基于MATLAB及小波變換的非平穩(wěn)隨機信號的消噪處理與分析

1.3萬字
自己原創(chuàng)的畢業(yè)論文，僅在本站獨家出售，重復(fù)率低，推薦下載使用

摘要：滾動軸承是旋轉(zhuǎn)機械中應(yīng)用最廣泛的機械零件，也是最易損壞的元件之一。旋轉(zhuǎn)機械的很多故障都與滾動軸承有關(guān)，軸承工作的好壞對機械的工作狀態(tài)有很大的影響，所以對軸承振蕩信號的故障診斷尤為重要，而對滾動軸承信號的消噪處理就是故障診斷的第一步，也是很關(guān)鍵的一步，消噪的好壞對下一步的特征值提取和故障信號分類有著很大的影響。
本文在MATLAB軟件平臺上，利用滾動軸承正常情況下的信號與高斯白噪聲信號疊加生成帶噪信號，用小波變換對帶噪信號進行了強制、默認閾值、給定軟閾值、給定硬閾值、自適應(yīng)閾值硬閾值、自適應(yīng)閾值軟閾值、啟發(fā)式閾值硬閾值、啟發(fā)式閾值軟閾值、閾值等于sqrt(2*log(length(x)))硬閾值、閾值等于sqrt(2*log(length(x)))軟閾值、用極大極小原理選擇閾值的硬閾值、用極大極小原理選擇閾值的軟閾值消噪、小波包硬消噪、小波包軟消噪總共十四種小波消噪方法來消噪，并提取滾動軸承正常情況下信號的均值和方差以及十四種消噪后的各自信號的均值和方差，將消噪后的信號的特征值與滾動軸承正常情況下的信號的特征值進行比較，觀察哪種更接近正常情況下信號的特征值，說明哪種消噪方法更好。同時本文還提取了滾動軸承正常情況下消噪后的信號，內(nèi)圈消噪后信號，外圈消噪后信號的特征值（均值，方差），并畫出這三種信號消噪后功率譜密度圖，結(jié)合均值，方差，功率譜密度圖判斷出了其中任何一個信號的故障類型。
通過特征值的比較發(fā)現(xiàn)啟發(fā)式閾值硬閾值消噪方法在對帶噪信號的消噪方法中是最好的方法，然后利用這種方法再對滾動軸承正常情況下的信號，內(nèi)圈故障信號，外圈故障信號進行消噪，再對三種消噪后的信號進行特征值的提取，比較三種消噪后信號的特征值，發(fā)現(xiàn)均值，方差都可以用來區(qū)分這三種信號，最后生成三種消噪后信號的功率譜密度圖，比較三者的功率譜密度圖發(fā)現(xiàn)也可以通過功率譜密度圖區(qū)分這三種信號，因此本文為大家提供了十四種信號的消噪方法，還提供了三種滾動軸承故障診斷方法。

關(guān)鍵詞：非平穩(wěn)隨機信號 滾動軸承信號 小波變換 消噪 MATLAB軟件

Based on MATLAB wavelet transform de-noising non-stationary random signal processing
Abstract Rolling is the widely used in rotating machinery mechanical parts, which is one of the most vulnerable components. Many failures of rotating machinery are concerned with rolling bearings, bearing good and bad work has a great influence on the mechanical working condition, which can lead to defective equipment generates abnormal vibration and noise, and even cause damage to the equipment, so the oscillation signal bearing fault diagnosis particularly important, while the noise canceling signal processing is the first step in rolling bearing fault diagnosis, but also a very crucial step, noise-canceling feature is good or bad value for the next fault signal extraction and final classification has a great impact.
In this paper, the MATLAB software platform, using the signal with a Gaussian white noise signal superimposed to generate noisy signals of rolling bearings under normal circumstances, using wavelet transform noisy signal forced default threshold, given the soft threshold, given the hard threshold, since Adaptive Threshold hard threshold, adaptive threshold soft threshold, the heuristic threshold hard threshold, the heuristic threshold soft threshold, the threshold is equal to sqrt (2 * log (length (x))) hard threshold, the threshold is equal to sqrt (2 * log (length ( x))) soft threshold, select the hard threshold threshold Minimax principle, choose soft threshold de-noising threshold Minimax theory, wavelet packet de-noising hard, soft wavelet packet wavelet de-noising total of fourteen kinds of consumer noise method to eliminate noise and extract the mean and variance of the mean and variance of the signal and noise cancellation after fourteen respective signals of rolling bearings under normal circumstances, the value of the characteristic features of the signal after de-noising and Rolling signal under normal circumstances value, which is more close to the characteristic values observed under normal circumstances, signal, indicating what kind of de-noising method is better. It also extracts the signal while the rolling bearing under normal circumstances, the signal characteristic value of the inner and outer signal (mean and variance), and draw the power spectral density of the three signals, combining the mean, variance, power spectral density determine which type of failure of any one of the signal.
Eigenvalues found by comparing the heuristic threshold hard threshold de-noising method for noisy signal de-noising method is the best method, and then use this method and then the signals of rolling bearings under normal circumstances, the fault signal inner and outer de-noising ring fault signal, then the signal is normal rolling bearings, the inner ring fault signal extracting outer fault signal characteristic value, comparison of three characteristic values of the original signal and found the mean, the variance can be used to distinguish between these three signal, and finally generate the power spectral density of three of the original signal, comparing the three power spectral density can be found in..