Comparative assessment of the frying efficiency of standard and low linolenic rapeseed oils: Principal Component Analysis (PCA)

In this research,the performance of regular rapeseed oil(RSO)and modified low-linolenic rapeseed oil(LLRO)during frying was assessed using a frying procedure that commonly found in fast-food restaurants.Key physicochemical attributes of these oils were investigated.RSO and LLRO differed for initial linolenic acid(12.21%vs.2.59%),linoleic acid(19.15%vs.24.73%).After 6 successive days frying period of French fries,the ratio of linoleic acid to palmitic acid dropped by 54.49%in RSO,higher than that in LLRO(51.54%).The increment in total oxidation value for LLRO(40.46 unit)was observed to be significantly lower than those of RSO(42.58 unit).The changes in carbonyl group value and iodine value throughout the frying trial were also lower in LLRO compared to RSO.The formation rate in total polar compounds for LLRO was 1.08%per frying day,lower than that of RSO(1.31%).In addition,the formation in color component and degradation in tocopherols were proportional to the frying time for two frying oils.Besides,a longer induction period was also observed in LLRO(8.87 h)compared to RSO(7.68 h)after frying period.Overall,LLRO exhibited the better frying stability,which was confirmed by principal component analysis(PCA).

Using deep neural networks coupled with principal component analysis for ore production forecasting at open-pit mines

Ore production is usually affected by multiple influencing inputs at open-pit mines.Nevertheless,the complex nonlinear relationships between these inputs and ore production remain unclear.This becomes even more challenging when training data(e.g.truck haulage information and weather conditions)are massive.In machine learning(ML)algorithms,deep neural network(DNN)is a superior method for processing nonlinear and massive data by adjusting the amount of neurons and hidden layers.This study adopted DNN to forecast ore production using truck haulage information and weather conditions at open-pit mines as training data.Before the prediction models were built,principal component analysis(PCA)was employed to reduce the data dimensionality and eliminate the multicollinearity among highly correlated input variables.To verify the superiority of DNN,three ANNs containing only one hidden layer and six traditional ML models were established as benchmark models.The DNN model with multiple hidden layers performed better than the ANN models with a single hidden layer.The DNN model outperformed the extensively applied benchmark models in predicting ore production.This can provide engineers and researchers with an accurate method to forecast ore production,which helps make sound budgetary decisions and mine planning at open-pit mines.

A Hybrid Optimization Approach of Single Point Incremental Sheet Forming of AISI 316L Stainless Steel Using Grey Relation Analysis Coupled with Principal Component Analysiss

We investigated the parametric optimization on incremental sheet forming of stainless steel using Grey Relational Analysis(GRA) coupled with Principal Component Analysis(PCA). AISI 316L stainless steel sheets were used to develop double wall angle pyramid with aid of tungsten carbide tool. GRA coupled with PCA was used to plan the experiment conditions. Control factors such as Tool Diameter(TD), Step Depth(SD), Bottom Wall Angle(BWA), Feed Rate(FR) and Spindle Speed(SS) on Top Wall Angle(TWA) and Top Wall Angle Surface Roughness(TWASR) have been studied. Wall angle increases with increasing tool diameter due to large contact area between tool and workpiece. As the step depth, feed rate and spindle speed increase,TWASR decreases with increasing tool diameter. As the step depth increasing, the hydrostatic stress is raised causing severe cracks in the deformed surface. Hence it was concluded that the proposed hybrid method was suitable for optimizing the factors and response.

A Modified Principal Component Analysis Method for Honeycomb Sandwich Panel Debonding Recognition Based on Distributed Optical Fiber Sensing Signals

The safety and integrity requirements of aerospace composite structures necessitate real-time health monitoring throughout their service life.To this end,distributed optical fiber sensors utilizing back Rayleigh scattering have been extensively deployed in structural health monitoring due to their advantages,such as lightweight and ease of embedding.However,identifying the precise location of damage from the optical fiber signals remains a critical challenge.In this paper,a novel approach which namely Modified Sliding Window Principal Component Analysis(MSWPCA)was proposed to facilitate automatic damage identification and localization via distributed optical fiber sensors.The proposed method is able to extract signal characteristics interfered by measurement noise to improve the accuracy of damage detection.Specifically,we applied the MSWPCA method to monitor and analyze the debonding propagation process in honeycomb sandwich panel structures.Our findings demonstrate that the training model exhibits high precision in detecting the location and size of honeycomb debonding,thereby facilitating reliable and efficient online assessment of the structural health state.

Robust Principal Component Analysis Integrating Sparse and Low-Rank Priors

Principal Component Analysis (PCA) is a widely used technique for data analysis and dimensionality reduction, but its sensitivity to feature scale and outliers limits its applicability. Robust Principal Component Analysis (RPCA) addresses these limitations by decomposing data into a low-rank matrix capturing the underlying structure and a sparse matrix identifying outliers, enhancing robustness against noise and outliers. This paper introduces a novel RPCA variant, Robust PCA Integrating Sparse and Low-rank Priors (RPCA-SL). Each prior targets a specific aspect of the data’s underlying structure and their combination allows for a more nuanced and accurate separation of the main data components from outliers and noise. Then RPCA-SL is solved by employing a proximal gradient algorithm for improved anomaly detection and data decomposition. Experimental results on simulation and real data demonstrate significant advancements.

Comparative Analysis of Differences among Northern,Jiangnan,and Lingnan Classical Private Gardens Using Principal Component Cluster Method

This paper investigates the design essence of Chinese classical private gardens,integrating their design elements and fundamental principles.It systematically analyzes the unique characteristics and differences among classical private gardens in the Northern,Jiangnan,and Lingnan regions.The study examines nine classical private gardens from Northern China,Jiangnan,and Lingnan by utilizing the advanced tool of principal component cluster analysis.Based on literature analysis and field research,273 variables were selected for principal component analysis,from which four components with higher contribution rates were chosen for further study.Subsequently,we employed clustering analysis techniques to compare the differences among the three types of gardens.The results reveal that the first principal component effectively highlights the differences between Jiangnan and Lingnan private gardens.The second principal component serves as the key to defining the types of Northern private gardens and distinguishing them from the other two types,and the third principal component indicates that Lingnan private gardens can be categorized into two distinct types as well.

基于KPCA-CNN-DBiGRU模型的短期负荷预测方法

本文针对已有神经网络模型在短期负荷预测中输入维度过高、预测误差较大等问题,提出了一种结合核主成分分析、卷积神经网络和深度双向门控循环单元的短期负荷预测方法。先运用核主成分分析法对原始高维输入变量进行降维,再通过卷积深度双向门控循环单元网络模型进行负荷预测。以第九届全国电工数学建模竞赛试题A题中的负荷数据作为实际算例,结果表明所提方法较降维之前预测误差大大降低,与已有预测方法相比也有大幅的误差降低。

Estimation of the Number of Collapsed Houses Damaged by Typhoon Based on Principal Components Analysis and Support Vector Machine

The evaluation model was established to estimate the number of houses collapsed during typhoon disaster for Zhejiang Province.The factor leading to disaster,the environment fostering disaster and the exposure of buildings were processed by Principal Component Analysis.The key factor was extracted to support input of vector machine model and to build an evaluation model;the historical fitting result kept in line with the fact.In the real evaluation of two typhoons landed in Zhejiang Province in 2008 and 2009,the coincidence of evaluating result and actual value proved the feasibility of this model.

基于KPCA-PSO-ELM算法的地表水化学需氧量紫外-可见吸收光谱检测研究

化学需氧量(COD)是水质检测重要指标之一,反映水体有机物含量。传统的COD化学检测方法存在操作繁琐,等待时间长,二次污染等缺点。紫外-可见吸收光谱法是目前水体化学需氧量检测中应用最为广泛的方法之一,具有检测快速、无污染等特点。为了满足地表水化学需氧量快速、实时、在线监测等要求,采用紫外-可见吸收光谱进行测量,提出了内核主成分分析(KPCA)结合粒子群优化极限学习机(PSO-ELM)预测模型,满足当前对地表水化学需氧量快速、实时监测的要求。对光谱进行Savitzky-Golay(SG)滤波以降低随机噪声的影响;用积分光谱代替原光谱,以降低信号波动带来的影响;再将得到的光谱信息归一化,消除不同光谱数据量纲的影响。将预处理后的数据利用KPCA算法将全光谱数据压缩为5个特征,有效解决光谱信息冗余的问题;采用PSO算法对ELM的权重和偏置进行优化极大提高了模型的精度。对217个河流、长江及支流、湖库等地表水样本按照7∶3随机划分成训练集和测试集,并进行建模测试,其中训练集拟合优度(R2)为0.930 2、均方根误差(RMSE)为0.363 0 mg·L^(-1)、测试集拟合优度R2为0.931 9、均方根误差(RMSE)为0.400 7 mg·L^(-1)。为了验证提出的基于KPCA全光谱数据压缩方法对预测模型的提升效果,分别对比了主成分分析(PCA)、连续投影算法(SPA)、套索回归(LASSO)等特征处理算法。PCA-PSO-ELM模型的RMSE为0.715 1 mg·L^(-1)、 SPA-PSO-ELM模型的RMSE为0.473 7 mg·L^(-1)、 LASSO-PSO-ELM模型的RMSE为0.412 6 mg·L^(-1), KPCA-PSO-ELM模型较上述三种模型,RMSE分别降低了78.46%、 18.22%、 2.97%,结果表明KPCA是一种高效的光谱降维算法,能够有效消除光谱冗余信息,提升模型预测精度。基于KPCA-PSO-ELM预测模型结合紫外-可见吸收光谱可以实现对地表水COD快速、实时检测,为在线COD检测场景提供方法支撑。

基于SMOTE-IKPCA-SeNet深度迁移学习的小批量生产质量预测研究

随着智能制造技术的发展和客户个性化需求的增加,多品种小批量生产方式逐渐成为制造业的主流。面向大批量生产、以统计过程控制为核心的质量管理方式并不适用于小批量生产。针对复杂生产过程存在参数多、非线性和交互作用的问题,提出利用深度迁移学习的方式将历史生产数据作为源域迁移至小样本目标产品数据进行质量预测。首先,通过合成少数类过采样技术(synthetic minority over-sampling technique,SMOTE)和改进的核主成分分析(improved kernel principal component analysis,IKPCA)算法筛选源域和目标域的可迁移特征,这不仅兼顾了特征重要性和可迁移性,还减少了“负迁移”,提高了模型泛化能力;然后,采用结合通道注意力机制的卷积神经网络SeNet构建基于深度迁移学习的质量预测模型。仿真结果表明,随着目标域样本的增加,所提方法的预测准确性明显优于广泛采用的支持向量机建模方法。同时,所提可迁移特征筛选方法显著提高了深度迁移学习的质量预测效果,为复杂的小批量生产过程质量保证提供了新方法。

基于聚类及PCA分析的羌活栽培区土壤质量评价

为了准确评价川产道地药材羌活栽培区耕作层土壤质量状况,分别采用聚类分析法(CA)和主成分分析法(PCA)构建栽培区耕作层土壤质量最小数据集(minimum data set,MDS),利用最小数据集土壤质量指数(soil quality index-CA,SQI-CA和SQI-PCA)和全量数据集土壤质量指数(SQI-T)评价川西北羌活栽培区耕作层土壤质量。结果表明:(1)羌活栽培区土壤有机质含量为(19.14±6.75)g·kg^(−1),处于中度贫瘠化水平;土壤速效氮、速效磷和速效钾含量较高,分别为(129.78±47.78)mg·kg^(−1)、(22.89±14.78)g·kg^(−1)和(159.87±97.87)mg·kg^(−1);土壤为中性土壤,pH均值为7.20±1.68。(2)基于不同数据集的土壤质量指数均值排序为SQI-T>SQI-PCA>SQI-CA,而SQI-PCA与SQI-T之间的Nash有效系数高于SQI-CA,相对偏差系数低于SQI-CA,说明基于主成分分析的最小数据集(MDS-PCA)评价效果更优,该数据集包括土壤容重、抗剪强度、有机质含量、饱和导水率、黏粒含量、pH、速效氮和砂粒含量共8个指标。(3)川西北羌活栽培区土壤质量指数SQI-PCA<0.33,表明该研究区耕作层土壤质量总体水平较差,主要体现在土壤紧实、有机质含量低,需要通过合理耕作、施肥和土壤改良等方式对耕作层土壤质量进行有效调控。研究结果可为川西北高原羌活栽培区土壤质量改良和生产适宜性调控提供参考,有利于川西北高原区中药材产区土壤可持续利用。

基于KPCA与KLPP及Wilks统计量的留兰香三维荧光数据特征提取与鉴别分析

为实现留兰香产地的快速鉴别,该研究提出了一种核主成分分析(kernel principal component analysis,KPCA)与核局部保持投影(kernel locality preserving projections,KLPP)及WilksΛ统计量序贯融合的特征波长提取策略,在此基础上鉴别5个产地的留兰香。首先,在采集5个产地300个留兰香样本的三维荧光数据后,运用三角形内插值法去除原始光谱中的瑞利散射和拉曼散射,并运用SG(Savitzky-Golay)对数据进行平滑预处理。然后,对预处理后的荧光光谱数据分别利用KPCA、KPCA+KLPP、KPCA+WilksΛ统计量、 KPCA+KLPP+WilksΛ统计量4种方法提取特征激发波长和特征发射波长。接着,按特征激发波长从小到大顺序将其对应的特征发射波长光谱值首尾相连转换成行向量;4种方法从300个样本中各得到1个300行的特征波长光谱值矩阵。再者,运用Fisher判别分析(fisher discriminant analysis,FDA)对特征波长光谱值矩阵进行数据可分性融合,生成可分性FD(fisher discriminant)变量。选取前4个累计判别能力达到99%的FD变量作为鉴别模型的输入向量。最后,用支持向量机(support vector machine,SVM)算法分析4个FD变量,分别得到对应于4种特征提取波长方法的FDA+SVM鉴别结果,其正确率分别为92.00%、96.00%、94.67%、100%。结果表明,所提出的KPCA+KLPP+WilksΛ统计量序贯融合的特征波长提取策略能够有效减少三维荧光光谱数据的冗余,并能表征原始荧光数据的信息特征,实现了5种留兰香产地的正确鉴别。该研究可为后续利用三维荧光光谱开展留兰香重要组分量化分析提供一定的基础。

基于RS-PCA-SVM的建筑项目安全预测模型

为了减少建筑项目安全事故的发生,文章提出一种基于RS-PCA-SVM建筑项目安全组合预测模型,采用粗糙集理论(rough set,RS)对数据进行属性约简,剔除交叉和冗余信息,降低输入变量维数和计算复杂度,减少训练时间;利用主成分分析(principal component analysis,PCA)法进行降维处理,除去贡献率较低的主成分,将剩余主成分作为支持向量机(support vector machine,SVM)的输入变量,并选择自适应权重粒子群优化算法(particle swarm optimization,PSO)优化SVM的参数,避免参数选择的盲目性。结果表明:该模型的平均预测准确率为93.78%,相比传统方法预测精度高、计算速度快。

基于双子空间PCA降维的脑力负荷分类

人类社会至今的飞速发展使得大量体力劳动被机械工程替代,工作者的任务重心也从体力劳动逐渐转变为脑力劳动,对操作者脑力负荷进行实时评估以增强工作效率在当下有着重大意义。目前人类对于脑力负荷评估共有3种方式,有研究表明,采用生物电信号进行脑力负荷分类效果较其余两种方法更客观。但脑电信号经过特征提取后维数极高,所需数据量和运算量巨大,需要对其进行降维。目前降维方面最广泛运用的两种算法为主成分分析(principal component analysis,PCA)和线性判别分析(linear discriminate analysis,LDA)。针对PCA的非监督性和LDA的特征冗余敏感性,提出一种二分类下基于双子空间主成分分析的降维算法,分别对不同类别的训练集数据进行主成分分析,并将所有训练集数据映射到生成的空间中,再次进行PCA-LDA降维,以此提高降维后数据的可分性。实验结果表明,双子空间PCA-LDA降维算法在二分类任务下测试集精度整体高于单子空间PCA-LDA算法,以此为脑力负荷分类领域和高维数据降维领域提供了新思路。

基于PCA-BP神经网络的巷道通风摩擦阻力系数预测模型

根据实测巷道通风摩擦阻力系数数据的特点,建立了主成分分析PCA-BP神经网络预测模型。采用PCA法对影响巷道通风摩擦阻力系数的支护类型、断面形状、巷道宽、巷道高、支护部分周边长、巷道断面积和巷道长度7个因素进行降维。将降维后因素的贡献率进行排序筛选,得到3个主成分指标(F_(1)、F_(2)和F_(3)),作为BP神经网络输入层的神经元。利用实测数据对PCA-BP神经网络模型进行训练和测试,并将测试结果与支持向量机回归(SVM)模型和BP神经网络模型的测试结果进行对比,结果显示:全因素的BP神经网络预测模型和SVM预测模型的平均精度分别为92.9420%、93.0235%,而PCA-BP预测模型的平均精度达到了96.4325%。PCA-BP神经网络模型不但简化了网络结构,更提高了网络的泛化能力,使预测误差更小、精度更高,为更准确地获得巷道通风摩擦阻力系数提供了一种有效的方法。

基于PCA-EWM两级特征融合和NGO-GRU的梁桥损伤诊断

为了提高损伤识别中单一指标对损伤的灵敏度和抗噪能力,基于模态应变能理论,提出联合主成分分析(principal component analysis,PCA)和熵权融合(entropy weight method,EWM)的两级特征融合方法,并使用北方苍鹰优化算法(northern goshawk optimization,NGO)结合门控循环单元(gated recurrent unit,GRU)进行桥梁损伤程度预测。首先,基于传统的模态应变能理论,构造出对角模态应变能比,由此衍生出对角模态应变能比变化率,对角模态应变能比耗散率,标准化对角模态应变能比差指标。其次,使用主成分分析实现指标内特征提取,熵权法融合指标间的特征,从而构造出加权决策指标(weighted decision index,WDI)。将单个模态应变能衍生指标输入到NGO-GRU混合神经网络中,损伤程度为输出,从而建立指标值与损伤程度之间的关系,进而实现损伤量化。通过三跨连续梁桥数值模型对所提出的方法进行验证,结果表明:加权决策指标具有良好的损伤定位能力和抗噪性,混合神经网络具有较高的损伤预测精度,预测准确率为91.14%。

基于PCA-ShapeDTW-QWGRU的分布式光伏集群短期功率预测

针对分布式光伏短期功率预测建立基于主成分分析、改进的动态时间规整算法与量子加权门控循环单元(PCAShapeDTW-QWGRU)的集群功率预测模型。针对集群划分不够精细、光伏电站数据蕴含的信息难以捕捉的问题,提出基于主成分分析结合密度聚类算法(PCA-OPTICS)的集群划分方法;针对目前选取代表电站与集群相似性较低的问题,提出基于改进的动态时间规整算法(ShapeDTW)的代表电站的选取方法,利用ShapeDTW度量相似性距离,选取最小值作为代表电站,并利用基于均方根传播梯度下降法优化的量子加权门控循环单元(RMSprop-QWGRU)模型进行预测;为了解决代表电站与集群功率的变换系数转换差异较大的问题,采用实时变换系数对代表电站进行集群功率值预测计算。实验结果表明,所提方法能有效提升光伏集群功率预测的精度。

基于粗糙集理论与PCA-APSO-SVM的沥青路面使用性能预测

针对传统沥青路面使用性能预测精度较低的问题,建立了基于粗糙集理论(rough set,RS)与主成分分析法(principal compoent analysis,PCA)-自适应粒子群算法(adaptive particle swarm optimization,APSO)-支持向量机(support vector machine,SVM)的沥青路面使用性能预测模型。基于沥青路面的时序指标与影响因素指标,建立了11个初始预测指标(包括前3年的路面使用性能、当量轴次、路龄、养护性质、坑槽率、修补率、年降水量、平均气温、日照时数);通过RS属性约减筛选出9个核心指标;利用PCA提取4个主成分,得到了基于4个主成分的数据集;将APSO引入到SVM中,对数据集进行训练,并优化了SVM模型参数;建立了路面使用性能的PCA-APSO-SVM预测模型,并以G6京藏高速甘肃境内某段道路为例,对路面使用性能进行预测。研究结果表明:PCA-APSO-SVM模型预测精度较PCA-PSO-SVM、APSO-SVM、PSO-SVM有较大提高,预测结果与实际情况更加符合,能为路面养护决策提供相关参考。

基于PCA/PSO的3T1R并联机构性能优化

根据少自由度并联机构应用广泛的优点,提出了一种3T1R并联机构,该机构具有构型简单、结构对称、定位精度高等特点,可应用于小范围的精密操作,或者是大范围的搬运、分拣以及喷涂等领域。基于方位特征方程的拓扑分析理论,对该并联机构完成了自由度种类以及数目的分析与验证;基于闭环矢量法完成了运动学模型建立,并通过位置正逆解算例验证了运动学的合理性。基于位置逆解方程利用极限边界搜索法分析了3T1R并联机构可达工作空间;通过速度分析建立了速度雅可比矩阵,并根据该矩阵分析机构的定位精度与可操作度性能指标。利用主成分分析(PCA)与粒子群算法(PSO)对3个性能指标进行优化设计,并对优化结果进行了分析,最终优化后可达工作空间体积从0.2933m3提高到0.4231m3,定位精度误差放大因子从15.5044减小至4.4308,可操作度指数从9.7027减小至1.3996。

改进PCA方法的牙颌特征提取和数据集构建

为了提高牙颌数字模型特征提取的准确性和牙齿分割的效率,提出一种几何变换和主成分分析相结合的坐标系标准化方法,并以离散曲率、法向量、形状直径函数和离散测地距离为基础特征对牙颌数字模型进行了特征提取,在此基础上进一步扩展构建了76个特征数据集。采用提出的坐标系标准化方法和数据集对上颌进行了牙齿分割实验。结果表明:改进主成分分析方法能够快速准确地实现牙颌数字模型坐标系对齐,能够准确识别牙齿的特征信息并作出标记,牙齿分割完整,平均分割准确率达到99.74%。基于改进主成分分析方法的牙颌模型特征提取方法能够极大地提高特征对牙齿的区分性,从而降低位姿对牙颌特征提取带来的负面影响,实现在特征数据较少的情况下准确分割牙齿,可为数字化口腔诊疗提供一定的参考。