基于改进GA-SVR算法的隧道工程三维弹塑性模型参数的智能辨识

将支持向量回归（SVR）算法引入岩土工程数值计算模型参数的辨识中可以充分发挥SVR算法的小样本、泛化性好和全局最优化的优点。但现阶段标准的SVR算法只能解决一维输出变量的回归问题，这就使其在反分析领域的应用受到限制。引入一种改进的SVR算法，这种算法通过将多维输出变量回归转化为多层标准一维输出变量回归来解决这个问题，并与十进制编码的遗传算法相结合，形成改进的GA-SVR算法，用遗传算法搜索最优的SVR模型参数以建立最优的待辨识参数与位移之间的非线性映射关系，然后用遗传算法进行待辨识参数的最优辨识。为对比这种改进GA-SVR算法的效果，将遗传算法与BP神经网络相结合，形成GA-BP算法且编制相应的计算程序。将这两种算法运用于同样的隧道工程三维弹塑性模型参数的智能辨识，数值算例表明改进的GA-SVR算法较GA-BP算法可以取得更高的辨识精度和更好的计算效率，可运用于类似岩土工程计算参数的辨识。     

基于SVR技术的混凝土强度预测

提出了一种基于支持向量机回归（SVR）模型的混凝土强度预测方法,通过SVR预测结果和Bolomey经验公式的计算结果相比较,证明基于SVR技术的混凝土强度预测模型精度要远高于传统的Bolomey经验公式。     

基于PCA特征提取的SVR地下水位动态预测方法

本文提出了一种基于PCA特征提取的SVR地下水位动态预测方法.PCA用来对地下水位样本值进行相关性分析和特征提取,将得到的最佳特征向量子集输入SVR模型,并通过网格搜索的方法进行SVR参数寻优,进行地下水位预测.通过地下水位的实例分析和UCI标准数据集对模型的有效性进行验证,结果表明该方法不仅降低了预测所需的计算工作量...     

Soft computing for solar radiation potential assessment in Algeria

ABSTRACT

Precise estimation of solar radiation is a highly required parameter for the design and assessment of solar energy applications. Over the past years, many machine learning techniques have been proposed in order to improve the forecasting performance using different input attributes. The aim of this study is the forecasting of one day ahead of horizontal global solar radiation using a set of meteorological and geographical inputs. In this respect, the Gaussian process regression methodology (GPR) and least-square support vector machine (LS-SVM) with different kernels are evaluated in order to select the most appropriate forecasting model. In order to assess the proposed models, the southern Algerian city, Ghardaia regions, was selected for this study. A historical data of five years (2013–2017) of meteorological data collected at Renewable Energies (URAER) in Ghardaia city are used. The achieved results demonstrate that all the proposed models give approximately similar results in terms of statistical indicators. In term of processing time, all the models showed acceptable computational efficiency with less computational costs of the GPR model among all machine learning models.     

基于 SVR 的工程建设项目快速投资估算方法研究

在建设项目前期,如何快速而准确地估算工程项目的造价,对项目的投资决策具有很大的意义。针对传统造价估算 方法的不足之处,采用 SPSS 统计分析软件进行工程造价指标的相关性分析及指标体系选取,将之作为输入变量,使用真实 案例训练集样本训练 SVR 模型并进行仿真模拟预测。为了验证提出的 SVR 模型的有效性,引入 BP 人工神经网络来进行预 测结果的对比验证。结果表明,SVR 模型得到的预测值平均绝对百分比误差约为 5%,拟合优度 R2高达 0.97,远小于 BPNN 模型的预测误差 14%,即提出的 SVR 估算模型要比 BP 人工神经网络预测模型具有更良好的泛化能力,预测精度更高,因 此其在工程项目前期投资估算实践中具有一定的现实意义。     

岩石力学性态预测的PSO-SVM模型

 传统的固体力学方法在描述岩石的各种地质因素与其力学性态之间的复杂非线性关系时存在困难。引入粒子群算法(PSO)对支持向量机(SVM)进行优化,提出岩石力学性态预测的粒子群优化支持向量机模型(PSO-SVM)。该模型利用SVM来建立岩石地质因素与力学性态之间的非线性关系;同时利用PSO对SVM参数进行全局寻优,避免人为选择参数的盲目性,从而提高模型的预测精度。将PSO-SVM应用到岩石压缩系数的预测中,并与传统的BP神经网络(BP-NN)进行对比分析。结果显示,PSO-SVM的预测精度较BP-NN有较大的提高,从而表明PSO-SVM在岩石力学性态预测中的可行性和有效性。     

Smart Artificial Firefly Colony Algorithm‐Based Support Vector Regression for Enhanced Forecasting in Civil Engineering

Advanced data mining techniques are potential tools for solving civil engineering (CE) problems. This study proposes a novel smart artificial firefly colony algorithm‐based support vector regression (SAFCA‐SVR) system that integrates firefly algorithm (FA), chaotic maps, adaptive inertia weight, Lévy flight, and least squares support vector regression (LS‐SVR). First, adaptive approach and randomization methods are incorporated in FA to construct a novel and highly effective metaheuristic algorithm for global optimization. The enhanced FA is then used to optimize parameters in LS‐SVR model. The proposed system is validated by comparing its performance with those of empirical methods and previous works via cross‐validation algorithm and hypothesis test through the real‐world engineering cases. Specifically, high‐performance concrete, resilient modulus of subgrade soils, and building cooling load are used as case studies. The SAFCA‐SVR achieved 8.8%–91.3% better error rates than those of previous works. Analytical results confirm that using the proposed hybrid system significantly improves the accuracy in solving CE problems.     

基于LMD-PSO-LSSVM组合模型的深基坑变形预测

变形是造成基坑安全隐患的重要因素。为准确预测基坑变形趋势,提出一种将局部均值分解(LMD)、粒子群优化算法(PSO)与最小二乘支持向量机(LSSVM)组合的深基坑变形预测模型。通过 LMD 将时序样本分解为多个分量,利用PSO优化后的LSSVM模型对各分量建立非线性基坑变形预测模型,最后采用滚动预测的方法对各分量进行预测并将结果叠加得到时序样本的预测值。通过实际工程进行模型预测与分析。结果表明:该模型不仅反映出基坑变形本质特征,而且预测精度明显提高,将其运用于基坑变形预测研究中具有较好的应用性和可靠性。     

一种基于补偿模糊神经网络的水质预测方法

在污水处理系统过程控制中,对水质变化规律进行预测是控制系统可靠、稳定运行的重要环节。介绍了基于模糊逻揖和神经网络的补偿神经网络(CFNN)及其学习算法,利用CFNN学习速度快、学习过程稳定、全局动态优化运算等特点,建立污水处理厂CFNN的水质预测模型。实例预测结果表明该模型对初始值的选择不敏感,具有很好的收敛性和预测精度,适合实际工程应用。     

Analysis of crowd jam in public buildings based on cusp-catastrophe theory

The partial differential equation (PDE) model is established based on the physical movement of the continuous crowd states, and has a strong dependence on the initial conditions. In addition, some characteristics of large crowd like the phenomena of discontinuous jumping in reality are hard to explain by partial differential equations (PDEs). The catastrophe theory, however, can explain these characteristics. The catastrophe model is based on complex systems, and does not rely on the initial conditions. A catastrophe model is presented in this paper to study the movement mechanism of crowd jam. The critical density and the critical velocity are also derived. Results of the analysis indicate that (1) the catastrophe model based on complex systems cannot only describe the phenomena of non-continuity in crowds, but also obtain the critical density and the critical velocity; (2) the catastrophe model does not require initial conditions, besides the problem solving process does not involve the time and location. This means the applicability of the catastrophe model can be used more broadly than PDE model. In other words, the model is better suited to be used in various applications; (3) the catastrophe model can forecast and control the jamming state by monitoring and adjusting the variables in practice. The model supports the decision of the management of emergency evacuation.     

Prediction of grey-catastrophe destabilization time of a granite residual soil slope under rainfall

The stability of a residual soil slope is a system of grey and white, and gradual change and sudden change. It is difficult to accurately forecast the instability time of a granite residual soil slope, because the factors affecting the instability of granite residual soil slopes are random and uncontrollable. Using the grey prediction method, the accumulation generation of data weakens the influence of random disturbance factors in the original sequence, and enhances the regularity of data. In this paper, we proposed a grey cusp catastrophe prediction model to calculate the instability time of granite residual soil slopes under rainfall. We took one-time cumulative transformation of the remote monitoring data of slope displacement, and then conducted polynomial regression fitting on the processed displacement data and monitoring time. The grey cusp catastrophe instability prediction model of granite residual soil slopes was established using the nonlinear dynamic catastrophic prediction theory. The critical instability time of the slope was predicted using the grey catastrophe instability prediction method presented in this paper. For comparison, we conducted exponential curve fitting on the displacement data and monitoring time, and then established the Saito instability forecasting method of a soil slope to predict the slope instability. Through the engineering application, and using the presented grey catastrophe slope instability prediction method, the slope instability process may be accurately predicted. The calculation results using the grey catastrophe slope instability prediction method presented in this paper are in good agreement with the actual situation. Whether it can be used for other slopes remains to be further studied.

     

Ensemble unit and AI techniques for prediction of rock strain

The behavior of rock masses is influenced by a variety of forces, with measurement of stress and strain playing the most critical roles in assessing deformation. The laboratory test for determining strain at each location within rock samples is expensive and difficult but rock strain data are important for predicting failure of rock material. Many researchers employ AI technology in order to solve these difficulties. AI algorithms such as gradient boosting machine (GBM), support vector regression (SVR), random forest (RF), and group method of data handling (GMDH) are used to efficiently estimate the strain at every point within a rock sample. Additionally, the ensemble unit (EnU) may be utilized to evaluate rock strain. In this study, 3000 experimental data are used for the purpose of prediction. The obtained strain values are then evaluated using various statistical parameters and compared to each other using EnU. Ranking analysis, stress-strain curve, Young’s modulus, Poisson’s ratio, actual vs. predicted curve, error matrix and the Akaike’s information criterion (AIC) values are used for comparing models. The GBM model achieved 98.16% and 99.98% prediction accuracy (in terms of values of R²) in the longitudinal and lateral dimensions, respectively, during the testing phase. The GBM model, based on the experimental data, has the potential to be a new option for engineers to use when assessing rock strain.     

Development of an England-wide indoor overheating and air pollution model using artificial neural networks

With the UK climate projected to warm in future decades, there is an increased research focus on the risks of indoor overheating. Energy-efficient building adaptations may modify a buildings risk of overheating and the infiltration of air pollution from outdoor sources. This paper presents the development of a national model of indoor overheating and air pollution, capable of modelling the existing and future building stocks, along with changes to the climate, outdoor air pollution levels, and occupant behaviour. The model presented is based on a large number of EnergyPlus simulations run in parallel. A metamodelling approach is used to create a model that estimates the indoor overheating and air pollution risks for the English housing stock. The performance of neural networks (NNs) is compared to a support vector regression (SVR) algorithm when forming the metamodel. NNs are shown to give almost a 50% better overall performance than SVR.     

Using extreme learning machines for short-term urban water demand forecasting

This study explores the ability of various machine learning methods to improve the accuracy of urban water demand forecasting for the city of Montreal (Canada). Artificial Neural Network (ANN), Support Vector Regression (SVR) and Extreme Learning Machine (ELM) models, in addition to a traditional model (Multiple linear regression, MLR) were developed to forecast urban water demand at lead times of 1 and 3 days. The use of models based on ELM in water demand forecasting has not previously been explored in much detail. Models were based on different combinations of the main input variables (e.g., daily maximum temperature, daily total precipitation and daily water demand), for which data were available for Montreal, Canada between 1999 and 2010. Based on the squared coefficient of determination, the root mean square error and an examination of the residuals, ELM models provided greater accuracy than MLR, ANN or SVR models in forecasting Montreal urban water demand for 1 day and 3 days ahead, and can be considered a promising method for short-term urban water demand forecasting.     

A pattern recognition technique for structural identification using observed vibration signals: Linear case studies

This and the companion article summarize linear and nonlinear structural identification (SI) methods using a pattern recognition technique, support vector regression (SVR). Signal processing plays a key role in the SI field, because observed data are often incomplete and contaminated by noise. Support vector regression (SVR) is a novel data processing technique that is superior in terms of its robustness, thus it has the potential to be applied for accurate and efficient structural identification. Three SVR-based methods employing the autoregression moving average (ARMA) time series, the high-order AR model, and the sub-structuring strategy are presented for linear structural parameter identification using observed vibration data. The SVR coefficient selection and incremental training algorithm have also been presented. Numerical evaluations demonstrate that the SVR-based methods identify structural parameters accurately. A five-floor structure shaking table test has also been conducted, and the observed data are used to verify experimentally the novel SVR technique for linear structural identification.     

Forecasting of daily air quality index in Delhi

As the impact of air pollutants on human health through ambient air address much attention in recent years, the air quality forecasting in terms of air pollution parameters becomes an important topic in environmental science. The Air Quality Index (AQI) can be estimated through a formula, based on comprehensive assessment of concentration of air pollutants, which can be used by government agencies to characterize the status of air quality at a given location. The present study aims to develop forecasting model for predicting daily AQI, which can be used as a basis of decision making processes. Firstly, the AQI has been estimated through a method used by US Environmental Protection Agency (USEPA) for different criteria pollutants as Respirable Suspended Particulate Matter (RSPM), Sulfur dioxide (SO₂), Nitrogen dioxide (NO₂) and Suspended Particulate Matter (SPM). However, the sub-index and breakpoint concentrations in the formula are made according to Indian National Ambient Air Quality Standard. Secondly, the daily AQI for each season is forecasted through three statistical models namely time series auto regressive integrated moving average (ARIMA) (model 1), principal component regression (PCR) (model 2) and combination of both (model 3) in Delhi. The performance of all three models are evaluated with the help of observed concentrations of pollutants, which reflects that model 3 agrees well with observed values, as compared to the values of model 1 and model 2. The same is supported by the statistical parameters also. The significance of meteorological parameters of model 3 has been assessed through principal component analysis (PCA), which indicates that daily rainfall, station level pressure, daily mean temperature, wind direction index are maximum explained in summer, monsoon, post-monsoon and winter respectively. Further, the variation of AQI during the weekends (holidays) and weekdays are found negligible. Therefore all the days of week are accounted same in the models.     

自适应时序模型在地下工程位移预报中的应用

自适应时序模型的基本原理就是将自适应滤波理论应用于自回归时序AR(n)模型中。该模型在一定程度上根据量测数据和估计结果自行调整模型参数，通过递推算法自动地对模型参数加以修正，使其接近某种最佳值，即便在尚不完全掌握序列特性的情况下也能得到满意的结果。通过对山东龙口洼里煤矿一回采巷道金属支架的收敛位移和北京地铁王一东区间隧道北正线中洞断面收敛位移进行自适应建模，预报结果表明，此方法可行，预报结果也令人满意。     

A comparative study of using two numerical strategies to simulate the biochemical processes in microbially induced calcite precipitation

In this study, we carried out a comparative study of two different numerical strategies for the modeling of the biogeochemical processes in microbially induced calcite precipitation (MICP) process. A simplified MICP model was used, which is based on the mass transport theory. Two numerical strategies, namely the operator splitting (OS) and the global implicit (GI) strategies, were adopted to solve the coupled reactive mass transport problems. These two strategies were compared in the aspects of numerical accuracy, convergence property and computational efficiency by solving the presented MICP model. To look more into the details of the model, sensitivity analysis of some important modeling parameters was also carried out in this paper.     

垃圾填埋场灾变过程的温度–渗流–应力–化学耦合效应研究

 针对当前垃圾填埋场灾变过程预测与控制的迫切需求,结合垃圾填埋场及其周围复杂而特殊的环境地质条件,从温度–渗流–应力–化学(T-H-M-C)多场耦合角度深入分析垃圾填埋场灾变过程的演化机制与开展多场耦合研究的必要性。提出填埋气体运移的微生物降解–温度–渗流(B-T-H)耦合模型、考虑好氧和厌氧微生物降解作用的垃圾渗沥液污染物迁移转化渗流–微生物降解–化学(H-B-C)耦合模型、复合衬垫系统污染物运移渗流–化学(H-C)耦合模型以及考虑热量变化和水蒸气迁移过程对开裂过程影响的填埋场封场覆盖系统干燥开裂温度–渗流–应力(T-H-M)耦合模型,为垃圾填埋场灾变过程的预测和安全性评价提供有效的分析手段。提出一套多场耦合测试分析方法与试验技术,开发集监测、控制与数据采集于一体的填埋场中污染物传输的多场耦合测试分析系统。形成一套填埋场污染物多参数远程同步监测方法与技术,研制集实时监测与视频监督于一体的垃圾填埋场污染物远程在线监督系统。针对多场耦合作用下封场覆盖系统开裂问题,提出新型环保的垃圾填埋场封场覆盖生态污泥腾发覆盖技术(EST)。上述研究成果可为垃圾填埋场灾变过程的预防与控制提供科学手段和技术支持,同时对于丰富和拓宽多场多相耦合理论的发展具有重要的理论意义和应用价值。     

新陈代谢GM（1,1）模型在河流水质预测中的应用

针对常规GM(1,1)模型存在的不足,运用灰色系统理论,建立了灰色新陈代谢GM(1,1)河流水质预测模型,对该模型的精度以及误差进行了分析,并利用该模型对某地区河流的水质进行了预测,预测结果显示:灰色新陈代谢GM(1,1)预测模型能够明显地提高预测精度,增加预测的可信度。