Evolution Of Model Parameters Research Articles

Application of a surrogate model for condition monitoring of a digital twin gas turbine

Condition monitoring technology plays a crucial role in ensuring the reliable operation of gas turbines. Digital twin has propelled condition monitoring research into a new phase. This paper established a surrogate model of gas turbines for condition monitoring based on Markov-projection approximation subspace tracking. Furthermore, it explores the application of surrogate model in developing digital twin for gas turbines. The study initially establishes a Markov matrix and acquires an observation vector, utilizing the framework of the linear model. Utilizing real-time measurement data of gas turbine, the signal subspace of the observation vector autocorrelation matrix is updated through the projection approximation subspace tracking. By aligning this signal subspace with the generalized observability matrix, the identification results of the surrogate model parameters are obtained online. Furthermore, a variable weight projection approximation subspace tracking method has been proposed to enhance the algorithm robustness. Simulation and real experiment demonstrate that the surrogate model output effectively tracks the real-time changes in gas turbine measurement data. When faults and degradation arise, condition monitoring can be achieved by analyzing the evolution of model parameters to obtain feedback information from the gas turbine. The proposed method maintains its robustness in the presence of impulsive noise. These features offer a novel approach for the development of gas turbine digital twin.

A unified stiffness model of rolling lobe air spring with nonlinear structural parameters and air pressure dependence of rubber bellows

The structural parameters of the rolling lobe air spring and the mechanical characteristic of rubber bellows are the key factors affecting the stiffness and mechanical characteristic of the rolling lobe air spring. Aiming at the prediction difficulties of structural parameters of the rolling lobe air spring with the composite curved contour piston and the modeling complexity of the hysteretic mechanical characteristic of rubber bellows under variable pressure conditions, the geometrical method is applied to derive the structural parameters models of the rolling lobe air spring with the composite curved contour piston. A new pressure factor is introduced and the Coulomb frictional pressure perturbation model and the fractional derivative Kelvin-Voigt pressure perturbation model are reconstructed to accurately describe the hysteretic mechanical characteristic of rubber bellows under variable pressure conditions. A unified pressure equation is constructed to characterize the evolution of model parameters under variable pressure conditions. Furthermore, a hysteretic mechanical characteristic pressure perturbation model (abbreviated as HMCPP model) of rubber bellows under variable pressure conditions is put forward. Finally, a unified stiffness model of the rolling lobe air spring including prediction models of nonlinear structural parameters and a HMCPP model of rubber bellows is built. Taking a certain type of rolling lobe air spring as the test sample A, the structural parameters tests and static/dynamic characteristic tests of sample A are carried out based on the MTS852.05 test bench, which verified the accuracy of the unified stiffness model of the rolling lobe air spring. The research results provide theoretical support for the mechanical characteristic matching and air pressure precise control of the rolling lobe air spring under variable pressure conditions.

Degradation of artificially corroded galvanized high-strength steel wires: Corrosion morphology and mechanical behavior

• Four typical pit morphologies exist on the corroded steel wires. • Pit depth and uniform corrosion depth are in constant competition while the competition tends to stabilize when corrosion loss is more than 10%. • Ductility degradation is greater than strength and the ductility degradation is related to the ratio of pit depth to residual radius. The corrosion morphology of artificially galvanized high-strength steel wires and its effect on mechanical behavior was presented. The corrosion morphology was obtained by 3D optical scanner. Four typical pit morphologies on corroded wires were identified, namely conical pit, hemispherical pit, combined pit and secondary pit. The probabilistic density function of cross-sectional parameters under different corrosion degree were modeled by normal distribution, and the evolution of model parameters was obtained. Based on the proposed evolution model, the maximum cross-sectional area loss was estimated by ‘3 σ rule’ of normal distribution function. Subsequently, the effect of corrosion pits on the mechanical behavior of corroded wires was discussed. It shows that the ductility degradation of corroded wires is much greater than that of strength, the ductility degradation is closely related to the ratio of pit depth to residual radius. Besides, it was found that the probabilistic density function of the ratio of pit depth to uniform corrosion depth tends to stabilize when the corrosion loss is more than 10%, which indicates that the surface of corroded wire with corrosion pits is uniformly thinning.

Transfer learning in medical image segmentation: New insights from analysis of the dynamics of model parameters and learned representations

We present a critical assessment of the role of transfer learning in training fully convolutional networks (FCNs) for medical image segmentation. We first show that although transfer learning reduces the training time on the target task, improvements in segmentation accuracy are highly task/data-dependent. Large improvements are observed only when the segmentation task is more challenging and the target training data is smaller. We shed light on these observations by investigating the impact of transfer learning on the evolution of model parameters and learned representations. We observe that convolutional filters change little during training and still look random at convergence. We further show that quite accurate FCNs can be built by freezing the encoder section of the network at random values and only training the decoder section. At least for medical image segmentation, this finding challenges the common belief that the encoder section needs to learn data/task-specific representations. We examine the evolution of FCN representations to gain a deeper insight into the effects of transfer learning on the training dynamics. Our analysis shows that although FCNs trained via transfer learning learn different representations than FCNs trained with random initialization, the variability among FCNs trained via transfer learning can be as high as that among FCNs trained with random initialization. Moreover, feature reuse is not restricted to the early encoder layers; rather, it can be more significant in deeper layers. These findings offer new insights and suggest alternative ways of training FCNs for medical image segmentation.

Effects of Internal Erosion on Parameters of Subloading Cam-Clay Model

Internal erosion is widely detected in both natural deposits and earthen structures and potentially causes severe disasters. Suffusion is one of the modes of internal erosion in which fine particles in the soil are washed out along with water flow through pores formed by coarse particles. Mechanical consequences of internal erosion, specifically, suffusion, are not well investigated in term of constitutive modelling. Also, most of the present constitutive models concerning suffusion are validated by DEM simulations, not by actual soil response observed in soil tests. In this paper, triaxial seepage tests followed by drained compression on soil with 35% initial fines content under 50 kPa, 100 kPa and 200 kPa mean stresses are studied to investigate the applicability of the existing soil model to internally eroded soils. The subloading Cam-clay model is used to simulate the mechanical behaviour of eroded soils. After confirming that the model can capture key features of uneroded specimens, the evolution of model parameters with erosion is examined by back analysis of the eroded specimens. From the simulation on the eroded specimens, evolutions of the slope of normal compression line and initial stress ratio with erosion are quantified. The changes of model parameters with erosion provide a useful reference for investigating the mechanical behaviour of granular materials subjected to suffusion.

A New Neural Mass Model Driven Method and Its Application in Early Epileptic Seizure Detection.

Despite numerous neural computational models proposed to explain physiological and pathological mechanisms of brain activity, a large gap remains between theory and application of the models. Building on the successful application of data-driven methods in epileptic seizure detection, we aim to build a bridge between data and models in this paper. We first propose a novel model-driven seizure detection method based on dynamic features in epileptic EEGs, where the rationale for dynamic features in epileptic EEGs can be clarified in theory by characterizing the variation of parameters of the model. Then we apply the proposed D&F-model-driven method to the problem of early epileptic seizure detection, where the evolution of model parameters selected and optimized by the proposed method is measured and used to detect the starting point of the seizure. Numerical results on two open EEG databases demonstrate that our proposed method does a good job of early epileptic seizure detection. The average detection sensitivity, false positive rate and early detection period attain 100%, 0.1/h, and 7.1s respectively. This paper provides a strategy to characterize EEG signals using a NMM-related method and the model parameters optimized by real EEG may then serve as features in their own right for early seizure detection. An useful attempt to early detect epileptic seizures by combining the neural mass model with data analysis.

An approach to identifying micro behavior: How banks’ strategies influence financial cycles

In this paper, we show that the values of parameters of a well-calibrated model are useful in detecting micro behavior. We use a calibration procedure suitable for validating agent-based models to show how the evolution of model parameters, obtained via a rolling window estimation, illustrates the evolution of agents’ strategies in response to different economic conditions. In this regard, we calibrate the well-known financial model of Brock and Hommes using three banking indices (i.e., the S&P SmallCap 600 Financials Index, the STOXX Europe 600 Banks, and the STOXX Asia/Pacific 600 Banks) running from 1994 to 2016. The choice of a spatially and temporally diversified dataset allows us to analyze differences and similarities in the behavior of banks belonging to the different macro areas, as well as to capture agents’ reaction to the several economic phases characterizing the time series investigated.

How Banks' Strategies Infuence Financial Cycles: An Approach to Identifying Micro Behavior

In this paper, we show that the values of parameters of a well-calibrated model are useful in detecting micro behavior. We use a calibration procedure suitable for validating agent-based models to show how the evolution of model parameters, obtained via a rolling window estimation, illustrates the evolution of agents' strategies in response to different economic conditions. In this regard, we calibrate the well-known financial model of Brock and Hommes using three banking indices (i.e., the S&P SmallCap 600 Financials Index, the STOXX Europe 600 Banks, and the STOXX Asia/Pacific 600 Banks) running from 1994 to 2016. The choice of a spatially and temporally diversified dataset allows us to analyze differences and similarities in the behavior of banks belonging to the different macro areas, as well as to capture agents' reaction to the several economic phases characterizing the time series investigated.

Calibration of a productivity model for the microalgae Dunaliella salina accounting for light and temperature

This study aimed to calibrate a productivity model for the algal species Dunaliella salina accounting for the impacts of light intensity and temperature. The calibration was performed by using a dedicated experimental set-up measuring short-term oxygen production rates at different light intensities. The rate of photosynthesis was shown to follow a typical Monod function of light intensity. The slope of photosynthesis-light curves at low light intensity was also shown to be independent on temperature and the evolution of model parameters with temperature obeyed relationships consistent with previous observations in the literature. Finally, the rate of respiration was shown to follow an exponential function of temperature. This good level of agreement with prior observations in the literature indirectly validates the experimental technique used for model calibration. The resulting model should therefore yield accurate productivity predictions in outdoor cultivation systems.

A self-updating model driven by a higher-order hidden Markov chain for temperature dynamics

We develop a model for the evolution of daily average temperatures (DATs) that could benefit the analysis of weather derivatives in finance and economics as well as the modelling of time series data in meteorology, hydrology and other branches of the sciences and engineering. Our focus is to capture the mean-reverting, seasonality, memory and stochastic properties of the temperature movement and other time series exhibiting such properties. To model both mean-reversion and stochasticity, a deseasonalised component is assumed to follow an Ornstein–Uhlenbeck (OU) process modulated by a higher-order hidden Markov chain, which takes into account short/long range dependence in the data. The seasonality part is modelled through a combination of linear and sinusoidal functions with appropriate coefficients and arguments. Furthermore, we put forward a parameter estimation approach that establishes recursive higher-order hidden Markov model (HOHMM) filtering algorithms customised for the regime-switching evolution of model parameters. Quantities that are functions of HOHMC characterise these filters. Utilising the EM method in conjunction with the change of measure technique, optimal and self-updating parameter estimates are obtained. We illustrate the numerical implementation of our model and estimation technique using a 4-year Toronto DATs data set compiled by the National Climatic Data Center. We perform pertinent model selection and validation diagnostics to assess the performance of our methodology. It is shown that a 2-state HOHMM-based model best captures the empirical characteristics of the temperature data under examination on the basis of various error-based and information-criterion metrics.

Characterising Lithium-Ion Battery Degradation through the Identification and Tracking of Electrochemical Battery Model Parameters

Lithium-ion (Li-ion) batteries undergo complex electrochemical and mechanical degradation. This complexity is pronounced in applications such as electric vehicles, where highly demanding cycles of operation and varying environmental conditions lead to non-trivial interactions of ageing stress factors. This work presents the framework for an ageing diagnostic tool based on identifying and then tracking the evolution of model parameters of a fundamental electrochemistry-based battery model from non-invasive voltage/current cycling tests. In addition to understanding the underlying mechanisms for degradation, the optimisation algorithm developed in this work allows for rapid parametrisation of the pseudo-two dimensional (P2D), Doyle-Fuller-Newman, battery model. This is achieved through exploiting the embedded symbolic manipulation capabilities and global optimisation methods within MapleSim. Results are presented that highlight the significant reductions in the computational resources required for solving systems of coupled non-linear partial differential equations.

A 1‐D morphodynamic model of postglacial valley incision

AbstractChilliwack River is typical of many Cordilleran valley river systems that have undergone dramatic Holocene degradation of valley fills that built up over the course of Pleistocene glaciation. Downstream controls on base level, mainly blockage of valleys by glaciers, led to aggradation of significant glaciofluvial and glaciolacustrine valley fills and fan deposits, subsequently incised by fluvial action. Models of such large‐scale, long‐term degradation present a number of important challenges since the evolution of model parameters, such as the rate of bedload transport and grain size characteristics, are governed by the nature of the deposit. Sediment sampling in the Chilliwack Valley reveals a complex sequence of very coarse to fine textural modes. We present a 1‐D numerical morphodynamic model for the river‐floodplain system tailored to conditions in the valley. The model is adapted to dynamically adjust channel width to optimize sediment transporting capacity and to integrate relict valley fill material as the channel incises through valley deposits. Sensitivity to model parameters is studied using four principal criteria: profile concavity, rate of downstream grain size fining, bed surface sand content, and the timescale to equilibrium. Model results indicate that rates of abrasion and coarsening of the grain size distributions exert the strongest controls on all of the interrelated model performance criteria. While there are a number of difficulties in satisfying all model criteria simultaneously, results indicate that 1‐D models of valley bottom sedimentary systems can provide a suitable framework for integrating results from sediment budget studies and chronologies of sediment evacuation established from dating.

Forecasting the Real Prices of Crude Oil: A Dynamic Model Averaging Approach

Forecasting oil prices has been of great interests for macroeconomists in the recent years. Our article contributes to this strand of the literature by using a dynamic model averaging (DMA) method to improve forecasting accuracy of real oil prices. The advantage of DMA is that the method combines models in a dynamic way using two forgetting factors to approximate the evolution of model parameters and model switching probabilities, respectively. Our empirical results show that DMA generates more accurate forecasts than the no-change forecasts at the relatively longer horizons. At a horizon of 12 months, the reduction of mean squared prediction error is as high as 30% and the accuracy of directional forecasts increases as high as 71%. It is also found that DMA performs better than Bayesian model averaging, the commonly-used mean combination of forecasts, and more sophisticated individual models such as a time-varying dimension model for the horizons of 3 and 12 months.

Modelling the liquidity premium on corporate bonds

AbstractThe liquidity premium on corporate bonds has been high on the agenda of Solvency regulators owing to its potential relationship to an additional discount factor on long-dated insurance liabilities. We analyse components of the credit spread as a function of standard bond characteristics during 2003–2014 on a daily basis by regression analyses, after introducing a new liquidity proxy. We derive daily distributions of illiquidity contributions to the credit spread at the individual bond level and find that liquidity premia were close to zero just before the financial crisis. We observe the time-varying nature of liquidity premia as well as a widening in the daily distribution in the years after the credit crunch. We find evidence to support higher liquidity premia, on average, on bonds of lower credit quality. The evolution of model parameters is economically intuitive and brings additional insight into investors’ behaviour. The frequent and bond-level estimation of liquidity premia, combined with few data restrictions makes the approach suitable for ALM modelling, especially when future work is directed towards arriving at forward-looking estimates at both the aggregate and bond-specific level.

Investigation of Long Time Beef and Veal Meat Behavior by Bioimpedance Spectroscopy for Meat Monitoring

Maturation state of meat influences its fibers structure and conductivity and leads to observable changes of its complex impedance. The relationship between the postmortem state and the change of meat impedance has been characterized to serve as a basis for developing an in vitro measurement method for the freshness of meat. Experimental measurements were carried out for long period of time of 14 days with circular penetrating multielectrodes on beef and veal muscles. The results have been evaluated by means of a modified Fricke model, which considers distributed electric parameters in the biological tissue and leads therefore to a good fitting performance. Beef meat shows a higher extracellular resistance and a higher membrane capacitance than veal due to the fact that beef muscle fibers are thicker and less juicy than veal muscle fibers. The evolution of model parameters is correlated with the expected behavior with aging. The parameter R e that represents extracellular fluids decreases considerably from day 6 to day 14 about 7%-25% depending on the muscle type. The capacitive element C that corresponds to the insulating cell membranes decreases with time from day 6 to day 14 about 5%-17% depending on the muscle type. The parameter R i that represents intracellular fluids increases from day 6 to day 14 about 3%-12% depending on the muscle type.

Parameter Trajectory Analysis to Identify Treatment Effects of Pharmacological Interventions

The field of medical systems biology aims to advance understanding of molecular mechanisms that drive disease progression and to translate this knowledge into therapies to effectively treat diseases. A challenging task is the investigation of long-term effects of a (pharmacological) treatment, to establish its applicability and to identify potential side effects. We present a new modeling approach, called Analysis of Dynamic Adaptations in Parameter Trajectories (ADAPT), to analyze the long-term effects of a pharmacological intervention. A concept of time-dependent evolution of model parameters is introduced to study the dynamics of molecular adaptations. The progression of these adaptations is predicted by identifying necessary dynamic changes in the model parameters to describe the transition between experimental data obtained during different stages of the treatment. The trajectories provide insight in the affected underlying biological systems and identify the molecular events that should be studied in more detail to unravel the mechanistic basis of treatment outcome. Modulating effects caused by interactions with the proteome and transcriptome levels, which are often less well understood, can be captured by the time-dependent descriptions of the parameters. ADAPT was employed to identify metabolic adaptations induced upon pharmacological activation of the liver X receptor (LXR), a potential drug target to treat or prevent atherosclerosis. The trajectories were investigated to study the cascade of adaptations. This provided a counter-intuitive insight concerning the function of scavenger receptor class B1 (SR-B1), a receptor that facilitates the hepatic uptake of cholesterol. Although activation of LXR promotes cholesterol efflux and -excretion, our computational analysis showed that the hepatic capacity to clear cholesterol was reduced upon prolonged treatment. This prediction was confirmed experimentally by immunoblotting measurements of SR-B1 in hepatic membranes. Next to the identification of potential unwanted side effects, we demonstrate how ADAPT can be used to design new target interventions to prevent these.

Applications of analysis of dynamic adaptations in parameter trajectories

Metabolic profiling in combination with pathway-based analyses and computational modelling are becoming increasingly important in clinical and preclinical research. Modelling multi-factorial, progressive diseases requires the integration of molecular data at the metabolome, proteome and transcriptome levels. Also the dynamic interaction of organs and tissues needs to be considered. The processes involved cover time scales that are several orders of magnitude different. We report applications of a computational approach to bridge the scales and different levels of biological detail. Analysis of dynamic adaptations in parameter trajectories (ADAPTs) aims to investigate phenotype transitions during disease development and after a therapeutic intervention. ADAPT is based on a time-dependent evolution of model parameters to describe the dynamics of metabolic adaptations. The progression of metabolic adaptations is predicted by identifying necessary dynamic changes in the model parameters to describe the transition between experimental data obtained during different stages. To get a better understanding of the concept, the ADAPT approach is illustrated in a theoretical study. Its application in research on progressive changes in lipoprotein metabolism is also discussed.

ACG: rapid inference of population history from recombining nucleotide sequences

BackgroundReconstruction of population history from genetic data often requires Monte Carlo integration over the genealogy of the samples. Among tools that perform such computations, few are able to consider genetic histories including recombination events, precluding their use on most alignments of nuclear DNA. Explicit consideration of recombinations requires modeling the history of the sequences with an Ancestral Recombination Graph (ARG) in place of a simple tree, which presents significant computational challenges.ResultsACG is an extensible desktop application that uses a Bayesian Markov chain Monte Carlo procedure to estimate the posterior likelihood of an evolutionary model conditional on an alignment of genetic data. The ancestry of the sequences is represented by an ARG, which is estimated from the data with other model parameters. Importantly, ACG computes the full, Felsenstein likelihood of the ARG, not a pairwise or composite likelihood. Several strategies are used to speed computations, and ACG is roughly 100x faster than a similar, recombination-aware program.ConclusionsModeling the ancestry of the sequences with an ARG allows ACG to estimate the evolutionary history of recombining nucleotide sequences. ACG can accurately estimate the posterior distribution of population parameters such as the (scaled) population size and recombination rate, as well as many aspects of the recombinant history, including the positions of recombination breakpoints, the distribution of time to most recent common ancestor along the sequence, and the non-recombining trees at individual sites. Multiple substitution models and population size models are provided. ACG also provides a richly informative graphical interface that allows users to view the evolution of model parameters and likelihoods in real time.

Study on the Relationship between Parameters of Deterioration Creep Model of Rock under Different Modeling Assumptions

Based on the analysis to the characteristics of two kinds of rock steady creep curves, the nBurgers model that can describe the uniaxial creep properties of rock is presented. Considering the external influences on rock deformation as instantaneous elastic deterioration defined by the reduction of instantaneous elastic modulus and long-term creep deterioration defined by the reduction of long-term creep modulus, kBurgers model with deterioration is presented too. On this basis, three-dimensional model considering the Poisson’s ratio and Bulk modulus as the constant are established separately, and the relationship between their parameters is obtained. Finally, According to a calculation example, the deterioration evolution of model parameters considering water content changing under the assumption of constant bulk modulus is analyzed. In addition, the deterioration evolution equations of instantaneous elastic and long-term creep under the assumption of constant bulk modulus are obtained. It is verified that conversion relationships of model parameters are reasonable.

Thermal aging model of InP/InGaAs/InP DHBT

This paper presents the measurement result and modeling of the storage accelerated aging tests performed on InP/InGaAs/InP DHBT. From the Gummel characteristics, we observe that the principle mode of device degradation results from the increase of base current and reduction in the current gain which comes from the base–emitter junction periphery. Topics covered include: (1) underlying physical mechanism of base current degradation; (2) choosing HICUM model LEVEL2 for the modeling purpose; (3) evolution of model parameters with stress time after the extraction of model parameter before aging and the description of the parameter drift with suitable equation; (4) implementation in compact electrical model allows to simulate the impact of device failure mechanisms on the circuit in operating conditions.