Spectral imaging becomes more and more interesting not only for agricultural use but also for industrial application. Especially wavelength in the near infrared (NIR) range can be used for materials classification. Today sorting systems for plastics are available in different variations, utilizing single-point spectroscopy and the different characteristics of plastics in the SWIR band. Sorting systems for paper and cardboard will have increased significance because better sorting can increase the price of the secondary material and reduce the need of chemicals in paper production. However, sorting paper qualities is a very difficult task due to the close similarities between the materials. The present work describes the development of an unique industrial inline material sorting system using spectral imaging technique focusing on classification for cellulose based materials such as pulp, paper and cardboard. It deals with the hardware requirements for industrial use of spectral imaging solutions as well as with adjustment and calibration techniques. Due to needed classification speed the software design and classification methods are described under this focus. To cope with the vast amount of spectral data and to implement a stable and reliable classification algorithm for different materials chemometric standard methods are used. The PCA is used to reduce data and obtain as much information of the samples's characteristics as possible by transforming the original multidimensional data-space into a space with lower dimensions. However PCA is no method to discriminate between classes, it allows to separate cellulose-based materials from plastics. For further discrimination an LDA-Algorithm is used. All chemometric methods need training data sets of well defined samples. To classify an unknown spectra, it is necessary to create models for the classes to be distinguished from each other inside the transformed data-space. Training spectra have to be carefully selected to represent the characteristics of a specific class best possible. The classification-tree uses an adapted KNN-algorithm. In order to avoid a serious bottleneck in processing-speed the continuous result space was converted into discrete space representation.
This paper will describe the distributed industrial inline application “broken roll detection”, which is placed in a really harsh industrial environment, with all aspects from the sensing base to algorithm, implementation and technology. In a seamless steel tube production the pipe shells produced in the punch bench are running through many roller stands (3-roll system) to get the final dimension. If one of the rolls is broken, structural voids near the surface are the consequence. So finding the structure voids on the tube means to find broken rolls. Since pipe shells are hot (approximately 900°C) after passing the rolls, temperature distribution on its surface is different when voids happen. This gives a good base for detecting such voids by watching the surface temperature by sensing the radiation at wavelengths from 0,7 to 1.1μm, which means that standard line scan cameras (3 x 2048 pixels, 10kHz line rate) can be used. Images of up to 600MB are the result for each imaged pipe shell. Evaluation of image data is done stepwise (in a pipeline) and on a separate channel for each camera with the objective to reduce data at each step. Images are detruncated, position-normalized, filtered, segmented and converted into object-descriptions that are sent to another PC for evaluating periodic occurrences. Once found such a periodic occurrence, the system signalizes it to the production line to stop the machine and repair the broken roll.
This work gives some practical, simulated and calculated design parameters for the detection of voids inside the material with active thermography for different void geometry, orientation and depths. Main goal is to find the limitations of detectability for different materials and voids, to help designers for test systems with: a quick estimation of the feasibility and to find the necessary camera parameters. The methods used (algebraic, numeric and practical) to find these values will be described.
For inline applications the so called square pulse technique is easy to automate and needs less power from the source, because energy can be brought into the probe for a longer time span. Further its strength (in relation to flash pulse technique) is to find voids deeper below the surface. Therefore all of the calculations and practical verifications will be done only with square pulse.
The finite difference calculations are used to get a quick approximation for the dimensioning parameters. Some hints how to work with this method and how to prevent errors will be given in this paper. Practical tests with artificial probes and known void properties will be done with some of the parameters to verify the calculated values.