本文選題：無損檢驗機器人 + 焊縫識別��； 參考：《深圳大學》2017年碩士論文

【摘要】：通常情況下,無損檢驗機器人的路徑規(guī)劃問題是通過人工標識或者位置要求嚴格的離線編程方式來解決的,而這兩種路徑規(guī)劃的方式常常會受到檢驗環(huán)境的限制而導致檢驗效率低下、精度不高,同時也無法完全做到智能檢驗。所以研究焊縫識別技術(shù)在克服無損檢驗機器人傳統(tǒng)路徑規(guī)劃的缺點、提升無損檢驗機器人的焊縫識別效率與精度、解決無損檢驗機器人的自動跟蹤難題、實現(xiàn)焊縫無損檢驗的智能化與自動化等方面有重要的意義。在無損檢驗機器人焊縫識別技術(shù)研究中,焊縫識別方法研究是其關(guān)鍵。本文針對無損檢驗機器人的焊縫識別問題,研究不同的焊縫識別方法,并將研究的不同焊縫識別方法構(gòu)成一套焊縫識別技術(shù)評價系統(tǒng),同時建立了相應的算法性能評價指標。針對不同的焊縫工件,無損檢驗機器人焊縫識別技術(shù)評價系統(tǒng)根據(jù)算法性能評價指標選擇一種合適的焊縫識別算法完成焊縫的識別。本文通過分析焊縫圖像的視覺特征,提出了基于圖像邊界特征的焊縫識別算法;本文通過分析焊縫圖像的統(tǒng)計特征,提出了基于圖像紋理特征的焊縫識別算法。本文在焊縫識別算法研究中既含有普通的圖像識別方法(如利用投影技術(shù)識別焊縫)又覆蓋復雜的圖像識別方法(如利用人工神經(jīng)網(wǎng)絡技術(shù)識別焊縫)。而且本文采用鋼管焊縫與相貫線焊縫兩種不同類型的焊縫對本文所提出的焊縫識別算法分別進行了仿真實驗驗證,實驗驗證的結(jié)果表明本文所提出的焊縫識別算法均可完成焊縫的識別。最后在Matlab平臺上基于本文所研究的焊縫識別方法搭建了一個焊縫識別算法網(wǎng)絡,構(gòu)建無損檢驗機器人焊縫識別技術(shù)評價系統(tǒng),并針對不同類型的焊縫分別進行了仿真實驗,實驗結(jié)果表明根據(jù)該評價系統(tǒng)中的性能評價指標可以從系統(tǒng)中選擇合適的焊縫識別算法實現(xiàn)焊縫的識別。本文對無損檢驗機器人的焊縫識別技術(shù)進行了理論研究和實驗測試,并根據(jù)提出的方法搭建實驗平臺,有助于提高無損檢驗機器人焊縫識別的精度與穩(wěn)定性、解決無損檢驗機器人自動跟蹤問題,為焊縫無損檢驗系統(tǒng)的智能性奠定理論基礎(chǔ)。同時,本文所研究的內(nèi)容也可為無損檢驗機器人的視覺引導技術(shù)研究提供理論指導。
[Abstract]:In general, the path planning problem of nondestructive testing robot is solved by manual marking or strict off-line programming. However, these two ways of path planning are often limited by the inspection environment, which leads to low efficiency and low accuracy, and it is also unable to achieve intelligent testing completely. Therefore, the research of weld seam identification technology in overcoming the shortcomings of traditional path planning of nondestructive inspection robot, improving the efficiency and precision of weld identification of non-destructive testing robot, solving the problem of automatic tracking of non-destructive testing robot. It is of great significance to realize the intelligence and automation of weld non-destructive testing. In the research of nondestructive inspection robot welding seam identification method is the key. In this paper, aiming at the problem of weld identification of nondestructive testing robot, different weld identification methods are studied, and a set of evaluation system of weld seam identification technology is constructed, and the corresponding algorithm performance evaluation index is established at the same time. For different weld parts, the evaluation system of nondestructive inspection robot weld identification technology selects a suitable weld identification algorithm according to the algorithm performance evaluation index to complete the weld seam identification. By analyzing the visual features of weld images, a weld recognition algorithm based on image boundary features is proposed, and a weld recognition algorithm based on image texture features is proposed by analyzing the statistical features of weld images. In this paper, there are not only common image recognition methods (such as using projection technology to identify weld seam) but also complicated image recognition methods (such as artificial neural network) in the research of weld seam recognition algorithm. In this paper, two different types of weld seam, steel pipe weld and intersecting line weld, are used to verify the proposed weld identification algorithm. The experimental results show that the proposed algorithm can be used to identify the weld. Finally, a weld recognition algorithm network is built on the Matlab platform based on the welding seam identification method studied in this paper, and a non-destructive inspection robot weld recognition technology evaluation system is constructed, and the simulation experiments are carried out for different types of weld seam respectively. The experimental results show that the proper weld recognition algorithm can be selected from the system according to the performance evaluation index of the evaluation system. In this paper, the welding seam recognition technology of the nondestructive testing robot is studied theoretically and experimentally, and the experimental platform is built according to the proposed method, which is helpful to improve the accuracy and stability of the weld seam identification of the nondestructive testing robot. The problem of automatic tracking of nondestructive testing robot is solved, which lays a theoretical foundation for the intelligence of weld nondestructive testing system. At the same time, the contents of this paper can also provide theoretical guidance for the study of visual guidance technology of NDT robots.
【學位授予單位】：深圳大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：TP391.41;TP242

【參考文獻】

相關(guān)期刊論文 前10條

1 孫浩益;梁冬泰;李國平;周善e，

本文編號：1800889