Prediction of uniaxial compressive strength and modulus of elasticity for Travertine samples using regression and artificial neural networks

Uniaxial Compressive Strength (UCS) and modulus of elasticity (E) are the most important rock parameters required and determined for rock mechanical studies in most civil and mining projects, in this study, two mathematical methods, regression analysis and Artificial Neural Networks (ANNs), were used to predict the uniaxial compressive strength and modulus of elasticity. The P-wave velocity, the point load index, the Schmidt hammer rebound number and porosity were used as inputs for both methods. The regression equations show that the relationship between P-wave velocity, point load index, Schmidt hammer rebound number and the porosity input sets with uniaxial compressive strength and modulus of elasticity under conditions of linear relations obtained coefficients of determination of (R2) of 0.64 and 0.56, respectively. ANNs were used to improve the regression results. The generalized regression and feed forward neural networks with two outputs (UCS and E) improved the coefficients of determination to more acceptable levels of 0.86 and 0.92 for UCS and to 0.77 and 0.82 for E. The results show that the proposed ANN methods could be applied as a new acceptable method for the prediction of uniaxial compressive strength and modulus of elasticity of intact rocks.     

Comparative evaluation of different statistical tools for the prediction of uniaxial compressive strength of rocks

In this study, uniaxial compressive strength (UCS), unit weight (UW), Brazilian tensile strength (BTS), Schmidt hardness (SHH), Shore hardness (SSH), point load index (Is₅₀) and P-wave velocity (V_p) properties were determined. To predict the UCS, simple regression (SRA), multiple regression (MRA), artificial neural network (ANN), adaptive neuro-fuzzy inference system (ANFIS) and genetic expression programming (GEP) have been utilized. The obtained UCS values were compared with the actual UCS values with the help of various graphs. Datasets were modeled using different methods and compared with each other. In the study where the performance indice PI_at was used to determine the best performing method, MRA method is the most successful method with a small difference. It is concluded that the mean PI_at equal to 2.46 for testing dataset suggests the superiority of the MRA, while these values are 2.44, 2.33, and 2.22 for GEP, ANFIS, and ANN techniques, respectively. The results pointed out that the MRA can be used for predicting UCS of rocks with higher capacity in comparison with others. According to the performance index assessment, the weakest model among the nine model is P7, while the most successful models are P2, P9, and P8, respectively.     

Shape ratio effects on the mechanical characteristics of rectangular prism rocks and isolated pillars under uniaxial compression

Isolated pillars in underground mines are subjected to uniaxial stress, and the load bearing cross-section of pillars is commonly rectangularly shaped. In addition, the uniaxial compression test (UCT) is widely used for determining the basic mechanical properties of rocks and revealing the mechanism of isolated pillar disasters under unidimensional stress. The shape effects of rock mechanical properties under uniaxial compression are mainly quantitatively reflected in the specific shape ratios of rocks. Therefore, it is necessary to study the detailed shape ratio effects on the mechanical properties of rectangular prism rock specimens and isolated pillars under uniaxial compressive stress. In this study, granite, marble and sandstone rectangular prism specimens with various height to width ratios (r) and width to thickness ratios (u) were prepared and tested. The study results show that r and u have a great influence on the bearing ability of rocks, and thin or high rocks have lower uniaxial compressive strength. Reducing the level of r can enhance the u effect on the strength of rocks, and increasing the level of u can enhance the r effect on the strength of rocks. The lateral strain on the thickness side of the rock specimen is larger than that on the width side, which implies that crack growth occurs easily on the thickness side. Considering r and u, a novel strength prediction model of isolated pillars was proposed based on the testing results, and the prediction model was used for the safety assessment of 179 isolated pillars in the Xianglu Mountain Tungsten Mine.     

Failure criterion for soft rocks considering intermediate principal stress

The significant differences between hard rocks(more brittle) and soft rocks(more ductile) may suggest the use of different failure criteria. A strength criterion for soft rocks that includes intermediate principal stress was proposed. The new criterion includes two independent parameters: the uniaxial compressive strength(rci), which can be obtained from common laboratory tests or indirectly estimated by alternative index tests in the laboratory or field; and f(joint), which is used to characterize the rock mass quality and can be easily estimated. The authors compared the predictive capabilities of the new criterion with other criteria using the database of soft rocks under two conditions: with and without triaxial data. For the estimation of triaxial and true-triaxial strengths, the new criterion generally produced a better fit. The proposed criterion is practical for an approximate first estimation of rock mass strength, even without triaxial data, as it balances accuracy(lower prediction error) and simplicity(fewer independent parameters).     

Objective Quantification of Small Motor Sound Quality Based on Psychoacoustic Metrics

With the widespread application of electrification and intelligence of automobiles, the number of electric devices with small DC motors in automobiles has gradually increased, and the interior of electric vehicles is quieter. The sound quality (SQ) of small motor directly affects the passenger experience. Therefore, the research on the SQ of small motor is of great significance. In this paper, the objective quantification of small motor sound quality was investigated based on traditional psychoacoustic metrics. The time-frequency characteristics of sound signal was analyzed to quantify the subjective perception caused by the sound of small motor. And a new psychoacoustic metrics of objective evaluation which were suitable for small motor SQ evaluation were proposed, namely specific loudness energy (SLE), specific prominence ratio index (SPRI), relative pitch exceedance (RPE) and tremolo index (TI). Then, two objective evaluation models of small motor SQ were established to characterize the multi-dimensional subjective perception attributes by using multiple linear regression (MLR) and support vector regression (SVR) respectively, which can be used for the prediction and evaluation of the small motor SQ. The results show that the prediction accuracy of the model established by SVR method was higher than that of MLR, and SVR has stronger robustness. The objective evaluation model of small motors SQ established in this study is of great importance for improving the sound quality of small motors.     

回归模型构建中变量选择的研究

回归模型建立过程中如何选择合适的解释变量以及如何确定每个解释变量的函数形式至关重要,它直接影响模型的拟合能力和预测能力.论文在传统的变量选择的评价准则和变量选择方法的基础上,探讨了如何在模型中包含的变量不确定的情况下,准确有效的进行变量的选择,并在变量选择的同时进行变量的变换.旨在提高模型预测的精确度.     

羟基多溴二苯醚生物活性的QSAR研究

采用VSMP方法从大量的分子结构描述符中筛选最优子集,用多元线性回归方法分别构建了羟基多溴二苯醚（hydroxylated polybrominated diphenyl ethers,OH-PBDEs）对细胞色素CYP19介导类固醇生成的抑制活性,甲状腺受体-β（thyroid receptors,TR-β）的激素活性以及与甲状腺素结合球蛋白（thyrox-ine-binding globulin,TBG）和甲状腺转运蛋白（transthyretin,TTR）之间结合能力的定量构效关系（quantita-tive structure-activity relationship,QSAR）模型。模型的留一法交叉验证（leave-one-out cross validation,LOO-CV）相关系数Q2和拟合相关系数R2均高于0.95和0.97,表明模型具有良好的稳健性、拟合能力和预测能力。QSAR模型表明OH-PBDEs的分子三维结构和芳香性是其活性的重要影响因素。另外,分子中溴原子数量和取代位置对OH-PBDE生物活性具有重要影响。     

基于人工神经网络含稀土元素熔敷金属力学性能预测

为了提高焊条的力学性能并缩短焊条研发周期,在E4301型焊条药皮配方基础上加入了CeO_2和稀土元素La,并对焊条进行了力学性能试验.对试验数据进行分析后发现,加入适量的稀土元素可以改善焊条的力学性能.利用典型BP和RBF神经网络分别建立力学性能预测模型.将焊条中的CeO_2、La、Si、Mn含量与焊接速度作为预测模型的输入变量,将熔敷金属的抗拉强度、下屈服强度、断后伸长率与热影响区平均硬度作为输出变量.结果表明,将BP和RBF神经网络用于对含稀土焊条力学性能的预测是可行的,且RBF神经网络模型的预测精度和效率要高于BP神经网络模型.     

高速公路事故预测模型

为了掌握高速公路未来的安全状况,通过有效地控制各种影响因素,减少交通事故,增进高速公路安全,在路段划分和影响因素分析的基础上,利用收集的多条高速公路数据建立了基于广义线性回归的高速公路事故预测模型,通过比较泊松、负二项、零堆积泊松和零堆积负二项4种概率分布模型回归的结果,最终确定了负二项分布形式的事故预测模型,并利用弹性分析的方法确定了模型中单个变量对事故的边际影响.研究表明:环境变量和交通流变量对事故的发生有较大影响.     

探索提高三维定量构效关系模型预测能力的方法

为了解决使用三维定量构效关系(three-dimensional quantitative structure-activity relationship,3D-QSAR)模型预测新化合物生物活性效果不理想的问题,建立了2种新的一致性模型.模型一是由多元线性回归(multiple linear regression,MLR)方法构建的加权一致性模型(weighted consensus modeling,WCM),该模型为每个子模型添加了各自的权重系数.模型二通过计算多个子模型预测值的平均值来构建平均一致性模型(average consensus modeling,ACM).研究结果表明,当交叉验证相关系数0.5q~2≤0.8时,一致性模型可以提高预测能力,而在q~20.8时不能提高3D-QSAR模型的预测能力.该方法可为提高模型预测能力和设计新型高活性抑制剂提供帮助.     

Effects of Polypropylene Fibers on the Physical and Mechanical Properties of Recycled Aggregate Concrete

The viability of using polypropylene fibers(PPF) in concrete was largely studied. Yet, few of the existing research studies investigated the effects of PPF on the properties of concrete containing recycled concrete aggregate(RCA). Mixes with different RCA replacement ratios and different PPF content were designed and tested. The test results showed that the addition of PPF did not change significantly the compressive strength and the density of the concrete, but slightly decreased its modulus of elasticity and Poisson's ratio. The drop in the splitting tensile strength and the flexural strength due to RCA inclusions was to a large extent compensated by the PPF addition. The water absorption decreased and the percent voids increased with increased PPF addition. Correlations between the RCA content, the PPF content and the properties of concrete were studied. Useful regression models were proposed to predict the properties of concrete in relevant ranges of RCA and PPF content.     

基于机器学习的高强钢焊接等截面箱型柱整体稳定性预测方法

目前,针对高强钢构件整体稳定性的研究多采用有限元建模或实验室试验方法,而基于机器学习的预测方法能够显著提升预测的准确性和便捷性。为了准确预测高强钢焊接等截面箱型柱的整体稳定性,提出使用纤维模型构建数据库并利用机器学习建立预测模型的方法。首先确定模型的输入输出参数,并通过纤维模型方法建立数据库;接着,选用常见的3种不同类型的机器学习模型和现有规范中的经验模型进行预测,并依据评价指标进行性能对比;最后,根据可解释算法分析机器学习模型的合理性。结果表明：大部分机器学习模型预测结果与试验结果吻合度略高于现有规范中的经验模型,其中,高斯过程回归模型对高强钢构件整体稳定性的预测表现最优;机器学习预测模型中各类参数对构件整体稳定性的影响趋势符合预期,验证了机器学习模型的合理性和可靠性;构件的正则化长细比对预测结果影响最大,而构件初始缺陷的影响相对最小。     

基于变量组合集群分析法的小麦蛋白质近红外光谱变量选择方法研究

为了解决小麦蛋白质的近红外光谱信息复杂、共线性严重及全光谱建模的预测能力不足等问题,采用一种新的变量选择方法——变量组合集群分析法(VCPA)对小麦蛋白质的近红外光谱进行特征波长选取.首先利用二进制矩阵采样策略(BMS)和指数衰减函数(EDF)删除无信息变量,优选小麦中蛋白质近红外特征波长,然后结合偏最小二乘法(PLS)建立预测模型.与其他变量选择方法相比,VCPA所选用的波长点最少,模型的预测能力最强,VCPA算法所采用的BMS变量采样策略弥补了蒙特卡洛采样方法的不足.研究结果表明,VCPA算法可以有效选择小麦蛋白质近红外光谱特征波长,提高预测模型的可靠性和适用性.     

Relationship between fatigue life of asphalt concrete and polypropylene/polyester fibers using artificial neural network and genetic algorithm

While various kinds of fibers are used to improve the hot mix asphalt(HMA) performance, a few works have been undertaken on the hybrid fiber-reinforced HMA. Therefore, the fatigue life of modified HMA samples using polypropylene and polyester fibers was evaluated and two models namely regression and artificial neural network(ANN) were used to predict the fatigue life based on the fibers parameters. As ANN contains many parameters such as the number of hidden layers which directly influence the prediction accuracy, genetic algorithm(GA) was used to solve optimization problem for ANN. Moreover, the trial and error method was used to optimize the GA parameters such as the population size. The comparison of the results obtained from regression and optimized ANN with GA shows that the two-hidden-layer ANN with two and five neurons in the first and second hidden layers, respectively, can predict the fatigue life of fiber-reinforced HMA with high accuracy(correlation coefficient of 0.96).     

Investigation of the influence of intermediate principal stress on the dynamic responses of rocks subjected to true triaxial stress state

Precisely understanding the dynamic mechanical properties and failure modes of rocks subjected to true triaxial stress state(σ_1 σ_2 σ_3, where σ_1, σ_2, and σ_3 are the major principal stress, intermediate principal stress, and minor principal stress, respectively) is essential to the safety of underground engineering. However, in the laboratory, it is difficult to maintain the constant true triaxial stress state of rocks during the dynamic testing process. Herein, a numerical servo triaxial Hopkinson bar(NSTHB) was developed to study the dynamic responses of rocks confronted with a true triaxial stress state, in which lateral stresses can maintain constant. The results indicate that the dynamic strength and elastic modulus of rocks increase with the rise of intermediate principal stress σ_2, while the dynamic elastic modulus is independent of the dynamic strain rate. Simulated acoustic emission distributions indicate that the intermediate principal stress σ_2 dramatically affects dynamic failure modes of triaxial confined rocks. As σ_2 increases, the failure pattern switches from a single diagonal shear zone into two parallel shear zones with a small slant. Moreover, a recent triaxial Hopkinson bar experimental system using three bar pairs is also numerically established, and the measuring discrepancies are identified between the two numerical bar systems. The proposed NSTHB system provides a controllable tool for studying the dynamic triaxial behavior of rocks.     

单轴压缩下横观各向同性岩石破裂过程 声发射特性的离散元模拟

采用离散单元法研究了横观各向同性岩石在不同层面倾角条件下的单轴压缩破坏过程及声发射特性。结果表明:由于层面倾角的变化,导致横观各向同性岩石破裂过程具有不同的声发射特性。声发射特性与应力存在一定的耦合关系,且声发射空间响应集中在2种岩石的交界面上。互层岩体由于层间力学属性不同,极有可能在交界面处引起应力集中,使得在交界面附近裂纹最先萌生,裂纹进一步向交界面两侧岩体中扩展,从而引起宏观裂纹的出现,最终引起岩体的破坏。随着层面倾角的增大,岩石的单轴抗压强度和弹性模量呈先减小后增大,层面倾角90°时的强度甚至超过了0°时的强度。     

A new group contribution-based method for estimation of flash point temperature of alkanes

Flash point is a primary property used to determine the fire and explosion hazards of a liquid. New group contribution-based models were presented for estimation of the flash point of alkanes by the use of multiple linear regression(MLR)and artificial neural network(ANN). This simple linear model shows a low average relative deviation(AARD) of 2.8% for a data set including 50(40 for training set and 10 for validation set) flash points. Furthermore, the predictive ability of the model was evaluated using LOO cross validation. The results demonstrate ANN model is clearly superior both in fitness and in prediction performance.ANN model has only the average absolute deviation of 2.9 K and the average relative deviation of 0.72%.     

一种考虑主应力空间的岩石非线性真三轴强度准则

为评估岩石的真三轴强度特性,首先根据常规三轴试验岩石强度的变化规律,提出基于偏应力极值的非线性强度准则,并与12种岩石试验强度和4种典型岩石强度准则进行对比,发现该强度准则的预测值与试验强度非常接近,相关性系数R~2均在0.98以上,平均相对误差(MAPE,E_(MAP))均小于4%(除7号岩石为6.83%);该强度准则、指数准则、H-B准则、MM-C准则和D-P准则对所有12种岩石预测E_(MAP)的平均值分别为2.32%、2.43%、5.28%、7.39%和13.74%,说明该强度准则能够很好地预测不同类型岩石的强度,其预测精度略优于指数准则(岩石力学界认为预测精度较高的强度准则),远好于其他3种强度准则。在上述所建常规三轴强度准则的基础上,通过引入中主应力参数和罗德应力参数,构建考虑中主应力效应的真三轴强度准则,并与8种岩石的真三轴试验强度进行对比,该强度准则很好地反映了大主应力随中主应力的增加呈先增大后减小的变化规律,所得R~2均在0.9以上,其中5种岩石的R~2大于0.96;除了13和14号岩石(E_(MAP)分别为7.79%和4.84%),其余6种岩石的平均相对误差E_(MAP)均小于4%,说明该强度准则能够较好地预测岩石的真三轴试验强度,很好地反映了中主应力效应,具有良好的普遍适用性。子午面和偏平面的应力空间特征也说明该强度准则很好地反映了岩石的静水压力效应和中主应力对大主应力的影响规律。     

基于GA-GRNN数据挖掘的SCR脱硝系统建模优化研究

火电厂选择性催化还原法(SCR)烟气脱硝系统是处理燃煤机组烟气排放NOx污染的主要途径,但该系统具有多输入变量、环境影响复杂、时变非线性等特征,因此建立准确的系统模型是SCR优化控制的基础。提出了一种融合遗传算法(GA)主元分析和广义回归神经网络(GRNN)数据挖掘的SCR系统建模方法。首先使用GA对运行数据进行变量选择优化计算;然后将最优变量作为GRNN的输入量,利用数据挖掘技术建立SCR系统数据模型。基于某电厂机组运行数据的实例分析表明,该方法建立的模型具有复杂度低、精度高、泛化能力强等优点。     

高性能混凝土的力学性能研究

为了弄清高性能混凝土的有关力学性能,本文对龄期为５６天、抗压强度６０至１００ＭＰａ（加硅粉或不加硅粉）的高性能混凝土进行了试验研究。分析并讨论了抗压强度随时间变化和受干燥影响的试验结果。通过２０个试件的试验测得了静力弹性模量和泊松比的试验值,并用回归方法导出了弹性模量的线性方程式。