Earticle

건물 외부에 설치된 각종 기상 측정센서에서 전송된 현재의 외부 기상조건과 일자별, 특수일별 건물 에너지 사용량과의 관계를 인공 지능기법으로 분석하고 학습을 통한 예측기능을 갖도록 함으로써, 일자별, 특수일별, 계절별 그리고 기상조건에 따른 익일 전력 사용 량을 예측하고 이에 따른 부하의 On/Off 우선순위를 결정하는 기능을 갖는 지능형 직접부하제어 시스템 구조를 설계한다.

The electric utility has the responsibility of reducing the impact of peaks on electricity demand and related costs. Therefore, they have introduced Direct Load Control System (DLCS) to automate the external control of shedding customer load that it controls. The existing DLCS have been operated only depend on On/Off signal from the electric utility. That kind of DLCS operating has been successfully used until now. But since the number of customer load participating in the DLC program are keep increasing, On/Off signal control from the electric utility is no longer meets the needs of many different kind of customers. Therefore, In this paper, the author suggest the design of direct load control system using weather sensors to meet the diversity of different customer needs.

스마트그리드에서 정확한 단기 부하 예측을 통한 자원의 이용 계획은 에너지 시스템 운영의 불확실성을 줄이고 운영 효율을 높이는데 있어서 매우 중요하다. 단기 부하 예측에 얕은 신경회로망을 포함한 다수의 머신 러닝 기법이 적용되어왔지만 예측 정확도의 개선이 요구되고 있다. 최근에는 컴퓨터 비전이나 음성인식 분야에서 심층 신경회로망의 뛰어난 연구 결과로 인해 심층 신경회로망을 단기 전력수요 예측에 적용해 예측 정확도를 개선하려는 시도가 주목 받고 있다. 본 논문에서는 일별 전력 부하 첨두치를 예측하기 위한 다층신경회로망 구조의 심층 신경회로망 모델을 제안한다. 제안된 심층 신경회로망은 층별 학습이 선행된 후 전체 모델의 학습이 이루어진다. 한국전력거래소에서 얻은 4년 동안의 일별 전력 수요 데이터를 사용, 하루 및 이틀 앞선 전력수요 첨두치를 예측하는 심층 신경회로망 모델을 구축하고 예측 정확도를 비교, 평가한다.

In smart grid an accurate load forecasting is crucial in planning resources, which aids in improving its operation efficiency and reducing the dynamic uncertainties of energy systems. Research in this area has included the use of shallow neural networks and other machine learning techniques to solve this problem. Recent researches in the field of computer vision and speech recognition, have shown great promise for Deep Neural Networks (DNN). To improve the performance of daily electric peak load forecasting the paper presents a new deep neural network model which has the architecture of two multi-layer neural networks being serially connected. The proposed network model is progressively pre-learned layer by layer ahead of learning the whole network. For both one day and two day ahead peak load forecasting the proposed models are trained and tested using four years of hourly load data obtained from the Korea Power Exchange (KPX).

In this paper, adjust the system parameters back-propagation algorithm based on fuzzy similarity interval type proposed by the fuzzy rule base to streamline redundant fuzzy sets, we can also merge with the means to reduce the number of redundant fuzzy rules, then singular value decomposition method is preferred fuzzy rules. The algorithm can effectively eliminate the adverse effects caused by redundant fuzzy rule, which improve the interpretability of fuzzy rules to reduce the computational complexity of the fuzzy reasoning process, and to improve the approximation accuracy of the system. Based on the long-term and short-term load power load characteristics analysis, to identify the influence of the load itself changes and related factors, gray system theory, neural network model and chaotic time series methods, models and methods for forecasting power load range were research. Examples verified, interval prediction has better precision, demonstrate the effectiveness of the interval prediction algorithm, the research results can be used in power market analysis and forecasting systems, power system operation and provide scientific basis for management decisions.

This paper presents the development of Short Term Load Forecasting (STLF) model using Artificial Neural Network (ANN). STLF is required for electric power planning and electricity market planning. The proposed model predicts the load demand of Connecticut in the U.S. using hourly historical electric load and weather data. For improving the load prediction accuracy, we consider two main issues that are seasons and weather factors. Each season has different load demand patterns, thus the weather factors are differently applied in each season. The proposed model uses the composited weather factor which consists of temperature and dew point. The temperature and dew point weather factors are selected through the correlation coefficient to obtain the meaningful data among the weather factors. The selected weather factors adjust the level of the pitch which is the predicted load demand of one day ahead. The proposed model improves the forecasting accuracy both in summer and winter.

In this paper, a new model, core vector regression (CVR) optimized by memetic algorithm (MA), is presented to predict electric daily load. Support vector regression (SVR) has obtained wide focus in recent years to solve nonlinear regression problems in many fields. However, it is limited on large scale dataset problem because of its high time and space complexity. Hence, CVR is proposed to improve the SVR on solving large scale dataset problem. Proper parameters selection of CVR model determines the complexity and accuracy of the model. In this paper, MA is proposed to optimize the parameters of CVR, which is called MA-CVR. Electric load is the time-dependent data which shows recurrent pattern weekly, seasonally and yearly. In this paper, we adopt MA optimization method and choose adaptive parameters dynamically based on time recurrent character of electric load data. Experimental results show that MA-CVR outperforms the existing model optimized by genetic algorithm which is called GA-CVR.

To evaluate short-term power load properly and efficiently, this paper proposes a modified DWT-QDE-LSSVM (Discrete wavelet transform (DWT) and least squares support vector machine (LSSVM) optimized by quantum differential evolution (QDE)) model combined with input selected. The load data series of the previous days are first decomposed into an approximation component and a detail component. Then LSSVM is built to model the approximation component and QDE algorithm is applied to overcome the problems faced by LSSVM in selecting parameters. In order to raise forecasting accuracy, this paper proposes the refinement of related factors. The empirical results show that the proposed DWT-QDE-LSSVM model is feasible and can satisfy the short-term load forecasting requirements in China.

The short-term load forecasting is an important method for security dispatching and economical operation in electric power system, and its prediction accuracy directly affects the operating reliability of the electric system. So the global optimization ability of particle swarm optimization (PSO) algorithm and classification prediction ability of support vector machine (SVM) are combined in order to realize the mutual supplement with each other's advantages in this paper. Firstly, the PSO algorithm is used to optimize the parameters of the SVM in order to obtain the optimal parameters of the SVM. Then a short-term load forecasting method based on combining the PSO and SVM according to the characteristics and influencing factors of short-term load forecasting is proposed. An actual power system in one region is applied to test and verify the short-term load forecasting method. The results show that the short-term load forecasting method takes on the good convergence and higher prediction precision.

Load forecasting plays a major role in planning and operation of a power system. Many techniques are available in the literature among these neural networks, linear multiple regression, regression trees, curve fitting and averaging models are the most popular because these models gives accurate solutions with very less tolerable Least Mean Absolute Percent Error(MAPE). In this paper a comparative study was made between these forecasting models and it was found that when compared to the four independent models, the averaging model i.e. combination of Curve Fitting, Regression Trees & Neural Network gives less MAPE. MATLAB programming results validates that averaging model gives better performance than individual models.

In order to accurately, fast and efficiently forecast the short-term load of power system, an improved particle swarm optimization algorithm is proposed to optimize the parameters of fuzzy radial basis function fuzzy neural network(FRBFNN) model in order to train the FRBFNN model for obtaining the optimized FRBFNN(IWPSRFN) method. In the proposed IWPSRFN method, the linear decreasing weight method is used to adjust the inertia weight of PSO algorithm. The global optimization ability of improved PSO algorithm is used to adjust the parameters of FRBFNN model by putting these parameters in the particle encoding, then the optimal values are found in the large number of viable solutions by continuous iteration of improved PSO algorithm. The found optimal values are regarded as the parameters of FRBFNN model to obtain the final IWPSRFN method for forecasting short-term load of power system. Finally, a certain region is selected to test the effectiveness of IWPSRFN method, the experiment results show that the improved PSO algorithm can effectively optimize the weights of FRBFNN and solve the slow convergence speed, and the IWPSRFN method can obtain the higher prediction accuracy and is an effective method for forecasting short-term load.

In this paper, we presented the performance of forecasting model and error correction will affect the accuracy of short-term load forecasting. Least squares support vector machines (LS-SVM) based on improved particle swarm optimization is selected as load forecasting model. Forecasting accuracy and generalization performance of LS-SVM depend on selection of its parameters greatly. Adaptive particle swarm optimization (APSO) based on fitness function was put forward to optimize the kernel parameter σ and regularization parameter γ of LS-SVM. Based on the optimized forecasting model, non-parametric error correction model is also presented by iterative method. The error forecasted by non-parametric model was used to update the forecasted load so as to improve the forecasting accuracy. Load data selected from some area in South China as training and forecasting data is used to analyze. Case study illustrates that the proposed forecasting model (NP-APSO-SVM) has more generalized performance and better forecasting accuracy compared with the method of standard SVM.

In this paper, the problem of short-term load forecasting is divided into load classification and forecasting. Load classification is needed to obtain meaningful load data as input to train forecasting models. To this end, k-NN and K-mean algorithms are presented. K-mean and k-NN algorithms can handle seasonal load classification and daily load classification, respectively. The classified load data are used to train forecasting models, which are Artificial Neural Networks, Simple Exponential Smoothing, and ARIMA models. As a real case study, we tried to forecast the electric power load of the Republic of Korea. A comparison between the classified and non-classified load forecasts demonstrates the efficiency of the proposed method.

Accurate Short Term Load Forecasting (STLF) is essential to the operating and planning for electricity supply industry. For increase accuracy of the STLF, we analyzed load patterns which are categorized by the weather-load relationship and the time-load relationship. The time-load relationship has typical patterns which show the concentrated load consumption shape under the specific time period. The weather-load relationship is identified by correlation between weather factors and load demand and used to adjust the weather weight for the load forecasting accuracy. This paper describes the analyzing of the relationships which are concern with load demand and proposed the improved an Artificial Neural Network (ANN) based non-linear model for 24-hour-ahead load forecasting.

Short-term load forecast plays an important role in the safe and economic operation of power system. Its prediction accuracy affects the power system's security, profit and quality directly. And meteorological factor is one of the key factors that affect the accuracy of load forecasting. In this paper, we put forward enterprise electric load forecasting method combined with grey relational degree algorithm and multivariate linear regression forecasting method. The example shows that this method can get better prediction accuracy.

The relationship between med-long term load forecasting and socio-economic indicators is very difficult to describe with an accurate mathematical model. The paper introduce data mining technology into the association analysis of China's electricity consumption growth, select many socio-economic indicators since 2000, constitute the relevant factors database, complement of a few missing data, and dig out a number of indicators closely related to the electricity consumption with cluster analysis, and the data of distortion indicators is corrected, thus, build a more scientific load forecasting model. Validate and test the correlation of electricity consumption and selected indicators by dynamic neural network time sequence tool. The results show that the prediction model has good convergence, and the effect is satisfactory.

Current grey clustering analysis methods have some defects. So, this paper proposes a prediction model based on improved grey clustering analysis. Firstly, it constructs the grey classical domain and the grey sector domain based on prediction subjects and data and according to relevant theory about grey clustering analysis. Secondly, it categorizes samples according to features of prediction subjects and confirms the analysis categories corresponding to the classical domain. Thirdly, based on the grey system theory, it constructs the grey correlation coefficient model and grey correlation degree model so as to obtain the weighed grey correlation degree. Thus, prediction subjects can be divided into proper category. Finally, power load forecasting in the power industry is taken as a case to prove that the model is reliable and has efficacy.

The medium and long term load forecasting is the basis of power planning, investment, production, scheduling and trade, which plays an important role in electric power safety and economic operation. In China, it has the increasing uncertainty and the uncertainty of random variation to forecast the medium and long term load. Thus we can regard it as a typical grey system. However, the traditional grey prediction method cannot be adapt to the needs of the load forecasting gradually. It need to be rich and perfect with the continuous improvement of power system complexity and power marketization degree. This paper studied the modelling mechanism of grey prediction model. Then we analyzed the problems existing in the model, including the boundary value problem, the background value structure problem and the least squares parameter identification problem. This paper put forward an optimization method to directly identify the boundary value x(0)(1), the developing coefficient a and grey coefficient b using ant colony algorithm according to the time response expression of GM(1,1) model, so that it established an optimized GM(1,1) prediction model based on ant colony algorithm. This model can fix the impact of boundary value, and also avoid the errors brought by the background value construction and the least squares parameter estimation. It can verify the effectiveness of the proposed optimization model through the load data simulation. And it can improve the prediction accuracy effectively.

본 논문에서는 레이더 시스템에서 탐지 거리 추정에 영향을 미치는 레이더 단면적의 크기에 대한 집적 방식을 비교 분석한다. 본 논문에서는 레이더 단면적의 크기에 따라 크기가 작을 경우 스웰링 케이스 1, 클 경우에는 스웰링 케이스 3의 탐지 확률을 레이더 방정식에 적용하여 탐지 거리를 추정하였다. 모의실험을 통해서 스웰링 케이스의 차이에 따른 코히런트 집적과 비 코히런트 집적을 비교 분석하였다. 비교 분석 결과, 비 코히런트 집적 방식이 추정 거리가 가장 우수하였고 코히런트 집적 방식은 스웰링 케이스를 적용한 탐지 거리 추정에 적합하지 않음을 알 수 있었다.

This paper comparatively analyze to integration case to have a influence detection range estimation about radar cross section in radar system. This paper estimate detection range used to probability of detection in radar equation that used to swerling case 1 in case of radar cross section is small and used to swerling case 3 in case of radar cross section is large. Through simulation, coherent integration and non-coherent integration about swerling case difference were comparatively analyzed. In the result of comparative analysis, non-coherent integration case is outstanding detection range and we known that coherent integration don't suitable for detection range estimation.

The load forecasting has been essential in planning and operation of power systems. The load composition rata is also needed to analyze power-systems - load flow calculation and system stability. This paper proposes the monthly peak load forecasting methods for load groups in residential class using load composition rate and electric consumption characteristics. The proposed methods were applied to a real-scale power system and the effectiveness was turned out.

This paper presents fuzzy clustering and wavelet transform analysis based technique for the industrial hourly load forecasting fur the purpose of peak demand control. Firstly, one year of historical load data were sorted and clustered into several groups using fuzzy clustering and then wavelet transform is adopted using the Biorthogonal mother wavelet in order to forecast the peak load of one hour ahead. The 5-level decomposition of the daily industrial load curve is implemented to consider the weather sensitive component of loads effectively. The wavelet coefficients associated with certain frequency and time localization is adjusted using the conventional multiple regression method and the components are reconstructed to predict the final loads through a five-scale synthesis technique. The outcome of the study clearly indicates that the proposed composite model of fuzzy clustering and wavelet transform approach can be used as an attractive and effective means for the industrial hourly peak load forecasting.

Recently, microgrid (MG) has been proposed as one of the most critical solutions for various energy problems. For the optimal and economic operation of MGs, it is very important to forecast the load profile. However, it is not easy to predict the load accurately since the load in a MG is small and highly variable. In this paper, we propose an artificial neural network (ANN) based method to predict the energy use in campus buildings in short-term time series from one hour up to one week. The proposed method analyzes and extracts the features from the historical data of load and temperature to generate the prediction of future energy consumption in the building based on sparsified K-means. To evaluate the performance of the proposed approach, historical load data in hourly resolution collected from the campus buildings were used. The experimental results show that the proposed approach outperforms the conventional forecasting methods.