨道判别的新方法;从模型训练充分性角度出发探讨了MSP-HONFIR滤波器模型训练集规模控制的依据.数值研究表明MSP-HONFIR滤波器模型的多步预测性能优于原有HONFIR滤波器模型. 关键词： 混沌 非线性自适应预测 Volterra滤波器模型 训练模式     

Magnetic resonance imaging in patients with unstable angina: comparison with acute myocardial infarction and normals

The role of magnetic resonance imaging in characterizing normal, ischemic and infarcted segments of myocardium was examined in 8 patients with unstable angina, 11 patients with acute myocardial infarction, and 7 patients with stable angina. Eleven normal volunteers were imaged for comparison. Myocardial segments in short axis magnetic resonance images were classified as normal or abnormal on the basis of perfusion changes observed in thallium-201 images in 22 patients and according to the electrocariographic localization of infarction in 4 patients. T2 relaxation time was measured in 57 myocardial segments with abnormal perfusion (24 with reversible and 33 with irreversible perfusion changes) and in 25 normally perfused segments. T2 measurements in normally perfused segments of patients with acute myocardial infarction, unstable angina and stable angina were within normal range derived from T2 measurements in 48 myocardial segments of 11 normal volunteers (42 +/- 10 ms). T2 in abnormal myocardial segments of patients with stable angina also was not significantly different from normal. T2 of abnormal segments in patients with unstable angina (64 +/- 14 in reversibly ischemic and 67 +/- 21 in the irreversibly ischemic segments) was prolonged when compared to normal (p less than 0.0001) and was not significantly different from T2 in abnormal segments of patients with acute myocardial infarction (62 +/- 18 for reversibly and 66 +/- 11 for irreversibly ischemic segments). The data indicate that T2 prolongation is not specific for acute myocardial infarction and may be observed in abnormally perfused segments of patients with unstable angina.(ABSTRACT TRUNCATED AT 250 WORDS)     

基于Hadoop的GA-BP网络在山洪预测中的研究

研究了山洪灾害监测预警系统中雨情数据的分布式存储和分布式预测。针对采集到的水文数据急剧增长和对预测精度和预报时效的要求不断提高,分别应用Hadoop分布式文件系统对数据进行分布式存储和 MapReduce框架结合遗传算法优化神经网络的权值和阈值进行分布式预测。采用基于BP神经网络的多因子山洪灾害雨量预测模型,结合遗传算法能够实现全局优化特点来优化神经网络的权值和阈值,并在数据并行处理过程中,采用了批处理和MapReduce工作流的方式,以误差和准确率来评估预测模型,解决了神经网络在处理海量数据时训练时间长等问题。实验表明,该方法可以在不影响准确度的前提下,大大缩短运行时间,提高预测效率。     

基于遗传算法-支持向量机的兔肝VX2肿瘤光谱鉴别

兔肝VX２肿瘤是一种快速生长的肿瘤模型,可以在多种器官如肝、肺、直肠等快速生长,常用于肿瘤研究。采用可见-近红外高光谱技术对四只兔子的兔肝VX２肿瘤和正常组织进行活体和离体的反射光谱检测,然后采用支持向量机分别实现了二分类(正常肝组织和肝VX2肿瘤组织)和四分类(未出血活体正常肝组织、未出血活体VX2肿瘤组织、出血离体正常肝组织和出血离体肝VX2肿瘤组织)。根据其光谱反射曲线的特征,选择了400～1 800 nm区间的数据为特征变量。为进一步提高分类准确率,分别采用5折交叉验证和遗传算法对支持向量机的核函数参数g和惩罚因子c进行了优化。其中5折交叉验证优化参数和分类结果为：二分类优化的惩罚参数c为4,核函数参数g为0.125 0,其校正集和预测集的准确率都达到了100%;四分类中优化出的参数c为8,g为0.121 1,其校正集和预测集的准确率分别达到了99.242 4%和93.333%。遗传算法优化参数和结果为：二分类中优化的参数c为0.845 6,g为0.062 5,其校正集和预测集的准确率同样都达到了100%;四分类中优化的参数c为5.5307,g为0.068 5,其校正集和预测集的准确率分别达到了99.242 4%和100%。结果显示两种优化方法都取得了很好的效果,遗传算法优化参数对四分类的分类更为精确。为进一步提升算法速度,采用间隔选取变量的方法来不断减少特征变量,最终每隔100 nm谱段选择一个变量,共选择14个谱段作为特征变量。采用遗传算法优化支持向量机参数并对其分类进行了研究,结果表明：二分类和四分类的校正集和预测集结果准确率均为99.242 4%,而且运行时间分别为11.4和20.0 s, 与选择全波段的运行时间：340.3和491.0 s相比, 说明多光谱技术可以进行肝VX2肿瘤组织和正常肝组织的鉴别,且分类准确率可达99%以上,而且运行时间缩短了很多。为未来多光谱技术在未来临床肿瘤诊断中实现肿瘤组织的快速实时在线检测和分类奠定了基础,显示出巨大的应用潜力。     

Unsupervised Anomaly Detection with Distillated Teacher-Student Network Ensemble

We address the problem of unsupervised anomaly detection for multivariate data. Traditional machine learning based anomaly detection algorithms rely on specific assumptions of normal patterns and fail to model complex feature interactions and relations. Recently, existing deep learning based methods are promising for extracting representations from complex features. These methods train an auxiliary task, e.g., reconstruction and prediction, on normal samples. They further assume that anomalies fail to perform well on the auxiliary task since they are never trained during the model optimization. However, the assumption does not always hold in practice. Deep models may also perform the auxiliary task well on anomalous samples, leading to the failure detection of anomalies. To effectively detect anomalies for multivariate data, this paper introduces a teacher-student distillation based framework Distillated Teacher-Student Network Ensemble (DTSNE). The paradigm of the teacher-student distillation is able to deal with high-dimensional complex features. In addition, an ensemble of student networks provides a better capability to avoid generalizing the auxiliary task performance on anomalous samples. To validate the effectiveness of our model, we conduct extensive experiments on real-world datasets. Experimental results show superior performance of DTSNE over competing methods. Analysis and discussion towards the behavior of our model are also provided in the experiment section.     

URLLC occasional large time delay prediction based on unbalanced regression and LSTM

Occasional large time delay is one of the bottlenecks for 5G applied to the industrial field. Accurate measurement, analysis, and prediction of communication network latency are of great significance. Due to the strong randomness and small number of samples in ultra-reliable and low latency communication (URLLC), the prediction task of occasional large time delay is very challenging. This paper proposes an URLLC occasional large time delay prediction method based on unbalanced regression algorithms and long–short term memory (LSTM). We first decompose the original sequence by using the variational mode decomposition (VMD) to obtain relatively stable component sequences. The parameters of the VMD are automatically optimized by grasshopper optimization algorithm (GOA). In order to improve the prediction effect of URLLC occasional large time delay, we introduce unbalanced regression algorithms. Before the VMD decomposition of data samples, the SMOGN is introduced to balance the number of large-time delay samples and common-time delay samples. Next we calculate the weight of each sample based on the rarity of each sample point and our proposed LDSWeight. Finally, LSTM is used to complete cost-sensitive learning. The experimental results show that the URLLC occasional large time delay prediction method proposed in this paper has better prediction accuracy than other methods.     

基于近红外光谱分析技术测定库尔勒香梨硬度

采用近红外(NIR)漫反射光谱法对新疆特色梨果库尔勒香梨的五种不同果（包括青头、粗皮、脱萼、宿萼、突顶果）的硬度进行测定。由于近红外光谱数据量大且原始光谱噪声明显、测定水果时散射严重等导致光谱建模时关键波长变量提取困难。以新疆库尔勒香梨为研究对象,为了有效地消除固体表面散射以及光程变化对NIR漫反射光谱的影响,首先采用标准正态变量变换（SNV）和多元散射校正（MSC）对库尔勒香梨的原始光谱进行预处理。为寻找适合近红外光谱检测库尔勒香梨硬度的最佳特征波长筛选方法,进行香梨近红外光谱的特征波长变量选择方法的比较与研究。研究比较了两种特征波长筛选方法对库尔勒香梨硬度偏最小二乘法（PLS）建模精度的影响。同时使用反向偏最小二乘（BiPLS）和遗传算法结合反向偏最小二乘（BiPLS-GA）在全光谱范围内筛选香梨硬度的特征波长变量,将校正均方根误差（RESMC）、预测均方根误差（RESMP）以及决定系数（R2）作为模型的评价标准,并最终确定最优波段选择方法及最佳预测模型。基于选择的特征波长变量建立的PLS模型（BiPLS-GA）与全光谱变量建立的PLS模型进行比较发现BiPLS-GA模型仅仅使用原始变量中6.6%的信息就获得了比全变量PLS模型更好的库尔勒香梨硬度的预测结果,其中R2,RMSEC和RMSEP分别为0.91,1.03和1.01。进一步与基于反向偏最小二乘算法（BiPLS）获得的特征变量建立的PLS模型比较发现,BiPLS-GA不仅可以去除原始光谱数据中的无信息变量,同时也能够对共线性的变量进行压缩去除,使得建模变量从301个减少到20个。极大地简化模型的同时有效地提高了模型的预测精准度和稳定性。因此该方法能够有效地用于近红外光谱数据变量的选择。证明了近红外光谱分析技术结合BiPLS-GA模型能够高效地选择出建模变量,去除与库尔勒香梨硬度无关的近红外光谱信息,显著地提高库尔勒香梨硬度定量模型的预测精度。这不仅为新疆地区特色梨果库尔勒香梨的快速、精确、无损优选分级提供一定的技术支持,同时也为基于近红外光谱分析技术预测水果内部品质的研究提供了参考。     

基于Vis/NIR光谱技术的酿酒葡萄成熟期间SSC预测研究

酿酒葡萄成熟度是确定葡萄采收期的重要品质指标,针对酿酒葡萄大田中成熟度检测难度大的问题,利用可见/近红外（Vis/NIR）光谱技术和化学计量学,研究了酿酒葡萄可溶性固形物含量（SSC）与光谱数据之间的内在联系。采用USB2000+光谱仪获取5种酿酒葡萄及其叶片在不同成熟时期的Vis/NIR光谱数据,通过OMNIC 8.0软件提取光谱数据,将化学值与光谱吸收率值通过TQ Analyst8.0软件建立模型。选取信噪比高的450~1 000 nm波段,利用PCA剔除异常光谱数据,将一阶导数（FD）、Savitzky-Golay卷积平滑（S-G）、多元散射校正（MSC）、标准正态变换（SNV）分别组合共4种方法用于光谱数据预处理。利用偏最小二乘（PLS）法分别建立了5种葡萄基于酿酒葡萄光谱数据的SSC预测模型,建立了5种葡萄基于冠层叶片光谱数据的SSC预测模型,对比了不同方式预处理后的建模效果,并选择最优预处理方式建模。最后用外部样本分别验证了SSC预测模型。结果表明,采用S-G平滑+FD+MSC的预处理方法时大多数预测模型性能达到最好。5种葡萄浆果校正集和验证集的R分别达到0.93和0.86以上,最高均方根误差分别为0.30和0.48,5种葡萄冠层叶片校正集和验证集的R分别达到0.73和0.65以上,最大均方根误差分别为0.95和0.75。5种葡萄浆果外部试验样本预测值与真实值间的平均RE最高为0.43%。基于酿酒葡萄浆果光谱的SSC预测模型具备良好的预测能力,优于基于酿酒葡萄冠层叶片光谱的SSC预测模型,SSC预测模型能够为酿酒葡萄成熟度评价研究提供理论参考。Vis/NIR光谱技术适用于在酿酒葡萄大田中快速、无损检测SSC。