基于带通滤波技术的数据分解方法与工程应用

发布时间：2018-01-07 14:38

本文关键词：基于带通滤波技术的数据分解方法与工程应用　出处：《浙江海洋学院》2015年硕士论文　论文类型：学位论文

【摘要】：随着海洋工程发展规模的迅速扩大,人们可以获取的各类海洋信息也日益增多,基于有限的海洋数据信息,准确预测海洋环境和海洋灾害成为海上工程建设的迫切需要。人们对海上结构安全性的重视程度日渐提高,也必然会对海洋信息的分析处理工作提出更高的要求。将不再满足模糊的定性分析,更希望获取准确的定量分析结果,数据处理速度上也由低效率的人工处理方式步入高效的自动或智能化处理。因此更加有效地处理、分析资料,消除各种测量误差和系统误差,进一步提高其使用价值,充分利用和发挥现有数据的使用价值,成为当下的重点研究内容。本文较为全面的介绍了当今几种常用的信号分析方法,客观的评价了这几种信号(数据)分析方法的优点和缺点,针对非平稳信号的特性,提出了2种新的信号(数据)分析方法:基于FFT和带通滤波技术的经验模态分解法(FB-EMD)和FFT带通滤波平滑方法。FB-EMD法主要沿用Huang-HHT经验模态分解和固有模态函数的提法,从信号分解思想来看,本文方法可以看作是HHT经验模态分解的改进型。FB-EMD法采用新的分解方法和分解准则,可以将平稳随机过程和非平稳随机过程分解成一系列两两正交的固有模态函数。经实验室数据验证,FB-EMD法可以较好地完成各种信号的分解。本文还对HHT和FB-EMD作定性比较,在分解简单Stokes 2阶波时,本文得到的FB-EMD法要明显的优于HHT的经验模态分解方法,本方法可以快速、准确地对信号进行有效的分析,并且分解结果完全符合信号的原始特征。秉承了HHT方法具有瞬时频率分析的特性,本文FB-EMD分解法也可以用于信号的瞬时频率分析,从理论上看,凡是HHT信号分析方法能解决的问题,FB-EMD方法都能够进行替代。后一种方法主要用于对波浪信号的平滑处理上,FFT平滑方法克服现有海浪平滑方法存在的问题,能对数据进行较好的平滑。与常用平滑方法相比,FFT平滑方法理论依据充分,工程上简单易用,平滑效果较好,非常适合于计算机自动化处理。这两种信号分析方法扩展了现有信号分析方法的应用范畴。可以期望,这两种信号分析方法会有更广阔的应用前景。
[Abstract]:With the rapid expansion of marine engineering scale, people can obtain more and more kinds of marine information, based on limited ocean data information. Accurate prediction of marine environment and marine disasters has become an urgent need for marine engineering construction. People pay more and more attention to the safety of marine structures. It is also bound to put forward higher requirements for the analysis and processing of marine information. It will no longer meet the needs of fuzzy qualitative analysis and more hope to obtain accurate quantitative analysis results. Data processing speed is also from inefficient manual processing to efficient automatic or intelligent processing, so more effective processing, analysis of data, eliminate all kinds of measurement errors and system errors. Further improve its use value, fully use and play the use of existing data value, become the focus of the current research. This paper more comprehensive introduction of the current several commonly used signal analysis methods. The advantages and disadvantages of these signal (data) analysis methods are evaluated objectively, aiming at the characteristics of non-stationary signals. Two new signal (data) analysis methods are proposed: empirical mode decomposition method based on FFT and bandpass filtering technique and FB-EMD). And FFT bandpass filter smoothing method. FB-EMD method mainly uses the Huang-HHT empirical mode decomposition and inherent mode function. From the viewpoint of signal decomposition, the method in this paper can be regarded as the improved. FB-EMD method of HHT empirical mode decomposition using new decomposition method and decomposition criterion. The stationary stochastic process and the non-stationary stochastic process can be decomposed into a series of natural mode functions of pairwise orthogonality, which are verified by laboratory data. The FB-EMD method can be used to decompose all kinds of signals. The qualitative comparison between HHT and FB-EMD is made in this paper, when the simple Stokes 2 wave is decomposed. The FB-EMD method obtained in this paper is obviously superior to the empirical mode decomposition method of HHT. This method can be used to analyze the signal quickly and accurately. And the decomposition results completely accord with the original characteristics of the signal. According to the HHT method has the characteristics of instantaneous frequency analysis, this FB-EMD decomposition method can also be used in the instantaneous frequency analysis of signals, theoretically. The FB-EMD method can be replaced by HHT signal analysis method, which is mainly used to smooth wave signal processing. FFT smoothing method overcomes the problems existing in the existing wave smoothing methods and can smooth the data better. Compared with the usual smoothing methods, the FFT smoothing method has sufficient theoretical basis and is simple and easy to use in engineering. The smoothing effect is good, which is very suitable for computer automation. These two signal analysis methods extend the scope of application of existing signal analysis methods and can be expected. These two signal analysis methods will have a wider application prospect.
【学位授予单位】：浙江海洋学院
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN713

【相似文献】