Ka-veng Yuen
University of Macau, Civil and Environmental Engineering, Faculty Member
- Prof. Yuen (PhD, Caltech) is the Dean of Graduate School and a Distinguished Professor at the University of Macau. Hi... moreProf. Yuen (PhD, Caltech) is the Dean of Graduate School and a Distinguished Professor at the University of Macau. His expertise includes Bayesian inference, model selection, system identification structural health monitoring, vibration control and reliability analysis.edit
Research Interests:
Research Interests:
Records of precipitation in Macau are collected from three local monitoring stations from 1993 to 2003. Chemical compositions of precipitation are analyzed to reveal the causes of wet deposition inMacau. The review of volume weighted... more
Records of precipitation in Macau are collected from three local monitoring stations from 1993 to 2003. Chemical compositions of precipitation are analyzed to reveal the causes of wet deposition inMacau. The review of volume weighted annual pH values over this period shows that the entire area of Macau 2 is subjected to acidic precipitation. The concentrations of three ionic species H+, SO42- and NO3- are analyzed statistically and their distributions are found to have positive skewness and kurtosis. Lognormal cumulative distribution function fits each sample cumulative distribution function for each ionic species well implying that their temporal behaviours are governed according to the lognormal distribution. Nonetheless, the distribution of each ionic species exhibits spatial variability across the stations. Therefore, it is suspected that the sources of wet acid deposition indifferent parts of Macau are different and further correlation analysis on the values of pH, sulfate, and...
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Abstract Extended Kalman filter (EKF) is widely adopted for state estimation and parametric identification of dynamical systems. In this algorithm, it is required to specify the covariance matrices of the process noise and measurement... more
Abstract Extended Kalman filter (EKF) is widely adopted for state estimation and parametric identification of dynamical systems. In this algorithm, it is required to specify the covariance matrices of the process noise and measurement noise based on prior knowledge. However, improper assignment of these noise covariance matrices leads to unreliable estimation and misleading uncertainty estimation on the system state and model parameters. Furthermore, it may induce diverging estimation. To resolve these problems, we propose a Bayesian probabilistic algorithm for online estimation of the noise parameters which are used to characterize the noise covariance matrices. There are three major appealing features of the proposed approach. First, it resolves the divergence problem in the conventional usage of EKF due to improper choice of the noise covariance matrices. Second, the proposed approach ensures the reliability of the uncertainty quantification. Finally, since the noise parameters are allowed to be time-varying, nonstationary process noise and/or measurement noise are explicitly taken into account. Examples using stationary/nonstationary response of linear/nonlinear time-varying dynamical systems are presented to demonstrate the efficacy of the proposed approach. Furthermore, comparison with the conventional usage of EKF will be provided to reveal the necessity of the proposed approach for reliable model updating and uncertainty quantification.
Research Interests:
Research Interests:
The soil–water characteristic curve (SWCC) of fine-grained soils is usually determined experimentally. In the design of mine waste covers and landfill liners, the unsaturated hydraulic conductivity function, k(h), is often derived... more
The soil–water characteristic curve (SWCC) of fine-grained soils is usually determined experimentally. In the design of mine waste covers and landfill liners, the unsaturated hydraulic conductivity function, k(h), is often derived theoretically from the measured SWCC. Implicit in these derivations is the transformation of the SWCC to a pore-size distribution (PSD), typically assumed to be constant and monomodal. However, PSD measurements of a clayey till compacted at various water contents after compaction, after flexible-wall permeability testing and before and after SWCC tests show that the PSD of the same material varies significantly under the stated physical conditions. Predictions of the SWCCs using PSDs measured both before and after the SWCC tests significantly underpredicted the values measured. By applying a simple transformation to the PSD to account for the scaling effect from the porosimetry samples (approximately 1 g dry weight) to the SWCC test samples (approximately 200 g dry weight), the ...
Research Interests:
AbstractThe pullout resistance of soil nail is a key parameter in soil nailing design. Pullout resistance is affected by many factors, such as overburden pressure, grouting pressure, soil dilation, and degree of saturation of soil.... more
AbstractThe pullout resistance of soil nail is a key parameter in soil nailing design. Pullout resistance is affected by many factors, such as overburden pressure, grouting pressure, soil dilation, and degree of saturation of soil. Because of the complexity of the pullout mechanism, some factors have not been well incorporated in the current soil nail design methods. In this study, Bayesian analysis is performed to investigate the relative importance of several key factors and to build a new design formula to estimate maximum pullout shear stress of grouted soil nails. By using a series of laboratory soil nail pullout test data, Bayesian analysis is performed to select a predictive formula with suitable complexity and to identify its parameters. It is found that the most important factors are the degree of saturation and the product of grouting pressure and overburden pressure. It is shown that the proposed optimal model exhibits significantly stronger correlation with measurements than the existing effec...
Research Interests:
Evaluation of the cyclic shear modulus of soils is a crucial but challenging task for many geotechnical earthquake engineering and soil dynamic issues. Improper determination of this property unnecessarily drives up design and maintenance... more
Evaluation of the cyclic shear modulus of soils is a crucial but challenging task for many geotechnical earthquake engineering and soil dynamic issues. Improper determination of this property unnecessarily drives up design and maintenance costs or even leads to the construction of unsafe structures. Due to the complexities involved in the direct measurement, empirical curves for estimating the cyclic shear modulus have been commonly adopted in practice for simplicity and economical considerations. However, a systematic and robust approach for formulating a reliable model and empirical curve for cyclic shear modulus prediction for clayey soils is still lacking. In this study, the Bayesian model class selection approach is utilized to identify the most significant soil parameters affecting the normalized cyclic shear modulus and a reliable predictive model for normally to moderately over-consolidated clays is proposed. Results show that the predictability and reliability of the proposed model out performs the well-known empirical models. Finally, a new design chart is established for practical usage.
Research Interests:
Research Interests:
The present study introduces the novel Bayesian approach to solve a difficult problem that modellers face when doing linear regression for air quality prediction; i.e. the non-uniqueness of the input variables and the functional forms to... more
The present study introduces the novel Bayesian approach to solve a difficult problem that modellers face when doing linear regression for air quality prediction; i.e. the non-uniqueness of the input variables and the functional forms to be selected. Using the historical information of ozone, nitrogen dioxide, respirable particulates and eight meteorological elements recorded during the high ozone seasons (May-October) in Macau between 2006 and 2007 as the training data, the Bayesian approach was applied for model selection from 16383 candidates. The model with the best efficiency-robustness balance selected is not the most complicated one. It was then examined against the most complicated model with the data between 2008 and 2009. Results show that the selected model yields better performance with the root-mean-squared error (RMSE) and the coefficient of determination (r2) equal to 19.18 µg/m3 and 0.81, respectively. In addition, it is capable to capture 87% of the ozone episodes w...
Research Interests:
The aim of this paper is to present a novel attempt for parametric estimation in the hydrostatic-season-time (HST) model. The empirical HST-model has been widely used for the analysis of different measurement data types on dams. The... more
The aim of this paper is to present a novel attempt for parametric estimation in the hydrostatic-season-time (HST) model. The empirical HST-model has been widely used for the analysis of different measurement data types on dams. The significance of individual parameters or their sub-groups for modelling the influence of the water level, air and water temperature, and irreversible deformations due to the ageing of the dam, depends on the structure itself. The process of finding an accurate HST-model for a given data set, which remains robust to outliers, cannot only be demanding but also time consuming. The Bayesian model class selection approach imposes a penalisation against overly complex model candidates and admits a selection of the most plausible HST-model according to the maximum value of model evidence provided by the data or relative plausibility within a set of model class candidates. The potential of Bayes interference and its efficiency in an HST-model are presented on geodetic time series as a result of a permanent monitoring system on a rock-fill embankment dam. The method offers high potential for engineers in the decision making process, whilst the HST-model can be promptly adapted to new information given by new measurements and can enhance the safety and reliability of dams.
Research Interests:
This study presents a method of predicting the soil water retention curve (SWRC) of a soil using a set of measured SWRC data from a soil with the same texture but different initial void ratio. The relationships of the volumetric water... more
This study presents a method of predicting the soil water retention curve (SWRC) of a soil using a set of measured SWRC data from a soil with the same texture but different initial void ratio. The relationships of the volumetric water contents and the matric suctions between two samples with different initial void ratios are established. An adjustment parameter (β) is introduced to express the relationships between the matric suctions of two soil samples. The parameter β is a function of the initial void ratio, matric suction or volumetric water content. The function can take different forms, resulting in different predictive models. The optimal predictive models of β are determined for coarse-grained and fine-grained soils using the Bayesian method. The optimal models of β are validated by comparing the estimated matric suction and measured data. The comparisons show that the proposed method produces more accurate SWRCs than do other models for both coarse-grained and fine-grained soils. Furthermore, the influence of the model parameters of β on the predicted matric suction and SWRC is evaluated using Latin Hypercube sampling. An uncertainty analysis shows that the reliability of the predicted SWRC decreases with decreasing water content in fine-grained soils, and the initial void ratio has no apparent influence on the reliability of the predicted SWRCs in coarse-grained and fine-grained soils.
Research Interests:
Research Interests:
Research Interests:
A probabilistic approach for damage detection is presented using noisy incomplete input and response measurements that is an extension of a Bayesian system identification approach developed by the authors. This situation may be... more
A probabilistic approach for damage detection is presented using noisy incomplete input and response measurements that is an extension of a Bayesian system identification approach developed by the authors. This situation may be encountered, for example, during low-level ambient vibrations when a structure is instrumented with accelerometers that measure the input ground motion and structural response at a few locations but the wind excitation is not measured. A substructuring approach is used for the parameterization of the mass and stiffness distributions. Damage is defined to be a reduction of the substructure stiffness parameters compared with those of the undamaged structure. By using the proposed probabilistic methodology, the probability of various damage levels in each substructure can be calculated based on the available data. A four-story benchmark building subjected to wind and ground shaking is considered in order to demonstrate the proposed approach.
Research Interests:
Research Interests:
Outlier detection is an important problem in statistics. In this paper, we introduce a novel concept of outlier probability for outlier detection and robust linear regression. First, the Mahalanobis distance is utilized to identify the... more
Outlier detection is an important problem in statistics. In this paper, we introduce a novel concept of outlier probability for outlier detection and robust linear regression. First, the Mahalanobis distance is utilized to identify the leverage points. By excluding the leverage points, the maximum trimmed likelihood estimation will be used to obtain an initial set of regular data points while the remaining and the leverage points are included in the initial suspicious data set. Then, each suspicious data point and the combinations are evaluated by a novel outlier probability that depends not only on the residuals but also the size of the data set. Incorporating the data size is important as it controls the probability of the existence of a data point (or a batch or data points) exceeding a given value of the normalized residual. This outlier probability is robust because it incorporates also the posterior uncertainty quantified using the Bayesian approach. Then, the data points with...
Research Interests:
The Pearl River Delta (PRD) region is located on the southeast coast of mainland China and it is an important economic hub. The high levels of particulate matter (PM) in the atmosphere, however, and poor visibility have become a complex... more
The Pearl River Delta (PRD) region is located on the southeast coast of mainland China and it is an important economic hub. The high levels of particulate matter (PM) in the atmosphere, however, and poor visibility have become a complex environmental problem for the region. Air quality modeling systems are useful to understand the temporal and spatial distribution of air pollution, making use of atmospheric emission data as inputs. Over the years, several atmospheric emission inventories have been developed for the Asia region. The main purpose of this work is to evaluate the performance of the air quality modeling system for simulating PM concentrations over the PRD using three atmospheric emission inventories (i.e., EDGAR, REAS and MIX) during a winter and a summer period. In general, there is a tendency to underestimate PM levels, but results based on the EDGAR emission inventory show slightly better accuracy. However, improvements in the spatial and temporal disaggregation of em...
Abstract This study was devoted to investigating stochastic model updating in a Bayesian inference framework based on a frequency response function (FRF) vector without any post-processing such as smoothing and windowing. The statistics... more
Abstract This study was devoted to investigating stochastic model updating in a Bayesian inference framework based on a frequency response function (FRF) vector without any post-processing such as smoothing and windowing. The statistics of raw FRFs were inferred with a multivariate complex-valued Gaussian ratio distribution. The likelihood function was formulated by embedding the theoretical FRFs that contained the model parameters to be updated in the class of the probability model of the raw FRFs. The Transitional Markov chain Monte Carlo (TMCMC) used to sample the posterior probability density function implies considerable computational toll because of the large batch of repetitive analyses of the forward model and the increasing expense of the likelihood function calculations with large-scale loop operations. The vectorized formula was derived analytically to avoid time-consuming loop operations involved in the likelihood function evaluation. Furthermore, a distributed parallel computing scheme was developed to allow the TMCMC stochastic simulation to run across multiple CPU cores on multiple computers in a network. The case studies demonstrated that the fast-computational scheme could exploit the availability of high-performance computing facilities to drastically reduce the time-to-solution. Finally, parametric analysis was utilized to illustrate the uncertainty propagation properties of the model parameters with the variations of the noise level, sampling time, and frequency bandwidth.
Research Interests:
Research Interests:
Abstract Reliable verification and evaluation of the mechanical properties of a layered composite ensemble are critical for industrially relevant applications, however it still remains an open engineering challenge. In this study, a fast... more
Abstract Reliable verification and evaluation of the mechanical properties of a layered composite ensemble are critical for industrially relevant applications, however it still remains an open engineering challenge. In this study, a fast Bayesian inference scheme based on multi-frequency single shot measurements of wave propagation characteristics is developed to overcome the limitations of ill-conditioning and non-uniqueness associated with the conventional approaches. A Transitional Markov chain Monte Carlo (TMCMC) algorithm is employed for the sampling process. A Wave and Finite Element (WFE)-assisted metamodeling scheme in lieu of expensive-to-evaluate explicit FE analysis is proposed to cope with the high computational cost involved in TMCMC sampling. For this, the Kriging predictor providing a surrogate mapping between the probability spaces of the model predictions for the wave characteristics and the mechanical properties in the likelihood evaluations is established based on the training outputs computed using a WFE forward solver, coupling periodic structure theory to conventional FE. The valuable uncertainty information of the prediction variance introduced by the use of a surrogate model is also properly taken into account when estimating the parameters’ posterior probability distribution by TMCMC. A numerical study as well as an experimental study are conducted to verify the computational efficiency and accuracy of the proposed methodology. Results show that the TMCMC algorithm in conjunction with the WFE forward solver-aided metamodeling can sample the posterior Probability Density Function (PDF) of the updated parameters at a very reasonable cost. This approach is capable of quantifying the uncertainties of recovered independent characteristics for each layer of the composite structure under investigation through fast and inexpensive experimental measurements on localized portions of the structure.
Research Interests:
AbstractEstimation on the uniaxial compressive strength (UCS) of rock is an important issue in geotechnical engineering. Empirical relation establishment for UCS estimation is particularly favorabl...
AbstractNatural fibers are environment-friendly and efficient for soil reinforcement. Many studies have reported the influences of fiber percentage on the shear strength of fiber reinforced soil. H...
Research Interests:
AbstractStructural health indicators, such as modal frequencies, have been commonly utilized to interpret the health condition of monitored structures. This study modeled the relationship between s...
Research Interests:
Research Interests:
Abstract Of late, various constitutive models have been proposed in the literature for the purpose of capturing the various complex physical mechanisms governing the creep behavior of soft soil. However, the more complex the model, the... more
Abstract Of late, various constitutive models have been proposed in the literature for the purpose of capturing the various complex physical mechanisms governing the creep behavior of soft soil. However, the more complex the model, the greater the number of associated uncertain parameters it has, and the less robustness it is. In this study, the Bayesian model class selection approach is applied to select the most plausible/suitable model describing the creep behavior of soft soil using laboratory measurements. In total, one elastic plastic (EP) model and eight elastic viscoplastic (EVP) models are investigated. To assess the performance of the different models in the prediction of creep behavior of soft soils, Bayesian model class selection is respectively performed using the oedometer test data from the intact samples of Vanttila clay and reconstituted samples of Hong Kong Marine Clay collected from the literature. All unknown model parameters are identified simultaneously by adopting the transitional Markov Chain Monte Carlo (TMCMC) method, and their uncertainty is quantified through the obtained posterior probability density functions (PDFs). The result shows that the proposed method is an excellent candidate for identifying the most plausible model and its associated parameters for different kinds of soft soils. The approach also provides uncertainty evaluation of the model prediction based on the given data.
Research Interests:
Research Interests:
In ground motion prediction, the key is to develop a suitable and reliable GMPE (ground motion prediction equation) characterizing the ground motion pattern of the target seismic region. There are two critical goals encountered in GMPE... more
In ground motion prediction, the key is to develop a suitable and reliable GMPE (ground motion prediction equation) characterizing the ground motion pattern of the target seismic region. There are two critical goals encountered in GMPE development. Proposing a suitable predictive formula applicable to target seismic region has attracted much of the attention in previous studies. On the other hand, dependence between prediction–error variance and ground motion data has been observed and the study on this kind of heterogeneous relation becomes an important task yet to be explored. In this article, a novel HEteRogeneous BAyesian Learning (HERBAL) approach is proposed for achieving these two goals simultaneously. The homogeneity assumption on error in the traditional learning approach is relaxed, so the proposed approach is applicable for more general heterogeneous cases. With the generalization made on the traditional Bayesian learning by embedding the derived closed form expression for error variance parameter optimization component into the hyperparameter optimization of ARD (automatic relevance determination) prior, the proposed learning approach is capable of performing continuous model training on a prescribed predictive formula with unknown error pattern. A database of strong ground motion records in the Tangshan region of China is obtained for the analysis. It is shown that the trained optimal model class by the proposed approach is promising as that, the trained optimal model class retains model simplicity of the predictive formula with capability on both robustness enhancement ground motion prediction and precise determination of the error pattern.
Research Interests:
Abstract The techniques of bias correction are commonly used for improving the performance of deterministic air quality forecasting systems. One issue not addressed in previous studies is how to select systematically and objectively the... more
Abstract The techniques of bias correction are commonly used for improving the performance of deterministic air quality forecasting systems. One issue not addressed in previous studies is how to select systematically and objectively the best correction model from a pool of candidates. In this study, a method that could evaluate the probabilities of all model candidates based on a set of training data is proposed to select the most accurate and robust model by finding the one with the maximum probability. The Bayesian method was applied to select bias correction models at 12 monitoring stations for improving the forecasts of daily averaged PM 10 concentrations given by the deterministic air quality forecasting system WRF-EURAD in Porto, Portugal. At each station, 4095 (=2 12 –1) correction model candidates were systematically formed by adopting different linear combinations of 12 input variables. Selection of the best model was processed based on one year of monitoring and WRF-EURAD data. Based on the 2012 data, the selected model at each station was found to have significantly higher probability than the other candidates, and it is also much simpler than the full model. These selected models were then used to correct the raw forecasts by WRF-EURAD in the following year. The corrected forecasts show significant improvement on the performance indicators ( RMSE by 35.8%, R by 58.5%, MFB by 68%, EDR by 38.3%, FAR by 51.8%, CSI by 30.8%) over the raw outputs of WRF-EURAD, confirming the success of the proposed technique.
Research Interests:
Modal frequency is an important indicator for structural health assessment. Previous studies have shown that this indicator is substantially affected by the fluctuation of ambient conditions, such as temperature and humidity. Therefore,... more
Modal frequency is an important indicator for structural health assessment. Previous studies have shown that this indicator is substantially affected by the fluctuation of ambient conditions, such as temperature and humidity. Therefore, recognizing the pattern between modal frequency and ambient conditions is necessary for reliable long-term structural health assessment. In this article, a novel machine-learning algorithm is proposed to automatically select relevance features in modal frequency-ambient condition pattern recognition based on structural dynamic response and ambient condition measurement. In contrast to the traditional feature selection approaches by examining a large number of combinations of extracted features, the proposed algorithm conducts continuous relevance feature selection by introducing a sophisticated hyperparameterization on the weight parameter vector controlling the relevancy of different features in the prediction model. The proposed algorithm is then u...
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
The undrained shear strength (su) of cohesive soils is a crucial parameter for many geotechnical engineering applications. Due to the complexities and uncertainties associated with laboratory and in situ tests, it is a challenging task to... more
The undrained shear strength (su) of cohesive soils is a crucial parameter for many geotechnical engineering applications. Due to the complexities and uncertainties associated with laboratory and in situ tests, it is a challenging task to obtain the undrained shear strength in a reliable and economical manner. In this study, a probabilistic model for the su of moderately overconsolidated clays is developed using the Bayesian model class selection approach. The model is based on a comprehensive geotechnical database compiled for this study with field measurements of field vane strength (su), plastic limit (PL), natural water content (Wn), liquid limit (LL), vertical effective overburden stress ($$\sigma_{\nu }^{\prime }$$σν′), preconsolidation pressure ($$\sigma_{\text{p}}^{\prime }$$σp′) and overconsolidated ratio (OCR). Comparison study shows that the proposed model is superior to some well-known empirical relationships for OC clays. The proposed probabilistic model not only provides reliable and economical estimation of su but also facilitates reliability-based analysis and design for performance-based engineering applications.
Research Interests:
This study compared two types of (offline vs. online) bias correction models applied to correct the mismatch between the predicted daily averaged PM10 concentration by the deterministic air quality forecasting system WRF-EURAD and the... more
This study compared two types of (offline vs. online) bias correction models applied to correct the mismatch between the predicted daily averaged PM10 concentration by the deterministic air quality forecasting system WRF-EURAD and the measured concentration at the air quality monitoring station in Porto, Portugal. The WRF-EURAD is a Eulerian system consisting of a Weather Research Forecasting (WRF) model and a European Air Pollution Dispersion (EURAD) model. Both bias correction models were linear statistical models developed with the same set of input variables. The major difference between the online or the offline models is the adaptiveness. While the coefficients of the offline bias correction model are fixed after training with the ordinary least squares, those in the online bias correction model are updated adaptively with the Kalman filter. Comparison of these bias correction models was made at an urban traffic station Senhora da Hora in Porto, Portugal within 2013. The fract...