复合材料科学与工程 ›› 2015, Vol. 0 ›› Issue (11): 15-19.

• 基础研究 • 上一篇    下一篇

基于模态应变能理论的风机叶片结构损伤辨别仿真

张鑫, 顾桂梅   

  1. 兰州交通大学自动化与电气工程学院,兰州 730070
  • 收稿日期:2015-06-03 出版日期:2015-11-28 发布日期:2021-09-14
  • 作者简介:张鑫(1991-),男,硕士,主要研究方向为风机叶片故障诊断与机器学习,zx901019@126.com。
  • 基金资助:
    兰州交通大学科技支撑基金(ZC2012008);甘肃省高等学校科研项目(42015274)

SIMULATION FOR STRUCTURE DAMAGE INDENTIFICATION OF WIND TURBINE BLADE BASED ON MODE STRAIN ENERGY THEORY

ZHANG Xin, GU Gui-mei   

  1. School of Automation & Electrical Engineering, Lanzhou Jiao Tong University, Lanzhou 730070, China
  • Received:2015-06-03 Online:2015-11-28 Published:2021-09-14

摘要: 为解决风机叶片在多种损伤工况下的结构损伤识别问题,采用一种模态应变能变化率和BP神经网络结合的分步判别方法。首先利用ANSYS软件建立风机叶片的结构模型,选取模态应变能变化率作为检测结构损伤的特征参数;然后分别在单损伤与多损伤工况下,通过比较损伤前后各单元的模态应变能变化初步定位损伤;最后通过BP神经网络精确判定损伤状态,辨别叶片的损伤位置和程度。结果表明,通过比较各单元模态应变能变化率,可以有效实现损伤的初步定位,利用BP神经网络可以准确诊断出叶片损伤位置和程度。这种方法对同一单元不同程度损伤、同程度多损伤等多种损伤工况均有良好的识别效果。

关键词: 风机叶片, 损伤识别, 模态应变能, BP神经网络

Abstract: To solve damage identification problem of wind turbine blade structure in different cases, a stepwise method based on unit mode strain energy change rate and BP neural network is used. Change of mode strain energy per unit is compared to judge damage preliminarily. A mapping model is built for reflecting the relationship between the characteristic parameter and damage state to indentify the location and degree of damage, and the mode strain energy change rate is treated as characteristic parameter characterized structure damage. The results show that unit mode strain energy can judge possible damage unit effectively, and BP neural network can indentify structural damage location and degree more accurately. The method has good recognition effect for many damage cases.

Key words: wind turbine blade structure, damage identification, mode strain energy, BP neural network

中图分类号: