Please use the DOI link to navigate to the full publications...
List of publications

Selfoptimization of a pickling line by fusing decentral Industry 4.0 components with online process models
Marcus J. Neuer, Moritz Loos, Detlef Sonnenschein et al. (to be announced), upcoming SIS2016 conference publication
A technique is presented to optimize the throughput of a pickling line by combining recent Industry 4.0 architecture with process models. Socalled autonomous agents act as virtual representatives of the coils and the processes. These smart objects have access to the surface inspection data after hot rolling and after the pickling line, as well as to other process and quality data. Using the information of the surface inspection system (SIS), virtual coils can predict the amount of a expected scale on the strip to estimate its best pickling speed and control their speed during pickling. The complexity of the involved models was tailored to deliver a good balance between precision and online applicability. It is shown, that the joint capabilities can be exploited to induce selfoptimization of the pickling process. Different logistic, physical and economic constraints of the optimization are considered and discussed. Objectives like higher throughput and increased pickling line speed are addressed. Our solution is applied at a steel production site in Germany, demonstrating potential technical impact of the selfoptimization approach. The latter reveals also the interplay of the different actors in the system. 
Dynamic reallocation and rescheduling of steel products using agents with strategical anticipation and virtual marketstructures
Marcus J. Neuer (BFI), Francesca Marchiori (CSM), Alexander Ebel (BFI), Nikolaos Matskanis (CETIC), Luca Piedemonti (CSM), Andreas Wolff (BFI) and Gael Mathis (ArcelorMittal), upcoming IFAC MMM conference publication (Proc. IFAC MMM 2016)
A multiagent system for the steel production is presented. It reallocates o spec products to existing alternative orders. The system consists of three main Industry 4.0 components: agents, serviceoriented architecture (SOA) and a semantic modelling of the process chain. Using economic principles like virtual marketplaces, the solution of the reallocation is reached by means of auctions, where product agents act as sellers and order agents act as bidders. A virtual price denes the key for the matching process between order and product. To incorporate postprocessing, product agents anticipate their possibly reachable future states with rstprinciple models. The matching between order and product characteristics is then based on the predicted product future and the order specication. As the postprocessing comes at a cost, logistics costs, administrative costs and transformation costs are also considered for constructing the virtual price. The presented system runs as online openloop prototype at a steel plant of ArcelorMittal and shows promising results for the reallocation task. 
Smart reallocation of steel products
Marcus J. Neuer (BFI), Alexander Ebel (BFI), Andreas Wolff (BFI), Francesca Marchiori (CSM), Nikolaos Matskanis (CETIC), Martin Rößiger (SIEMENS) and Gael Mathis (ArcelorMittal) in Stahl & Eisen 11/2015, journal publication of ESTAD Talk (see below).
A system is presented to dynamically reallocate steel products, which do not meet the requirements of their original order. Typically, these products have to be sold on a less profitable secondary market. With the novel approach, they can be allocated to an alternative primary order. For hot rolled coils, the reallocation also considers to further downstream processing (cold rolling, hot dip galvanization etc). 
ModelBased, Analytical MaximumLikelihood Deconvolution for CZT Detectors
Marcus J. Neuer and Elmar Jacobs, IEEE Nuclear Science Symposium and Medical Imaging Conference 2015, San Diego
An analytical response function is presented that is based on a perturbed PearsonIV model for the peak shape and additional models for the first four moments of the distribution, mean, variance, skewness and kurtosis. The parametrisation of these moment models is acquired by using an evolutionary ensemble algorithm, delivering corresponding energy dependent functions for each of the moments. Using the relationship between the moments and the actual peak shape, this analytical equation is introduced into the MaximumLikelihood inversion, delivering a formula for deconvolving spectra. Through the parametrisation, the technique can be tailored to any CZT detector. Without having to store a large response matrix, the analytical deconvolution can be easily deployed also on smaller devices, with less computational performance. The technique is validated on a Raspberry Pi connected to a muSPEC500 device from Ritec, featuring a $300 mm^3$ CZT crystal. Example measurements for cesium Cs137, cobalt Co60, europium Eu152 and uranium U238 are deconvolved and show a significant capability of resolving peaks. 
Cognitive RTree for Stabilizing Temperature and Load Induced Gain Shifts of Scintillation Detectors
Marcus J. Neuer and Elmar Jacobs, IEEE Nuclear Science Symposium and Medical Imaging Conference 2015, San Diego
A stabilization concept based on a selflearning RTree index method is presented and demonstrated with measurements from a 1.5’’x1.5’’ cerium bromide detector. The concept uses a cognitive filter, a digital filter for nuclear signals that continuously updates itself to the current temperature by adjusting the filter components. The RTree combines the information from this cognitive filter together with (a) data about the temperature gradient, (b) the current load on the detector in terms of counts per second and (c) the current gain shift, which is determined from the spectrum. This technique consequently belongs to the socalled supervised learning algorithms, because the source is known in advance. The method is characterised by two operational phases. First a training in an industrial grade climate chamber and with selected strong radiation fields are conducted, which is a common procedure for producing spectroscopic equipment, building a base set of data points in the RTree. Second, the RTree learning is not stopped here. It continues during the whole instrument lifetime. Each time a manual calibration is launched with a known (preselected) source, all data for adding new training information is available and the RTree is updated. The instrument learns while being in the field. Tests with a cerium bromide and a sodium iodide detector are shown for a prototype system and for a complete commercial radioisotope identification device. Limits of the stabilisation are determined. 
Forecasting the direction of incoming radiation based on fusion of gyroscopic and spectroscopic data
Marcus J. Neuer, Christian Henke and Elmar Jacobs, ArchivX
A method is shown to estimate the position of a radioactive source by fusing the data from a sodium iodide detector and a gyroscope while panning the detector. Based on simple geometry considerations, the search motion of a detector is modelled as angular panning. Correlation of both sensor streams is shown as a predictor for the direction of the incoming radiation. The method also allows for a rough distance classification, where a distant source or a homogenous radiation field can be distinguished from a single source near to the detector. 
Raising economic efficiency of steel products by a smart reallocation respecting different process routes
Marcus J. Neuer, Alexander Ebel, Andreas Wolff, Francesca Marchiori, Martin Rößiger, Nikos Matskanis and Gael Mathis, European Steel Technology and Application Days (ESTAD), Düsseldorf, 2015
Publication from the EURFCS project I2MSTEEL, a collaboration of Centro Sviluppo Materiali (CSM, Italy), Siemens (Germany), Cetic (Belgium), VDEhBetriebsforschungsinstitut (BFI, Germany) and ArcelorMittal (France)Over the last decade, the quality of European steel has improved significantly, ensuring reliable products that withstand the highly competitive nature of nowadays market. Still, a notable amount of coils of high quality, that do not meet the demands of their original order need to be sold via secondary channels to gain a reasonable profit. On the other hand, many of these products actually fulfill the specifications of one or more very different orders within the order book of the steel manufacturer. Reallocation of such a product to one of these more suitable orders would be the most economic solution. A socalled agentbased technique is presented to perform this reallocation. It takes into account the production route of a hot rolled coil and predicts the potential posterior transformation steps needed to meet the demands of the new order. This order could e.g. require cold rolling and eventual galvanization processes. Here, also different plant locations are regarded as a hot rolled coil often travels from one plant to another one for further processing. The additional effort for reallocation is quantified by a set of key cost factors for administrative, logistics and transformation related expenses. Based on these factors, the benefit for attributing the coil to a new order is determined. Then, from multiple different orders, the best one is selected. This leads to a significant increase in economic efficiency. 
Evolutionary ensembles that learn spectroscopic characteristics of scintillation and CZT detectors
Marcus J. Neuer, Nikolai Teofilov, Yong Kong and Elmar Jacobs, IEEE Nuclear Science Symposium and Medical Imaging Conference 2014, Seattle
A method is described to automatically generate spectrum reference data for radioisotope identification devices respecting a detectors physical individuality. It extracts the peak shape and nonproportionality characteristics of scintillation and CZT detectors. The representation of these quantities is done with evolutionary ensembles, groups of Ndimensional autonomous points, which are propagated within a constrained space. Each ensemble member is used as parametrical input for describing peak shape and position within a simulation framework based on Geant4. Each subsequent generation of the ensemble iteratively converges the simulation result towards an optimised match with the measurement. Examples for the scintillator show that the shape convergence is straightforward due to the gaussianity of the peak, while the correction of the nonproportionality is within the quantity of up to 10%. Contrarily, our CZT example yielded nearly no nonproportionality along the energy scale, but required a complex, multiparametrical shape definition with learning curves for kurtosis, skewness and resolution to establish an adequately peak reproduction. A metric is presented to calculate the distance between the experimental data and the calculated result. The described system is suited to establish a production line with a fully automatised acquisition of spectral characteristics to support the deployment of detector individual reference data for nuclide identification instrumentation. 
A Cognitive Filter to Stabilize Peak Positions and Widths of a Scintillation Detector and to Determine Its Material
Elmar Jacbos, Christian Henke and Marcus J. Neuer, IEEE Nuclear Science Symposium and Medical Imaging Conference 2014, Seattle
Part II of a paper series on cognitive filteringA digital filter is presented that adapts automatically to the shape of a nuclear signal to stabilise against temperature induced peak shifting. The filter is called cognitive as it extracts the information about the exponential decays and determines the scintillation material intrinsically. The latter is done with a pole deconvolution approach. Based on the knowledge of the decays, the peak position in the pulseheight spectrum is stabilised due to the relationship between exponential decay and temperature. To achieve this, the peak positions are experimentally acquired within a climate chamber measurement and later fed into a learning artificial neural network. This network has additionally access to the decay times, the temperature and the speed and direction of the temperature. The filter is called every two minutes, automatically collecting a series of 500 suited signals and performing an update of the stabilisation. In tests, the temperature induced peak shifting was corrected within a 0.2% boundary. An additional output result of the filter is the material type of the scintillator. For the most materials, temperature curves are known and the assessment of the exponential decay is shown to lead straightforwardly to a determination of the material. Also the hot plugging of two materials is possible, yielding a recovery time of approximately four minutes for the system to adapt to the new material. 
A cognitive filter to automatically determine scintillation detector materials and to control their spectroscopic resolution during temperature changes
Marcus J. Neuer and Elmar Jacobs, IEEE Transactions on Nuclear Science, Volume 61, Issue 3, June 2014
Part I of a paper series on cognitive filteringThe shape of the nuclear signal from scintillators contains information about the detection material from which it originated and its current temperature. A digital filter for this type of signals is presented that automatically extracts both of these pieces of information from the shape of the signals and controls the temperature dependency of the resolution by continuous adaption to this shape. Hereby, nuclear signals are apriori modelled by an allpole filter. The concept combines a deconvolution approach based on this model with a leastmeansquares iteration. Beneath the iteration, there is a continuous assessment of the signal shape where the meansquared distance between the assumed shape and measured shape acts as control parameter for the resolution. From the information stored inside the dynamic filter parameters, the material characteristics are retrieved to identify the detector material. An implementation into a multichannelanalyser is shown and the technique is verified under realworld environmental conditions. 
Spectral identification of a 90Sr source in the presence of masking nuclides using MaximumLikelihood deconvolution
Marcus J. Neuer, Nuclear Instruments and Methods in Physics Research A, Volume 728, p. 7380, 2013
Abstract: A technique for the spectral identification of strontium90 is shown, utilising a MaximumLikelihood deconvolution. Different deconvolution approaches are discussed and summarised. Based on the intensity distribution of the beta emission and Geant4 simulations, a combined response matrix is derived, tailored to the β detection process in sodium iodide detectors. It includes scattering effects and attenuation by applying a base material decomposition extracted from Geant4 simulations with a CAD model for a realistic detector system. Inversion results of measurements show the agreement between deconvolution and reconstruction. A detailed investigation with additional masking sources like 40K, 226Ra and 131I shows that a contamination of strontium can be found in the presence of these nuisance sources. Identification algorithms for strontium are presented based on the derived technique. For the implementation of blind identification, an exemplary masking ratio is calculated.
DOI: 10.1016/j.nima.2013.06.013 
Methods and databases for identifying nuclides
Marcus J. Neuer, Yong Kong, Ralf Lentering, Jürgen Stein, Patent, Pub. No.: WO/2012/072103, Int. App. No.:PCT/EP2010/068448
A method is described to include various material interaction processes into the identification technique of modern radiation equipment. With that approach, it becomes possible to identify both, extremely shielded and heavily scattered sources. Moreover, after identifying the source by its spectrum, contributions of shielding and scattering are quantified and a forecast is provided whether the source is e.g. inside a human begin (for potential medical treatment) or not. The latter is important to decide the threat level. 
Comparison of different Cs2LiYCl6:Ce crystals: Energy resolution and pulse shape dependences on temperature
Cristina Plettner, Falko Scherwinski, Guntram Pausch, Ralf Lentering, Yong Kong, Achim Kreuels, Marcus Neuer and Jürgen Stein in Nuclear Science Symposium and Medical Imaging Conference (NSS/MIC), Anaheim, 2012
Abstract: Portable radiation detection systems will greatly benefit from the use of the same detector for gamma and neutrons, on the roadmap towards miniaturization, provided such devices are capable of robust discrimination of gammas against neutrons. One of the most promising and yet commercially available inorganic scintillator, delivering different pulse shapes for thermal neutrons and gammas is the Cs2LiYCl6:Ce (CLYC) [1,2]. The detection for thermal neutrons follows through the 6Li(n,alpha)3H capture reaction, and with the appropriate levels of 6Li enrichment, which have currently reached 95%, the CL YC crystals could become very efficient neutron detectors. In order to successfully integrate this new scintillator in our digital devices, the variability of the energy resolution across various crystals, the dependence of the energy resolution on energy, shaping time and temperature, the pulse shape parameters variability across the crystals and the thermal neutron sensitivity were investigated.DOI: 10.1109/NSSMIC.2012.6551454 
Towards design and optimization of scintillationdetector systems: A MonteCarlo simulation framework
Yong Kong, Guntram Pausch, Katja Roemer, Marcus Neuer, Cristina Plettner, Ralf Lentering and Jürgen Stein, Nuclear Science Symposium Conference Record (NSS/MIC), Knoxville, 2010
Abstract: A complete simulation framework that combines Monte Carlo simulation with the digitalsignalprocessing algorithms developed inhouse is presented for designing and optimizing scintillation detector systems. The framework covers basic components of a practical scintillation detector system and includes all relevant physical processes. With the complete simulation individual physical factors that can affect the performance of a detector system, such as detector geometry, surface finishing, decay time and scintillation yield of scintillator as well as responses of photodetector and frontend electronics, can be taken into account. As a demonstration of the simulation framework, the light collections of plastic (EJ200), NaI(Tl) and BGO scintillators with different surface types have been simulated as a function of the reflectivity of a diffuse reflector. Optical simulations were further performed to investigate the homogeneity of light transportation and detection in plastic scintillators with different surface finishing properties and coupled to one or two PMT(s). Finally, detector signals simulated for a NaI(Tl) detector have been compared with measured mean pulses. The simulated signal can be further digitized and used as input of digital signal processing.DOI: 10.1109/NSSMIC.2010.5873953 
Linearization of Gamma Energy Spectra in ScintillatorBased Commercial Instruments
Yong Kong, Guntram Pausch, Katja Roemer, Achim Kreuels, Claus Herbach, Marcus Neuer, Ralf Lentering and Jürgen Stein, IEEE Transactions on Nuclear Science, Volume 57, Issue 3, 2010
This paper presents a novel technique developed for linearizing the energy spectra of radiation detectors in commercial radioisotope identification devices. Based on few spectrum measurements with standard radionuclide sources, this method allows generation of individual nonlinear calibration functions at minimum expense in the routine instrument setup. Instead of fitting peak positions, the measured raw data are compared with simulated spectrum templates, and local gain factors providing the best correspondence are taken as reference points for the calibration function. This approach avoids the problem of fitting multiple peaks with intensity ratios influenced by absorbing layers and assures an accuracy of 1% in the energy range of 30 keV to 3 MeV.DOI: 10.1109/TNS.2009.2033684 
Surveillance of nuclear threats using multiple, autonomous detection units
Marcus J. Neuer, Kai Ruhnau, Arne Ruhnau, Ralf Lentering, Guntram Pausch, Frank Platte and Jürgen Stein, Nuclear Science Symposium Conference Record (NSS/MIC), Dresden, 2008
Abstract: A detection systems is presented that involves multiple autonomous detectors, which are merged to a net like structure. Each unit is equipped with a stabilized and linearized multichannel analyzer for the measurement of gamma spectra and optionally a He3 neutron detector. The flexible hard and software architecture allows the simple integration of a variable number of detectors, even with different volumes. With this technique, spectroscopical screening of large areas, e.g. airports or comparably sensitive objects, can be established by an individual combination of these detectors. The general outline of the nuclide identification algorithm is presented. In the context of medical isotopes, the importance of correct reference data that incorporates the scattering effects of human bodies is clarified. A special focus is drawn to the localization of nuclear sources. The model, corresponding simulations and its applications are depicted. Different tests and verification measures are described that refer to the nuclide identification and the source localization as well as to the spectral quality of the detectors. A finished real application is presented that combines the tested algorithms and has proven its reliability on different occasions.DOI: 10.1109/NSSMIC.2008.4775056 
A technique for estimating detection limits of radionuclide identifying detectors by means of computer simulations
ClausM. Herbach, Yong Kong, Ralf Lentering, Marcus Neuer, Guntram Pausch, Cristina Plettner, Kai Ruhnau and Jürgen Stein, Nuclear Science Symposium Conference Record (NSS/MIC), Dresden, 2008
A simulation technique has been developed to study the performance of nuclide identifying gamma detector units operating under a variety of different conditions. The studies are related to nuclide identification based on a template matching algorithm which has been well established, in particular, for the analysis of low statistics measurements. Results are presented for handheld devices equipped with NaI(Tl) and LaBr3(Ce3+) scintillation detectors, respectivelyDOI: 10.1109/NSSMIC.2008.4775218 
Pitch angle scattering and effective collision frequency caused by stochastic magnetic fields
Marcus Neuer and KarlH. Spatschek, AIP Physics of Plasmas 15, 022304, 2008
The motion of charged particles in a plasma with stochastic magnetic field lines is investigated on the basis of the socalled ALangevin equation. In contrast to the wellknown perpendicular (to an external magnetic field)transport here the parallel transport is considered. The two cases of a strong guiding field and a weak mean magnetic field respectively are investigated. The ALangevin equation is solved under the assumption that the Lagrangiancorrelation function for the magnetic field fluctuations is related to the Eulerian correlation function (in Gaussian form) via the Corrsin approximation. The latter is justified for small Kubo numbers. Generalizations to large Kubo numbers are discussed. It is shown that the magnetic field fluctuations are the source of diffusive motion along a strong guiding field even if binary collisions are absent. The case of small (or vanishing) mean fields is also investigated in the quasilinear limit. An effective collision frequency is derived which is consistent with the result for perpendicular quasilinear diffusion in strong guiding fields. The analytical results are checked by numerical simulations of the ALangevin equation.DOI: 10.1063/1.2844436 
Radiation detector signal processing using sampling kernels without bandlimiting constraints
Jürgen Stein, Marcus Neuer, Claus Herbach, Guntram Pausch and Kai Ruhnau, Nuclear Science Symposium Conference Record (NSS/MIC), Honolulu, 2007
For the development of digital signal processing systems for fast scintillation detectors we comprehensively study the modeling of nuclear signals, deconvolution of detector pulses and signal sampling. Applications for new scintillators with light decay times of a few nanoseconds demand suitable low power digital systems running at lowest possible sampling rates. We are interested in accurate subnanosecond timing and optimal energy resolution. The generalized nonbandlimiting sampling theorems allow filter structures with lower than Nyquist sampling rates for certain signals, where the classical sampling theorem fails. Recently it was shown that, by using a sampling rate greater or equal to the rate of innovation, it is possible to reconstruct certain restricted signals uniquely. The class of sampling kernels that can be used contains transfer functions with rational Fourier transforms. We introduce a physically realizable sampling scheme combined with a deconvolution timing filter algorithm for radiation detector signals. The novel architecture achieves subsampling rate timing accuracy together with optimal energy resolution and a high throughput for LaBr3:Ce scintillation detector system.DOI: 10.1109/NSSMIC.2007.4436303 
Diffusion of test particles in stochastic magnetic fields in the percolative regime
Marcus Neuer and KarlH. Spatschek, Physical Review E 74, 036401, 2006
For stochastic magnetic flux functions with percolative contours the test particle transport is investigated. The calculations make use of the stochastic Liouville approach. They start from the socalled ALangevin equations, including stochastic magnetic field components and binary collisions. Using the decorrelation trajectory method, a relation between the Lagrangian velocity correlation function and the Eulerian magnetic field correlation is derived and introduced into the GreenKubo formalism. Finite Larmor radius effects are included. Interesting results are presented in the percolation regime corresponding to high Kubo numbers. Previous results are found to be limiting cases for small Kubo numbers. For different percolative scenarios the diffusion is analyzed and strong influences of the percolative structures on the transport scaling are found. The finite Larmor radius effects are discussed in detail. Numerical simulations of the ALangevin equation confirm the semianalytical predictions.DOI: 10.1103/PhysRevE.74.036401 
Diffusion of test particles in stochastic magnetic fields for small Kubo numbers
Marcus Neuer and KarlH. Spatschek, Physical Review E 73, 026404, 2006
Motion of charged particles in a collisional plasma with stochastic magnetic field lines is investigated on the basis of the socalled ALangevin equation. Compared to the previously used VLangevin model, here finite Larmor radius effects are taken into account. The ALangevin equation is solved under the assumption that the Lagrangian correlation function for the magnetic field fluctuations is related to the Eulerian correlation function (in Gaussian form) via the Corrsin approximation. The latter is justified for small Kubo numbers. The velocity correlation function, being averaged with respect to the stochastic variables including collisions, leads to an implicit differential equation for the mean square displacement. From the latter, different transport regimes, including the wellknown RechesterRosenbluth diffusion coefficient, are derived. Finite Larmor radius contributions show a decrease of the diffusion coefficient compared to the guiding center limit. The case of small (or vanishing) mean fields is also discussed.DOI: 10.1103/PhysRevE.73.026404 
Finite Larmor Radius contributions to anomalous transport in plasmas with stochastic magnetic fields
Marcus Neuer, Dissertation, 2004
Anomalous transport of charged particles in a collisional stochastic plasma is investigated on the basis of the ALangevin equation. The latter is a stochastic differential equation describing the motion of a particle that experiences collisions and a stochastic magnetic field. Contrary to previously used guiding center models, here finite Larmor radius effects are taken into account. Two different approximation methods are applied to obtain the Lagrangian velocity correlation function from the solution of the ALangevin equation, distinguished in terms of a dimensionless parameter called the Kubo number, which measures the degree of magnetic turbulence. The first one, the Corrsin approximation, is a widely used straightforward technique to relate the Lagrangian correlation, which is determined in the comoving frame of reference, with the common, spatially dependend Eulerian correlation. Differential equations for the mean square diplacement and the diffusion coefficient are obtained from the GreenKubo formalism. The quasilinear limit is discussed and estimates for the Larmor radius effects are presented. The wellknown RechesterRosenbluth regime is derived and the influence of the Larmor radius is discussed in detail. Additionally the KadomtsevPogutse regimes are regarded. A new regime, the vanishing mean field regime is introduced and predictions for the transport quantities are presented. The Corrsin approximation is restricted to small Kubo numbers. A second, more elaborated method called the decorrelation trajectory approximation is applied. This method includes influences of the magnetic structure and covers the percolative regime. Correlation functions derived with this method have no limitations in the range of the Kubo number. An analytical relation between Corrsin and decorrelation trajectory approximation is presented. Percolative magnetic structures corresponding to high Kubo numbers are investigated and a significant influence of the flux structure and the Larmor radii is identified. The analytical and numerical results are verified with a Monte Carlo simulation of the ALangevin equation.Universität und Landesbibliothek Düsseldorf 
Chirped solitons as attractors for short light pulses
Marcus Neuer, Zhonghao Li and KarlH. Spatschek, Physical Review E 70, 056605, 2004
Nonlinear chirped pulse solutions are shown to exist as stable attractors for short light pulses in driven and damped systems. The attractors are determined for systems of different complexity, from simple gain and damping modelings up to the inclusion of higherorder dispersion, Raman processes, and delayed nonlinear responses. The chirped attractors, their stability, as well as the attractor basins can be determined analytically. The analytical predictions are in excellent agreement with numerical simulations.DOI: 10.1103/PhysRevE.70.056605