High performance concrete compressive strength forecasting using ensemble models based on discrete wavelet transform

This paper investigates the use of wavelet ensemble models for high performance concrete (HPC) compressive strength forecasting. More specifically, we incorporate bagging and gradient boosting methods in building artificial neural networks (ANN) ensembles (bagged artificial neural networks (BANN) and gradient boosted artificial neural networks (GBANN)), first. Coefficient of determination (R²), mean absolute error (MAE) and the root mean squared error (RMSE) statics are used for performance evaluation of proposed predictive models. Empirical results show that ensemble models (R²_BANN=0.9278, R²_GBANN=0.9270) are superior to a conventional ANN model (R²_ANN=0.9088). Then, we use the coupling of discrete wavelet transform (DWT) and ANN ensembles for enhancing the prediction accuracy. The study concludes that DWT is an effective tool for increasing the accuracy of the ANN ensembles (R²_WBANN=0.9397, R²_WGBANN=0.9528).     

Ensemble data mining modeling in corrosion of concrete sewer: A comparative study of network-based (MLPNN & RBFNN) and tree-based (RF,CHAID, & CART) models

This research aims to evaluate ensemble learning (bagging, boosting, and modified bagging) potential in predicting microbially induced concrete corrosion in sewer systems from the data mining (DM) perspective. Particular focus is laid on ensemble techniques for network-based DM methods, including multi-layer perceptron neural network (MLPNN) and radial basis function neural network (RBFNN) as well as tree-based DM methods, such as chi-square automatic interaction detector (CHAID), classification and regression tree (CART), and random forests (RF). Hence, an interdisciplinary approach is presented by combining findings from material sciences and hydrochemistry as well as data mining analyses to predict concrete corrosion. The effective factors on concrete corrosion such as time, gas temperature, gas-phase H₂S concentration, relative humidity, pH, and exposure phase are considered as the models’ inputs. All 433 datasets are randomly selected to construct an individual model and twenty component models of boosting, bagging, and modified bagging based on training, validating, and testing for each DM base learners. Considering some model performance indices, (e.g., Root mean square error, RMSE; mean absolute percentage error, MAPE; correlation coefficient, r) the best ensemble predictive models are selected. The results obtained indicate that the prediction ability of the random forests DM model is superior to the other ensemble learners, followed by the ensemble Bag-CHAID method. On average, the ensemble tree-based models acted better than the ensemble network-based models; nevertheless, it was also found that taking the advantages of ensemble learning would enhance the general performance of individual DM models by more than 10%.     

Assessing neural networks with wavelet denoising and regression models in predicting diel dynamics of eddy covariance-measured latent and sensible heat fluxes and evapotranspiration

Eddy covariance (EC)-measured data were used to develop multiple nonlinear regression (MNLR) models of latent (LE) and sensible heat (H _s) fluxes, and micrometeorological station-measured actual evapotranspiration (ET). Discrete wavelet transform (DWT) with symmlets (sym10), coiflets (coif10), and daubechies (db10) was used to decompose time series signals of LE, H _s, and ET into frequency components in order to feed denoised output data into 26 artificial neural networks (ANNs) with different learning algorithms, based on independent validation-derived values of coefficient of determination (r ²), root mean square error (RMSE), mean absolute error (MAE), wavelet neural networks (WNNs) with coif10-1 and db10-1 outperformed ANNs, and MNLR models. The best ones out of 26 WNNs appeared to be multilayer perceptrons (MLPs) for LE and H _s, and time-delay network (TDNN) for ET, while the best ones out of 26 ANNs were determined as TDNN for LE, MLP for H _s, and generalized feedforward network (GFF) for ET. The combination of batch mode and Levenberg–Marquardt algorithm was adopted in the ANNs and WNNs more frequently and generated better accuracy metrics than the combinations of online mode and Momentum algorithm, and batch mode and Momentum algorithm.     

A spatio-temporal prediction of NDVI based on precipitation: an application for grazing management in the arid and semi-arid grasslands

ABSTRACT

A method for predicting the dynamic spatio-temporal variations of the normalized difference vegetation index (NDVI) based on precipitation is proposed using combined nonlinear autoregressive with exogenous input (NARX) networks and artificial neural networks (ANNs). The proposed method is validated by applying to predict the spatio-temporal NDVI for the Hulunbuir grassland located in Inner Mongolia, China. The results show the good predictive ability for the spatio-temporal variations of NDVI with the mean absolute percentage error of 11.59%, mean absolute error of 7.11 × 10^?2 and root mean square error of 8.06 × 10^?2, respectively. The approach presented in the paper can be further used as the guidance to reduce the occurrence of overgrazing in the arid and semi-arid grasslands.     

Modeling NOx emissions from coal-fired utility boilers using support vector regression with ant colony optimization

Modeling NO_x emissions from coal fired utility boiler is critical to develop a predictive emissions monitoring system (PEMS) and to implement combustion optimization software package for low NO_x combustion. This paper presents an efficient NO_x emissions model based on support vector regression (SVR), and compares its performance with traditional modeling techniques, i.e., back propagation (BPNN) and generalized regression (GRNN) neural networks. A large number of NO_x emissions data from an actual power plant, was employed to train and validate the SVR model as well as two neural networks models. Moreover, an ant colony optimization (ACO) based technique was proposed to select the generalization parameter C and Gaussian kernel parameter γ. The focus is on the predictive accuracy and time response characteristics of the SVR model. Results show that ACO optimization algorithm can automatically obtain the optimal parameters, C and γ, of the SVR model with very high predictive accuracy. The predicted NO_x emissions from the SVR model, by comparing with the BPNN model, were in good agreement with those measured, and were comparable to those estimated from the GRNN model. Time response of establishing the optimum SVR model was in scale of minutes, which is suitable for on-line and real-time modeling NO_x emissions from coal-fired utility boilers.     

Error measures for fuzzy linear regression: Monte Carlo simulation approach

The focus of this study is to use Monte Carlo method in fuzzy linear regression. The purpose of the study is to figure out the appropriate error measures for the estimation of fuzzy linear regression model parameters with Monte Carlo method. Since model parameters are estimated without any mathematical programming or heavy fuzzy arithmetic operations in fuzzy linear regression with Monte Carlo method. In the literature, only two error measures (E₁ and E₂) are available for the estimation of fuzzy linear regression model parameters. Additionally, accuracy of available error measures under the Monte Carlo procedure has not been evaluated. In this article, mean square error, mean percentage error, mean absolute percentage error, and symmetric mean absolute percentage error are proposed for the estimation of fuzzy linear regression model parameters with Monte Carlo method. Moreover, estimation accuracies of existing and proposed error measures are explored. Error measures are compared to each other in terms of estimation accuracy; hence, this study demonstrates that the best error measures to estimate fuzzy linear regression model parameters with Monte Carlo method are proved to be E₁, E₂, and the mean square error. One the other hand, the worst one can be given as the mean percentage error. These results would be useful to enrich the studies that have already focused on fuzzy linear regression models.     

Integrating imaging spectroscopy and neural networks to map grass quality in the Kruger National Park, South Africa

A new integrated approach, involving continuum-removed absorption features, the red edge position and neural networks, is developed and applied to map grass nitrogen concentration in an African savanna rangeland. Nitrogen, which largely determines the nutritional quality of grasslands, is commonly the most limiting nutrient for grazers. Therefore, the remote sensing of foliar nitrogen concentration in savanna rangelands is important for an improved understanding of the distribution and feeding patterns of wildlife. Continuum removal was applied on two absorption features located in the visible (R_550-757) and the SWIR (R_2015-2199) from an atmospherically corrected HYMAP MKI image. A feature selection algorithm was used to select wavelength variables from the absorption features. Selected band depths from the absorption features as well as the red edge position (REP) were input into a backpropagation neural network. The best-trained neural network was used to map nitrogen concentration over the whole study area. Results indicate that the new integrated approach could explain 60% of the variation in savanna grass nitrogen concentration on an independent test data set, with a root mean square error (rmse) of 0.13 (±8.30% of the mean observed nitrogen concentration). This result is better compared to the result obtained using multiple linear regression, which yielded an R² of 38%, with a RMSE of 0.16 (±10.30% of the mean observed nitrogen concentration) on an independent test data set. The study demonstrates the potential of airborne hyperspectral data and neural networks to estimate and ultimately to map nitrogen concentration in the mixed species environments of Southern Africa.     

The dielectric properties prediction of the vegetation depending on the moisture content using the deep neural network model

In this paper, dielectric properties of citrus leaves are predicted with long short‐term memory (LSTM) which is one of the well‐known deep neural network (DNN) models and real‐time measurements for any moisture content (MC) values in the range of 4.90 to 7.05 GHz at a fixed temperature of 24°C for microwave applications, as a novelty. Firstly, S‐parameters of samples are measured with WR‐159 waveguide and Waveguide Transmission Line Method. In addition, the MCs of samples depending on their weights are calculated. Thus, the dataset depending on various MC and frequency is obtained with the measurement results to both training and testing the DNN model. Secondly, a total of 4000 datasets are obtained, 80% of which is used for training, and 20% for testing. The proposed DNN model consists of four inputs (f, MC, S₁₁, and S₂₁) and two outputs (ε′ and ε″). Finally, the dielectric parameters for the desired MC and f are displayed with the graphical user interface in real‐time. Success criteria for the prediction such as mean absolute error, root mean squared error, mean absolute percentage error, and R‐squared are calculated. The results indicated that there is good agreement between the measured and predicted ones. R‐squared are calculated as 0.962 and 0.968 for ε′ and ε″, respectively.     

Neural predictive monitoring and a comparison of frequentist and Bayesian approaches

Neural state classification (NSC) is a recently proposed method for runtime predictive monitoring of hybrid automata (HA) using deep neural networks (DNNs). NSC trains a DNN as an approximate reachability predictor that labels an HA state x as positive if an unsafe state is reachable from x within a given time bound, and labels x as negative otherwise. NSC predictors have very high accuracy, yet are prone to prediction errors that can negatively impact reliability. To overcome this limitation, we present neural predictive monitoring (NPM), a technique that complements NSC predictions with estimates of the predictive uncertainty. These measures yield principled criteria for the rejection of predictions likely to be incorrect, without knowing the true reachability values. We also present an active learning method that significantly reduces the NSC predictor’s error rate and the percentage of rejected predictions. We develop two versions of NPM based, respectively, on the use of frequentist and Bayesian techniques to learn the predictor and the rejection rule. Both versions are highly efficient, with computation times on the order of milliseconds, and effective, managing in our experimental evaluation to successfully reject almost all incorrect predictions. In our experiments on a benchmark suite of six hybrid systems, we found that the frequentist approach consistently outperforms the Bayesian one. We also observed that the Bayesian approach is less practical, requiring a careful and problem-specific choice of hyperparameters.

     

Short-term Maharashtra state electrical power load prediction with special emphasis on seasonal changes using a novel focused time lagged recurrent neural network based on time delay neural network model

In this paper, the parameter-wise optimization training process is implemented to achieve an optimal configuration of focused time lagged recurrent neural network (FTLRNN) models by embedding the gamma, laguarre, and multi-channel tapped delay line memory structure. The aim is to examine the prediction ability of the proposed models in order to predict one-day-ahead electric power load simultaneously as usual to oppose 1–24 h forecast in sequel with a special emphasis on seasonal changes over a year. An improved delta-bar-delta algorithm is used to accelerate the training of neural networks and to improve the stability of the convergence.Experimental results indicate that the FTLRNN with time delay neural network (TDNN) clearly outperformed the gamma and laguarre based short-term memory structure in various performance metrics such as mean square error (MSE), normalized MSE, correlation coefficient (r) and mean absolute percentage error (MAPE) during evaluation process. Empirical results show that the proposed dynamic NN model consistently performs well on daily, weekly, and monthly average basis in terms of prediction accuracy. It is noticed from the literature review that an optimally configured FTLRNN with multi-channel tapped delay line memory structure is not currently available to solve short-term electrical power load prediction. The proposed method gives acceptable errors in all seasons, months and on daily basis. The average prediction error on three weeks is obtained as low as 1.67%.     

Multi-objective evolutionary optimization of polynomial neural networks for modelling and prediction of explosive cutting process

In this paper, evolutionary algorithms (EAs) are deployed for multi-objective Pareto optimal design of group method of data handling (GMDH)-type neural networks which have been used for modelling an explosive cutting process using some input–output experimental data. In this way, multi-objective EAs (non-dominated sorting genetic algorithm, NSGA-II) with a new diversity-preserving mechanism are used for Pareto optimization of such GMDH-type neural networks. The important conflicting objectives of GMDH-type neural networks that are considered in this work are, namely, training error (TE), prediction error (PE), and number of neurons (N) of such neural networks. Different pairs of theses objective functions are selected for 2-objective optimization processes. Therefore, optimal Pareto fronts of such models are obtained in each case which exhibit the trade-off between the corresponding pair of conflicting objectives and, thus, provide different non-dominated optimal choices of GMDH-type neural networks models for explosive cutting process. Moreover, all the three objectives are considered in a 3-objective optimization process, which consequently leads to some more non-dominated choices of GMDH-type models representing the trade-offs among the training error, prediction error, and number of neurons (complexity of network), simultaneously. The overlay graphs of these Pareto fronts also reveal that the 3-objective results include those of the 2-objective results and, thus, provide more optimal choices for the multi-objective design of GMDH-type neural networks in terms of minimum training error, minimum prediction error, and minimum complexity.     

Prediction of unconfined compressive strength of rock surrounding a roadway using artificial neural network

The unconfined compressive strength (UCS) of rocks is an important design parameter in rock engineering and geotechnics, which is required and determined for rock mechanical studies in mining and civil projects. This parameter is usually determined through a laboratory UCS test. Since the preparation of high-quality samples is difficult, expensive and time consuming for laboratory tests, development of predictive models for determining the mechanical properties of rocks seems to be essential in rock engineering. In this study, an attempt was made to develop an artificial neural network (ANN) and multivariable regression analysis (MVRA) models in order to predict UCS of rock surrounding a roadway. For this, a database of laboratory tests was prepared, which includes rock type, Schmidt hardness, density and porosity as input parameters and UCS as output parameter. To make a database (including 93 datasets), different rock samples, ranging from weak to very strong types, are used. To compare the performance of developed models, determination coefficient (R ²), variance account for (VAF), mean absolute error (E _a) and mean relative error (E _r) indices between predicted and measured values were calculated. Based on this comparison, it was concluded that performance of the ANN model is considerably better than the MVRA model. Further, a sensitivity analysis shows that rock density and Schmidt hardness were recognized as the most effective parameters, whereas porosity was considered as the least effective input parameter on the ANN model output (UCS) in this study.     

Computer-aided prediction of the Al2O3 nanoparticles’ effects on tensile strength and percentage of water absorption of concrete specimens

In the present paper, two models based on artificial neural networks and genetic programming for predicting split tensile strength and percentage of water absorption of concretes containing Al₂O₃ nanoparticles have been developed at different ages of curing. For purpose of building these models, training and testing using experimental results for 144 specimens produced with 16 different mixture proportions were conducted. The data used in the multilayer feed-forward neural networks models and input variables of genetic programming models are arranged in a format of eight input parameters that cover the cement content, nanoparticle content, aggregate type, water content, the amount of superplasticizer, the type of curing medium, Age of curing and number of testing try. According to these input parameters, in the neural networks and genetic programming models, the split tensile strength and percentage of water absorption values of concretes containing Al₂O₃ nanoparticles were predicted. The training and testing results in the neural network and genetic programming models have shown that every two models have strong potential for predicting the split tensile strength and percentage of water absorption values of concretes containing Al₂O₃ nanoparticles. It has been found that NN and GEP models will be valid within the ranges of variables. In neural networks model, as the training and testing ended when minimum error norm of network gained, the best results were obtained, and in genetic programming model, when 4 gens was selected to construct the model, the best results were acquired. Although neural network have predicted better results, genetic programming is able to predict reasonable values with a simpler method rather than neural network.     

Fast error estimates for indirect measurements: applications to pavement engineering

In many real-life situations, we have the following problem: we want to know the value of some characteristicy that is difficult to measure directly (e.g., lifetime of a pavement, efficiency of an engine, etc.). To estimatey, we must know the relationship betweeny and some directly measurable physical quantitiesx ₁,...,x _n . From this relationship, we extract an algorithmf that allows us, givenx _i , to computey: y=f(x ₁, ...,x _n ). So, we measurex _i , apply an algorithmf, and get the desired estimate. Existing algorithms for error estimate (interval mathematics, Monte-Carlo methods, numerical differentiation, etc.) require computation time that is several times larger than the time necessary to computey=f(x ₁, ...,x _n ). So, if an algorithmf is already time-consuming, error estimates will take too long. In many cases, this algorithmf consists of two parts: first, we usex _i to determine the parametersz _k of a model that describes the measured object, and second, we use these parameters to estimatey. The most time-consuming part is findingz _k ; this is done by solving a system of non-linear equations; usually least squares method is used. We show that for suchf, one can estimate errors repeating this time-consuming part off only once. So, we can compute bothy and an error estimate fory with practically no increase in total computation time. As an example of this methodology, we give pavement lifetime estimates.     

Nonlinear system modeling and predictive control using the RBF nets-based quasi-linear ARX model

On the basis of the single-input single-output (SISO) RBF-ARX model proposed in previous works [Peng, H., et al. (2003b). Stability analysis of the RBF-ARX model based nonlinear predictive control. In Proceedings of the ECC2003; Peng, H., et al. (2003c). Modeling and control of nonlinear nitrogen oxide decomposition process. In Proceedings of the CDC’03; Peng, H., et al. (2004). RBF-ARX model based nonlinear system modeling and predictive control with application to a NO_x decomposition process. Control Engineering Practice, 12, 191–203; Peng, H., et al. (2007). Nonlinear predictive control using neural nets-based local linearization ARX model—Stability and industrial application. IEEE Transactions on Control Systems Technology, 15, 130–143] the multi-input multi-output (MIMO) RBF-ARX model and its state-space representation are derived to describe the dynamics of a class of multivariable nonlinear systems whose working-point varies with time and which may be linearized around the working-point. The proposed MIMO RBF-ARX model has a basic regression-model structure that is analogous to the linear ARX model structure, and the elements of its regression matrices are composed of Gaussian radial basis function (RBF) neural networks that are dependent on the working-point state of the current system. An off-line estimation approach to parameters and orders of the MIMO RBF-ARX model is presented, and, on the basis of the estimated MIMO RBF-ARX model, a predictive control strategy is designed to control the underlying nonlinear system. A case study on a simulator of a thermal power plant is also given to illustrate the effectiveness of the nonlinear modeling and control method proposed in this paper.     

The experimental investigation of the effects of uncoated,PVD- and CVD-coated cemented carbide inserts and cutting parameters on surface roughness in CNC turning and its prediction using artificial neural networks

In this study the machining of AISI 1030 steel (i.e. orthogonal cutting) uncoated, PVD- and CVD-coated cemented carbide insert with different feed rates of 0.25, 0.30, 0.35, 0.40 and 0.45 mm/rev with the cutting speeds of 100, 200 and 300 m/min by keeping depth of cuts constant (i.e. 2 mm), without using cooling liquids has been accomplished. The surface roughness effects of coating method, coating material, cutting speed and feed rate on the workpiece have been investigated. Among the cutting tools—with 200 mm/min cutting speed and 0.25 mm/rev feed rate—the TiN coated with PVD method has provided 2.16 μm, TiAlN coated with PVD method has provided 2.3 μm, AlTiN coated with PVD method has provided 2.46 μm surface roughness values, respectively. While the uncoated cutting tool with the cutting speed of 100 m/min and 0.25 mm/rev feed rate has yielded the surface roughness value of 2.45 μm. Afterwards, these experimental studies were executed on artificial neural networks (ANN). The training and test data of the ANNs have been prepared using experimental patterns for the surface roughness. In the input layer of the ANNs, the coating tools, feed rate (f) and cutting speed (V) values are used while at the output layer the surface roughness values are used. They are used to train and test multilayered, hierarchically connected and directed networks with varying numbers of the hidden layers using back-propagation scaled conjugate gradient (SCG) and Levenberg–Marquardt (LM) algorithms with the logistic sigmoid transfer function. The experimental values and ANN predictions are compared by statistical error analyzing methods. It is shown that the SCG model with nine neurons in the hidden layer has produced absolute fraction of variance (R²) values about 0.99985 for the training data, and 0.99983 for the test data; root mean square error (RMSE) values are smaller than 0.00265; and mean error percentage (MEP) are about 1.13458 and 1.88698 for the training and test data, respectively. Therefore, the surface roughness value has been determined by the ANN with an acceptable accuracy.     

Neural network committee to predict the AMEn of poultry feedstuffs

A committee of neural networks is the aggregation of two or more neural networks for making overall predictions that are supposedly more accurate than those obtained by the individual networks. The objective of this paper was to assign some uncertainty over the predictions of neural networks, using a network committee to estimate the nitrogen-corrected metabolizable energy (AMEn) values of the energetic and protein concentrate feedstuffs for broilers. The dataset used to implement each expert network contains 568 experimental results. Another dataset with 48 bioassay results was used as test data. From several implemented multilayer perceptrons, the networks that presented the best generalization performance were selected to constitute the committee. The percentage of correct predictions was used as the criterion to compare committees that contained different numbers of networks. The highest probability density intervals were obtained for each feedstuff in the test data in this comparison. The estimator that ensured more accurate predictions was selected. The highest accuracy for predicting the AMEn values of concentrate feedstuffs for broilers was achieved by a committee with 1,000 networks with the use of the mode of the empirical distribution obtained from 1,000 estimated values of the AMEn. The accuracy of the models was evaluated based on their values of error measures between the observed and predicted values, in which the mode of the empirical distribution presented lower values of mean squared error (MSE = 45,285.43), mean absolute deviation (MAD = 177.66) and mean absolute percentage error (MAPE = 5.97 %) compared to the mean and the median.     

Prediction of the moment capacity of reinforced concrete slabs in fire using artificial neural networks

In this study, the application of artificial neural networks (ANN) to predict the ultimate moment capacity of reinforced concrete (RC) slabs in fire is investigated. An ANN model is built, trained and tested using 294 data for slabs exposed to fire. The data used in the ANN model consists of seven input parameters, which are the distance from the extreme fiber in tension to the centroid of the steel on the tension side of the slab (d′), the effective depth (d), the ratio of previous parameters (d′/d), the area of reinforcement on the tension face of the slab (A_s), the fire exposure time (t), the compressive strength of the concrete (f_cd), and the yield strength of the reinforcement (f_yd). It is shown that ANN model predicts the ultimate moment capacity (M_u) of RC slabs in fire with high degree of accuracy within the range of input parameters considered. The moment capacities predicted by ANN are in line with the results provided by the ultimate moment capacity equation. These results are important as ANN model alleviates the problem of computational complexity in determining M_u.