This diploma thesis approaches the topic of data fusion from the hardware as well as the algorithmic perspective. In order to acquire data suitable for the merging process, two measurement systems are introduced featuring the opportunity to synchronously image a target with two measurement heads. Both systems allow non-destructive contact-free terahertz materials inspection with the means of the FMCW measurement principle. One scanning unit, the FusionHead, makes use of the radiation’s polarization to allow collinear inspection on a target volume element with both measurement heads. This makes the data inherently suitable for the data fusion process. The second acquisition system covered in this work is the Compact system, enabling the user to image a whole component in raster scan movements obtaining a complete 3D image. Besides featuring two measurement heads, the Compact system is equipped with a control software, which synchronizes the data acquisition, raster movements, data analysis and the correct image visualization.
The two measurement heads utilized in both systems feature adjacent frequency ranges, which are determined in the course of this thesis, and lead to diﬀerent resolutions and penetration depths into the target. In order to make use of both their beneﬁts and further increase the imaging resolution, the two heads’ data is fused. A set of requirements, necessary for the fusion process, is deﬁned and the data is processed to fulﬁll them. Furthermore, a set of gap ﬁlling algorithms is developed and implemented, which allow to ﬁll the gap between the measurement heads’ frequency ranges. The possible algorithms are tested with respect to imaging resolution and noise. The best performing procedures are put to further testing of thickness measurement in the real-life environment of an industry project.
Diese Diplomarbeit beleuchtet das Thema der Datenfusion, sowohl aus der technischen also auch aus der algorithmischen Perspektive. Um für den Fusionsprozess nutzbare Daten aufzuzeichnen, werden zwei Messsysteme vorgestellt. Beide bieten die Möglichkeit der zeitgleichen Vermessung eines Zielobjekts mit zwei Messköpfen. So befähigen sie zur nicht-invasiven kontaktlosen Materialinspektion mit Terahertz-Strahlung, basierend auf dem FMCW Prinzip. Eine der Scaneinheiten, der Fusion-Head, macht sich die unterschiedliche Polarisation der Strahlung zunutze, um die kollineare Untersuchung eines Volumenelementes des Zielobjekts zu ermöglichen. Dadurch sind die aufgenommenen Daten schon inhärent geeignet für den Fusionsprozess. Die zweite Einheit, die in dieser Arbeit behandelt wird, ist das Compact system, welches die 3D-Abbildung eines ganzen Bauteils durch Rasterbewegungen ermöglicht. Es ist ausgerüstet mit zwei Messköpfen und einer Steuerungssoftware, die die Datenakquise, die Rasterbewegungen sowie die Analyse und die Visualisierung der Daten synchronisiert.
Beide Messköpfe, die in den vorgestellten Systemen zum Einsatz kommen, haben angrenzende Frequenzspektren, die im Zuge dieser Arbeit genauer bestimmt werden und zu unterschiedlichen Auﬂösungen und Eindringtiefen der jeweiligen Strahlung führen. Um die Vorteile beider Bereiche zu nutzen, und gleichzeitig die Auﬂösung der Systeme weiter zu erhöhen, werden die gemessenen Daten fusioniert. Dazu muss eine Reihe von Anforderungen bestimmt und die Daten so prozessiert werden, dass sie diese erfüllen. Des Weiteren wird eine Reihe von Fusionsalgorithmen entwickelt und implementiert, die die vorhandene Lücke im Frequenzspektrum zwischen den Daten der zwei Messköpfen schließt. Die einzelnen Vorgehensweisen werden auf Originaldaten angewandt auf ihre Auswirkung auf Auﬂösung und Rauschen der entstehenden Bilder getestet. Die performantesten Algorithmen werden weiterentwickelt und im Zuge eines Industrieprojekts bei der Untersuchung realer Bauteile evaluiert.
1 Introduction and Motivation.. 1
2 Theoretical Framework.. 5
2.1 Frequency-Modulated Continuous-Wave Radar.. 6
2.2 Technical Requirements.. 13
2.2.1 Varicap Diode.. 13
2.2.2 Voltage Controlled Oscillator.. 15
2.2.3 Frequency Multiplier.. 18
2.3 FMCW Setup.. 19
2.4 Resolution and Error Considerations.. 21
2.4.1 Theoretical Resolution.. 21
2.4.2 Range of Unambiguousness.. 24
2.4.3 Phase Distortion and Sidelobes.. 24
2.4.4 Doppler Eﬀect.. 25
3 Measurement Systems.. 27
3.1 Setup of the Compact Measurement System.. 29
3.1.1 Measurement Heads.. 33
3.1.2 Transition Axes.. 34
3.2 Compact Control Software.. 37
3.2.1 VCO Control Voltage.. 39
3.2.2 Data and Position Acquisition.. 40
3.2.3 Data Processing and Saving.. 42
3.3 Measurement Head Featuring Sensor Fusion.. 47
3.3.1 Design.. 48
3.3.2 Acquisition Software.. 51
4 Data Fusion.. 53
4.1 VCO Frequency Range Determination.. 55
4.2 Data Fusion Algorithms.. 58
4.3 Fusion Software Realization.. 65
5 Measurement Results.. 67
5.1 Compact System.. 68
5.2 Comparison of the Fusion Algorithms.. 71
5.3 Thickness Measurement Performance.. 79
6 Conclusion and Outlook.. 85
1. Introduction and Motivation
The terahertz spectral region derives its name from the frequency range of the electromagnetic waves it is consisted of. This range reaches from 0.1 THz to 10 THz (1 THz = 1012 Hz) corresponding to wavelengths of 3 mm to 30 μm, respectively. As seen in ﬁgure 1.1, it is located in between the infrared and microwave region.
Figure 1.1: Terahertz in the electromagnetic spectrum
[Abb. in dieser Leseprobe nicht enthalten]
Due to the high technological requirements, which are necessary to generate and detect broadband terahertz radiation, research in this area began decades later than in the surrounding spectral regions. Thanks to technological innovations like femtosecond lasers and photoconducting antennas, the usage of terahertz radiation is possible in a wide range of ﬁelds nowadays: in biomedicine the absorption of terahertz radiation by water is used for the spectroscopy of molecules . Security applications of terahertz such as detection of drugs, explosives and weapons are rapidly evolving and already in operation . Even the ultra fast wireless transmission of data by means of terahertz electromagnetic waves is feasible . In the context of this work, however, terahertz radiation is utilized to research for applications in quality control and non-destructive testing.
There are many reasons for the variety of use cases for terahertz technology, the main ones being the following:
a.) Terahertz radiation has a comparably low photon energy. While diﬀerent testing methods using X-Ray radiation ionize biological tissues, terahertz radiation is harmless for living cells.
b.) Many materials including explosives and chemicals, as well as biological agents, can be identiﬁed by their characteristic terahertz spectrum, serving as a ﬁngerprint.
c.) Terahertz radiation has the ability to penetrate dielectric materials like plastics, paper and ceramics, that are opaque for visible light. Since air is a propagation medium of terahertz radiation, imaging in this spectral region can be performed contactless, which is a great advantage over ultrasound testing methods. Furthermore, polar liquids like water strongly absorb terahertz radiation. This can be utilized to detect water inclusions or moisture diﬀerences in the materials stated above.
Terahertz radiation can be generated by means of either optical or electronic methods. Optical systems mostly feature pulsed femtosecond lasers and photoconductive antennas to generate short terahertz pulses, whereas electronic systems make use of voltage controlled oscillators and frequency multipliers to produce continuous electromagnetic waves in the terahertz domain. These signals are then typically emitted from a horn antenna towards the target. A detector is either placed behind the object of interest, to measure the transmitted signal or a receiver acquires the reﬂected radiation. Analyzing the time delay or detected intensity variation compared to the emitted signal, one can draw conclusions regarding the target’s inner structure and composition.
Figure 1.2: Terahertz imaging in the automotive industry
[Abb. in dieser Leseprobe nicht enthalten]
Examples for the usage of terahertz technologies in the "Center for Materials Characterization and Testing" at Fraunhofer Institute for Industrial Mathematics (ITWM) can be seen in ﬁgures 1.2 and 1.3. Figure 1.2 illustrates a typical measurement head used for example in the automotive industry. In this case it is mounted on a robot arm to measure the thickness of multiple paint layers during the fabrication process. Figure 1.3 displays the detection of inclusions and faults in dielectric materials. The upper image shows a test object made out of ﬁberglass reinforced plastic in a honeycomb structure with artiﬁcially installed defects invisible to the human eye. In the lower images the defects are made visible by scanning the object with terahertz radiation of diﬀerent frequencies and printing an XY-cut at a depth of −20 mm of the resulting 3D image.
Figure 1.3: top: Fiberglass test object with invisible built-in defects
left: XY-cut of the resulting 3D image measured with radiation reaching from 110 GHz to 170 GHz. Defects are clearly visible from the plot of the normalized reﬂected power on the color axis in dB.
righ t: Another terahertz image, scanned with a bandwidth of 230 GHz to 320 GHz. The resolution has increased signiﬁcantly.
[Abb. in dieser Leseprobe nicht enthalten]
In our work-group electronic measurement systems for contactless non-destructive testing of component surfaces and thicknesses are developed. In practice, the imaging process is often performed by a single transceiver, which emits terahertz rays and detects their reﬂections at the same time. In this way, depth information of one volume element, a so called voxel, is retrieved. A complete 3D image of the target is formed by scanning it voxel by voxel and combining the successively acquired information.
Measurement heads ranging from 70 GHz up to 850 GHz with spectral ranges diﬀering between 30 GHz and 90 GHz are in use. The choice of these parameters, especially their frequency range, has a big impact on the system’s imaging performance. Low frequency terahertz radiation penetrates deep into most materials, allowing their transceivers to image on a greater depth range than the measurement heads emitting higher frequency terahertz waves. The latter though, as they come with shorter wavelengths and typically a higher frequency range than the former, assure an improved image resolution, as depicted in ﬁgure 1.3.
The goal of this work is to make use of both these beneﬁts, receiving the best resolution possible on an unaltered high penetration depth, as well as ensuring fast and eﬃcient scanning, by combining two measurement heads and using data fusion algorithms to merge their data. Furthermore, a handy setup for the two measurement heads will be equipped with hard- and software, making the system a portable robust device.
Structure of this Work
This diploma thesis is subdivided into six chapters. Having giving a short introduction and an overview of its main motivation in this ﬁrst chapter, the second chapter will provide the reader with the theoretical background this work is based on. The basic working principle of a frequency-modulated continuous-wave radar is followed by the functionality of the technical components it is composed of as well as their setup. Furthermore, its limitations and possible sources of error are considered from a theoretical point of view.
Chapter three outlines the setup of the two measurement systems used in the context of this work. The most important components are introduced, providing an insight into their control and mode of operation. Moreover, the measurement software of the entire systems is broken down and described in detail.
The fourth chapter motivates the application of data fusion, as well as stating the requirements and challenges it comes with. These include a detailed analysis of the measurement hardware in use, as well as the development of a set of algorithms to perform the merging process and their implementation in software.
In the ﬁfth chapter the measurement results and images acquired with the diﬀerent systems are analyzed and discussed with respect to noise and resolution. The diﬀerent fusion algorithms are contrasted and the best undergo further testing. Furthermore, the quality of the fused signal is compared to a simulated signal theoretically to be expected after the fusion process. As a ﬁnal proof of concept, the data acquired at the production site of an industry partner is processed with the software developed in the course of this thesis. The resulting images are presented at the end of chapter ﬁve.
The last chapter summarizes the ﬁndings of this work, as well as giving an outlook on possible further projects, which can be pursued based on this thesis.
2. Theoretical Framework
In this chapter an introduction to the theory of the frequency-modulated continuous wave radar (FMCW) will be given. The ﬁrst section includes the physical and mathematical foundations of the main working principle of FMCW. This principle allows a distance measurement of targets with a surface reﬂecting terahertz radiation such as metal or dielectric material. In the second section the technical essentials necessary to build up an FMCW system and their mode of operation will be examined thoroughly. Thereafter a possible FMCW setup is presented, displaying how the discussed principles and methods can be implemented by the introduced hardware. The last section is dedicated to performance considerations as well as physical limitations and possible error sources resulting from non-ideal hardware.
2.1 Frequency-Modulated Continuous-Wave Radar
The acronym FMCW radar stands for frequency-modulated continuous-wave radio dete ction and ranging, a certain kind of radar system, which emits electromagnetic waves onto a target and detects their reﬂection or transmission. In contrast to pulsed radar systems, emitting pulsed beams of radiation, continuous wave radars (CW) send out radiation continuously. If the output radiation’s frequency is constant, the velocity of moving objects can be measured by making use of the Doppler eﬀect. In order to also acquire location information, one requires to know the time diﬀerence between the transmitted and the received signal. Therefore, the reﬂected waves have to be identiﬁable by their frequency. For the FMCW radar this is achieved by periodically modulating the output signal’s frequency and multiplying the reﬂected signal and the current signal in a mixer. From the frequency diﬀerence fb (for be at frequency), extracted with the help of a Fourier transform, one can determine the wave run-time and ﬁnally, calculate the distance between antenna and reﬂector .
A very important prerequisite for the operation of an FMCW radar is the constraint that the frequency modulation period Tsweep (for sweep time) has to be much longer than the expected transit time of the signal, so that at any time, the frequency of the reﬂected signal can be attributed uniquely to its time of origin. As a result Tsweep deﬁnes the range of unambiguousness of the system as elaborated in chapter 2.4.2. Further limitations concern the refractive index of the target material: in order to measure the distance in reﬂection, the material must not be completely transparent or absorbing to the radiation in use. If the thickness of the target or its inner structure is to be observed as well, moreover, the reﬂectivity cannot be too high, otherwise, there will not be enough radiation penetrating into the target. The same applies for scattering eﬀects, which reduce the signal amplitude of the measurement signal and should therefore be minimized. Apart from that, other physical properties of the target have very little inﬂuence on the measurement result. The most important beneﬁt of FMCW measurement compared to diﬀerent mechanical or electronic measurement procedures though, is the fact that it can be performed contact-free, allowing it to be applied on sensitive surfaces that are opaque to visible light. A drawback of FMCW is the comparably high asset cost of expensive high frequency components.
Since data acquisition and analysis of FMCW radars are performed fast and real-time imaging is possible with a good resolution, a live supervision of production lines is feasible . Electronic FMCW radars are used for distance measurement and braking assistance in high- and middle-class cars  or serve as radio altimeters, measuring the ground distance of planes during take-oﬀ and landing . In heavy industries FMCW radars are applied in tanks, to indicate the ﬁlling level of liquids and bulk goods . Due to their characteristics FMCW radars and measurement systems play an important role in non-destructive testing. An example of FMCW radar utilizing terahertz radiation is the detection of faults in ceramics or dielectric materials such as airplane parts .
Closer insight into the main working principle of FMCW radars will be given below.
- Quote paper
- Henrik May (Author), 2017, Design, Setup and Characterization of an Electronic Terahertz Measurement System Featuring Sensor Fusion, Munich, GRIN Verlag, https://www.grin.com/document/425832