EMD及其改进算法在水工结构振动信号处理中的应用

发布时间：2018-10-15 08:25

【摘要】：水工结构振动信号在输送和获取的过程中,容易受到环境激励的高频白噪声和低频水流噪声的干扰,通常表现为低信噪比、非平稳随机信号,结构振动特征信息完全淹没在强噪声中,难以精确识别其模态信息,从而影响判断结构健康状况及振动危害评价的精度。因此,需采取有效的信号分析方法对实测数据降噪处理,以获取结构振动信号的优势特征信息。本文针对水工结构振动信号非平稳性和特征信息被强噪声淹没的实际问题,以EMD算法的自身特点及其不断发展完善为线索,全面探索不同阶段EMD算法在水工结构振动信号处理中的应用,研究其在水工结构信号处理中的特点及优势,以期得到较好适应于水工结构信号处理的方法,实现强噪声背景下泄流结构工作特性有效信息的提取,为结构下一步健康诊断工作提供帮助。本文所做的主要工作和得到的结论如下:1、为探索EMD算法在水工结构振动信号处理中的应用,针对水工结构振动信号的特点,介绍了一种联合运用小波阈值与EMD算法对水工结构振动信号进行降噪的新思路。仿真信号计算结果表明,小波阈值与EMD联合滤波降噪是相对优越的降噪方法。拉西瓦拱坝工程实例计算结果表明该方法可以有效的完成降噪的任务,准确得到坝体的振动信息和优势频率,为大坝的在线监测与安全健康诊断提供帮助。2、充分发挥正交化经验模态分解的优点,介绍了一种基于奇异值分解(SVD)和改进EMD联合的水工结构振动信号特征信息提取方法。该方法通过SVD将振动信号中的高频噪声滤除,并通过正交化EMD将低频水流噪声滤除,实现信号的二次滤波,最终得到水工结构的工作振动特征信息。仿真信号计算结果表明该方法的正确性,结合三峡5号坝段泄流振动实测数据,运用该方法进行坝体特征信息提取,并与ERA辨识结果进行比较,说明该方法在水工结构振动信息分析中的优越性,具有良好的降噪能力和工程实用性,可为水工结构在线监测和安全运行提供帮助。3、详细介绍了CEEMDAN算法和排列熵的工作原理,并充分发挥二者优势,提出了基于CEEMDAN和排列熵联合提取水工结构特征信息的方法。通过构造仿真数据,对比CEEMDAN算法、SVD以及CEEMDAN-PE-SVD算法三者的降噪结果,计算结果表明CEEMDAN-PE-SVD方法能够有效地滤除信号中的干扰成分,还原信号的优势特征频率,具有较高的提取精度,属于更优越的信号降噪方法。将该方法应用于三峡重力坝泄流工程,表明该方法能够精确提取结构的工作特征信息,抗噪性强,实用性强,具有极佳的应用前景。4、针对水工结构振动信号的特点,以EMD算法的不断改进发展为线索,研究不同阶段EMD算法的自身特点及其在水工结构信号处理中的应用。研究结果表明经验模态分解可以很好地应用于水工结构振动信号处理中,可为解决水工结构振动信号处理提供新思路。
[Abstract]:The vibration signals of hydraulic structures are easily disturbed by high frequency white noise and low frequency water flow noise in the process of conveying and obtaining the vibration signals of hydraulic structures, which usually appear as low signal-to-noise ratio (SNR) and non-stationary random signals. The vibration characteristic information of the structure is completely submerged in the strong noise, so it is difficult to identify the modal information accurately, thus affecting the accuracy of judging the health condition of the structure and the evaluation of the vibration hazard. Therefore, it is necessary to adopt effective signal analysis method to reduce the noise of the measured data in order to obtain the advantage characteristic information of the structural vibration signal. Aiming at the practical problem that the vibration signal of hydraulic structure is not stationary and characteristic information is submerged by strong noise, this paper takes the characteristic of EMD algorithm and its continuous development and perfection as the clue. This paper probes into the application of EMD algorithm in different stages in the vibration signal processing of hydraulic structures, studies its characteristics and advantages in the signal processing of hydraulic structures, in order to obtain a better method suitable for the signal processing of hydraulic structures. The effective information extraction of the working characteristics of the discharge structure under the strong noise background is realized, which provides the help for the next health diagnosis of the structure. The main work and conclusions obtained in this paper are as follows: 1. In order to explore the application of EMD algorithm in vibration signal processing of hydraulic structures, the characteristics of vibration signals of hydraulic structures are discussed. A new method for noise reduction of hydraulic structure vibration signal using wavelet threshold and EMD algorithm is introduced. The simulation results show that the wavelet threshold combined with EMD filtering is a relatively superior denoising method. The result of practical example of Laxiwa arch dam project shows that the method can effectively accomplish the task of noise reduction and accurately obtain the vibration information and dominant frequency of the dam body. In this paper, the advantages of orthogonal empirical mode decomposition (EMD) are brought into full play. A method based on singular value decomposition (SVD) and improved EMD is introduced to extract the characteristic information of vibration signals of hydraulic structures. In this method, the high frequency noise in the vibration signal is filtered by SVD, and the low frequency water flow noise is filtered by orthogonal EMD to realize the secondary filtering of the signal. Finally, the working vibration characteristic information of hydraulic structure is obtained. The result of simulation signal calculation shows that the method is correct. Combining with the measured data of discharge vibration of dam section 5 of the three Gorges Dam, the method is used to extract the characteristic information of the dam body, and the result is compared with the result of ERA identification. The advantages of this method in the vibration information analysis of hydraulic structures are illustrated. The method has good noise reduction ability and engineering practicability. It can provide help for on-line monitoring and safe operation of hydraulic structures. 3. The CEEMDAN algorithm and the working principle of permutation entropy are introduced in detail. Based on CEEMDAN and permutation entropy, the method of extracting the characteristic information of hydraulic structure is put forward. By constructing the simulation data and comparing the noise reduction results of CEEMDAN algorithm, SVD algorithm and CEEMDAN-PE-SVD algorithm, the results show that the CEEMDAN-PE-SVD method can effectively filter the interference components in the signal, restore the dominant characteristic frequency of the signal, and have a high extraction accuracy. It belongs to better signal denoising method. The method is applied to the discharge project of the three Gorges Gravity Dam. It shows that the method can extract the working characteristic information of the structure accurately, has strong anti-noise, strong practicability, and has excellent application prospect. 4, aiming at the characteristics of the vibration signal of hydraulic structure, Based on the continuous improvement and development of EMD algorithm, the characteristics of EMD algorithm in different stages and its application in hydraulic structure signal processing are studied. The results show that the empirical mode decomposition can be well applied to the vibration signal processing of hydraulic structures and can provide a new idea for solving the vibration signal processing of hydraulic structures.
【学位授予单位】：华北水利水电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TV312

【参考文献】