Einsatz hyperspektraler Sensoren zur Überwachung intelligenter LED Lichtquellen eine Fallstudie

Transcription

1 Einsatz hyperspektraler Sensoren zur Überwachung intelligenter LED Lichtquellen eine Fallstudie R. Nestler, R. Jahn, K. H. Franke, S. Junger, D. Gäbler 21. Workshop Farbbildverarbeitung Koblenz, Zentrum für Bild und Signalverarbeitung (ZBS) e.v. ilmenau.de Technische Universität Ilmenau ilmenau.de/gdv Fraunhofer IIS X FAB Semiconductor Foundries AG Dr. Rico Nestler ZBS e.v. Ilmenau / TU Ilmenau 21. Workshop FarbBV, Koblenz, , Dr. Rico Nestler

2 Agenda 1. Motivation for monitored LED based lighting systems 2. Multi channel sensor elements for lighting applications 3. Considerations about feasibility 4. Summary & Outlook 21. Workshop FarbBV, Koblenz, , Dr. Rico Nestler 2/23

3 1. Motivation for monitored LED based lighting systems LED Lightings Efficiency and Costs Light output and ( spectral ) diversity Versatility for different lighting tasks Customizable light effects, which result from specific spectral light compositions Origin: Premosys GmbH physiologically colorimetric (affects color quality, visual comfort, CRI), biologically (affects attention, wellbeing, HCT) or for a specific technical recognition task (affects contrasts) 21. Workshop FarbBV, Koblenz, , Dr. Rico Nestler 3/23

4 1. Motivation for monitored LED based lighting systems LED Lightings Color quality in general lighting or color critical applications Difficult relation between luminous efficiency and spectral characteristic spectral defined LED lightings: only hybrid, but dynamically relative spectral radiant flux Normlicht D65 hybled D nm 21. Workshop FarbBV, Koblenz, , Dr. Rico Nestler 4/23

5 1. Motivation for monitored LED based lighting systems LED Lightings Drift and aging behavior depends on LED type, specimen and operating conditions Radiant flux / spectral radiant flux = f (Time, Temperature, Operating Point) Only in / at operating point approximated intensity drift especially critical for high power LEDs Temperature Operating Point relative spectral radiant flux rel. to relative spectral radiant flux 21. Workshop FarbBV, Koblenz, , Dr. Rico Nestler 5/23

6 1. Motivation for monitored LED based lighting systems WhysensorymonitoredLED Lightings? Stable LED based light syntheses require or Preselection, pre aging and characterization of used LEDs expensive stabilization and tracking of electrical operating points costly stabilization of temperature Continuous sensor monitoring and sensor based adjustment of light synthesises degrees of freedom of the measurement greater or equal the degrees of freedom of the light synthesis spectral orapproximately spectral color stimulus detection 21. Workshop FarbBV, Koblenz, , Dr. Rico Nestler 6/23

7 2. Multi channel sensor elements for lighting applications Requirements from the perspective of the measuring task relative spectral radiant flux White points characterizing spectral light color stimuli colorimetrically, strong observer metameric level, unsuitable D65 F7 OSTAR* CIE-xy Color Chart y White MCC DC: Points D65, vs. F7, OSTAR* max 10.9 x Characterizing of spectral lights by an approximately spectral measurement, less metameric Mehrbereichsansatz 21. Workshop FarbBV, Koblenz, , Dr. Rico Nestler 7/23

8 2. Multi channel sensor elements for lighting applications Hyperspectral sensors more than 3 broadbanded channels Channel sensitivities more applicative, energetic or technological oriented than colorimetric spectral or colorimetric measurement based on spectral estimation spectral current density / IIS simulated Plasmon channels Plasmonic multispectral sensor Origin: Fraunhofer IIS 21. Workshop FarbBV, Koblenz, , Dr. Rico Nestler 8/23

9 2. Multi channel sensor elements for lighting applications Hyperspectral sensors more than 3 broadbanded channels Channel sensitivities more applicative, energetic or technological oriented than colorimetric spectral or colorimetric measurement based on spectral estimation spectral current density / IIS simulated Plasmon channels Spectral estimation is an inverse, ill posed problem Color stimulihavetobedescribedwith less degrees of freedom than the spectral domain. Stabilization (regularization) with knowledge about the measurement problem: specific spectral lighting situations or causal lights or parametric models of the lights 21. Workshop FarbBV, Koblenz, , Dr. Rico Nestler

10 2. Multi channel sensor elements for lighting applications Requirements from the perspective spectral estimation clear detection of significant changes of spectral color stimuli within sensor responses broadbanded overlapping channel sensitivities channelwise opposite edges in spectral bands with significant spectral signature changes monotonically ascending / descending sensitivities, where narrowbanded spectral signature changes exist Stability of the inverse mapping to external tolerances Robustness of the inverse mapping to internal tolerances best conditioned sensor configuration 21. Workshop FarbBV, Koblenz, , Dr. Rico Nestler 10/23

11 2. Multi channel sensor elements for lighting applications Requirements from the perspective spectral estimation clear detection of significant changes of spectral color stimuli within sensor responses broadbanded overlapping channel sensitivities simulated 0.32 Plasmon channels channelwise opposite edges in spectral bands with significant 0.24 spectral signature changes spectral current density / monotonically ascending / descending sensitivities, where narrowbanded spectral signature changes exist 8 Stability of the inverse mapping to external tolerances Robustness of the inverse mapping to internal tolerances best conditioned sensor configuration Opposite consequences for the design of find a compromise 21. Workshop FarbBV, Koblenz, , Dr. Rico Nestler

12 2. Multi channel sensor elements for lighting applications Condition based approach for optimal sensor configurations spectral input (spectral color stimulus) device matrix sensor response Condition describes the reaction of output variables, here, to changes in the input variables,.italso able to determine the stability of spectral estimates. the robustness of spectral estimates. Condition number maximal relative change of maximal relative change of 21. Workshop FarbBV, Koblenz, , Dr. Rico Nestler 12/23

13 2. Multi channel sensor elements for lighting applications Condition based approach for optimal sensor configurations an appropriate measure should be able to evaluate the stability of the inverse problem modified condition number maximal singular value minimal singular value spectral estimation inverse linear operator (possibly nonlinear) sensor response Characteristics worst case estimation of real condition 1 minimal if is orthogonal, non overlapping sensitivities maximal if has fully linearly dependent sensitivities equivalent for the behavior of the Moore Penrose Inverse 21. Workshop FarbBV, Koblenz, , Dr. Rico Nestler 13/23

14 2. Multi channel sensor elements for lighting applications Condition based approach for optimal sensor configurations (rows of the) device matrix , device matrix 0.5 1, Workshop FarbBV, Koblenz, , Dr. Rico Nestler 14/23

15 2. Multi channel sensor elements for lighting applications Condition based approach for optimal sensor configurations 10, 1 device matrix ,2633 0,4 0.2 device matrix plasmonic sensor nm all channels: 28,9118 without : 13, Workshop FarbBV, Koblenz, , Dr. Rico Nestler 15/23

16 2. Multi channel sensor elements for lighting applications Stability due to condition optimal sensor configuration channel light synthesis OSTAR channel plasmonic sensor relative spectral radiant flux spectral current density / iterative PI Reconstructions from sensor responses with random changes, 1%channel value Workshop FarbBV, Koblenz, , Dr. Rico Nestler 16/23

17 2. Multi channel sensor elements for lighting applications Stability due to condition optimal sensor configuration channel light synthesis OSTAR 0.4 condition based 8 channel sensor relative spectral radiant flux spectral current density / iterative PI Reconstructions from sensor responses with random changes, 1%channel value Workshop FarbBV, Koblenz, , Dr. Rico Nestler

18 3. Considerations about feasibility The feasibility is mainly influenced of internal tolerances of the measuring problem Technologically related manufacturing tolerances Field behavior, primary tolerances of the channel sensitivities Possibilities to deal with it secondary tolerances of the target sensor response spectral current density / 0.4 Reference simulated specimen Workshop FarbBV, Koblenz, , Dr. Rico Nestler 18/23

19 3. Considerations about feasibility Inverse problems are strong sensitive against inner and outer tolerances observable (secondary) tolerances are strong color stimulus dependent (mod. condition number) condition based device configuration additional regularization with global constraints to the Inverse or to the shape characteristics of spectral reconstructions Calibration at secondary level, z.b.! only with references to the measuring task calibration approach directly influenced by the condition of sensor device 21. Workshop FarbBV, Koblenz, , Dr. Rico Nestler 19/23

20 spectral current density / 3. Considerations about feasibility Effect of secondary level sensor calibration Reference simulated specimen Sensor response [ , , , , , , , , ] relative spectral radiant flux,. Calibrationstep Estimationstep Sensor response with calibration [ , , , , , , , , ] Reference sensor response [ , , , , , , , , ] 21. Workshop FarbBV, Koblenz, , Dr. Rico Nestler 20/23

21 4. Summary & Outlook Project FEEDLED 1. Components for innovative, self monitoring lighting solutions with new sensors, integrated measurement data processing, new optical light mixing principles economically and technically efficient scalable for different requirements 2. Sensor monitoring of spectral light characteristics 3. Compensation of aging and drift of all lighting components through integrated sensory control Plasmonic multispectral sensor Origin: Fraunhofer IIS 21. Workshop FarbBV, Koblenz, , Dr. Rico Nestler 21/23

22 4. Summary & Outlook FEEDLED Approach 21. Workshop FarbBV, Koblenz, , Dr. Rico Nestler 22/23

23 4. Summary & Outlook Thank you for your attention! The shown results are part of a currently running joint project Feedback system for intelligent LED based lightings (FEEDLED), which is anchored in the BMBF "Smart Lighting funding program. 21. Workshop FarbBV, Koblenz, , Dr. Rico Nestler 23/23