Inspecting flank wear on cutting tools

The current production tendency is the improvement of both production performance and quality levels in order to reduce costs and to avoid scrap. In the industrial manufacture flexible production systems with high performance and quality characteristics are required. Hence,  the antiquated quality assurance method by measuring the specification conformity of a product at the end of the production line is replaced by a preventive quality strategy with inline-metrology. Milling and turning are very  common processes  in industry today. Therefore, process monitoring of these machining processes has become of crucial importance to optimise production in view of quality and costs. Monitoring methods focus on the inspection of important process parameters, such as cutting forces, temperature and tool wear. Tool wear is usually the most relevant parameter inspected, as it has direct influence on the final product quality, the machine tool performance and the tool lifetime.

The flank wear is the most referred tool wear parameter in the monitoring of machining processes – it allows to estimate the cutting tool’s lifetime and to control the production process. Generally, the tool wear inspection is performed in  three  different  ways. First, a statistical evaluation, based on estimated or FEM (Finite Element Method) simulated lifetime intervals is possible. Second, process signals like the cutting forces or the acoustic emission can be used for a wear analysis. These indirect techniques try to evaluate the tool wear by inspecting the process data, which have a tight relationship with the tool wear. Third, a direct measurement on the cutting edge can be performed by using optical sensors.

Tool edge detection

In order to measure the tool wear,  it is only necessary to precede tool area within the acquired image. The tool area could be separated from image background by finding the top and side tool edges in the full-illuminated image. To detect both edges, two ROIs (regions of interest) on each edge are predefined. The locations of the ROIs in the image are calculated based on tool type information (e.g. radius, length). After applying the Canny edge detector to the image areas defined by ROIs, image pixels on tool borders are extracted. By fitting the sequence of detected points along the tool borders with a line function, the top and side tool edges are determined.

Wear type classification

On the base of the extracted outer contour of the wear region, the classification of tool wear Source: R. Schmitt1, Y. Cai and A. Pavim, Laboratory for Machine Tools and Production Engineering, RWTH Aachen University, Aachen, German. ACEEE Int. J. on Control System and Instrumentation, Vol. 03, No. 01, Feb 2012type is performed. A neural network based method is developed because of its ability to solve and generalise non-linear classification problems. This work currently focuses on the two most important cases:  flank wear and tool breakage. To build a distinctive description of the tool wear, the following features are extracted from the segmented wear region, which are tested as inputs for the neural network.

? image statistics: average, maximum, minimum, standard deviation

? surface texture: average variance of gray values of the segmented area, which analyses image textures of wear region

? Canny analysis: Canny filtering, which characterises the high-frequency image details, such as edges and wear textures

? histogram: which describes the brightness of the wear region

? FourierZoom Lebron XIII 13