• 대한전기학회논문지
• /
• 제45권4호
• /
• pp.602-611
• /
• 1996

In this paper, radial basis function networks with two hidden layers, which employ the K-means clustering method and the hierarchical training, are proposed for improving the short-term predictability of chaotic time series. Furthermore the recursive training method of radial basis function network using the recursive modified Gram-Schmidt algorithm is proposed for the purpose. In addition, the radial basis function networks trained by the proposed training methods are compared with the X.D. He A Lapedes's model and the radial basis function network by nonrecursive training method. Through this comparison, an improved radial basis function network for predicting chaotic time series is presented. (author). 17 refs., 8 figs., 3 tabs.

• 설비공학논문집
• /
• 제12권6호
• /
• pp.609-615
• /
• 2000

The energy conservation is one of the most important issues in recent years. Especially, the energy conservation through improved control strategies is one of the most highly possible area to be implemented in the near future. The energy conservation of the ice storage system can be accomplished through the improved control strategies. A real time building load prediction algorithm was developed. The expected highest and the lowest outdoor temperature of the next day were used to estimate the next day outdoor temperature profile. The measured dry bulb temperature and the measured building load were used to estimate system parameters by using the on-line weighted recursive least square method. The estimated hourly outdoor temperatures and the estimated hourly system parameters were used to predict the next day hourly building loads. In order to see the effectiveness of the building load prediction algorithm, two different types of building models were selected and analysed. The simulation results show less than 1% in error for the prediction of the next day building loads. Therefore, this algorithm may successfully be used for the development of improved control algorithms of the ice storage system.

• 한국건설관리학회논문집
• /
• 제22권3호
• /
• pp.21-30
• /
• 2021

본 연구는 딥러닝의 반복적 예측방식을 활용하여 5개월의 철근 가격 예측방법을 제안한다. 이 방식은 입력데이터의 특성을 모두 단기예측하여 원 데이터에 추가하고, 추가된 데이터로 다음의 시점을 예측하는 과정을 반복하여 장기 예측한다. 본 연구에서 제시하는 방식으로 1개월에서 5개월까지 예측한 철근 가격의 예측 평균 정확도는 약 97.24%이다. 제안된 방식을 통해 인간의 경험과 판단을 통한 가격 추정방법의 체계성을 보완하여 기존의 방식보다 정확한 비용계획이 가능할 것으로 사료된다. 또 철근 이외의 건축재료를 비롯하여 시계열 데이터로 가격을 장기예측하는 연구에서 본 연구에서 제시한 방법이 활용될 수 있을 것으로 기대한다.

• 대한전기학회논문지:전력기술부문A
• /
• 제48권7호
• /
• pp.898-902
• /
• 1999

In this paper, we proposed an algorithm for estimating parameters of nonlinear continuous-discrete state-space system. This algorithm uses the conventional extended Kalman filter(EKF) for estimating state variables, and modifies the recursive prediction error method for parameter estimation of the nonlinear system. Simulation results for both linear and nonlinear measurements under the environment of process and measurement noises show a convincing performance of the proposed algorithm.

• Journal of information and communication convergence engineering
• /
• 제1권4호
• /
• pp.189-193
• /
• 2003

A new algorithm is presented for adaptive notch filtering of narrow band or sine signals for embedded among broad band noise. The notch filter is implemented as a constrained infinite impulse response filter with a minimal number of parameters, Based on the recursive prediction error (RPE) method, the algorithm has the advantages of the fast convergence, accurate results and initial estimate of filter coefficient and its covariance is revealed. A convergence criterion is also developed. By using the information of the noise-to-signal power, the algorithm can self-adjust its initial filter coefficient estimate and its covariance to ensure convergence.

• 무역학회지
• /
• 제46권6호
• /
• pp.191-209
• /
• 2021

In this paper, various time series analysis models and machine learning models are presented for long-term prediction of export growth rate, and the prediction performance is compared and reviewed by RMSE and MAE. Export growth rate is one of the major economic indicators to evaluate the economic status. And It is also used to predict economic forecast. The export growth rate may have a negative (-) value as well as a positive (+) value. Therefore, Instead of using the ReLU function, which is often used for time series prediction of deep learning models, the PReLU function, which can have a negative (-) value as an output value, was used as the activation function of deep learning models. The time series prediction performance of each model for three types of data was compared and reviewed. The forecast data of long-term prediction of export growth rate was deduced by three forecast methods such as a fixed forecast method, a recursive forecast method and a rolling forecast method. As a result of the forecast, the traditional time series analysis model, ARDL, showed excellent performance, but as the time period of learning data increases, the performance of machine learning models including LSTM was relatively improved.

• 한국콘텐츠학회논문지
• /
• 제20권12호
• /
• pp.49-57
• /
• 2020

본 연구는 우리나라 수출 분야 산업의 경쟁력을 나타내는 부가가치율을 다양한 머신러닝 기법을 활용하여 예측하였다. 아울러, 예측의 정확성 및 안정성을 높이기 위하여 머신러닝 기법 예측값들에 예측조합 기법을 적용하였다. 특히, 본 연구는 산업별 부가가치율에 영향을 주는 다양한 변수를 고려하기 위하여 재귀적특성제거 방법을 사용하여 주요 변수를 선별한 후 머신러닝 기법에 적용함으로써 예측과정의 효율성을 높였다. 분석결과, 예측조합 방법에 따른 예측값은 머신러닝 기법 예측값들보다 실제의 산업 부가가치율에 근접한 것으로 나타났다. 또한, 머신러닝 기법의 예측값들이 큰 변동성을 보이는 것과 달리 예측조합 기법은 안정적인 예측값을 나타내었다.

• 전기학회논문지
• /
• 제67권3호
• /
• pp.427-432
• /
• 2018

This paper uses a recursive least squares method to estimate the projectile motion trajectory of an object in real time. The equations of motion of the object are obtained considering the air resistance which occurs in the actual experiment environment. Because these equations consider air resistance, parameter estimation of nonlinear terms is required. However, nonlinear recursive least squares estimation is not suitable for estimating trajectory of projectile in that it requires a lot of computation time. Therefore, parameter estimation for real-time trajectory prediction is performed by recursive least square estimation after using Taylor series expansion to approximate nonlinear terms to polynomials. The proposed method is verified through experiments by using VICON Bonita motion capture system which can get three dimensional coordinates of projectile. The results indicate that proposed method is more accurate than linear Kalman filter method based on the equations of motion of projectile that does not consider air resistance.

• 대한의용생체공학회:학술대회논문집
• /
• 대한의용생체공학회 1995년도 춘계학술대회
• /
• pp.265-269
• /
• 1995

In this study, we present a training method of radial basis function networks based on recursive modified Gram-Schmidt algorithm for single step prediction of chaotic time series. With single step predictions of Mackey-Glass time series and alpha-rhythm EEG which has chaotic characteristics, the radial basis function network trained by this method is compared with one trained by a classical non-recursive method and the radial basis function model proposed by X.D. He and A. Lapedes. The results show the effectiveness of the training method.

• 한국군사과학기술학회지
• /
• 제19권2호
• /
• pp.177-184
• /
• 2016

This paper presents a strike velocity estimation using the recursive Bayesian filter that operates both correction and prediction models to probabilistically remove noises of sensors and accurately estimate the strike velocity during the real-time experiments. Four different types of bullets such as 5.56 mm M193, 7.62 mm M80, 5.45 mm 7N10 and 7.62 mm MSC were used to validate the proposed method. Compared to the existing method, the proposed method statistically results in higher stability of the strike velocity estimation as well as its reliability for the ballistic limit velocity computation.