CNC lathe tool wear analysis using image processing and stray flux

Tornalanmış yüzey imgeleri gri ton ortalama değerlerinin takım aşınması ile ilişkisinin incelenmesi

“…Her ne kadar şu ana kadar literatürde ve özel sektör girişimlerinde yüksek hızlı görüntü yakalama ile gerçekleştirilen gerçek zamanlı bir takım izleme yöntemi geliştirilmemiş olsa bile tornalanmış yüzey görüntüleri üzerinden takım aşınmasını yansıtabilecek öznitelikler ve karar verme algoritmaları birçok araştırmacı tarafından yukarıda bahsedilen sinyal tabanlı aşınma takip yöntemlerine alternatif olarak araştırılmaktadır. Günümüz literatüründe; kenar bulma algoritmaları [7], histogram [8], dalgacık (wavelet) [9], gri seviye eş oluşum matrisi [10], Voronoi mozaikleme [11] ve Fourier dönüşümü [12] gibi çeşitli imge/doku analiz yöntemleri yüzey imgelerinden aşınmayı yansıtabilecek özniteliklerin çıkarılması için kullanılmıştır.…”

Methodology for Tool Wear Detection in CNC Machines Based on Fusion Flux Current of Motor and Image Workpieces

“…Jaen-Cuellar et al [19] reported the use of stray flux and current for the detection of three levels of wear with individual variation of two machining parameters for aluminum 6061 turning with coated carbide tools by means of feature extraction, dimensionality reduction and classification with an ANN, achieving a top performance of 94.4%. Diaz-Saldaña et al [20] employed stray flux signals and image analyses of the surface finish to identify three levels of wear on coated carbide tools with variation in cutting speed in aluminum 6061 turning, obtaining a good differentiation between all the conditions when applying histogram peak counts to the images and feature extraction on the stray flux signals.…”

“…Several literature reviews have identified the main techniques, tools and trends for processing [21][22][23][24]. Firstly, signals are processed directly in the time domain [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18] or techniques or transforms are used for their analyses in the frequency [3,5,[18][19][20] or time-frequency [19,20] domains. From here, several techniques are used to obtain features that allow the analysis to be carried out in a better way, such as the use of statistical indicators [3][4][5][6]8,14,[18][19][20], time or time-frequency transforms for direct feature extraction [3][4][5]19,20] and, in some cases, methods for the selection of the most appropriate features or dimensionality reduction such as heuristic techniques [5,6,22], linear discriminant analysis (LDA) [19] or principal component analysis (PCA).…”

“…The use of intelligent techniques is of great value when processing large amounts of information generated by fused sensors, allowing a more adequate manipulation and interpretation of the data. Another point to note is the implementation of the methodologies, since they can be performed online, allowing process monitoring [3][4][5][7][8][9][10][11][12][14][15][16][17][18][19], or offline once the machining of a part has been carried out [6,13,20,26]. This is an important aspect, since in methodologies that use vision systems or images, the machining process must be stopped for the acquisition of images of the tool or they may be taken after the process; however, this depends on the needs and planning of the process.…”