Model Structure Determination Research Articles

Evaluation Model of Teachers' Teaching Ability Based on Improved Random Forest with Grey Relation Projection

The development and improvement of teachers' teaching ability is the guarantee to continuously improve the overall teaching ability and school running level of universities. Therefore, it is of great significance to scientifically evaluate teachers' teaching ability. Aiming at the problems of traditional algorithms, such as weak generalization performance and difficult determination of parameters and model structure, a random forest algorithm is introduced into the field of teaching ability prediction. At an equal time, the ordinary grey correlation algorithm is accelerated through the usage of the projection principle, and instructor instructing capacity assessment mannequin based totally on grey projection expanded random woodland algorithm is proposed. The grey correlation degree judgment matrix is used to represent the correlation between historical samples and influencing factors, and the weight of influencing factors is established by the direct weight method to weight the judgment matrix. The random woodland algorithm is used to set up the prediction model, the grey projection is used to display screen the pattern set education model, and ultimately, the characteristic vector is entered to whole the prediction. The experimental results show that the new method has high prediction accuracy, robustness, and effectiveness. It not only enriches the evaluation methods of teachers' teaching ability but also provides a quantitative evaluation model reference for teachers' teaching ability evaluation.

Development of High-Strength Concrete Mixed Design System Using Artificial Intelligence

The quality inspection of high-strength concrete construction sites consists of a compressive strength test that is considered the most important, but this can be confirmed through a compressive strength test after 28 days of high-strength concrete application. Therefore, it is of paramount importance to ship high-quality products to ready-mixed concrete factories by increasing the reliability of the mixed design that affects high-strength concrete production. In addition, there is a need to develop an efficient management system for mixed design that determines high-strength concrete quality by measuring the mixing ratio of materials in the ready-mixed concrete factory production stage. This study used matrix laboratory(MATLAB) using Deep learning, a language that performs mathematics and engineering calculations based on matrices, and presented a mixed design model by adjusting the strength through input and output variables, learning data collection, model structure determination, learning error, and repetition results. The predicted mean value of 40 MPa was measured at 40.75 MPa, showing a difference of 0.75 MPa and 40 MPa, and the error rate was confirmed to be 4.13%. And the predicted mean value of 55 MPa was measured as 55.55 MPa, showing a difference between 55 MPa and 0.55 MPa, and the error rate was confirmed to be 1.73%. Through this study, the reliability of high-strength concrete quality management is secured by applying a high-strength concrete mixed design system using artificial intelligence(AI) and adjusting it in connection with all fields of the production process.

Hindcasting and Forecasting Total Suspended Sediment Concentrations Using a NARX Neural Network

Estimating and forecasting suspended sediments concentrations in streams constitutes a valuable asset for sustainable land management. This research presents the development of a non-linear autoregressive exogenous neural network (NARX) for forecasting sediment concentrations at the exit of Francia Creek watershed (Valparaiso, Chile). Details are presented on input data selection, data splitting, selection of model architecture, determination of model structure, NARX training (optimization of model parameters), and model validation (hindcasting and forecasting). The study explored if the developed artificial neural network model is valid for forecasting daily suspended sediment concentrations for a complete year, capturing seasonal trends, and maximum and baseflow concentrations. Francia Creek watershed covers approximately 3.24 km2. Land cover within the catchment consists mainly of native and exotic vegetation, eroded soil, and urban areas. Input data consisting of precipitation and stream flow time-series were fed to a NARX network for forecasting daily suspended sediments (SST) concentrations for years 2013–2014, and hindcasting for years 2008–2010. Training of the network was performed with daily SST, precipitation, and flow data from years 2012 and 2013. The resulting NARX net consisted of an open-loop, 12-node hidden layer, 100 iterations, using Bayesian regularization backpropagation. Hindcasting of daily and monthly SST concentrations for years 2008 through 2010 was successful. Daily SST concentrations for years 2013 and 2014 were forecasted successfully for baseflow conditions (R2 = 0.73, NS = 0.71, and Kling-Gupta efficiency index (K-G) = 0.84). Forecasting daily SST concentrations for year 2014 was within acceptable statistical fit and error margins (R2 = 0.53, NS = 0.47, K-G = 0.60, d = 0.82). Forecasting of monthly maximum SST concentrations for the two-year period (2013 and 2014) was also successful (R2 = 0.69, NS = 0.60, K-G = 0.54, d = 0.84).

Identification of dynamic models for a wave energy converter from experimental data

This paper addresses the issue of hydrodynamic model identification from recorded tank test data, for a prototype wave energy device. The study focusses on nonlinear Kolmogorov–Gabor polynomial models, with linear models also used as a baseline reference. Six different experimental data sets are employed for model identification and validation, all derived from a JONSWAP input sea state. Compared to identification on numerical data, this study shows that the determination of model structure and orders is not so straightforward, but that consistent and useful computationally efficient models can be obtained. For the particular tests undertaken, in which the prototype device generally behaves as a wave follower, the nonlinear models only show very marginal performance improvement over the linear ones.

Determination of Model Structure from Flight Test with Generalized Additive Models

The determination of an appropriate structure for an aerodynamic model developed from flight-test data can be challenging, particularly if a physical interpretation of the model is required. This paper introduces generalized additive models (GAMs) as an efficient and effective means to this end. GAMs provide an intuitive method to determine the structure by allowing the various aerodynamic effects to be separated into an additive model. Individual effects can be modeled by a number of nonlinear smooth functions. The suitability of this method is demonstrated on the example of a simple pitching moment coefficient model that is developed using flight-test data. The ability to separate and visualize effects together with a number of statistical indicators, the relatively low computational costs, and the availability of efficient software packages make GAMs attractive for practical applications. For the work presented in this paper, the “mgcv” package of the open-source statistical software R is used.

Identification of systems with slowly sampled outputs using LPV model

Identification of systems with slowly sampled output is studied. A linear parameter varying (LPV) model with multi-model structure is used to solve the problem. The output error (OE) method is used to estimate model parameters. Firstly, the local models and weighting functions are estimated separately using optimization methods. Then, a relaxation iteration method is developed to refine the parameters of the total model. For LPV model structure determination, an engineering approach is proposed that combines process knowledge with the so-called final output error criteria (FOE). The method is verified using both simulation data and industrial data. In the industrial case study, the LPV models give more accurate prediction of product qualities than that of a linear dynamic model and that of a static nonlinear model; the result also indicates the necessity of using test signals in soft-sensor development.

On the application of interval PCA to process monitoring: A robust strategy for sensor FDI with new efficient control statistics

In principal component analysis (PCA) based fault detection and isolation (FDI), the sensor data uncertainties cause a significant difficulty in control decision making, which evokes and increases the number of false alarms and imprecise decisions. In its standard form, PCA makes no distinction between data points and the associated measurement errors which vary depending on experimental conditions. A recent and robust solution consists in capturing the variability of the multivariate observations by symbolic data analysis methods, precisely, using interval-valued variables and interval PCA methods. The key idea is to extend the methodology of conventional PCA based statistical process monitoring to handle interval-valued data. In this paper, we compare four most known interval PCA methods and investigate their use for diagnosis purpose. Based on reconstruction principle used in the classical PCA approach, an interval reconstruction is proposed and a new criterion is derived for the determination of the interval PCA model structure (number of retained principal components). The monitoring routine includes the generation of interval residuals for fault detection, and the application of the extended interval reconstruction principle for fault isolation. We also introduce a new set of Shewhart type interval statistics, based on the conventional PCA statistics, that are more suited to interval data-set case with a better overall performances. The implementation of the proposed sensor FDI methods are illustrated using a simulation example and applied to distillation column process benchmark.

Overcoming the lack of kinetic information in biochemical reactions networks

A main aspect in computational modelling of biological systems is the determination of model structure and model parameters. Due to economical and technical reasons, only part of these details are well characterized, while the rest are unknown. To deal with this difficulty, many reverse engineering and parameter estimation methods have been proposed in the literature, however these methods often need an amount of experimental data not always available. In this paper we propose an alternative approach, which overcomes model indetermination solving an Optimization Problem (OP) with an objective function that, similarly to Flux Balance Analysis, is derived from an empirical biological knowledge and does not require large amounts of data. The system behaviour is described by a set of Ordinary Differential Equations (ODE). Model indetermination is resolved selecting time-varying coefficients that maximize/ minimize the objective function at each ODE integration step. Moreover, to facilitate the modelling phase we provide a graphical formalism, based on Petri Nets, which can be used to derive the corresponding ODEs and OP. Finally, the approach is illustrated on a case study focused on cancer metabolism.

The algorithms of structural identification of static processes with an expert in the regression analysis

The method of mathematical modeling of processes (systems) based on passive surveillance of its indicators and their relationship has been considered in the article. Simulation has been considered as one of the tools for the researching of the process structure when the process is represented by a number of inter-related indicators. The aim of the research is to develop methods of mathematical modeling of complex processes (systems) simplifies the task of selecting predictors for parametric identification model. Proposed method of structural modeling of static processes is based on the concept of the graph coherency. The algorithms of model structure forming has been offered. The modeling procedure has been performed with involvement of expert. The features of the model structure determination has been taken into account. The set of process indicators has been divided into two disjoint non-empty subsets − a subset of the independent parameters − predictors and output indicators − functions. The graph of relations parameters of the process has been built on the graph of correlation structure parameters in accordance with algorithms. Has been considered the case where the introduced parameters of quality has been represented by multivariate regression equation and its ultimate best value should correspond to the lowest value. Novelty. The relation of preference has been clarified in the problem of determining the mathematical model structure using correlation and regression analysis with demand positive indicators. Algorithms form the predictor parameters with expert has been developed. Significance. The approach and the algorithms proposed in the paper, allow the use of mathematical modeling methods of complex systems in the context of structural uncertainty.

Simultaneous structure and parameter identification of multivariate systems by matrix decomposition

Structure determination and parameter identification of multivariate systems are crucial but rather difficult issues in system identification. Due to the explosive growth of process data along with the scale increase of industrial processes, directional links between variables of such complex processes are often undistinguishable, which is indispensable to model structure determination but is often assumed to be known beforehand in most identification methods. In this article, a new modelling approach is developed to simultaneously estimate the model parameters and structures (including model orders as well as the directional links between different process variables) of multivariate systems. A vector auto-regressive (VAR) form is utilized as the model formulation in this algorithm. The key technique lies in constructing an interleaved information matrix with respect to a multiple model structure formulated for the VAR representation. Then by utilizing the upper diagonal factorization, all the parameter estimates of all path models with orders from zero to m, as well as the corresponding cost function values, can be obtained simultaneously. The effectiveness of the proposed method is demonstrated via a numerical example and a distillation column system.

Investigation Progresses and Applications of Fractional Derivative Model in Geotechnical Engineering

Over the past couple of decades, as a new mathematical tool for addressing a number of tough problems, fractional calculus has been gaining a continually increasing interest in diverse scientific fields, including geotechnical engineering due primarily to geotechnical rheology phenomenon. Unlike the classical constitutive models in which simulation analysis gradually fails to meet the reasonable accuracy of requirement, the fractional derivative models have shown the merits of hereditary phenomena with long memory. Additionally, it is traced that the fractional derivative model is one of the most effective and accurate approaches to describe the rheology phenomenon. In relation to this, an overview aimed first at model structure and parameter determination in combination with application cases based on fractional calculus was provided. Furthermore, this review paper shed light on the practical application aspects of deformation analysis of circular tunnel, rheological settlement of subgrade, and relevant loess researches subjected to the achievements acquired in geotechnical engineering. Finally, concluding remarks and important future investigation directions were pointed out.

Frequency-Domain Method for Automated Simulation Updates Based on Flight Data

A statistically based frequency-domain method for using flight data to update aerodynamic data tables used in flight simulators is developed and demonstrated. The capability to accurately and automatically update a simulation to incorporate nonlinear and unsteady aerodynamic effects embodied in the flight data is demonstrated. Useful features of this novel frequency-domain approach are highlighted, including computational efficiency, noise rejection, automated model structure determination in the frequency domain for nonlinear and unsteady aerodynamics, and the ability to generate simulation updates without air flow angle measurements in the flight data. An F-16 nonlinear simulation is used to demonstrate the technique.

Unsteady aerodynamic modeling at high angles of attack using support vector machines

Accurate aerodynamic models are the basis of flight simulation and control law design. Mathematically modeling unsteady aerodynamics at high angles of attack bears great difficulties in model structure determination and parameter estimation due to little understanding of the flow mechanism. Support vector machines (SVMs) based on statistical learning theory provide a novel tool for nonlinear system modeling. The work presented here examines the feasibility of applying SVMs to high angle-of-attack unsteady aerodynamic modeling field. Mainly, after a review of SVMs, several issues associated with unsteady aerodynamic modeling by use of SVMs are discussed in detail, such as selection of input variables, selection of output variables and determination of SVM parameters. The least squares SVM (LS-SVM) models are set up from certain dynamic wind tunnel test data of a delta wing and an aircraft configuration, and then used to predict the aerodynamic responses in other tests. The predictions are in good agreement with the test data, which indicates the satisfying learning and generalization performance of LS-SVMs.

Real-Time Monitoring Technology of Plaster-Based Three-Dimensional Printing System for Quality Determination of Model Structure

Real-Time Monitoring Technology of Plaster-Based Three-Dimensional Printing System for Quality Determination of Model Structure

Feature extraction and classification of load dynamic characteristics based on lifting wavelet packet transform in power system load modeling

Load dynamic characteristics classification and synthesis is the main approach to solve the problem of load time-variation. The basis and prerequisite of load dynamic characteristics classification is load dynamic characteristics feature extraction. Load model parameter space or the model response space gained by a standard voltage excitation is usually selected as the feature vector space in current practice of load dynamic characteristics feature extraction. However, both methods need to determine the load model structure and identify the model parameters. It would increase not only calculation error but also calculation time in the process of load model structure determination and parameter identification. Then the accuracy of the final classification results would be affected. It is reasonable and scientific to extract feature vector space of load dynamic characteristics directly from the measured response space. In this paper, a feature extraction method based on lifting wavelet packet transform is proposed for load dynamic characteristics classification. The load measured current response data is decomposed and reconstructed, then the wavelet packet coefficients can be extracted to construct energy moment based feature vector. On this basis, the load dynamic characteristics classification can be realized using fuzzy c-means (FCM) method. Finally, the validity and practicality of the proposed method have been proved by feature extraction and classification of dynamic simulation data acquired using Matlab/Simulink and field measurement data. Compared with traditional wavelet packet transform, the lifting wavelet packet transform has shown advantages both in computational speed and reconstruction error and can improve the accuracy of load dynamic characteristics classification.

Model structure selection in single-index-coefficient regression models

Single-index-coefficient regression models (SICRM) have been proposed and used in the literature for avoiding the “curse of dimensionality”. However, there is no efficient model structure determination methodology for the SICRM. This may cause a tendency to use models that are much larger than required. In this paper, we propose a new procedure for model structure determination in the SICRM; that is, the penalized estimating equations (PEE) for variable selection that combines the “delete-one-component” method and the smoothly clipped absolute deviation penalty. The proposed PEE method can simultaneously identify significant variables of the index and estimate the nonzero coefficients of the index parameters. We also further study testing for nonparametric index-coefficient functions. Asymptotic properties for the proposed estimation procedure have been established. Under the appropriate conditions, we demonstrate that the proposed estimators have the oracle properties. Monte Carlo simulation studies are conducted to assess the finite sample performance of the proposed methods. A real example is analyzed as an illustration.

Model Estimation and Classification Via Model Structure Determination

In model estimation, we often face problems with unknown parameters in the candidate models. This paper proposes the model structure determination (MSD) for model estimation with unknown parameters. We start with the problem of model order selection and decompose the probability density function (PDF) into the information provided by the data about the model parameters and that of the model structure. The factor that depends on the model parameters is approximated using a minimax procedure, and the MSD depends on the model structure only. It is shown that the MSD is equivalent to the exponentially embedded family (EEF) for model order selection under some conditions. Finally, we apply the MSD to a classification problem where we have partial knowledge about the parameters, and simulation results show that it outperforms the pseudo-maximum-likelihood (pseudo-ML) rule.

A General Nonlinear System Identification Method Based upon Time-Varying Trend of Instantaneous Frequencies and Amplitudes

Nonlinear phenomena are widely encountered in practical applications. The presence of nonlinearity makes a system exhibit different dynamic behaviours from its linear counterpart. A typical example is that free-vibration frequencies of nonlinear systems are amplitude-dependent, resulting in the variation of free-vibration frequencies with time. Therefore, Fourier spectrum which has been widely used in linear systems cannot completely represent dynamic characteristics of nonlinear systems. On the contrary, the time-varying trend of instantaneous frequencies and vibration amplitudes can completely capture dynamic characteristics of nonlinear systems. In this study, a general system identification method based on curving-fitting the time-varying trend of instantaneous frequencies and amplitudes is proposed for nonlinear systems. Herein Hilbert transform is employed to extract instantaneous frequencies and amplitudes from the measured responses. By taking the relationships between instantaneous vibration characteristics and structural physical parameters as regression models, structural physical parameters can be estimated by a least-squares estimation or an optimization method. The combination of the instantaneous information in time domain and frequency domain leads to high accuracy of identification results. The determination of model structure of nonlinearities is solved by assuming a generalized model which involves as many types of nonlinearities as possible. The proposed method has been demonstrated by numerical simulations.

Application of aerodynamic model structure determination to UAV data

AbstractThis paper concerns aircraft system identification and, in particular, the process of aerodynamic model structure determination. Its application to experimental data from unmanned aerial vehicles (UAVs) is also described. The procedure can be particularly useful for determining an aerodynamic model for aircraft with unconventional airframe configurations, which some unmanned aircraft tend to have. Two model structure determination techniques are outlined. The first is the well-established stepwise regression method, while the second is an adaptation of an existing frequency response approach which instead utilises maximum likelihood estimation. Example applications of the methods are presented for two data sources. The first is a set of UAV flight test data and the second is data recorded from dynamic wind tunnel tests on a UAV configuration. For both examples, the model structures determined using stepwise regression and maximum likelihood analysis matched one another, suggesting that the maximum likelihood approach and the chosen thresholds for its statistical metrics were reliable for the data being analysed.

Gene expression model (in)validation by Fourier analysis

BackgroundThe determination of the right model structure describing a gene regulation network and the identification of its parameters are major goals in systems biology. The task is often hampered by the lack of relevant experimental data with sufficiently low noise level, but the subset of genes whose concentration levels exhibit an oscillatory behavior in time can readily be analyzed on the basis of their Fourier spectrum, known to turn complex signals into few relatively noise-free parameters. Such genes therefore offer opportunities of understanding gene regulation quantitatively.ResultsFourier analysis is applied to data on gene expression levels in mouse liver cells that oscillate according to the circadian rhythm. Several model structures in the form of linear and nonlinear differential equations are matched to the data and it is shown that although the considered models can reproduce many features of the oscillatory patterns, some can be excluded on the basis of Fourier analysis without appeal to prior knowledge of regulatory pathways. A systematic method for testing models is also proposed based on measuring the effects of variations in gene copy-number on the expression levels of coupled genes.ConclusionsFourier analysis is a technique that is well-adapted to the study of biological oscillators and can be used instead or in addition to conventional modeling techniques. Its usefulness will increase as more high-resolution data become available.