基于多模态数据融合的电厂风机叶片故障检测方法研究
Research on Fault Detection Method for Power Plant Fan Blades Based on Multimodal Data Fusion
张家玉1
作者信息
- 1. 国家电投集团协鑫滨海发电有限公司,江苏 盐城 224500
- 折叠
摘要
常规的电厂风机叶片故障检测方法以检测数据不平衡问题检测为主,部分微弱的故障数据未被整合在数据集中,影响了最终故障检测的准确性,因此,设计了基于多模态数据融合的电厂风机叶片故障检测方法.提取电厂风机叶片声信号异常特征,随机截取风机叶片的异常声信号,确保故障检测的准确性.基于多模态数据融合构建风机叶片故障检测模型,分析运行状态下的叶片模态,以满足叶片故障检测的准确性需求.设定电厂风机叶片故障随机共振参数,将叶片频率范围、噪声强度、随机共振时间常数等方面考虑在内,以实现最佳的故障检测效果.采用对比实验,验证了该方法的故障检测准确性更高,能够应用于实际生活.
Abstract
Conventional fault detection methods for wind turbine blades in power plants mainly focus on detecting imbalanced data,and some weak fault data is not integrated into the dataset,which affects the accuracy of final fault detection,therefore,a fault detection method for power plant fan blades based on multimodal data fusion was designed.Extract abnormal characteristics of sound signals from wind turbine blades in power plants,randomly extract abnormal sound signals from wind turbine blades,and ensure the accuracy of fault detection.Building a fault detection model for wind turbine blades based on multimodal data fusion,analyzing the blade modes under operating conditions,in order to meet the accuracy requirements of blade fault detection.Set random resonance parameters for power plant fan blade faults,taking into account blade frequency range,noise intensity,random resonance time constant,etc.,to achieve the best fault detection effect.Through comparative experiments,it was verified that the fault detection accuracy of this method is higher and can be applied in practical life.
关键词
多模态数据融合/电厂/风机/叶片故障/检测方法Key words
multimodal data fusion/power plants/fan/blade failure/detection method引用本文复制引用
出版年
2024
NETL
NSTL
万方数据