• Journal of KIISE:Computing Practices and Letters
• /
• v.15 no.7
• /
• pp.469-477
• /
• 2009

In mobile systems, energy efficiency is critical to extend battery life. Therefore, power consumption should be taken into account to develop software in addition to performance, Efficient software design in power and performance is possible if accurate power prediction is accomplished during the execution of software, In this paper, power estimation model is developed using statistical analysis, The proposed model analyzes processor behavior Quantitatively using the data of performance monitoring events and power consumption collected by executing various benchmark programs, And then representative hardware events on power consumption are selected using hierarchical clustering, The power prediction model is established by regression analysis in which the selected events are independent variables and power is a response variable, The proposed model is applied to a PXA320 mobile processor based on Intel XScale architecture and shows average estimation error within 4% of the actual measured power consumption of the processor.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.11
• /
• pp.4028-4042
• /
• 2021

Aiming at the problem of software defect prediction difficulty caused by insufficient software defect marker samples and unbalanced classification, a semi-supervised software defect prediction model based on a tri-training algorithm was proposed by combining feature normalization, over-sampling technology, and a Tri-training algorithm. First, the feature normalization method is used to smooth the feature data to eliminate the influence of too large or too small feature values on the model's classification performance. Secondly, the oversampling method is used to expand and sample the data, which solves the unbalanced classification of labelled samples. Finally, the Tri-training algorithm performs machine learning on the training samples and establishes a defect prediction model. The novelty of this model is that it can effectively combine feature normalization, oversampling techniques, and the Tri-training algorithm to solve both the under-labelled sample and class imbalance problems. Simulation experiments using the NASA software defect prediction dataset show that the proposed method outperforms four existing supervised and semi-supervised learning in terms of Precision, Recall, and F-Measure values.

• Journal of Advanced Navigation Technology
• /
• v.22 no.5
• /
• pp.478-484
• /
• 2018

This study predicts solar radiation, solar radiation, and solar power generation using hourly weather data such as temperature, precipitation, wind direction, wind speed, humidity, cloudiness, sunshine and solar radiation. I/O pattern in supervised learning is the most important factor in prediction, but it must be determined by repeated experiments because humans have to decide. This study proposed four input and output patterns for solar and sunrise prediction. In addition, we predicted solar power generation using the predicted solar and solar radiation data and power generation data of Youngam solar power plant in Jeollanamdo. As a experiment result, the model 4 showed the best prediction results in the sunshine and solar radiation prediction, and the RMSE of sunshine was 1.5 times and the sunshine RMSE was 3 times less than that of model 1. As a experiment result of solar power generation prediction, the best prediction result was obtained for model 4 as well as sunshine and solar radiation, and the RMSE was reduced by 2.7 times less than that of model 1.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.34 no.1A
• /
• pp.27-33
• /
• 2009

This paper proposes transmit power control techniques with fading channel prediction scheme based on recurrent neural network for high-speed mobile communication systems. The operation result of recurrent neural network which is derived interpretively solves complexity problems of neural network circuit, and channel gain of multiple transmit antenna is derived with maximum ratio combining(MRC) by using the operation result, and this channel gain control transmit power of each antenna. simulation results show that proposed method has a outstanding performance compared to method that is not to be controlled power based on channel prediction. Most of legacy studies are for robust receive technique of fading signals or channel prediction of fading signals limited low-speed mobility, but in open loop Power control, proposed channel prediction method decrease system complexity with removal of fading effect in transmitter.

• Journal of Digital Convergence
• /
• v.12 no.3
• /
• pp.237-242
• /
• 2014

Recently, wind energy is expanding to combination of computing to forecast of wind power generation as well as intelligent of wind powerturbine. Wind power is rise and fall depending on weather conditions and difficult to predict the output for efficient power production. Wind power is need to reliably linked technology in order to efficient power generation. In this paper, distributed power generation forecasts to enhance the predicted and actual power generation in order to minimize the difference between the power of distributed power short-term prediction model is designed. The proposed model for prediction of short-term combining the physical models and statistical models were produced in a physical model of the predicted value predicted by the lattice points within the branch prediction to extract the value of a physical model by applying the estimated value of a statistical model for estimating power generation final gas phase produces a predicted value. Also, the proposed model in real-time National Weather Service forecast for medium-term and real-time observations used as input data to perform the short-term prediction models.

• Journal of the Korean Society for Precision Engineering
• /
• v.19 no.8
• /
• pp.71-76
• /
• 2002

In this study low cycle fatigue (LCF) behavior of 12Cr steel at high temperature are described. Secondly, comparisons between predicted lives and experimental lives are made for the several sample life prediction models. Two minute hold period in either tension or compression reduce the number of cycles to failure by about a factor of two. Twenty minute hold periods in compression lead to shorter lives than 2 minute hold periods in compression. Experiments showed that life predictions from classical phenomenological models have limitations. More LCF experiments should be pursued to gain understanding of the physical damage mechanisms and to allow the development of physically-based models which can enhance the accuracy of the predictions of components. From a design point-of-view, life prediction has been judged acceptable for these particular loading conditions but extrapolations to thermo-mechanical fatigue loading, for example, require more sophisticated models including physical damage mechanisms.

• Journal of the Korean Society of Systems Engineering
• /
• v.16 no.1
• /
• pp.58-67
• /
• 2020

A big data processing method to predict solar power generation using systems engineering approach is developed in this work. For developing analytical method, linear model (LM), support vector machine (SVN), and artificial neural network (ANN) technique are chosen. As evaluation indices, the cross-correlation and the mean square root of prediction error (RMSEP) are used. From multi-variable comparison test, it was found that ANN methodology provides the highest correlation and the lowest RMSEP.

• Proceedings of the Korean Society for Technology of Plasticity Conference
• /
• 2001.10a
• /
• pp.121-124
• /
• 2001

On-line prediction model which calculate roll force, roll power and forward slip of continuous hot strip rolling was built based on the results of plane strait rigid-viscoplastic finite element process model. Using the integrated FE process model, a series of finite element simulation was conducted over the process variables, and the influence of various process conditions on non-dimensional parameters was inspected. The prediction accuracy of the proposed on-line model under front and back tension is examined through comparison with predictions from a finite element process model over the various process conditions. In addition, we examined the validity of the on-line prediction model through comparison with roll force of experiment in hot rolling.

• Journal of the Korean Data and Information Science Society
• /
• v.19 no.4
• /
• pp.1305-1325
• /
• 2008

As index option markets grow recently, many analysts and investors become interested in forecasting the volatility of KOSPI 200 Index to achieve portfolio's goal from the point of financial risk management and asset evaluation. To serve this purpose, we introduce NN and SVM integrated with other financial series models such as GARCH, EGARCH, and EWMA. Moreover, according to the empirical test, Integrating NN with GARCH or EWMA models improves prediction power in terms of the precision and the direction of the volatility of KOSPI 200 index. However, integrating SVM with financial series models doesn't improve greatly the prediction power. In summary, SVM-EGARCH was the best in terms of predicting the direction of the volatility and NN-GARCH was the best in terms of the prediction precision. We conclude with advantages of the integration process and the need for integrating models to enhance the prediction power.

• Journal of information and communication convergence engineering
• /
• v.21 no.1
• /
• pp.82-89
• /
• 2023

Various machine-learning models may yield high predictive power for massive time series for time series prediction. However, these models are prone to instability in terms of computational cost because of the high dimensionality of the feature space and nonoptimized hyperparameter settings. Considering the potential risk that model training with a high-dimensional feature set can be time-consuming, we evaluate a feature-importance-based feature selection method to derive a tradeoff between predictive power and computational cost for time series prediction. We used two machine learning techniques for performance evaluation to generate prediction models from a retail sales dataset. First, we ranked the features using impurity- and Local Interpretable Model-agnostic Explanations (LIME) -based feature importance measures in the prediction models. Then, the recursive feature elimination method was applied to eliminate unimportant features sequentially. Consequently, we obtained a subset of features that could lead to reduced model training time while preserving acceptable model performance.