Suggestion of advanced regression model on friction angle of fault gouge in South Korea

Although friction characteristics of fault gouge are important to understand reactivation of fault,behavior of earthquake,and mechanism of slope failure,analysis results of fault gouge have low accuracy mostly than those of soils or rocks due to its high heterogeneity and low strength.Therefore,to improve the accuracy of analysis results,we conducted simple regression analysis and structural equation model analysis and selected major influential factors of friction characteristics among many factors,and then we deduced advanced regression model with the highest coefficient of determination(R^(2)) via multiple regression analysis.Whereas most coefficients of determination in simple regression analysis are below0.3-0.4,coefficient of determination in multiple regression analysis is remarkably large as 0.657.

基于LARS-泛Kriging模型的气动建模及翼型优化设计

以超临界翼型为例,基于泛Kriging模型与最小角回归算法建立了气动特性关于几何外形与攻角的代理模型。分析了代理模型的精度,并通过特征选择过程研究了不同参数对气动特性的影响。采用基于代理模型的全局优化方法进行了翼型阻力优化设计,并与原始翼型、直接优化结果、普通Kriging模型优化结果进行了对比。结果表明:LARS-泛Kriging模型可识别气动特性的重要影响参数,建模精度满足工程需求;采用LARS-泛Kriging模型优化相比直接CFD计算优化及初始普通Kriging模型优化所需CFD计算次数分别减少91.0%、65.6%,计算效率分别提高98.5%、94.0%,优化结果相似。

基于LASSO-LARS的软件复杂性度量属性特征选择研究

针对软件可靠性早期预测中软件复杂性度量属性维数灾难问题,提出了一种基于最小绝对值压缩与选择方法(The Least Absolute Shrinkage and Select Operator,LASSO)和最小角回归(Least Angle Regression,LARS)算法的软件复杂性度量属性特征选择方法。该方法筛选掉一些对早期预测结果影响较小的软件复杂性度量属性,得到与早期预测关系最为密切的关键属性子集。首先分析了LASSO回归方法的特点及其在特征选择中的应用,然后对LARS算法进行了修正,使其可以解决LASSO方法所涉及的问题,得到相关的复杂性度量属性子集。最后结合学习向量量化(Learning Vector Quantization,LVQ)神经网络进行软件可靠性早期预测,并基于十折交叉方法进行实验。通过与传统特征选择方法相比较,证明所提方法可以显著提高软件可靠性早期预测精度。

Influence of Tire Slip Angle on Adhesion CoefficientCharacteristic of Road Vehicles

Tire slip angle has a great influence on peak values of both logitudinal and lateral adhesion coefficients as well as optimal slip ratios. Regression equations based on the experi- mental data from several countries are given to describe their dependencies. The peak value of longitudiinal adhesion coefficient has a linear relationship with tire slip angle, but peak value of lateral has a complicated relationship; the optimal slip ratio of longitudinal has an exponent function relationship, the optimal slip ratio of lateral almost has nothinng to do with tire slipangle.

踝关节骨折术后踝穴冠状位角度波动与关节功能恢复的关系

背景:踝关节骨折术后的踝穴形态指标会出现相应变化,而其中踝穴冠状位角度变化与踝关节的功能恢复之间存在一定的关联,既往研究大多探讨患者术后踝穴高度恢复,故研究踝关节骨折术后踝穴冠状位角度的波动与关节功能恢复的相关性具有一定参考意义。目的:探讨踝关节骨折术后踝穴冠状位角度变化对关节功能恢复的影响。方法:纳入进行手术治疗的86例踝关节骨折患者为研究对象,根据末次随访美国骨科足踝外科协会踝-后足评分结果分为优良组(n=45)和可差组(n=41),比较两组患者的一般资料。根据踝关节的踝穴位及侧位X射线片对比术后患者患侧和健侧的踝穴形态指标:包括踝穴宽度和深度、冠状位角度及矢状位角度,以及上述指标患侧和健侧之间的差值,并进一步进行组内比较分析。构建联合模型,利用Cox回归分析评估冠状位角度波动与关节功能恢复情况的关系。利用最小绝对收缩选择算子回归法和多因素Logistic回归分析影响患者关节功能恢复的危险因素。按踝穴冠状位角度从低到高等分为5分位数组(Q1-Q5),比较5组患者的临床资料特征,并采用多因素Logistic回归分析踝穴冠状位角度变化与关节功能恢复不良风险的相关性,建立限制性立方样条Logistic回归模型分析其剂量-反应关系。建立回归方程y=1-1/(1+e^(-z))预测模型并进行验证。结果与结论:①患者患侧的踝穴宽度和深度、冠状位角度及矢状位角度均显著大于健侧(P均<0.05);且与优良组患者相比,可差组患者踝穴宽度差值、深度差值、冠状位角度差值及矢状位角度差值均更大(P<0.05);②联合模型显示无论冠状位角度是否在正常范围内,冠状位角度每纵向增加1˚,关节功能恢复不良的风险增加3%;③最小绝对收缩选择算子回归和多因素Logistics回归分析显示,调整潜在混杂因素后发现年龄过大、早期功能锻炼缺乏、无跟骨牵引、内固定物未取出、术后并发症、踝冠状位角度增大均为影响患者关节功能恢复的独立危险因素(P<0.05);④术后3个月内的冠状位角度与关节功能恢复效果不良风险存在独立相关性(OR=1.57,95%CI:1.38-1.76,P=0.002),且术后各时间段冠状位角度从低到高5分位数组趋势性检验差异有统计学意义(P趋势<0.001);⑤限制性立方样条模型分析显示,男性或女性的踝穴冠状位角度变化与关节功能恢复效果不良风险均不存在非线性关系;经Bootstrap自抽样,预测模型区分度、准确度较好;⑥提示踝关节骨折术后踝穴冠状位角度减小对关节功能恢复起到明显的促进作用,因此在踝关节骨折术后重点检测踝穴冠状位角度有助于了解患者关节功能恢复情况。

基于LARS特征选择的风电机组故障诊断的研究

在风力发电机组运行维护过程中,对设备故障诊断的要求越来越高,随着近年来大数据的广泛应用,其对风力发电的影响也越来越大,许多研究人员基于大数据展开了相关工作。在利用风电机组SACDA数据进行故障诊断的过程中,所用的特征量不同,解决问题的效果会受到影响。为了提高风电机组故障诊断的精准性,须要对其所用到的故障特征进行选择。文章提出了用最小角回归(Least Angle Regression, LARS)方法来对特征向量进行选择,针对这些被选出的特征向量用HMM(Hidden Markov Model)建立故障模型。利用某风场的运行数据进行验证,结果表明,文章提出的基于HMM-LARS方法建立的模型对故障类型具有较好的识别效果。

基于LARS-SVR的电影总票房预测模型研究

考虑影响票房的各种因素,使用最小角回归(LARS)算法进行因素选取,再利用支持向量回归(SVR)算法对所选因素建立预测模型。结果显示:电影热度、电影制式、上映时间、影片类型、演员是影响一部影片总票房的主要因素,且本文提出的LARS-SVR模型既通过变量选择避免了SVR出现的过拟合现象,还保持了与LARS、逐步回归相当的拟合效果,预测误差也远小于后面两个模型。此研究结果可为电影制片方、宣传营销方及院线经营者提供一些决策参考。

The Study of Relationships between the Damage of Visual Function and Hemorrheology, Ocular Rheography, as well as Other Related Factors in Patients with Primary Open Angle Glaucoma (POAG)

Researches of glaucoma visual function damage, hemorrheololgy, ocular rheography and other related multiplex factors, with computed multifactorial stepwise regresion analysis, indicate that the elevation of intraocular pressure (IOP) is not the only factor to induce visual impairment. POAG patients are shown to have markedly reduced diastolic purfussion pressure in ophthalmic artery, besides prolonged filling time of the retinal artery and vein, diminished erythrocyte deformability and increased platele...

Regression Equations for Predicting the Effect of Tine Width on Draught and Soil Translocation in Moderately Fine Textured Soil

Experiments were conducted in an indoor soil bin filled with sandy clay loam soil. Tests were carried out with tillage tines to study the effect tool width on soil disturbance and draught. Depth of operation was held constant at 35 mm and then at 70 mm while speed was varied at three levels of 1.0, 3.6 and 9.0 km/h. The widths of the tines tested were 10, 20, 31, 40, 51, 88, 126, 163 and 200 mm. The cone penetration resistance of the soil varied from 400 to 600 kPa. Draught was measured with a load cell while soil disturbance was measured with a profile meter and meter rule. Draught increased at a decreasing rate with tine width. Quadratic models best fitted the data points with high R2 values. The increase in draught was affected by the forward speed since higher draught values were obtained at higher speed. Results show that the parameters of soil disturbance increased with increase in tine width, except height of ridge （hr）, which did not show any specific trend. The specific draught was highest （10.63 N/cm） with tine T20 while Tine T1 had the least specific draught of 5.2 N/cm.

Variable Selection via Biased Estimators in the Linear Regression Model

Least Absolute Shrinkage and Selection Operator (LASSO) is used for variable selection as well as for handling the multicollinearity problem simultaneously in the linear regression model. LASSO produces estimates having high variance if the number of predictors is higher than the number of observations and if high multicollinearity exists among the predictor variables. To handle this problem, Elastic Net (ENet) estimator was introduced by combining LASSO and Ridge estimator (RE). The solutions of LASSO and ENet have been obtained using Least Angle Regression (LARS) and LARS-EN algorithms, respectively. In this article, we proposed an alternative algorithm to overcome the issues in LASSO that can be combined LASSO with other exiting biased estimators namely Almost Unbiased Ridge Estimator (AURE), Liu Estimator (LE), Almost Unbiased Liu Estimator (AULE), Principal Component Regression Estimator (PCRE), r-k class estimator and r-d class estimator. Further, we examine the performance of the proposed algorithm using a Monte-Carlo simulation study and real-world examples. The results showed that the LARS-rk and LARS-rd algorithms,?which are combined LASSO with r-k class estimator and r-d class estimator,?outperformed other algorithms under the moderated and severe multicollinearity.

The Way We Stand: A Sequential Case Study on Foot Angle

This study quantifies the significance of foot angle as it relates to common biometric data to assess its use in the clinical realm. There are two objectives to this study;first to determine the relationship foot angle has with ten surveyed biometric data points (age, weight, height, biological makeup, shoe size, orthotic use, physical activity level, competitive sports history, lower extremity injury history, and reason for visiting doctor), and second to determine the average left and right foot angles of the population studied. The duration of this sequential case study with a sample size of 73 was December 9, 2021, to April 7, 2022. Participants were fully informed of the measurement steps, the purpose of the survey, and the goal of the project. Participants were asked to walk down the hallway as a distraction to collect the most accurate data on the participants’ stance. This study was performed at the Arizona Institute of Motion, a clinical office and private practice. Participants had the option to decline participation in the study. Correlational calculations and linear regressions were performed to support or reject the existence of a relationship between foot angle and the biometric data. It was found that the average left foot angle was 26.35 degrees, and the average right foot angle was 26.54 degrees. It was found that there was a strong positive correlation between left and right foot angles, and a weak positive relationship with both foot angles and weight, height, and shoe size. The optimized linear regression model had an adjusted r2 value of 0.549 for left foot angle against right foot angle, shoe size, and height, and an adjusted r2 value of 0.522 for right foot angle against left foot angle, shoe size, and height. Foot angle had no other relationships with the other biometric data and had limited quantitative significance. Additional study is required to confirm its numerical importance, but foot angle may continue to be used in clinical settings in conjunction with gait analysis and qualitative assessments for musculoskeletal function. Foot angle has limited quantitative significance but may maintain a high potential for clinical utility when used in conjunction with gait analysis and qualitative assessments.

外置耗能角钢的摇摆-自复位桥梁双柱墩地震易损性分析

为评估设置耗能角钢的摇摆-自复位桥梁双柱墩的抗震能力,本文建立该结构的有限元分析模型,分别定义和量化双柱墩不同损伤状态和损伤指标。本文从太平洋地震工程研究中心选取16条近断层地震动,基于增量动力分析,得到该结构的地震响应。基于最小二乘法,对得到的增量动力分析曲线进行对数回归分析,计算结构不同损伤状态的超越概率及易损性曲线。研究表明:当地面峰值加速度为0.1 g时,外置耗能角钢与耗能钢筋屈服的超越概率分别为34%和22%;当峰值加速度为0.4 g时,外置耗能角钢与耗能钢筋失效的超越概率分别为6%和11%,而预应力筋未屈服;当峰值加速度为1.0 g时,外置耗能角钢和耗能钢筋失效的概率分别达到85%和91%,桥墩发生倒塌的概率为46%。

新拌混凝土离散元参数的测量与标定

随着计算机技术的发展,可以采用仿真的方法来研究新拌混凝土的流变性能.离散元方法适合于新拌混凝土的大变形流动.颗粒的物性参数和接触参数的设定是模拟结果真实可靠的关键.在本研究中,将混凝土分为机制石和砂浆两相.首先测量了物性参数,包含密度、恢复系数、静摩擦系数和滚动摩擦系数.使用Hertz-Mindlin(no slip)接触模型表示粗骨料-边界和粗骨料-粗骨料之间的相互作用,并通过休止角的实验进行了验证;采用Hertz-Mindlin with JKR接触模型来描述粗骨料-砂浆、砂浆-砂浆、砂浆-边界之间的相互作用,用坍落度实验对JKR参数进行了标定,并采用响应曲面法确定了最佳的参数组合值.最后通过L型箱试验对新拌混凝土的离散元仿真方法进行了验证.

基于CFD的喷射角变化对沉积特性影响的研究

水平喷杆喷雾喷幅较宽,作业效率高、用户量较多,但底盘运动等因素引起的喷杆振动成为雾滴流失的主要原因。为此,基于CFD并通过喷头物理规格建立喷雾流场,研究水平喷杆发生摆动时实时调整喷射源旋转角度对雾滴沉积水平的影响。模拟结果表明:喷杆下摆时逆时针旋转喷射源;喷杆上扬时顺时针旋转喷射源,沉积效果相比不改变喷射源旋转角均趋于优化,两种调整最佳效果分别为雾滴直径DV.9值43.2μm、沉积密度108粒/cm 2和雾滴粒径DV.9值47.3μm、沉积密度118粒/cm 2。将此变化过程视为线性变化可知,实时调节喷射源旋转角度对提高雾滴沉积效果具有积极作用。通过计算各状态雾滴的沉积比例并进行回归分析可知,喷杆产生较大摆角时雾滴沉积效果直接受喷射源旋转角度影响。研究结果可为精量喷雾控制相关研究提供理论依据和数据参考。

基于改进最小角回归算法的Hammerstein模型辨识

针对一类未知时滞和阶次的Hammerstein模型的辨识问题,本文提出一种基于绝对角度停止准则最小角回归(AS-LAR)的稀疏辨识方法,该方法可以同时辨识出Hammerstein模型的时滞、阶次和参数.首先,通过引入最大非线性阶次和输入回归长度,将系统表示成具有稀疏参数向量的高维辨识模型;然后,提出一种绝对角度停止准则,对最小角回归算法进行改进,并基于改进的AS-LAR算法获得稀疏参数向量的估计;最后,基于参数向量稀疏结构,估计出系统的时滞和阶次,并从估计的参数向量中提取和分离出系统线性部分和非线性部分的参数估计值.数值仿真和水箱实例结果表明,提出的辨识方法有效,且与其它辨识方法相比,具有估计精度高、计算量小、速度快等特点.

石灰改良粉质黏土力学强度试验

不良土质在长跨度公路工程中普遍存在,为探究石灰对粉质黏土路基力学特性的改善效果,通过42组室内试验结合灰色关联度分析,研究了石灰掺入比、压实度和龄期对标准养护组及浸水组改良土抗压强度的影响,并提出了改良土抗压强度与3个影响因素之间的回归方程。同时,通过调整石灰掺入比及养护条件,分析了改良土抗剪强度的变化规律。结果表明:压实度和石灰掺入比对标准养护组及浸水组改良土的无侧限抗压强度影响最大;随着石灰掺入比的增加,改良土的黏聚力显著提高,而摩擦角的增幅不明显,最佳石灰掺入比为7%。

线弹性迭代的等肢角钢输电塔极限承载力研究

为了规避现有的构件截面广义屈服函数表达形式繁琐的缺点,同时弥补函数不等比例加载的不足,以逐步回归分析为基础,建立了等肢角钢简化的齐次广义屈服函数,提出了一种高效线弹性迭代方法,用以分析输电塔在构件强度方面的极限承载力。首先,利用最小二乘法建立等肢角钢齐次广义屈服函数;然后通过逐步回归分析,逐个剔除函数中影响不显著的各项,简化上述齐次广义屈服函数,同时保证其较高的拟合精度;最后,基于弹性模量调整策略,提出了一种高效线弹性迭代方法,用以等肢角钢输电塔的极限承载力分析。结果表明:所建立的简化形式的齐次广义屈服函数,不仅精度高,而且项数少,克服了现有传统的广义屈服函数表达形式繁琐、不等比例加载,以及由此带来的计算结果不稳定等问题。通过对比传统方法,验证了所述方法在分析输电塔构件强度方面的极限承载力具有高精度和高计算效率的特点。

基于GPR-UKF的天文测角/测速组合导航方法

在以太阳作为目标源的天文测速导航中,多普勒速度量测量存在较多野值误差,严重影响导航精度。对此,提出一种基于高斯过程回归与无迹卡尔曼滤波(Gaussian process regression and unscented Kalman filtering,GPR-UKF)的野值检测修复方法,建立速度量测量的动态预测模型。此外,还针对不同参数对模型精度的影响进行研究。经仿真验证,所提方法效果显著优于传统野值处理方法。

基于回归分析的刮板取料机俯仰拉力曲线拟合研究

文中针对刮板式取料机中俯仰机构钢丝绳拉力安全值确定简单、无法全面、准确评估刮板取料设备运行安全系数的问题,对常用俯仰角度以及其对应的拉力值进行回归分析,计算俯仰角度及各角度对应拉力的曲线回归方程,并检测该曲线的拟合程度和显著性。以该方程为约束建立钢丝绳俯仰拉力保护的数学模型,通过将空载状态下或重载状态下取料机的钢丝绳理论拉力保护曲线设置为安全检测程序,结合控制软件的运行逻辑对设备的启动、下俯、行车进行限制,确保该设备运行的安全性。

一种基于L_1范数正则化的回声状态网络

针对回声状态网络存在的病态解以及模型规模控制问题,本文提出一种基于L1范数正则化的改进回声状态网络.该方法通过在目标函数中添加L1范数惩罚项,提高模型求解的数值稳定性,同时借助于L1范数正则化的特征选择能力,控制网络的复杂程度,防止出现过拟合.对于L1范数正则化的求解,采用最小角回归算法计算正则化路径,通过贝叶斯信息准则进行模型选择,避免估计正则化参数.将模型应用于人造数据和实际数据的时间序列预测中,仿真结果证明了本文方法的有效性和实用性.