• Journal of Electrical Engineering and Technology
• /
• v.12 no.6
• /
• pp.2177-2186
• /
• 2017

The goal of this paper is to provide the specific forecasting steps and to explain how to design the forecasting architecture and training data sets to forecast very short-term wind power when the numerical weather prediction (NWP) is unavailable, and when the sampling periods of the wind power and training data are different. We forecast the very short-term wind power every 15 minutes starting two hours after receiving the most recent measurements up to 40 hours for a total of 38 hours, without using the NWP data but using the historical weather data. Generally, the NWP works as a predictor and can be converted to wind power forecasts through machine learning-based forecasting algorithms. Without the NWP, we can still build the predictor by shifting the historical weather data and apply the machine learning-based algorithms to the shifted weather data. In this process, the sampling intervals of the weather and wind power data are unified. To verify our approaches, we participated in the 2017 wind power forecasting competition held by the European Energy Market conference and ranked sixth. We have shown that the wind power can be accurately forecasted through the data shifting although the NWP is unavailable.

• Journal of Intelligence and Information Systems
• /
• v.15 no.4
• /
• pp.99-121
• /
• 2009

Recently, there have been many studies that predict corporate bankruptcy using data mining techniques. Although various data mining techniques have been investigated, some researchers have tried to combine the results of each data mining technique in order to improve classification performance. In this study, we classify 4 types of data mining techniques via their characteristics and select representative techniques of each type then combine them using a genetic algorithm. The genetic algorithm may find optimal or near-optimal solution because it is a global optimization technique. This study compares the results of single models, typical combination models, and the proposed integration model using the genetic algorithm.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.39 no.1
• /
• pp.105-115
• /
• 2016

Ensemble classification involves combining individually trained classifiers to yield more accurate prediction, compared with individual models. Ensemble techniques are very useful for improving the generalization ability of classifiers. The random subspace ensemble technique is a simple but effective method for constructing ensemble classifiers; it involves randomly drawing some of the features from each classifier in the ensemble. The instance selection technique involves selecting critical instances while deleting and removing irrelevant and noisy instances from the original dataset. The instance selection and random subspace methods are both well known in the field of data mining and have proven to be very effective in many applications. However, few studies have focused on integrating the instance selection and random subspace methods. Therefore, this study proposed a new hybrid ensemble model that integrates instance selection and random subspace techniques using genetic algorithms (GAs) to improve the performance of a random subspace ensemble model. GAs are used to select optimal (or near optimal) instances, which are used as input data for the random subspace ensemble model. The proposed model was applied to both Kaggle credit data and corporate credit data, and the results were compared with those of other models to investigate performance in terms of classification accuracy, levels of diversity, and average classification rates of base classifiers in the ensemble. The experimental results demonstrated that the proposed model outperformed other models including the single model, the instance selection model, and the original random subspace ensemble model.

• Journal of the Korean Society of Safety
• /
• v.7 no.2
• /
• pp.71-79
• /
• 1992

This paper presents a least-square algorithms of lattice structures and their use for adaptive prediction of time series generated from the dynamic system. As the view point of adaptive prediction, a new method of Identification of dynamic characteristics by means of estimating the parameters of linear auto regressive model is proposed. The fast convergence of adaptive lattice algorithms is seen to be due to the orthogonalization and decoupling properties of the lattice. The superiority of the least-square lattice is verified by computer simulation, then predictor coefficients are computed from the linear sequential time data. For the application to the dynamic characteristic analysis of unknown system, the transfer function of ideal system represented in frquency domain and the estimated one obtained by predicted coefficients are compared. Using the proposed method, the damping ratio and the natural frequency of a dynamic structure subjected to random excitations can be estimated. It is expected that this method will be widely applicable to other technical dynamic problem in which estimation of damping ratio and fundamental vibration modes are required.

• Journal of Electrical Engineering and Technology
• /
• v.11 no.4
• /
• pp.974-986
• /
• 2016

Reversible image watermarking, a type of digital data hiding, is capable of recovering the original image and extracting the hidden message with precision. A number of reversible algorithms have been proposed to achieve a high embedding capacity and a low distortion. While numerous algorithms for the achievement of a favorable performance regarding a small embedding capacity exist, the main goal of this paper is the achievement of a more favorable performance regarding a larger embedding capacity and a lower distortion. This paper therefore proposes a reversible data hiding algorithm for which a novel piecewise 2D auto-regression (P2AR) predictor that is based on a rhombus-embedding scheme is used. In addition, a minimum description length (MDL) approach is applied to remove the outlier pixels from a training set so that the effect of a multiple linear regression can be maximized. The experiment results demonstrate that the performance of the proposed method is superior to those of previous methods.

• Korean Journal of Remote Sensing
• /
• v.35 no.1
• /
• pp.93-115
• /
• 2019

The availability of high-resolution laser scanning data and advanced machine learning algorithms has enabled an accurate potential rockfall source identification. However, the presence of other mass movements, such as landslides within the same region of interest, poses additional challenges to this task. Thus, this research presents a method based on an integration of Gaussian mixture model (GMM) and ensemble artificial neural network (bagging ANN [BANN]) for automatic detection of potential rockfall sources at Kinta Valley area, Malaysia. The GMM was utilised to determine slope angle thresholds of various geomorphological units. Different algorithms(ANN, support vector machine [SVM] and k nearest neighbour [kNN]) were individually tested with various ensemble models (bagging, voting and boosting). Grid search method was adopted to optimise the hyperparameters of the investigated base models. The proposed model achieves excellent results with success and prediction accuracies at 95% and 94%, respectively. In addition, this technique has achieved excellent accuracies (ROC = 95%) over other methods used. Moreover, the proposed model has achieved the optimal prediction accuracies (92%) on the basis of testing data, thereby indicating that the model can be generalised and replicated in different regions, and the proposed method can be applied to various landslide studies.

• Steel and Composite Structures
• /
• v.39 no.1
• /
• pp.81-93
• /
• 2021

Splitting tensile strength (STS) is an important mechanical parameter of concrete. This study offers novel methodologies for the early prediction of this parameter. Artificial neural network (ANN), which is a leading predictive method, is synthesized with two metaheuristic algorithms, namely atom search optimization (ASO) and equilibrium optimizer (EO) to achieve an optimal tuning of the weights and biases. The models are applied to data collected from the published literature. The sensitivity of the ASO and EO to the population size is first investigated, and then, proper configurations of the ASO-NN and EO-NN are compared to the conventional ANN. Evaluating the prediction results revealed the excellent efficiency of EO in optimizing the ANN. Accuracy improvements attained by this algorithm were 13.26 and 11.41% in terms of root mean square error and mean absolute error, respectively. Moreover, it raised the correlation from 0.89958 to 0.92722. This is while the results of the conventional ANN were slightly better than ASO-NN. The EO was also a faster optimizer than ASO. Based on these findings, the combination of the ANN and EO can be an efficient non-destructive tool for predicting the STS.

• Computers and Concrete
• /
• v.30 no.3
• /
• pp.225-236
• /
• 2022

In this article, Multi-Gene Genetic Programming (MGGP) is proposed for the estimation of the compressive strength of concrete. MGGP is known to be a powerful algorithm able to find a relationship between certain input space features and a desired output vector. With respect to most conventional machine learning algorithms, which are often used as "black boxes" that do not provide a mathematical formulation of the output-input relationship, MGGP is able to identify a closed-form formula for the input-output relationship. In the study presented in this article, MGPP was used to predict the compressive strength of plain concrete, concrete with fly ash, and concrete with furnace slag. A formula was extracted for each mixture and the performance and the accuracy of the predictions were compared to the results of Artificial Neural Network (ANN) and Extreme Learning Machine (ELM) algorithms, which are conventional and well-established machine learning techniques. The results of the study showed that MGGP can achieve a desirable performance, as the coefficients of determination for plain concrete, concrete with ash, and concrete with slag from the testing phase were equal to 0.928, 0.906, 0.890, respectively. In addition, it was found that MGGP outperforms ELM in all cases and its' accuracy is slightly less than ANN's accuracy. However, MGGP models are practical and easy-to-use since they extract closed-form formulas that may be implemented and used for the prediction of compressive strength.

• Smart Structures and Systems
• /
• v.29 no.3
• /
• pp.433-444
• /
• 2022

Concrete is a most utilized material in the construction industry that have main components. The strength of concrete can be improved by adding some admixtures. Evaluating the impact of fly ash (FA) and silica fume (SF) on the long-term compressive strength (CS) of concrete provokes to find the significant parameters in predicting the CS, which could be useful in the practical works and would be extensible in the future analysis. In this study, to evaluate the effective parameters in predicting the CS of concrete containing admixtures in the long-term and present a fitted equation, the multivariate adaptive regression splines (MARS) method has been used, which could find a relationship between independent and dependent variables. Next, for optimizing the output equation, biogeography-based optimization (BBO), particle swarm optimization (PSO), and hybrid PSOBBO methods have been utilized to find the most optimal conclusions. It could be concluded that for CS predictions in the long-term, all proposed models have the coefficient of determination (R²) larger than 0.9243. Furthermore, MARS-PSOBBO could be offered as the best model to predict CS between three hybrid algorithms accurately.

• Smart Structures and Systems
• /
• v.30 no.2
• /
• pp.195-207
• /
• 2022

Computational drawbacks associated with regular predictive models have motivated engineers to use hybrid techniques in dealing with complex engineering tasks like simulating the compressive strength of concrete (CSC). This study evaluates the efficiency of tree potential metaheuristic schemes, namely shuffled complex evolution (SCE), multi-verse optimizer (MVO), and beetle antennae search (BAS) for optimizing the performance of a multi-layer perceptron (MLP) system. The models are fed by the information of 1030 concrete specimens (where the amount of cement, blast furnace slag (BFS), fly ash (FA1), water, superplasticizer (SP), coarse aggregate (CA), and fine aggregate (FA2) are taken as independent factors). The results of the ensembles are compared to unreinforced MLP to examine improvements resulted from the incorporation of the SCE, MVO, and BAS. It was shown that these algorithms can considerably enhance the training and prediction accuracy of the MLP. Overall, the proposed models are capable of presenting an early, inexpensive, and reliable prediction of the CSC. Due to the higher accuracy of the BAS-based model, a predictive formula is extracted from this algorithm.