Prediction of Sewer Condition Grade Using Support Vector Machines

Assessing the condition of sewer networks is an important asset management approach. However, because of high inspection costs and limited budget, only a small proportion of sewer systems may be inspected. Tools are therefore required to help target inspection efforts and to extract maximum value from the condition data collected. Owing to the difficulty in modeling the complexities of sewer condition deterioration, there has been interest in the application of artificial intelligence-based techniques such as artificial neural networks to develop models that can infer an unknown structural condition based on data from sewers that have been inspected. To this end, this study investigates the use of support vector machine (SVM) models to predict the condition of sewers. The results of model testing showed that the SVM achieves good predictive performance. With access to a representative set of training data, the SVM modeling approach can therefore be used to allocate a condition grade to sewer assets with reasonable confidence and thus identify high risk sewer assets for subsequent inspection.     

Objective Site Characterization Using Clustering of Piezocone Data

Cluster analysis is a statistical method for grouping similar mathematical data sets and is used herein for delineating geostratigraphy from piezocone penetration test data. In terms of site characterization, clustering is an improvement over other statistical methods because no preliminary estimation of the inherent groups within the analyzed data is needed, and no overlapping is permitted between identified clusters. Clustering can accommodate single or multivariables and no data filtering is required. Its application to defining stratigraphic interfaces is illustrated using five case studies with layered profiles. Clustering is able to detect major changes within the stratigraphy not apparent by visually examining the trends of piezocone data or by available cone soil classification methods.     

Support Vector Machines Approach to HMA Stiffness Prediction

The application of artificial intelligence (AI) techniques to engineering has increased tremendously over the last decade. Support vector machine (SVM) is one efficient AI technique based on statistical learning theory. This paper explores the SVM approach to model the mechanical behavior of hot-mix asphalt (HMA) owing to high degree of complexity and uncertainty inherent in HMA modeling. The dynamic modulus (|E?|), among HMA mechanical property parameters, not only is important for HMA pavement design but also in determining HMA pavement performance associated with pavement response. Previously employed approaches for development of the predictive |E?| models concentrated on multivariate regression analysis of database. In this paper, SVM-based |E?| prediction models were developed using the latest comprehensive |E?| database containing 7,400 data points from 346 HMA mixtures. The developed SVM models were compared with the existing multivariate regression-based |E?| model as well as the artificial neural networks (ANN) based |E?| models developed recently by the writers. The prediction performance of SVM model is better than multivariate regression-based model and comparable to the ANN. Fewer constraints in SVM compared to ANN can make it a promising alternative considering the availability of limited and nonrepresentative data frequently encountered in construction materials characterization.     

Prediction of Soil Composition from CPT Data Using General Regression Neural Network

Soil type is typically inferred from the information collected during a cone penetration test (CPT) using one of the many available soil classification methods. In this study, a general regression neural network (GRNN) was developed for predicting soil composition from CPT data. Measured values of cone resistance and sleeve friction obtained from CPT soundings, together with grain-size distribution results of soil samples retrieved from adjacent standard penetration test boreholes, were used to train and test the network. The trained GRNN model was tested by presenting it with new, previously unseen CPT data, and the model predictions were compared with the reference particle-size distribution and the results of two existing CPT soil classification methods. The profiles of soil composition estimated by the GRNN generally compare very well with the actual grain-size distribution profiles, and overall the neural network had an 86% success rate at classifying soils as coarse grained or fine grained.     

基于EEMD穿墙雷达人的运动模式识别

根据穿墙雷达动目标探测中人的运动多普勒信号属于非线性、非平稳信号的特点,分别采用经验模式分解(EMD)和整体平均经验模式分解(EEMD)将人.5种运动的多普勒信号分解为一系列本征模式函数(IMF).采用支持向量机(SVM)学习算法,将两种方法分解后的各IMF能量占总能量的百分比作为支持向量机分类器的特征向量进行模式识别,分析了特征向量维数对识别率的影响,比较了EMD和EEMD的识别率.EEMD能够消除EMD存在的模式混合问题,识别率更高,达到94%以上.     

Modeling Pile Capacity Using Support Vector Machines and Generalized Regression Neural Network

This note investigates the potential of support vector machines based regression approach to model the static pile capacity from dynamic stress-wave data. A data set of 105 prestressed precast high strength concrete spun pipe piles is used. Radial basis function and polynomial kernel based support vector machines were used to model the total pile capacity and results were compared with a generalized regression neural network approach. A total of 81 data set were used to train, whereas the remaining 24 data sets were used to test the created model. A correlation coefficient value of 0.977 was achieved by generalized regression neural network in comparison to values of 0.967 and 0.964 achieved by radial basis function and polynomial kernel based support vector machines, respectively. Results suggest an improved performance by generalized regression neural network based approach in comparison to support vector machines but polynomial kernel based support vector machines provide a linear relationship to predict total pile capacity using stress-wave data.     

Revised Multilinear Regression Equations for Prediction of Lateral Spread Displacement

In 1992 and 1995, Bartlett and Youd introduced empirical equations for the prediction of lateral spread displacement; these equations have gained wide use in engineering practice. The equations were developed from the multilinear regression (MLR) of a large case history database. This study corrects and updates the original analysis. Corrections and modifications include: (1) Bartlett and Youd erroneously overestimated measured displacements for lateral spreads generated by the 1983 Nihonkai-Chubu, Japan earthquake; those errors are corrected herein. (2) Several sites were deleted where boundary shear impeded free lateral displacement. (3) Data were added from three additional earthquakes. (4) The functional form of the mean-grain-size term was modified from (D5015) to log(D5015+0.1?mm) to produce improved prediction of displacements for coarse-grained granular sites. (5) The functional form of the model was changed from log(R) to log(R?), where R? is a function of the magnitude of the earthquake, to prevent unrealistic overprediction of displacements when R becomes small. The revised data were re-regressed to generate new MLR equations. The new equations are recommended for engineering practice.     

基于LS-SVM的粘结NdFeB永磁体磁性能预测

基于粘结NdFeB永磁体制备工艺优化实验,建立了一个最小二乘支持向量机(LS- SVM)算法模型用于工艺参数的优化.以粘结剂含量、固化温度、固化时间以及单位压制力大小四个工艺参数为影响因数,以剩余磁感应强度Br、矫顽力Hcj;和最大磁能积(BH)m为影响对象,通过最小二乘支持向量机算法模型建立起影响因素与影响对象之间的复杂的非线形关系.针对多影响对象,提出了一种γ和σ选择算法;以均匀设计试验结果为样本进行训练,用训练好的模型进行预测.结果表明,LS - SVM模型的实验结果与预测结果吻合良好,二者相对误差很小,对比ANN模型预测结果,LS - SVM模型具有更高的精度和运算速度,具有很好的实用性.     

Prediction of Remaining Service Life of Bridge Decks Using Machine Learning

This paper demonstrates that it is feasible to use either the k-nearest-neighbor instance-based learning (IBL) technique or the inductive learning (IL) technique for engineering applications in classification and prediction problems such as estimating the remaining service life of bridge decks. It is shown that IBL is more efficient than IL: The best achieved percentages of correctly classified instances are 50% as generated by k-nearest-neighbor IBL and 41.8% when generated by the C4.5/IL learning algorithm. From a machine learning (ML) standpoint both these values are considered low, but this is attributed to the fact that the deterioration model used to compute the remaining service life turned out to be inadequate. It is based on a methodology developed under the Strategic Highway Research Program (SHRP) for life-cost analysis of concrete bridges relative to reinforcement corrosion. Actual bridge deck surveys were obtained from the Kansas Department of Transportation that include the type of attributes needed for the SHRP methodology. The experimentation with the ML algorithms reported here also describes the experience one may go through when faced with an imperfect model, or with incomplete data or missing attributes.     

Mining of Existing Data for Cement-Solidified Wastes Using Neural Networks

This paper summarizes the results of an investigation into the use of neural networks to analyze data collected from the literature regarding the interaction of wastes and hydraulic binders in, and final properties of, cement-solidified wastes. Neural network models were constructed for prediction of the effects of contaminants on setting time, unconfined compressive strength, and leachate pH. It was found that construction of successful models was possible, with prediction errors approaching experimental error, and that modeling was useful for generalizing about the relative effects of the input variables on the outputs using the results from the different studies. The work has shown that the potential for practical implementation of models of this type in prediction of key properties related to long-term behavior, and/or formulation design in waste treatment facilities clearly exists, but more detailed definition of the data space by experimentation, with more complete harmonization of methods and reporting of experimental results, will be necessary to develop reliable commercial models.     

基于支持向量机的高炉向凉、向热炉况预测

高炉冶炼过程中炉温是影响技术经济指标的关键参数,保持合理的炉温是高炉稳定顺行的关键因素。采用某炼铁厂在线采集的数据,通过核主元分析对建模数据进行预处理,根据相关系数选定模型参数,确定参数对炉温的滞后时间,基于支持向量机建立了高炉向凉、向热预测诊断模型。通过实例验证,该模型具有很高的精度。     

Simulation and Prediction of Alkalinity in Sintering Process Based on Grey Least Squares Support Vector Machine

The prediction of the alkalinity is difficult during the sintering process. Whether or not the level of the alkalinity of sintering process is successful is directly related to the quality of sinter. There is no very good method for predicting the alkalinity by now owing to the high complexity, high nonlinearity, strong coupling, high time delay, and etc. Therefore, a new technique, the grey squares support machine, was introduced. The grey support vector machine model of the alkalinity enabled the development of new equation and algorithm to predict the alkalinity. During modelling, the fluctuation of data sequence was weakened by the grey theory and the support vector machine was capable of processing nonlinear adaptable information, and the grey support vector machine has a combination of those advantages. The results revealed that the alkalinity of sinter could be accurately predicted using this model by reference to small sample and information. The experimental results showed that the grey support vector machine model was effective and practical owing to the advantages of high precision, less samples required, and simple calculation.     

Quantifying the Effect of Rock Strength Criteria on Minimum Drilling Mud Weight Prediction Using Polyaxial Rock Strength Test Data

In engineering practice, a linear poroelasticity stress model in combination with a rock strength criterion is commonly used to determine a minimum mud weight for stable well drilling. Rock strength criterion therefore plays a key role in minimum mud weight prediction. There are a variety of rock strength criteria available in the literature. It is well known that all those criteria fall into two categories: intermediate principal stress dependent (σ2 dependent) criteria and intermediate principal stress independent (σ2 independent) criteria. To identify if a specific rock failure is σ2 dependent or σ2 independent, a polyaxial (true triaxial) rock strength test is essential. Similarly, to study the effect of rock strength criteria on wellbore stability and minimum drilling mud weight prediction, polyaxial rock strength test data are most useful. In this paper, we present a systematic approach to quantify the effect of three most commonly used rock strength criteria on minimum drilling mud weight prediction using polyaxial rock strength test data for Yuubari shale and Dunham dolomite.     

Optimizing the Prediction Accuracy of Concrete Compressive Strength Based on a Comparison of Data-Mining Techniques

This study attempts to optimize the prediction accuracy of the compressive strength of high-performance concrete (HPC) by comparing data-mining methods. Modeling the dynamics of HPC, which is a highly complex composite material, is extremely challenging. Concrete compressive strength is also a highly nonlinear function of ingredients. Several studies have independently shown that concrete strength is determined not only by the water-to-cement ratio but also by additive materials. The compressive strength of HPC is a function of all concrete content, including cement, fly ash, blast-furnace slag, water, superplasticizer, age, and coarse and fine aggregate. The quantitative analyses in this study were performed by using five different data-mining methods: two machine learning models (artificial neural networks and support vector machines), one statistical model (multiple regression), and two metaclassifier models (multiple additive regression trees and bagging regression trees). The methods were developed and tested against a data set derived from 17 concrete strength test laboratories. The cross-validation of unbiased estimates of the prediction models for performance comparison purposes indicated that multiple additive regression tree (MART) was superior in prediction accuracy, training time, and aversion to overfitting. Analytical results suggested that MART-based modeling is effective for predicting the compressive strength of varying HPC age.     

Generalization of ETo ANN Models through Data Supplanting

This paper describes the application of artificial neural networks (ANNs) for estimating reference evapotranspiration (ETo) as a function of local maximum and minimum air temperatures as well as exogenous relative humidity and reference evapotranspiration in different continental contexts of the autonomous Valencia region, on the Spanish Mediterranean coast. The development of new and more precise models for ETo prediction from minimum climatic data is required, since the application of existing methods that provide acceptable results is limited to those places where large amounts of reliable climatic data are available. The Penman-Monteith model for ETo prediction, proposed by the FAO as the sole standard method for ETo estimation, was used to provide the ANN targets for the training and testing processes. Concerning models which demand scant climatic inputs, the proposed model provides performances with lower associated errors than the currently existing temperature-based models, which only consider local data.     

Endpoint Prediction of EAF Based on Multiple Support Vector Machines

 The endpoint parameters are very important to the process of EAF steel making, but their on line measurement is difficult. The soft sensor technology is widely used for the prediction of endpoint parameters. Based on the analysis of the smelting process of EAF and the advantages of support vector machines, a soft sensor model for predicting the endpoint parameters was built using multiple support vector machines (MSVM). In this model, the input space was divided by subtractive clustering and a sub model based on LS SVM was built in each sub space. To decrease the correlation among the sub models and to improve the accuracy and robustness of the model, the sub models were combined by Principal Components Regression. The accuracy of the soft sensor model is perfectly improved. The simulation result demonstrates the practicability and efficiency of the MSVM model for the endpoint prediction of EAF.     

基于灰狼优化支持向量机回归与SHAP值的锡冶炼能耗预测

锡冶炼过程综合能源消耗占整个锡生产过程90%,存在很大节能潜力。针对锡冶炼过程综合能耗机理模型难以建立、导致预测准确度不高的问题,提出灰狼优化的支持向量机回归(GWO-SVR)模型用于锡冶炼过程综合能耗的预测,并以某锡冶炼厂为例,将所提模型与SVR、RF(随机森林)、BP(反向传播神经网络)、LR(线性回归)模型进行比较。结果表明,GWO-SVR模型可获得最理想的预测结果,在预测精度上相比于其他机器学习算法有着巨大优势。此外,使用SHAP值从全局解释和单样本解释两个方面解释所建立的GWO-SVR模型,可视化特征对输出的贡献,增加了GWO-SVR的可解释性,并以此制定可靠的节能策略。     

基于支持向量机挖掘不一致事例隐含的异常信息

基于支持向量机,提出一种挖掘粗集信息表中不一致事例背后隐藏某种有价值信息的算法,即不一致是由于错误引起,还是由于误差引起,抑或是由于缺少属性引起,并提出一些排除不一致的方案和算法.     

基于支持向量机回归的铸坯表面目标温度设定

提出了一种多输入多输出支持向量机回归算法,利用冶金技术人员计算的目标温度设定表,设定实时二冷区铸坯表面目标温度。200 mm×1 534 mm 16Mn钢板坯连铸试验结果表明,在训练样本相同时,支持向量机训练时间为3.2 s,预测目标温度误差为±1℃,BP神经网络训练时间为23.5 s,预测目标温度误差为±2℃,多输入多输出支持向量机回归算法优于BP神经网络算法,能够根据工艺变化情况,实时改变目标温度,为实现连铸动态控制提供了条件,有助于提高铸坯的质量。     

Durability Prediction for GFRP Reinforcing Bars Using Short-Term Data of Accelerated Aging Tests

This paper presents a procedure based on the Arrhenius relation to predict the long-term behavior of glass fiber-reinforced polymer (GFRP) bars in concrete structures, based on short-term data from accelerated aging tests. GFRP reinforcing bars were exposed to simulated concrete pore solutions at 20, 40, and 60°C. The tensile strengths of the bars determined before and after exposure were considered a measure of the durability performance of the specimens. Based on the short-term data, a detailed procedure is developed and verified to predict the long-term durability performance of GFRP bars. A modified Arrhenius analysis is included in the procedure to evaluate the validity of accelerated aging tests before the prediction is made. The accelerated test and prediction procedure used in this study can be a reliable method to evaluate the durability performance of FRP composites exposed to solutions or in contact with concrete.