【摘要】：隨著機器視覺技術(shù)的發(fā)展,越來越多的產(chǎn)品質(zhì)量檢測采用數(shù)字圖像處理技術(shù)進行分析與識別,能夠極大地提高生產(chǎn)的自動化程度。為實現(xiàn)工廠生產(chǎn)的故障指示器產(chǎn)品質(zhì)量智能檢測,本文基于卷積神經(jīng)網(wǎng)絡(luò)的故障指示器狀態(tài)識別進行研究,能夠有效地解決產(chǎn)品生產(chǎn)過程中的產(chǎn)品質(zhì)量檢測問題,實現(xiàn)工業(yè)生產(chǎn)自動化、智能化、綠色化和高效化。本文針對故障指示器狀態(tài)智能識別任務(wù),分別從構(gòu)建識別系統(tǒng)、改進卷積神經(jīng)網(wǎng)絡(luò)(Convolutional Neural Networks,CNN)模型、實驗驗證等方面進行了研究。通過分析故障指示器產(chǎn)品檢測場景,構(gòu)建了圖像采集系統(tǒng)并采集了故障指示器的原始視頻圖像。設(shè)計了識別算法流程,并實驗驗證了將CNN直接用到故障指示器原始圖片上進行狀態(tài)識別的可行性,同時分析實驗結(jié)果,找出了傳統(tǒng)CNN在此任務(wù)中存在的問題,啟發(fā)了后續(xù)對原始圖片的處理和CNN的改進工作。而后,針對現(xiàn)實場景中模糊、光照不均勻、色偏的故障指示器圖片,本文對圖片進行濾波、增強和高光消除等預(yù)處理,減少了各種因素對識別的影響,進一步采用基于閾值、邊緣檢測和聚類的方式對圖像進行分割實驗,接著對圖片進行平移、旋轉(zhuǎn)、縮放等數(shù)據(jù)擴充方式增大數(shù)據(jù)量,提升小樣本對訓練卷積神經(jīng)網(wǎng)絡(luò)的識別性能。針對傳統(tǒng)CNN模型魯棒性問題,本文改進網(wǎng)絡(luò)結(jié)構(gòu),對網(wǎng)絡(luò)加入尺度估計,提出了多尺度卷積神經(jīng)網(wǎng)絡(luò)模型,通過實驗驗證了該方法的魯棒性;針對傳統(tǒng)的CNN的收斂時間長,識別率低的問題,分析已收斂的CNN各核函數(shù)之間存在很大的相關(guān)性,提出了小波變換初始化第一層核函數(shù)的方法,實驗表明該方法既縮短了網(wǎng)絡(luò)收斂時間,又提高了識別率;將上述兩種改進方法的結(jié)合起來發(fā)揮了各自優(yōu)勢,與傳統(tǒng)的卷積神經(jīng)網(wǎng)絡(luò)相比,識別率提高7.28%,最終達到96.32%。最后,作為CNN算法的前沿技術(shù),集檢測與識別于一體的Faster R-CNN模型也被應(yīng)用到了本文的故障指示器狀態(tài)識別任務(wù)中,實驗結(jié)果表明,Faster R-CNN技術(shù)能夠有效解決故障指示器狀態(tài)識別問題。
[Abstract]:With the development of machine vision technology, more and more product quality detection uses digital image processing technology to analyze and identify, which can greatly improve the degree of automation of production. In order to realize the intelligent detection of the product quality of the fault indicator in the factory, this paper studies the status recognition of the fault indicator based on the convolution neural network, which can effectively solve the problem of the product quality detection in the process of product production. Realize industrial production automation, intelligence, green and high efficiency. In this paper, the task of intelligent recognition of fault indicator states is studied from the aspects of constructing recognition system, improving the Convolutional Neural Network (CNN) model, and experimental verification. By analyzing the detection scene of the fault indicator, the image acquisition system is constructed and the original video image of the fault indicator is collected. The recognition algorithm flow is designed, and the feasibility of using CNN directly on the original image of the fault indicator is verified. At the same time, the experimental results are analyzed, and the problems existing in the task of traditional CNN are found out. Inspired the subsequent processing of the original images and CNN improvements. Then, aiming at the fault indicator picture of fuzzy, uneven illumination and color deviation in the real scene, this paper preprocesses the image, such as filtering, enhancement and highlight elimination, which reduces the influence of various factors on the recognition, and further adopts the threshold based method. The methods of edge detection and clustering are used to segment the image, and then the image is expanded by translation, rotation, zoom and so on, which increases the amount of data and improves the recognition performance of the small sample to the training convolutional neural network. Aiming at the robustness problem of traditional CNN model, this paper improves the network structure, proposes a multi-scale convolution neural network model for network scale estimation, and verifies the robustness of this method through experiments, aiming at the long convergence time of traditional CNN. The problem of low recognition rate is analyzed. The correlation between the convergent CNN kernel functions is analyzed. A wavelet transform method is proposed to initialize the first layer kernel function. The experiments show that this method not only shortens the convergence time of the network, but also improves the recognition rate. Compared with the traditional convolution neural network, the recognition rate is increased by 7.28%, and finally reaches 96.3232%. Finally, as the cutting-edge technology of CNN algorithm, the Faster R-CNN model, which integrates detection and recognition, is also applied to the task of fault indicator status recognition in this paper. The experimental results show that the Faster R-CNN technique can effectively solve the problem of fault indicator state identification.
【學位授予單位】：華北電力大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：TP391.41;TP183

【參考文獻】

相關(guān)期刊論文 前10條

1 范榮;;基于卷積神經(jīng)網(wǎng)絡(luò)的服裝種類識別[J];現(xiàn)代計算機(專業(yè)版);2016年09期

2 姜磊;趙漢理;吳承文;;基于卷積神經(jīng)網(wǎng)絡(luò)的鞋類圖像檢索研究[J];現(xiàn)代計算機(專業(yè)版);2016年06期

3 孫志遠;魯成祥;史忠植;馬剛;;深度學習研究與進展[J];計算機科學;2016年02期

4 焦李成;楊淑媛;劉芳;王士剛;馮志璽;;神經(jīng)網(wǎng)絡(luò)七十年:回顧與展望[J];計算機學報;2016年08期

5 余萍;趙繼生;張潔;;基于非線性修正函數(shù)的卷積神經(jīng)網(wǎng)絡(luò)圖像識別研究[J];科學技術(shù)與工程;2015年34期

6 陽武;李倩;趙繼生;高強;余萍;;深度信念網(wǎng)絡(luò)在故障指示器檢測中的應(yīng)用[J];傳感器與微系統(tǒng);2015年07期

7 曉雷;;打造“智慧工廠” 設(shè)備裝備大有可為[J];中國設(shè)備工程;2015年01期

8 尹寶才;王文通;王立春;;深度學習研究綜述[J];北京工業(yè)大學學報;2015年01期

9 丁純;李君揚;;德國“工業(yè)4.0”:內(nèi)容、動因與前景及其啟示[J];德國研究;2014年04期

10 黃敦華;李勇;;物料工況監(jiān)測機器視覺系統(tǒng)應(yīng)用發(fā)展與探究[J];機電產(chǎn)品開發(fā)與創(chuàng)新;2012年06期

相關(guān)博士學位論文 前2條

1 郭輝;基于機器視覺的蜜柚品質(zhì)檢測方法研究[D];中國農(nóng)業(yè)大學;2015年

2 吳彰良;油封表面缺陷自動在線圖像檢測關(guān)鍵技術(shù)研究[D];天津大學;2013年

相關(guān)碩士學位論文 前3條

1 王一冰;基于機器視覺的玻璃瓶缺陷檢測與抓取系統(tǒng)研究[D];長春工業(yè)大學;2016年

2 楊建華;基于線結(jié)構(gòu)光視覺傳感器的三維表面測量系統(tǒng)關(guān)鍵技術(shù)研究[D];廣東工業(yè)大學;2013年

3 徐俊成;基于機器視覺的精密注塑零件的識別與檢測[D];北京化工大學;2011年

，

本文編號：2137631