Artificial intelligence- integrated dual-spectral laser probe gun for molten bath analysis

Abstract

The present invention relates to a dual-spectral artificial intelligence (AI)-integrated laser probe device (200). The device (200) comprises a dual-spectral sensing module (8) comprising an ultraviolet spectrometer (3) configured to collect emission spectra of an elemental composition, and an infrared spectrometer (3) configured to detect vibrational spectra of molecular compounds, a tunable laser auto-focus module comprising a nanosecond-pulse solid-state Nd: YAG laser source, a robust thermal-protection module configured to provide a refractory coating of YSZ combined with an aerogel insulator, an AI-enhanced chemometric processing module (2) comprising a PLSR is combined with ensemble machine learning model configured to perform real-time spectral fusion of UV and IR spectra data, a display interface (4) is configured to present real-time composition, slag-chemistry ratios, and operational alerts to assist real-time process data, wherein the device (200) is configured to provide real-time molten-bath composition analysis of immersion and remote sensing.

Description

FIELD OF THE INVENTION

[0001] The present invention relates to a laser probe device. More specifically, thepresent invention is related to a dual-spectral artificial intelligence (AI)-integratedlaser probe device and method for real-time molten-bath composition analysis.BACKGROUND OF THE INVENTION

[0002] Real-time compositional analysis of liquid steel is a crucial step in modernsteelmaking, enabling tighter control over alloying, deoxidation, desulfurization, andoverall product quality. Traditionally, for chemical composition analysis of liquidiron, steel, slag, etc., during processing in the steel industry, manual sampling andlaboratory analysis via optical emission spectroscopy or wet chemistry have beenperformed. This process provides accurate results but introduces a time delay ofseveral minutes.

[0003] However, the entire composition analysis process took more than 3-4 hours.These delays can lead to inefficiencies, off-spec heats, or additional correctivetreatments. In addition, proper steps towards quality control are quite difficult as theintegrated steel plant / processing steps are linked one with another.

[0004] Moreover, based on the working principle of laser-induced breakdownspectroscopy (LIBS), Optical emission spectroscopy (OES), found few literatures butprimarily based on review articles and also not exactly related to current planning. Inaddition, some lab scale experiments are performed for sensors development andindividual elements identifications but lack of use in practical applications like in steelindustries.

[0005] Various prior studies have reviewed the applications for compositionalanalysis of liquid steel. For example, in the first literature, [1] S.W. Hudson, J.Craparo, R. De Saro, D. Apelian, Applications of Laser-Induced BreakdownSpectroscopy (LIBS) in Molten Metal Processing, Metall. Mater. Trans. B. 48 (2017)2731-2742. https: / / ui.adsabs.harvard.edu / abs / 2017MMTB...48.2731H / abstract.

[0006] In another literature, [2] J.D. Pedarnig, S. Trautner, S. Grunberger, N.Giannakaris, S. Eschlbock-Fuchs, J. Hofstadler, Review of Element Analysis ofIndustrial Materials by In-Line Laser-Induced Breakdown Spectroscopy (LIBS),Appl. Sci. 11 (2021) 9274. https: / / www.mdpi.com / 2076-3417 / 11 / 19 / 9274.

[0007] In another literature, [3] A. Bengtson, Laser Induced Breakdown Spectroscopycompared with conventional plasma optical emission techniques for the analysis ofmetals - A review of applications and analytical performance, Spectrochim. Acta PartB At. Spectrosc. 134 (2017) 123-132.https: / / ui.adsabs.harvard.edu / abs / 2017AcSpB.134..123B / abstract.

[0008] Some prior art / patent searching relating to this invention are as follows:

[0009] The US patent number 10640846 discloses an ultrasonic degassing method formolten metals such as aluminum or copper, where an ultrasonic device with anelongated probe is operated directly in the melt while a purging gas such as argon ornitrogen is introduced near the probe tip to efficiently remove dissolved gases likehydrogen and associated impurities. The device integrates an ultrasonic transducer,probe assembly, and gas delivery system, allowing flexible configurations such asmultiple probes or varying outlet positions, thereby enhancing melt quality andrefining efficiency. The process ensures uniform acoustic cavitation throughout themolten bath, leading to finer gas bubble dispersion and improved degassing kinetics.Additionally, the modular design enables adaptation for different furnace sizes andalloys, supporting both batch and continuous casting operations.

[0010] Moreover, the other prior art CN201266169Y discloses an online detector formolten steel quality based on laser spark spectroscopy, designed to overcome thelimitations of conventional composition detection methods by enabling rapid real-timemonitoring of multiple trace elements such as C, S, P, Cr, Ni, and Si in a steelmakingfurnace. The system consists of an optical probe with plano-convex mirrors housed ina vertical cylindrical shell, a lifting unit for probe positioning, an LSS unit with a laserand vacuum ultraviolet spectrometer, and a signal processing control unit. It furtherincorporates circulating water pipelines for cooling, argon gas pipelines for shielding,and vacuum pumps for stable operation, ensuring accurate, efficient, and continuousdetection of molten steel composition. The integrated control system enablesautomated calibration and data correction, enhancing measurement precision underharsh high-temperature conditions. Furthermore, the compact and robust structuraldesign allows easy integration into existing production lines for continuous,unattended process monitoring.

[0011] Moreover, the other prior art WO2004001394A2 discloses a method andapparatus for continuous monitoring of elements in molten materials using laserinduced breakdown spectroscopy, reducing or eliminating the need for discretesampling and laboratory analysis. By directly analyzing the molten material in realtime, the system enables immediate adjustments to processing conditions, improvingefficiency, precision, and overall product quality. The approach integrates an elementanalysis system coupled directly to the melt, ensuring faster and more accuratedetermination of elemental composition during processing. The design includesoptical fibers and focusing lenses optimized for high-temperature environments,maintaining signal integrity and minimizing contamination. Additionally, the systemsupports multi-element detection with automated spectral calibration, enablingadaptive process control for diverse metallurgical applications.

[0012] Therefore, there is a need to develop a device and method that overcome theselimitations, real-time and in-situ measurement technologies like laser-inducedbreakdown spectroscopy (LIBS), Optical emission spectroscopy (OES) withimmersion probes, additions of other advanced sensors, and, more importantly, byadding a database of all probable elements and compounds and the use of advancedartificial intelligence-machine learning (AI-ML) are promising towards thedevelopment.

[0013] In a nutshell, there is a need to develop a device and method that works on AIML model which provides immediate feedback for process control, processmodification according to the requirements, reducing material waste, energyconsumption, and production costs while ensuring that steel meets stringent qualitystandards.OBJECTIVE OF THE INVENTION

[0014] The main objective of the present invention is to develop a device whichprovides real time composition analysis to provide fast, in-situ analysis of molten iron,steel, and slag composition at a preset temperatures within a period of time.

[0015] Yet another objective of the present invention is to enhance the efficiency ofprocesses like dephosphorization by providing accurate and real-time data, includingphosphorus levels as low as requirements.

[0016] Yet another objective of the present invention is to enable automated processadjustments, such as lance height and oxygen blowing duration, by wirelesslycommunicating real-time data to a central steel plant control module which minimizesreagent waste and reduces over-blowing.SUMMARY OF THE INVENTION

[0017] In an aspect of the present invention, a dual-spectral artificial intelligence(AI)-integrated laser probe device and method for real-time molten-bath compositionanalysis is disclosed.

[0018] In one embodiment of the present invention, a dual-spectral artificialintelligence (AI)-integrated laser probe device comprises a dual-spectral sensingmodule comprising an ultraviolet spectrometer configured to collect emission spectraof an elemental composition which is generated by a laser-induced breakdownspectroscopy (LIBS). The dual-spectral sensing module operates in both immersionmode and remote-sensing mode which is activated via a switch. The elementalcomposition comprising a molten steel, molten iron, and slag.

[0019] In one embodiment of the present invention, the dual-spectral sensingmodule further comprises an infrared spectrometer configured to detect vibrationalspectra of molecular compounds.

[0020] In another embodiment of the present invention, the device further comprisesa tunable laser auto-focus module comprising a nanosecond-pulse solid-stateneodymium-doped yttrium aluminum garnet (Nd: YAG) laser source. The tunablelaser auto-focus module adjusts focal depth based on optical feedback from slagtransparency, surface turbulence, and foam thickness.

[0021] In another embodiment of the present invention, the device further comprisesa harmonic-generator configured to generate ultraviolet (UV) and infrared (IR) laserwavelengths, and an auto-focus mechanism configured to dynamically adjust focaldepth based on molten-bath conditions. The molten-bath conditions comprising slagfoam thickness, surface turbulence, and temperature variations, optimizing signalacquisition. The harmonic generator comprises a Type-I LBO second-harmonicgenerator and a Type-II LBO sum frequency generator for generating about 300-400nanometers deep-UV pulses wavelength.

[0022] In another embodiment of the present invention, the device further comprisesan optical circulator configured to provide distinct pathways of outgoing laser pulsesand incoming emission spectra, a robust thermal-protection module configured toprovide a refractory coating of yttrium-stabilized zirconia (YSZ) combined with anaerogel insulator. The refractory coating comprises about 0.5-2 millimeters thick YSZwith an external aerogel layer for thermal insulation.

[0023] In another embodiment of the present invention, the device further comprisesan artificial-intelligence (AI)-enhanced chemometric processing module comprising apartial least squares regression (PLSR) is combined with ensemble machine learningmodel configured to perform real-time spectral fusion of UV and IR spectra data.

[0024] In another embodiment of the present invention, the device further comprisesa wireless communication module combined with a steel plant's control room moduleconfigured to automate adjustments in process parameters based on real-time readingsof the data. The process parameters comprising lance height, oxygen-blowing durationduring basic oxygen furnace operation.

[0025] In another embodiment of the present invention, the device further comprisesa display interface is configured to present real-time composition, slag-chemistryratios, and operational alerts to assist real-time process data. The device (200) isconfigured to provide real-time molten-bath composition analysis of immersion andremote sensing.

[0026] In one embodiment of the present invention, a method for real-time moltenbath composition analysis is disclosed.

[0027] In one embodiment of the present invention, the first step of the methodinvolves directing tunable ultraviolet (UV) and infrared (IR) laser wavelengths towardthe molten bath via a tunable laser auto-focus module.

[0028] In one embodiment of the present invention, the followed steps of the methodinvolves generating an emission spectra of an elemental composition and molecularvibrational spectra from the molten bath, capturing the emitted spectra via a dualspectral sensing module comprising UV and IR spectrometers, and processing theresultant spectra via an artificial-intelligence model which performs chemometricmethods by using partial least squares regression (PLSR), and applies ensemblemachine-learning model to perform real-time spectral fusion of UV and IR spectradata.

[0029] In one embodiment of the present invention, the followed steps of the methodinvolves determining concentrations of elemental composition and molecularcompound, displaying real-time process data on a handheld interface, and transmittingthe results wirelessly with a steel plant's control room module to provide automaticadjustment of process parameters based on real-time readings of the data. The overallanalytical cycle completes in less than about five seconds.

[0030] This together with the other aspects of the present invention along with thevarious features of novelty that characterize the present disclosure is pointed out withparticularity.

[0031] For a better understanding of the present disclosure, its operating advantages,and the specified objective attained by its uses, reference should be made to theaccompanying descriptive matter in which there are illustrated exemplaryembodiments of the present invention.DESCRIPTION OF THE DRAWINGS

[0001] The advantages and features of the present invention will become betterunderstood with reference to the following detailed description taken in conjunctionwith the accompanying drawings, in which:

[0002] Fig. 1 represents a flow chart showing a method for real-time molten-bathcomposition analysis, according to various embodiments of the present invention;

[0003] Fig. 2 represents an isometric side view of a dual-spectral artificial intelligence(AI)-integrated laser probe device, according to various embodiments of the presentinvention;

[0004] Fig. 3 represents a schematic view of a detailed laser and optical path,according to various embodiments of the present invention;

[0005] Fig. 4 represents a schematic view of an immersion probe gun, according tovarious embodiments of the present invention;

[0006] Fig. 5 represents an overall laser source module, according to variousembodiments of the present invention;

[0007] Fig. 6 represents a schematic view of a laser optics and wavelength-generationsystem, according to various embodiments of the present invention; and

[0008] Fig. 7 represents an isometric view of a structural layout of the AI-integrateddual-spectral laser probe device, according to various embodiments of the presentinvention.

[0009] Like numerals denote like elements throughout the figures.DESCRIPTION OF THE INVENTION

[0010] The exemplary embodiments described herein detail for illustrative purposesare subjected to many variations. It should be emphasized, however, that the presentinvention is not limited to as disclosed.

[0011] It is understood that various omissions and substitutions of equivalents arecontemplated as circumstances may suggest or render expedient, but these areintended to cover the application or implementation without departing from the spiritor scope of the present invention.

[0012] Specifically, the following terms have the meanings indicated below.

[0013] The terms "a" and "an" herein do not denote a limitation of quantity but ratherdenote the presence of at least one of the referenced items.

[0014] The terms "having", "comprising", "including", and variations thereof signifythe presence of a component.

[0015] More specifically, the technical terms used herein are to be understood ascommonly known by those skilled in the field.

[0016] The inventive aspects of the invention along with various components andengineering involved will now be explained with reference to Figures 1- 7 herein.

[0017] In an aspect of the present invention, dual-spectral artificial intelligence (AI)-integrated laser probe device (200) and method (100) for real-time molten-bathcomposition analysis is disclosed.

[0018] Referring to Fig. 1, a flowchart is shown representing steps of real-timemolten-bath composition analysis.

[0019] Referring to Fig. 1, the method (100) begins with step 102, which involvesdirecting tunable ultraviolet (UV) and infrared (IR) laser wavelengths toward themolten bath via a tunable laser auto-focus module.

[0020] Referring to Fig. 1, the method (100) begins with step 104, which involves anemission spectra of an elemental composition and molecular vibrational spectra fromthe molten bath.

[0021] Referring to Fig. 1, the method (100) begins with step 106, which involvescapturing the emitted spectra via a dual-spectral sensing module (8) comprising UVand IR spectrometers (3).

[0022] Referring to Fig. 1, the method (100) begins with step 108, which involvesprocessing the resultant spectra via an artificial-intelligence model which performschemometric methods by using partial least squares regression (PLSR), and appliesensemble machine-learning model to perform real-time spectral fusion of UV and IRspectra data.

[0023] Referring to Fig. 1, the method (100) begins with step 110, which involvesdetermining concentrations of elemental composition and molecular compound.

[0024] Referring to Fig. 1, the method (100) begins with step 112, which involvesdisplaying real-time process data on a handheld interface.

[0025] Referring to Fig. 1, the method (100) begins with step 114, which involvestransmitting the results wirelessly with a steel plant's control room module to provideautomatic adjustment of process parameters based on real-time readings of the data.The overall analytical cycle completes in less than about five seconds. The processparameters comprising lance height, oxygen-blowing duration during basic oxygenfurnace operation.

[0026] In another embodiment of the present invention, dual-spectral artificialintelligence (AI)-integrated laser probe device (200) is disclosed.

[0027] Referring to Fig. 2, a dual-spectral AI-integrated laser probe device (200)comprises a dual-spectral sensing module (8) comprising an ultraviolet spectrometer(3) configured to collect emission spectra of an elemental composition which isgenerated by a laser-induced breakdown spectroscopy (LIBS).

[0028] Moreover, the dual-spectral sensing module (8) operates in both immersionmode and remote-sensing mode which is activated via a switch (6), and the elementalcomposition comprising a molten steel, molten iron, and slag.

[0029] Moreover, the dual-spectral sensing module (8) comprising further comprisesan infrared spectrometer (3) configured to detect vibrational spectra of molecularcompounds.

[0030] Referring to Fig. 2, the device (200) further comprises a tunable laser autofocus module comprising a nanosecond-pulse solid-state neodymium-doped yttriumaluminum garnet (Nd: YAG) laser source, and a harmonic-generator configured togenerate ultraviolet (UV) and infrared (IR) laser wavelengths. The harmonic generatorcomprises a Type-I LBO second-harmonic generator and a Type-II LBO sumfrequency generator for generating about 300-400 nanometers deep-UV pulseswavelength.

[0031] Moreover, the tunable laser auto-focus module is configured to adjust focaldepth based on optical feedback from slag transparency, surface turbulence, and foamthickness.

[0032] Furthermore, the tunable laser auto-focus module comprises an auto-focusmechanism configured to dynamically adjust focal depth based on molten-bathconditions. The molten-bath conditions comprising slag foam thickness, surfaceturbulence, and temperature variations, optimizing signal acquisition.

[0033] Referring to Fig. 2, the device (200) further comprises an optical circulatorconfigured to provide distinct pathways of outgoing laser pulses and incomingemission spectra.

[0034] Referring to Fig. 2, the device (200) further comprises a robust thermalprotection module configured to provide a refractory coating of yttrium-stabilizedzirconia (YSZ) combined with an aerogel insulator. The refractory coating comprisesabout 0.5-2 millimeters thick YSZ with an external aerogel layer for thermalinsulation.

[0035] Referring to Fig. 2, the device (200) further comprises an artificialintelligence (AI)-enhanced chemometric processing module (2) comprising a partialleast squares regression (PLSR) is combined with ensemble machine learning modelconfigured to perform real-time spectral fusion of UV and IR spectra data.

[0036] Referring to Fig. 2, the device (200) further comprises a wirelesscommunication module combined with a steel plant's control room module configuredto automate adjustments in process parameters based on real-time readings of the data.

[0037] Referring to Fig. 2, the device (200) further comprises a display interface (4)is configured to present real-time composition, slag-chemistry ratios, and operationalalerts to assist real-time process data. The device (200) is configured to provide realtime molten-bath composition analysis of immersion and remote sensing.

[0038] Referring to Fig. 3 illustrates a detailed laser and optical path comprising ofmirrors (9, 12), output coupler (10), high refractor (11), diode and cathode units (14and 15), anode assemblies (13, 18), Nd:YAG crystal (17), pockets cell crystal (KDP)(19), polarizer (21), and coupler (16). Fig. 3 demonstrates the internal active Qswitching mechanism and generation of high-energy laser pulses.

[0039] Referring to Fig. 4 illustrates an immersion probe gun (23) with YSZ ceramicand aerogel insulation (22), including probe connectors used for the optical circulatorsystem Probe 1 (24), Probe 2 (25), and Probe 3 (26).

[0040] Referring to Fig. 5 illustrates a complete laser source module (27) with theharmonic generation setup, including the type-I LBO second harmonic generator (28),type-II LBO sum frequency generator (29), optical filter AR-355 nm (30), opticalparametric oscillator (31), input coupler (32), BBO crystal (33), switchable outputcoupler (34), in-path SHG for UV (35), and IR / UV optical filter switch (36), deep-UVfilter (37), mid-IR filter (38), short-wavelength resonant coupler (39), and longwavelength resonant coupler (40). The figure highlights how dual spectral UV and IRbeams are generated, tuned, and delivered.

[0041] Referring to Fig. 6 illustrates a schematic view of the laser optics andwavelength-generation system used inside the probe device (200). Fig. 6 begins withthe laser source (27) followed by nonlinear optical components, second harmonicgenerator (28), sum frequency generator (29), and optical filters (30) which convertabout 1000-1100 nm fundamental wavelength into UV or IR output data. The beamthen enters the optical parametric oscillator (31) through an input coupler (32), wherea Type-II BBO crystal (33) and switchable output coupler (34) generate tunable signaland idler wavelengths.

[0042] Moreover, modules such as the SHG (35), optical filter switch (36), deep-UVfilter (37), and mid-IR filter (38) configure the beam for dual-spectral sensing. Also,Fig. 6 illustrates the resonant cavity layout, including the short- and long-wavelengthcouplers (39, 40) used for switching between UV and IR operation.

[0043] Referring to Fig. 7 illustrates a structural layout of the AI-integrated dualspectral laser probe device (200). Fig. 7 discloses a tubular, mesh-type reinforcedframe enclosing the internal electronic and optical components. A liquid nitrogen inlet(41) and outlet (42) are positioned along the upper side of the frame to supply activecooling to the immersion section. Inside the housing, the arrangement of the AIprocessing unit, spectrometer (3), optics, laser module, and beam delivery path isvisible.

[0044] Moreover, Fig. 7 also illustrates a cut-section side view of the handheld gun,including the handle, display interface (4), remote-sensing switch (6), and internalrouting of optical fibers, laser resonators, and cooling pipes.

[0045] In one embodiment of the present invention, a laser probe device (200) iscapable of immersion into molten iron as well as remote sensing, steel or slagenvironments at about 1200 degree Celsius(0C)- 17000C. The device (200) comprisesdual-spectral sensing i.e., infrared vibrational and ultraviolet elemental spectroscopywith AI-enhanced chemometric modelling such as partial least squares regression(PLSR) which is combined with ensemble machine learning.

[0046] Moreover, the device (200) comprises a tunable laser auto-focus module topenetrate slag foams, paired with refractory-coated immersion shields and activeliquid nitrogen (N2) gas cooling for thermal protection. Also, real-time spectral fusionalgorithms are selectively utilizing IR versus UV data or combined signals to optimizemolecular compound and elemental quantification.

[0047] Furthermore, the device (200) comprises a wireless communication modulewith a steel plant's control room module enable automated feedback control such aslance height, oxygen blowing duration etc. based on detected phosphorous levels andslag chemistry ratios, delivering enhanced dephosphorization efficiency and reducedover-blowing.

[0048] Moreover, the overall analysis process operated with less than five (<5)second detection time and includes ambient interference mitigation via shieldedphotodiodes and arc compensation. Also, the molten steel and slag compositionmonitoring traditionally relies on grab sampling or fixed optical emission or X-rayfluorescence systems, which involve delays, batch processing, and immobilizedinfrastructure. Real time in-line measurements remain challenging due to extremetemperatures, slag foaming, turbulence, and refractory wear.

[0049] Furthermore, current portable device (200) offers elemental or compoundanalysis, not fused dual-spectral insight. Chemometric methods such as PLSR existbut suffer from calibration drift and limited robustness under dynamic conditions,ensemble ML methods promise greater accuracy but have not been integrated inportable molten-bath tools. Feedback automation in dephosphorization is often basedon indirect process sensors rather than direct molten-phase analytics.

[0050] However, the present invention addresses these limitations by combining bothspectral modalities i.e., IR and UV for structural and elemental detection,implementing AI fusion to select or merge modalities based on slag transparency andtemperature, and deploying a tunable laser focusing mechanism that adapts to slagdepth and optical distortion. Also, thermal resilience is ensured via refractory coatingand argon gas cooling, and wireless connectivity enables closed-loop control byadjusting process parameters in near real-time.

[0051] Moreover, the device's thresholds such as temperature, viscosity, detectiontime, elemental sensitivity are tailored for extreme steelmaking conditions. Also, thedevice (200) offers faster, more accurate, in-situ molten metal and slag analysis,supporting automated steel process control with enhanced efficiency and reducedreagent waste.

[0052] In another embodiment of the present invention, the device (200) comprisesa dual-spectral sensing module (8) combined with a mid-IR spectrometer (3) forcomputer fingerprinting and a deep-UV spectrometer (3) for atomic signatures, asmart spectral fusion engine driven by weighted PLSR ensemble models refined bymachine learning, a laser-focused penetration system with dynamic depth adjustment,and a robust thermal protection system using yttrium-stabilized zirconia (YZS)ceramics, aerogel insulation, and argon jet cooling.

[0053] Moreover, the device (200) comprises a wireless process interface for realtime integration with steel plant control systems, delivering within 5 seconds detectioncycles and precision in slag component measurement across extreme viscosities andtemperature. An actively Q-switched diode pulsed solid state neodymium-dopedyttrium aluminum garnet (Nd: YAG) laser source which is used to generate laserpulses of wavelength of about 1000-1100 nanometers.

[0054] In another embodiment of the present invention, a hollow cylindrical diode isused to pump the solid gain medium i.e. the neodymium (III) ion (Nd3+) doped yttriumaluminum garnet (YAG) crystal rod to generate the characteristic laser beam.Integrated with an electro-optic modulator i.e. pockets cell, where non-linear optical(NLO) KD*P (Potassium Dideuterium Phosphate) crystals serve as the gain medium,for active Q-switch which stores energy from the continuous beam and emit Q-factormodulated ultra-short high energy pulses. The optical arrangements serve as the laserresonator. The output coupler allows nanosecond pulses of short pulse duration ofabout 3-6 nanoseconds (ns), 200-300 hertz (Hz) repetition rate and about 500-1000millijoule (mJ) energy pulses at about 1064 nanometers (nm).LASER PROBE GUN

[0055] Coated Immersion with Cooling facility: The dual-spectral laser probe gunis going to develop for direct immersion into molten steel and slag environments ofabout 1200°C-1700°C using a refractory-coated shield made primarily of yttriumstabilized zirconia (YSZ) ceramics combined with aerogel insulation, with coatingthickness typically ranging of ab out 0.5 to 2 millimeters (mm) to provide robustthermal and mechanical protection while maintaining probe maneuverability.

[0056] Moreover, this coating ensures high thermal stability and chemical inertnessat operating temperatures, preventing degradation of the optics and sensors. Thermalprotection is further enhanced by an active cooling system: a liquid nitrogen flowingdirects to the immersion tip, creating a protective barrier that reduces slag adhesion,limits oxidations, and cools the probe surface. Liquid N2 flow also extended withinthe device (200) to keep internal electronics and AI processing modules (2) at safeoperating temperatures. Together, these features enable stable, real-time dual-spectralUV and infrared measurements by mitigating ambient interference, thermal drift, andmechanical wear during immersion in extreme molten bath conditions.

[0057] Furthermore, this integrated approach ensures precise, rapid (<5 s) elementaland compound analysis with enhanced durability and reliability, supporting automatedsteelmaking control. As demonstrated, a detachable aerogel coated YSZ immersionrod of length approximately 600-700 mm is connected to the nozzle of device (200)for immersive spectroscopy. Also, thin grooves inside the wall of immersion are servefor liquid cooling. A closed-loop active indirect liquid N2 based cooling system isemployed for the device (200). The cylindrically meshed radiative coils and heatexchangers ensure fast and efficient cooling of the device (200).

[0058] Optics & Beam delivery system: The dual-spectral laser Probe gun employsan integrated optics and beam delivery system coupled with a dual- spectrometer (3)assembly to enable real-time, in-situ analysis of molten steel and slag. The device(200) combines ultraviolet i.e., UV: 200-400 nm and mid-infrared i.e., IR: 2.5-25micrometer (μm) spectroscopy for simultaneous elemental and molecularcharacterization.

[0059] Moreover, tunable beam-splitting and dichroic optics direct or fuse the UVand IR beams within a compact, thermally protected optical path, ensuring precisespectral routing and alignment. A tunable laser auto-focus module or smart auto-focusmechanism is capable of dynamically adjusts focal depth to compensate for slag foamthickness, surface turbulence, and temperature variations, optimizing signalacquisition.

[0060] Furthermore, the UV channel, based on laser-induced breakdownspectroscopy (LIBS) captures atomic emissions from elements such as phosphorus,Sulphur, manganese, and silicon, while the IR spectrometer (3) detects molecularvibrational signatures of compounds such as FeO, Fe2O3, CaO, Al2O3, MnO, MgO,SiO2, and P2O5 for slag composition analysis.

[0061] Moreover, an AI-driven spectral fusion engine integrates data from bothchannels using advanced regression and ensemble models, enhancing analyticalrobustness under variable process conditions. The entire system, enclosed within arefractory-coated, nitrogen-cooled housing, maintains optical stability and enablessub-5-second spectral response with wireless connectivity to plant control systems forautomated process optimization. A system of optics is used for generating the thirdharmonics of about 1000-1100 nm pulses at about 350-360 nm.

[0062] Furthermore, this consists of a second harmonic generator, and then a sumfrequency generator. The second harmonics is generated by Type-I cut NLO LBO(Lithium Triborate) crystal which stores two photons of same phase and energy andreleases a photon with double the energy and frequency. The sum frequency generatorhas a Type-II cut NLO LBO crystal which adds energy of two same phase photons.

[0063] Moreover, these phases are manipulated by polarizers and λ / 2 or λ / 4 waveplates. An optical parametric oscillator (OPO) is pulsed by the 355 nm pulses to giveshort and long wavelength outputs. It consists of a type-II cut NLO BBO (β-bariumborate) crystal which generates emitted two photons of lesser energies namely signaland idler. The resonating cavity is switchable between idler (long wavelength) andsignal (short wavelength) resonant using the couplers.

[0064] Furthermore, the size of resonating cavity determines the energy of outputpulses. For generating deep UV, a second harmonic generator (SHG) of Type-I cutNLO BBO will be in-path to convert pulses of visible wavelength (signal) into UV,whereas, for mid IR pulses the SHG will be out-of-path to allow the idler pulses. AnIR and UV switching optical filter ensures delivery of monochromatic pulses. Anoptical circulator is ensuring the distinct pathways of outgoing laser pulses andincoming emission spectra by acting as a laser diode. In the demonstration, the photonsentering in probe 1 will be exiting through probe 2 and photons entering through probe2 will be exiting through probe 3.

[0065] AI processing model: The AI processing module (2) in the dual-spectral laserprobe device (200) functions as the central analytical intelligence, processing realtime spectral data from both the infrared (IR) and ultraviolet (UV) spectrometers (3)to predict the physical and chemical composition of molten metal, slag, and steelphases including Fe, C, Mn, Cu, P, S, Ni, Si, Al2O3, FeO, CaO, MnO, MgO, Fe2O3,P2O5, SiO2, etc.

[0066] Moreover, the module comprises a hybrid AI-ML engine which combinespartial least squares regression (PLSR) for linear chemometric calibration withensemble machine learning algorithms e.g., Random Forest, Gradient Boosting, andNeural Networks to model nonlinear temperature-dependent spectral responses.

[0067] Furthermore, the AI module dynamically selects or fuses IR vibrational andUV elemental data based on signal clarity, slag transparency, and bath conditions,referencing an integrated materials property database containing thermodynamicconstants, optical coefficients, atomic emission lines, and vibrational fingerprints fromthe periodic table. Through iterative model optimization and error correction loops,the AI continuously refines predictions, compensates for arc noise, turbulence, andcalibration drift, and outputs elemental concentrations and slag chemistry ratios withabout <5 percent (%) relative error.

[0068] These results are wirelessly transmitted to the steel plant's control module forreal-time feedback actions (e.g., adjusting oxygen blowing or lance height), ensuringoptimized dephosphorization, minimized reagent use, and enhanced steel purity. Asystem on chips (SoC) is used for AI processing module (2) for computing largelibraries and operating hardware and electronics. A touch enabled light-emitting diode(LED) display is used for user interface for specialized optimization of elementalanalyses.

[0069] Dual Spectrometer: The dual-spectral module in the handheld laser probedevice (200) combines ultraviolet (UV) and infrared (IR) spectroscopy to deliver realtime, in-situ analysis of molten steel and slag. The UV spectrometer (3) of about 200-400 nm operates on principles similar to laser-induced breakdown spectroscopy(LIBS), capturing atomic emission lines from excited elements like phosphorus,Sulphur, manganese, and silicon, enabling precise elemental quantification even attrace levels.

[0070] Moreover, simultaneously, the mid-infrared spectrometer (3) of about 2.5-25μm detects molecular vibrational signatures of compounds such as FeO, SiO2, P2O5etc., and providing structural insights into slag chemistry. A tunable laser auto-focusmodule dynamically adjusts penetration depth through slag foams, ensuring optimalsignal acquisition. An AI-driven spectral fusion engine, combining partial leastsquares regression (PLSR) with ensemble machine learning models, intelligentlyweighs or merges UV and IR data based on slag opacity, temperature, and spectralclarity, enhancing robustness and accuracy under variable conditions.

[0071] Furthermore, this dual-modality approach, integrated within a thermallyprotected, argon-cooled probe, enables sub-5-second detection cycles and wirelessintegration with steel plant control systems, supporting automated feedback controlfor improved dephosphorization efficiency and reduced process variability.

[0072] Remote sensing facility: The remote sensing facility operates in twocomplementary modes immersion and ray-based (non-contact) sensing to enable realtime analysis of molten iron, steel, or slag under extreme conditions. In immersionmode, the probe tip, protected by refractory coating and N2 gas cooling, is directlyinserted into the molten bath where a dual-spectral laser such as infrared andultraviolet which generates molecular and atomic spectra that are collected andprocessed by the optics and spectrometer (3).

[0073] Moreover, the AI-based chemometric engine fuses IR vibrational data forcompound identification with UV emission data for elemental quantification,producing composition and temperature readings within five seconds. In remotesensing mode, activated via a switch (6), the same laser and optics system operatesfrom a safe distance either from the furnace sidewall or above the molten surface usingauto-focus optics and adaptive beam delivery to penetrate slag foam to sub surface orsurface layers.

[0074] Furthermore, reflected and re-emitted spectra are analyzed similarly, with AIdynamically selecting IR or UV channels depending on slag transparency andtemperature. This hybrid configuration allows the probe to automatically switch (6)between immersion and remote sensing based on optical feedback and thermalconditions, ensuring continuous, accurate in-line monitoring of molten metal and slagcomposition. A wireless transmission of results to the steel plant control moduleenables closed-loop process automation for optimized dephosphorization and oxygenblowing efficiency, offering a robust, portable alternative to conventional delayedlaboratory sampling methods.

[0075] Moreover, the remote sensing facility comprises several key componentsworking together to enable accurate spectral analysis of molten metal. A pulsed lasersource generates high-energy laser pulses that create micro-plasma on the moltensurface, while beam delivery and focusing optics direct and concentrating the beamover long distances to achieve the required energy density. The spectrometer (3) thencollects and analyses emissions from spectral regions, reducing interferences andenhancing element-specific sensitivity. An intensified charge-coupled device (ICCD)detector, with high sensitivity and time-gated capability, captures faint spectral linesfrom the rapidly cooling plasma while suppressing intense background radiation fromthe molten bath.

[0076] Furthermore, all optical and electronic components are enclosed within arobust remote housing that shields them from extreme heat, dust, and corrosive fumes.Additionally, an optional inert gas purging system, typically using argon, clears fumesand slag from the line of sight, minimizes atmospheric interference, and preventsoxidation on the molten surface, ensuring more stable and repeatable measurements.

[0077] Display interface: The display interface (4) of the dual-spectral laser probedevice (200) functions as the real-time interface that presents AI-processed analyticalresults obtained from dual-spectral sensing of molten iron, steel, or slag. The displayreceives processed data from the AI processing unit, which fuses infrared (IR) andultraviolet (UV) spectral signals through chemometric and ensemble machine learningmodels to quantify both elemental and compound compositions within about 4-5seconds.

[0078] Moreover, the display shows critical process parameters such asconcentrations of FeO, CaO, SiO2, and phosphorus, basicity ratios, slag transparency,and system status indicators like temperature, calibration health, and signalconfidence. It also provides operational alerts and recommendations e.g., extendingoxygen blowing or adjusting lance height to assist real-time process control.

[0079] Furthermore, with a high-brightness, heat-resistant organic light emittingdiode (OLED) or liquid-crystal display (LCD) interface, the display remains visiblein harsh furnace environments and supports both immersion and remote sensingmodes. Additionally, it confirms wireless transmission of analytical results to thecentral steel plant control system, ensuring closed-loop automation and optimizeddephosphorization efficiency.

[0080] Process Digitalization: The process digitalization of the dual-spectral laserprobe device (200) functions through seamless integration with the steel plant's digitalcontrol architecture. The handheld probe, upon immersion into molten metal or slag,acquires real-time infrared and ultraviolet spectra which are processed on board usingAI-driven chemometric models.

[0081] Moreover, the AI processing module (2) executes data fusion algorithms tointerpret compound and elemental compositions within seconds. The processed datais then wirelessly transmitted via industrial IoT protocols e.g., OPC-UA or MQTT tothe central steelmaking control system or Level 2 automation network. In the controlroom, this information is visualized on process dashboards and integrated withsupervisory control logic to enable automatic adjustments such as oxygen lancepositioning, blowing rate, or flux additions.

[0082] Furthermore, continuous feedback loops ensure dynamic optimization ofdephosphorization, desulfurization, and slag formation processes. The digitalconnection enables real-time decision-making, predictive maintenance alerts, and dataarchiving for process analytics, effectively linking field-level molten bath sensing withplant-level process digitalization and closed-loop control.Numbering Description1 Cooling in & out system2 Artificial Intelligence (AI) processing unit3 Spectrometer4 Result display section5 Handle for immersion sensing6 Switch for remote sensing7 Optics & beam delivery system8 Dual spectral laser9 1st mirror10 Output Coupler11 High refractor12 2nd mirror13 Anode14 Diode15 Cathode16 Coupler17 Nd:YAG Crystal18 Anode19 Pockets Cell Crystal (KDP)20 Cathode21 Polarizer22 Yttrium-stabilized zirconia (YSZ) ceramic coating combined with aerogelinsulator23 Immersion probe gun24 Probe 1 for optical circulator25 Probe 2 for optical circulator26 Probe 3 for optical circulator27 Laser Source28 Second Harmonic generator (type 1 LBO)29 Sum frequency generator (Type 2 LBO)30 Optical filter (AR @ 355 nm)31 Optical parametric oscillator (dual resonant cavity)32 Input coupler33 Type 2 BBO (non-collinear phase matching)34 Output coupler switchable for signal and idler resonant35 SHG (in path for UV output)36 Optical filter switch for UV and IR37 Deep UV filter38 Mid IR filter39 Short wavelength resonant coupler40 Long wavelength resonant coupler41 Liquid N2 gas flow in42 Liquid N2 gas flow outAdvantageous effects of the present invention

[0083] The main of the present invention that the device give an extra hand to theworking expert in steel industry for instant, quick and real time compositional analysisof the steel or slags.

[0084] Moreover, the present invention performs a real-time analysis which providesaccurate compositional results within about 5 seconds, thereby reducing analysis timefrom hours to seconds and allowing immediate process correction.

[0085] Moreover, the present invention performs enhanced precision, in which AIdriven data fusion minimizes temperature- and slag-related interferences, therebymaintaining compositional accuracy within ±3-5 % and temperature fluctuation below±10 °C.

[0086] Moreover, the present invention enables comprehensive detection whichsimultaneously quantifies both elemental and compound species, enabling a completeunderstanding of the molten bath chemistry that conventional methods lack.

[0087] Moreover, the present invention provides dual-mode flexibility whichoperates in both immersive (direct contact) and non-immersive (remote sensing)modes, ensuring uninterrupted monitoring across different furnace or slag conditions.

[0088] Moreover, the present invention provides thermal and mechanical resilience,in which the YSZ ceramic coating with aerogel insulation and cryogenic coolingsafeguards internal optics and electronics against extreme heat and slag corrosion.

[0089] In a nutshell, the device and method of the present invention overcomes thedrawbacks discussed in the conventional techniques and provided a cost effective andefficient way of operating real-time molten-bath composition analysis.

[0090] The foregoing descriptions of specific embodiment of the present inventionhave been presented for purposes of description. They are not intended to beexhaustive or to limit the present invention to the precise forms disclosed, andobviously many modifications and variations are possible in light of the aboveteaching.

[0091] Further, the embodiments were chosen and described in order to best explainthe principles of the present invention and its practical application and thereby enableothers skilled in the art to best utilize the present invention and various embodimentswith various modifications as are suited to the particular use contemplated.

[0092] It is understood that various omissions and substitutions of equivalents arecontemplated as circumstances may suggest or render expedient, but such omissionsand substitutions are intended to cover the application or implementation withoutdeparting from the spirit or scope of the present invention.

Claims

1. A dual-spectral artificial intelligence (AI)-integrated laser probe device (200), wherein the device (200) comprising: a dual-spectral sensing module (8) comprising: an ultraviolet spectrometer (3) configured to collect emission spectra of an elemental composition which is generated by a laser-induced breakdown spectroscopy (LIBS), and an infrared spectrometer (3) configured to detect vibrational spectra of molecular compounds; a tunable laser auto-focus module comprising: a nanosecond-pulse solid-state neodymium-doped yttrium aluminum garnet (Nd: YAG) laser source, and a harmonic-generator configured to generate ultraviolet (UV) and infrared (IR) laser wavelengths, and an auto-focus mechanism configured to dynamically adjust focal depth based on molten-bath conditions; an optical circulator configured to provide distinct pathways of outgoing laser pulses and incoming emission spectra; a robust thermal-protection module configured to provide a refractory coating of yttrium-stabilized zirconia (YSZ) combined with an aerogel insulator; an artificial-intelligence (AI)-enhanced chemometric processing module (2) comprising a partial least squares regression (PLSR) is combined with ensemble machine learning model configured to perform real-time spectral fusion of UV and IR spectra data; a wireless communication module combined with a steel plant's control room module configured to automate adjustments in process parameters based on realtime readings of the data; and a display interface (4) is configured to present real-time composition, slagchemistry ratios, and operational alerts to assist real-time process data, wherein the device (200) is configured to provide real-time molten-bath composition analysis of immersion and remote sensing.

2. The device (200) as claimed in claim 1, wherein the tunable laser auto-focus module adjusts focal depth based on optical feedback from slag transparency, surface turbulence, and foam thickness.

3. The device (200) as claimed in claim 1, wherein the elemental composition comprising a molten steel, molten iron, and slag.

4. The device (200) as claimed in claim 1, wherein the molten-bath conditions comprising slag foam thickness, surface turbulence, and temperature variations, optimizing signal acquisition.

5. The device (200) as claimed in claim 1, wherein the process parameters comprising lance height, oxygen-blowing duration during basic oxygen furnace operation.

6. The device (200) as claimed in claim 1, wherein the refractory coating comprises about 0.5-2 millimeters thick YSZ with an external aerogel layer for thermal insulation.

7. The device (200) as claimed in claim 1, wherein the dual-spectral sensing module (8) operates in both immersion mode and remote-sensing mode which is activated via a switch (6).

8. The device (200) as claimed in claim 1, wherein the harmonic generator comprises a Type-I LBO second-harmonic generator and a Type-II LBO sum frequency generator for generating about 300-400 nanometers deep-UV pulses wavelength.

9. A method (100) for real-time molten-bath composition analysis, wherein the method (100) comprising the steps of: directing (at step 102) tunable ultraviolet (UV) and infrared (IR) laser wavelengths toward the molten bath via a tunable laser auto-focus module; generating (at step 104) an emission spectra of an elemental composition and molecular vibrational spectra from the molten bath; capturing (at step 106) the emitted win via a dualispectral sensing module 18) comprising (IV and IR spectrometers 13); processing IN step 108) the resultant spectra via an artifcial-intelligence model which perfonns chemometric methods by using partial least squares regression (P1.SR I. and applies ensemble machine-learning model to perform real-time spectral fusion of 11V and IR spectra data; determining (at step 110) concentrations of elemental composition and molecular compound; displaying (at step 112) realaime process data on a handheld interface; and transmitting (at step 114) the result. wirelessly with a steel plants control room module to provide automatic adjustment of prams parameters based on real-time readings of the data, wherein the overall analytical cycle complete in less than about live seconds.