基于徑向基網(wǎng)絡(luò)的鋼板表面缺陷檢測(cè)算法研究



摘  要

作為鋼鐵工業(yè)的主要產(chǎn)品形式之一, 鋼板已經(jīng)成為汽車、機(jī)械制造、化工、航空航天和造船等工業(yè)不可缺少的原材料, 因此其表面質(zhì)量的優(yōu)劣將直接影響這些產(chǎn)品的最終性能和質(zhì)量。在鋼板制造過程中, 由于連鑄鋼坯、軋制設(shè)備、加工工藝等多方面的原因, 導(dǎo)致鋼板表面出現(xiàn)裂紋、氧化皮、結(jié)疤、輥印、刮傷、空洞、針眼、折疊、表面分層等不同類型的表面缺陷, 這些缺陷不僅影響產(chǎn)品的外觀, 更嚴(yán)重地降低了產(chǎn)品的抗腐蝕性、耐磨性和疲勞強(qiáng)度等性能。

神經(jīng)網(wǎng)絡(luò)理論作為一門新興學(xué)科, 近年來被廣泛應(yīng)用于鋼板的表面缺陷檢測(cè)中, 但大部分都是基于BP 神經(jīng)網(wǎng)絡(luò)進(jìn)行識(shí)別的, BP 網(wǎng)絡(luò)的缺點(diǎn)是對(duì)于干擾量非常敏感、隱含層節(jié)點(diǎn)數(shù)需要經(jīng)過多次嘗試、學(xué)習(xí)速度慢且易陷入局部極小點(diǎn)。

本文使用RBF 神經(jīng)網(wǎng)絡(luò)( RBFNN)對(duì)鋼板缺陷數(shù)據(jù)集進(jìn)行計(jì)算機(jī)仿真,將缺陷識(shí)別結(jié)果與期望結(jié)果進(jìn)行對(duì)比。 由于其結(jié)構(gòu)簡(jiǎn)單,每個(gè)隱含層節(jié)點(diǎn)映射函數(shù)是不同的非線性函數(shù),所以可以保證快速的學(xué)習(xí)速度, 并且是全局最優(yōu)。

關(guān)鍵詞：徑向基函數(shù)（RBF）；人工神經(jīng)網(wǎng)絡(luò)；鋼板表面缺陷；缺陷識(shí)別

 


 


 


 


 


 




Abstract

As one of the major forms of production in steel industry，steel sheet has now become a kind of indispensable material for many industries like automobile，mechanical ma nufacturing，chemistry，aviation，astr0navigation，ship building，etc．Its surface quality affects the final production’S perform ance and quality. In steel manufacturing process, because continuous casting billet, rolling equipment, processing technology and so on various causes, the steel plate surface occurrence crackle, descaling, scarring and roller and scratches, hollow, needle, folding, surface layer etc of different types of surface flaws that not only affect the product's appearance, the more severely reduced products corrosion resistance, abrasion resistance and fatigue strength of performance.

Neural network theory, as a new subject, widely used in recent years in the steel sheet surface defects detection, but mostly based on BP neural network to identify, BP network drawback is for interference quantity quite sensitive, hidden nodes several need after many attempts, the slow learning speed and easy into the local minimum points.

The paper uses RBF neural network (RBFNN) to recognize the fault dataset of iron sheets using computer simulations. The recognized results are compared with the expected outputs. Due to its simple structure, a variety of mapping functions, and rapid learning speed, RBFNN can achieve a global optimization.


Keywords: Radial Basis Function (RBF); Artificial Neural Network; Surface Defect; Defect Recognition


 


 




目  錄

第一章 緒 論 1

1.1  課題的研究目的 1

1.2  課題研究的意義 1

1.3  課題研究的內(nèi)容與方法 2

第二章 模式識(shí)別技術(shù)與人工神經(jīng)網(wǎng)絡(luò)介紹 4

2.1  模式識(shí)別技術(shù) 4

2.1.1  模式識(shí)別的基本概念 4

2.1.2  識(shí)別對(duì)象的特征及其提取 4

2.1.3  模式識(shí)別分類 5

2.2  人工神經(jīng)網(wǎng)絡(luò)簡(jiǎn)介 6

2.2.1  人工神經(jīng)網(wǎng)絡(luò)模型 6

2.2.2  人工神經(jīng)網(wǎng)絡(luò)發(fā)展史 7

2.2.3  人工神經(jīng)網(wǎng)絡(luò)的分類及學(xué)習(xí)規(guī)則 8

2.2.4  人工神經(jīng)網(wǎng)絡(luò)的應(yīng)用 12

第三章 基于徑向基網(wǎng)絡(luò)的鋼板表面缺陷檢測(cè)仿真實(shí)現(xiàn) 14

3.1  徑向基網(wǎng)絡(luò)基本介紹 14

3.1.1  徑向基網(wǎng)絡(luò)模型 14

3.1.2  徑向基網(wǎng)絡(luò)的結(jié)構(gòu)與特點(diǎn) 15

3.2  MATLAB神經(jīng)網(wǎng)絡(luò)工具箱簡(jiǎn)介 17

3.3  基于徑向基網(wǎng)絡(luò)的鋼板表面缺陷檢測(cè)仿真實(shí)現(xiàn) 18

3.3.1  鋼板表面缺陷檢測(cè)過程設(shè)計(jì) 18

3.3.2 仿真結(jié)果分析 27

結(jié)論 29

參考文獻(xiàn) 30

致  謝 31

附  錄 32

1.程序矩陣 32

2．程序清單 34

3．實(shí)驗(yàn)流圖 36