본 발명은, 선박의 성능을 나타내는 각종 성능계수를 추정하기 위한 장치 및 방법에 관한 것으로, 본 발명에 따르면, 선박의 주요 제원을 입력으로 하여 성능을 예측하는 기존의 방법들은 대부분 선형을 대표하는 변수로만 성능을 예측함으로 인해 미세한 형상의 변화를 반영하기 어려운 한계가 있었던 종래기술의 통계기반 선박 성능 추정방법들의 문제점을 해결하기 위해, 딥러닝(Deep Learning) 및 컨벌루션 신경망(Convolutional Neural Network ; CNN) 기술을 활용하여, 선박의 형상정보(오프셋 단면정보)를 나타내는 오프셋 데이터를 이미지 데이터로 변환하고, 이미지 데이터에 선박의 주요 제원변수와 속도변수, 벌브형상정보를 매칭하여 2차원 CNN 또는 3차원 CNN을 적용하여 학습을 수행하며, 이때, 필요에 따라 과적합을 방지하기 위한 드롭아웃(dropout)과 정규화(L2 regularization)를 추가하는 것에 의해, 임의의 선형의 오프셋 데이터와 주요 제원변수, 선수 벌브변수, 속력변수를 입력으로 하여 원하는 성능변수가 출력될 수 있도록 구성되는 컨벌루션 신경망을 이용한 선박 성능계수 추정시스템 및 방법이 제공된다.

    The present invention relates to an apparatus and a method for estimating various kinds of performance coefficients indicating the performance of a ship. According to the present invention, in order to solve the problems of statistics-based ship performance estimation methods of a conventional technique in which conventional methods using the main specifications of a ship as an input so as to estimate performance are limited in that it is difficult to reflect minute changes in shape since performance is predicted only with variables representing mostly ship type, provided are a system and a method for estimating a ship performance coefficient by using a convolutional neural network (CNN), the system and the method: converting offset data indicating shape information (offset cross-section information) about a ship into image data by using deep learning and CNN technologies; matching a main specification variable, a speed variable, and bulb shape information about the ship with the image data, thereby performing training by applying a two-dimensional or three-dimensional CNN to the matching result; and adding dropout and L2 regularization in order to prevent overfitting as necessary, and thus desired performance variables can be output using, as inputs, offset data, the main specification variable, a bow bulb variable, and the speed variable of an arbitrary ship type.


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    System and method for estimating ship performance coefficient using convolutional neural network


    Weitere Titelangaben:

    컨벌루션 신경망을 이용한 선박 성능계수 추정시스템 및 방법


    Beteiligte:
    KIM YOO CHUL (Autor:in) / KIM MYOUNG SOO (Autor:in) / HWANG SEUNG HYUN (Autor:in) / YEON SEONG MO (Autor:in) / LEE YOUNG YEON (Autor:in) / KIM KWANG SOO (Autor:in)

    Erscheinungsdatum :

    2024-01-09


    Medientyp :

    Patent


    Format :

    Elektronische Ressource


    Sprache :

    Koreanisch


    Klassifikation :

    IPC:    B63B Schiffe oder sonstige Wasserfahrzeuge , SHIPS OR OTHER WATERBORNE VESSELS / G06F ELECTRIC DIGITAL DATA PROCESSING , Elektrische digitale Datenverarbeitung / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen



    SYSTEM AND METHOD FOR ESTIMATING SHIP PERFORMANCE COEFFICIENT BY USING CONVOLUTIONAL NEURAL NETWORK

    KIM YOO CHUL / KIM MYOUNG SOO / HWANG SEUNG HYUN et al. | Europäisches Patentamt | 2024

    Freier Zugriff

    Visual Explainable Convolutional Neural Network for Aerodynamic Coefficient Prediction

    Yanxuan Zhao / Chengwen Zhong / Fang Wang et al. | DOAJ | 2022

    Freier Zugriff

    Ship traffic flow prediction method based on improved graph convolutional neural network

    LIU WEN / ZHAN YANG / LIANG MAOHAN et al. | Europäisches Patentamt | 2022

    Freier Zugriff

    Application of Convolutional Neural Network to Predict Airfoil Lift Coefficient

    Zhang, Yao / Sung, Woong Je / Mavris, Dimitri N. | AIAA | 2018


    An Improved Deep Convolutional Neural Network to Predict Airfoil Lift Coefficient

    Yu, Boping / Xie, Liang / Wang, Fuxin | British Library Conference Proceedings | 2020