• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>Materials Science amp;amp;amp; Technology Conference and Exhibition >APPLICATION OF ARTIFICIAL NEURAL NETWORKS TO LOW CYCLE FATIGUE AND CREEP DATA PROCESSING FOR POWER PLANT MATERIALS
【24h】

APPLICATION OF ARTIFICIAL NEURAL NETWORKS TO LOW CYCLE FATIGUE AND CREEP DATA PROCESSING FOR POWER PLANT MATERIALS

机译：人工神经网络在发电厂材料中将人工神经网络在低循环疲劳和蠕变数据处理中的应用

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

In recent years, interest in artificial intelligence has been increasing, and efforts are being made to create synergy by utilizing artificial intelligence technology in the industrial field. Artificial neural networks, a field of artificial intelligence, simulate the neuron action of the brain and can provide an excellent solution to the problems of regression analysis or classification. The authors use an artificial neural network technique for material analysis data processing in order to overcome limitations of regression analysis using existing linear or quadratic polynomials for physical property data analysis. The first field of application is low-cycle fatigue test data processing. For the design of power generation components, design curves for component material properties are required. Among these is the fatigue design curve, which is necessary to ensure integrity against cyclic loading. In order to secure a Coffin-Manson design curve in low cycle fatigue testing, tests are carried out under various temperatures and stress ratios, and low cycle fatigue design curves for the tested conditions are obtained. However, to simulate a wide variety of design situations, fatigue design curves at arbitrary temperature and stress ratio conditions, which have not been tested, are also required. To solve this problem, an artificial neural network technique is proposed which can derive a general fatigue design curve equation for arbitrary conditions from a fatigue diagram obtained under specific conditions. The second area of use for neural network techniques is creep test data processing. Creep is a phenomenon that worsens with time in a high temperature environment, causing deformation and eventually destruction. Creep tests take a long time, and, therefore, there are still test specimens under test when calculating the design data. These long-time test data are important data for evaluating the long-term reliability of plant operation, but they are difficult to utilize. Survival analysis is applied to the data from specimens still being tested, and is limited to linear analysis. In this study, an artificial neural network technique was applied to an existing survival analysis method to extend the range of analysis and to improve the reliability of analysis so that a creep rupture line can be used even in nonlinear cases.
机译：近年来，对人工智能的兴趣一直在增加，正在通过在工业领域的人工智能技术利用人工智能技术来创造协同作用。人工神经网络，一种人工智能领域，模拟大脑的神经元作用，可以为回归分析或分类问题提供优异的解决方案。作者使用用于材料分析数据处理的人工神经网络技术，以克服使用现有线性或二次多项式进行物理性质数据分析的回归分析的限制。第一个应用领域是低周期疲劳测试数据处理。对于发电组件的设计，需要用于组件材料特性的设计曲线。其中是疲劳设计曲线，这是确保循环载荷的完整性所必需的。为了确保低循环疲劳测试中的棺材 - 曼森设计曲线，在各种温度和应力比下进行测试，获得用于测试条件的低循环疲劳设计曲线。然而，为了模拟各种各样的设计情况，还需要在尚未测试的任意温度和应力比条件下的疲劳设计曲线。为了解决这个问题，提出了一种人工神经网络技术，其可以从特定条件下获得的疲劳图中导出任意条件的一般疲劳设计曲线方程。用于神经网络技术的第二个使用领域是蠕变测试数据处理。蠕变是一种在高温环境中随时间恶化的现象，引起变形并最终破坏。蠕变测试需要很长时间，因此，在计算设计数据时仍有测试标本。这些长时间测试数据是评估工厂操作长期可靠性的重要数据，但它们难以利用。存活分析应用于仍在测试的标本中的数据，并限于线性分析。在该研究中，将人工神经网络技术应用于现有的存活分析方法以延长分析范围，提高分析的可靠性，从而即使在非线性情况下也可以使用蠕变破裂线。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号