• Title/Summary/Keyword: 신경 근사치

Search Result 17, Processing Time 0.024 seconds

Water Quality Forecasting at Gongju station in Geum River using Neural Network Model (신경망 모형을 적용한 금강 공주지점의 수질예측)

  • An, Sang-Jin;Yeon, In-Seong;Han, Yang-Su;Lee, Jae-Gyeong
    • Journal of Korea Water Resources Association
    • /
    • v.34 no.6
    • /
    • pp.701-711
    • /
    • 2001
  • Forecasting of water quality variation is not an easy process due to the complicated nature of various water quality factors and their interrelationships. The objective of this study is to test the applicability of neural network models to the forecasting of the water quality at Gongju station in Geum River. This is done by forecasting monthly water qualities such as DO, BOD, and TN, and comparing with those obtained by ARIMA model. The neural network models of this study use BP(Back Propagation) algorithm for training. In order to improve the performance of the training, the models are tested in three different styles ; MANN model which uses the Moment-Adaptive learning rate method, LMNN model which uses the Levenberg-Marquardt method, and MNN model which separates the hidden layers for judgement factors from the hidden layers for water quality data. the results show that the forecasted water qualities are reasonably close to the observed data. And the MNN model shows the best results among the three models tested

  • PDF

Performance Analysis of Mulitilayer Neural Net Claddifiers Using Simulated Pattern-Generating Processes (모의 패턴생성 프로세스를 이용한 다단신경망분류기의 성능분석)

  • Park, Dong-Seon
    • The Transactions of the Korea Information Processing Society
    • /
    • v.4 no.2
    • /
    • pp.456-464
    • /
    • 1997
  • We describe a random prcess model that prvides sets of patterms whth prcisely contrlolled within-class varia-bility and between-class distinctions.We used these pattems in a simulation study wity the back-propagation netwoek to chracterize its perfotmance as we varied the process-controlling parameters,the statistical differences between the processes,and the random noise on the patterns.Our results indicated that grneralized statistical difference between the processes genrating the patterns provided a good predictor of the difficulty of the clssi-fication problem. Also we analyzed the performance of the Bayes classifier whith the maximum-likeihood cri-terion and we compared the performance of the neural network to that of the Bayes classifier.We found that the performance of neural network was intermediate between that of the simulated and theoretical Bayes classifier.

  • PDF

Neuro-controller for Broadcast Lighting LED to Express xy Chromaticity Coordinates (xy 색도좌표 표현을 위한 방송 조명용 LED 신경망 제어기)

  • Park, Sung-Chan;Park, Jin-Hyun
    • Journal of the Korea Institute of Information and Communication Engineering
    • /
    • v.24 no.6
    • /
    • pp.706-713
    • /
    • 2020
  • To control the LED lighting for broadcasting, LED current control using tri-stimulus values is used for RGB LEDs. For the convenience of control, this control is approximated as a linear function or used as an appropriate value through trial and error. Also, it is not suitable for broadcast lighting because it does not use a diffuser plate applied for mixing sufficient light and color required for actual it. In this study, a neural network with excellent nonlinear function approximation is used as a control method for LED panels for broadcast lighting. We intend to implement an LED panels controller suitable for the desired chromaticity coordinates and dimming values of intensity. As a result of the performance evaluation, the errors of the xy chromaticity coordinates are mostly ±0.02 and the acceptable range of ANSI C78.377A was satisfied. The average errors of the xy chromaticity coordinate are xerror=0.0044 and yerror=0.0030, respectively, and we confirmed the superiority and stable performance of the proposed algorithm.

Application for Prediction of Crown Settlements Using RMR in Weathering Rock Tunnels (RMR을 이용한 풍화암 터널의 천단침하량 예측 평가)

  • Kim, Young-Su;Kim, Dae-Man
    • Journal of the Korean Geotechnical Society
    • /
    • v.25 no.10
    • /
    • pp.67-76
    • /
    • 2009
  • Statistical analysis was performed using a series of data on RMR, RMR* and crown settlements collected from sites of weathering rock tunnels in Korea. The crown settlements were predicted by recurrence analysis, exponential function, and artificial neural network (ANN) using collected in-situ data. The result of the prediction fitted well compared to the measured settlement in the order of ANN, exponential function, and recurrence analysis. The range of crown settlement predicted by recurrence analysis widely scattered and promised larger settlement than the measured. Also in all method, the predicted value by RMR well matched compared to the measured settlement predicted by RMR*.

Optimization of Stock Trading System based on Multi-Agent Q-Learning Framework (다중 에이전트 Q-학습 구조에 기반한 주식 매매 시스템의 최적화)

  • Kim, Yu-Seop;Lee, Jae-Won;Lee, Jong-Woo
    • The KIPS Transactions:PartB
    • /
    • v.11B no.2
    • /
    • pp.207-212
    • /
    • 2004
  • This paper presents a reinforcement learning framework for stock trading systems. Trading system parameters are optimized by Q-learning algorithm and neural networks are adopted for value approximation. In this framework, cooperative multiple agents are used to efficiently integrate global trend prediction and local trading strategy for obtaining better trading performance. Agents Communicate With Others Sharing training episodes and learned policies, while keeping the overall scheme of conventional Q-learning. Experimental results on KOSPI 200 show that a trading system based on the proposed framework outperforms the market average and makes appreciable profits. Furthermore, in view of risk management, the system is superior to a system trained by supervised learning.

Application Examples of Daecheong Dam for Efficient Water Management Based on Integrated Water Management (통합물관리 기반 효율적 물관리를 위한 대청댐 실무적용 사례)

  • Kang, Kwon-Su;Heo, Jun-Haeng
    • Proceedings of the Korea Water Resources Association Conference
    • /
    • 2017.05a
    • /
    • pp.85-85
    • /
    • 2017
  • 효율적 물관리란 거대한 물순환 과정에서 인간이 편안한 삶을 사는데 필요한 물의 이용효율을 극대화하는 것이다. 과거의 물관리는 이원화된 수량과 수질관리, 수량중심에서는 용수공급과 홍수조절이 주요한 관심사였다. 현재는 과거의 물관리에 친수와 환경을 더한 복잡한 분야로 확대되고 있다. 통합물관리란 물을 최적으로 관리하기 위해 물관리 이해당사자간의 소통과 물 기술의 고도화를 기반으로 기존에 분산된 물관리 구성요소들(시설 정보, 수량 수질 등)을 권역적으로 관리하는 것을 말한다. 본 연구에서는 대청댐 방류에 따른 금강 하류부의 홍수추적을 위해 수행한 댐하류 소유역별 강우량 빈도분석 과정, 용담댐 방류를 고려한 대청댐 홍수도달시간 검토, Poincare Section과 신경망기법을 이용한 수문자료 예측, 추계학적 다변량 해석과 다변량 신경망해석에 의한 대청댐 유입량 산정과정, 보조여수로 건설에 따른 주여수로와 보조여수로간의 연계운영방안, 단계(관심, 주의, 경계, 심각)를 고려한 대청댐 확보수위 산정, 저수지 중장기 운영계획 수립과 댐 운영 기준수위를 결정하기 위해 누가차분방식으로 적용되는 갈수기 유입량 빈도분석에 대한 실무적용 사례를 소개하고자 한다. 강우량 빈도분석 과정은 L-모멘트방법(Hosking과 Wallis, 1993)을 적용하였고, 홍수도달시간 검토는 평균유속, 하류 수위상승 기점 영향검토, 수리학적 모형(FLDWAV, Progressive lag method 등)을 활용하였다. 카오스 이론을 도입하여 대청댐 수문자료의 상관성 검토 및 추계학적 모형을 이용한 모의발생을 유도하여 수문자료 예측을 시행하였다. 추계학적 모형과 신경망모형 연구의 대상은 대청댐으로, 시계열 자료는 댐의 월강우량, 월유입량, 최고기온, 평균기온, 최소기온, 습도, 증발량 등의 자료를 기반으로 하였다. 적용기간은 1981~2009년의 자료를 이용하여 2010년 1월부터 12월까지 12개월 동안의 월유입량을 예측하였다. 수문자료 해석의 기본이 되는 약 30년간의 자료를 이용하여 분석을 실시하였다. 대청댐의 유입량 예측을 위해 적용된 모형으로는 추계학적 모형인 ARMA모형, TF모형, TFN 모형 등이 적용되었고, 또한 신경망 모형의 종류인 다층 퍼셉트론, PCA모형 등을 활용하여 실측치와 가장 가깝게 근사화시키는 방법론을 찾고자 하였다. 또한, 기존여수로와 보조여수로 연계운영을 위해 3차원 수치해석을 통한 댐하류 안정성 검토 및 확보수위 산정을 통해 단계(관심, 주의, 경계, 심각)별로 대처가 가능한 수위를 산정하였다.

  • PDF

A Study on the Air Pollution Monitoring Network Algorithm Using Deep Learning (심층신경망 모델을 이용한 대기오염망 자료확정 알고리즘 연구)

  • Lee, Seon-Woo;Yang, Ho-Jun;Lee, Mun-Hyung;Choi, Jung-Moo;Yun, Se-Hwan;Kwon, Jang-Woo;Park, Ji-Hoon;Jung, Dong-Hee;Shin, Hye-Jung
    • Journal of Convergence for Information Technology
    • /
    • v.11 no.11
    • /
    • pp.57-65
    • /
    • 2021
  • We propose a novel method to detect abnormal data of specific symptoms using deep learning in air pollution measurement system. Existing methods generally detect abnomal data by classifying data showing unusual patterns different from the existing time series data. However, these approaches have limitations in detecting specific symptoms. In this paper, we use DeepLab V3+ model mainly used for foreground segmentation of images, whose structure has been changed to handle one-dimensional data. Instead of images, the model receives time-series data from multiple sensors and can detect data showing specific symptoms. In addition, we improve model's performance by reducing the complexity of noisy form time series data by using 'piecewise aggregation approximation'. Through the experimental results, it can be confirmed that anomaly data detection can be performed successfully.