Intelligent welding control system for prefabricated box girder frame

Abstract

The application relates to the technical field of intelligent welding control, and discloses an intelligent welding control system for a prefabricated box girder frame, which comprises an alternating magnetic field module, a laser excitation module, an interference receiving module, a phase-locked extraction module and a main control module. The alternating magnetic field module generates a magnetic field in a molten pool area and outputs a reference signal, and drives a solid-liquid interface to produce micro oscillation; the laser excitation module excites ultrasonic waves, and the interference receiving module converts ultrasonic echoes carrying oscillation characteristics into voltage signals; the phase-locked extraction module carries out phase-locked demodulation by using the reference signal and extracts time-of-flight data; and the main control module calculates actual penetration depth by combining the data with a sound velocity distribution model. The application filters out environmental interference through phase-locked demodulation, improves calculation precision by using sound velocity compensation, adjusts electromagnetic lifting force generated by the alternating magnetic field when there is a risk of burning through, realizes closed-loop intervention and prevents welding defects.

Description

Technical Field

[0001] This invention relates to the field of intelligent welding control technology, specifically to an intelligent welding control system for prefabricated box girder frames. Background Technology

[0002] In the manufacturing process of precast box girder frames, welding quality directly affects the mechanical properties and safety stability of the overall structure. Among these, precise control of weld penetration depth is crucial to ensuring joint strength. Current online detection and control of weld penetration depth mostly rely on ultrasonic reflection methods to obtain the penetration state, but this method still has significant shortcomings in practical applications.

[0003] Existing technologies for extracting ultrasonic echoes from the bottom of the molten pool face challenges. High-intensity arc noise and photoelectric coupling interference at the welding site easily drown out weak ultrasonic echo signals, making effective extraction of ultrasonic time-of-flight characteristics difficult and limiting the overall system's anti-interference capability. Furthermore, welding heat input creates a non-uniform transient temperature gradient field within the base material, causing nonlinear changes in the sound velocity as ultrasound propagates in a variable-temperature medium. Current measurement schemes often use constant sound velocity or single empirical formulas for compensation, lacking dynamic reconstruction of the one-dimensional temperature distribution profile. This results in discrepancies between the sound velocity model and the actual physical state, leading to lower accuracy in calculating the actual weld depth.

[0004] Furthermore, existing control systems use a static threshold for the thickness of the underlying substrate when assessing welding defect risks, failing to consider the decrease in the load-bearing capacity of the base material due to transient heat accumulation. When facing critical conditions, the system can only passively respond by adjusting external welding parameters, lacking physical intervention methods to actively counteract the gravitational load of the liquid metal deep within the molten pool. This makes it difficult to effectively avoid burn-through or incomplete penetration defects even under complex working conditions. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an intelligent welding control system for prefabricated box girder frames, which solves the problem that arc noise and transient nonlinear temperature fields lead to low accuracy in online weld penetration detection, thus easily causing defects such as incomplete penetration or burn-through.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an intelligent welding control system for precast box girder frames, comprising:

[0007] An alternating magnetic field module, deployed in the peripheral area of ​​the welding torch, includes an array of orthogonal electromagnetic coils and an excitation current generator connected to the array of orthogonal electromagnetic coils. It is used to generate an alternating magnetic field in the weld pool area and output a reference signal with a corresponding characteristic frequency.

[0008] The laser excitation module is integrated at the front end of the welding torch. The optical path of the emitting end is aligned with the surface of the base material in the solidification zone to emit a pulsed laser beam and generate broadband ultrasonic waves.

[0009] An interference receiving module is arranged adjacent to a laser excitation module. The optical receiving end is aligned with the ultrasonic excitation area of ​​the laser excitation module. It is used to convert the transient surface displacement caused by the ultrasonic echo generated after the broadband ultrasonic wave is reflected into the original interference voltage signal.

[0010] The phase-locked extraction module receives the reference reference signal and uses the reference reference signal to perform quadrature phase-locked demodulation processing on the original interference voltage signal, locking and extracting the corresponding time-of-flight data.

[0011] The main control module calculates the current actual melting depth based on flight time data and system built-in parameters, compares the actual melting depth with the target melting depth data to obtain the deviation value, and outputs control parameter adjustment commands to the alternating magnetic field module.

[0012] Furthermore, an alternating magnetic field module excites a volumetric Lorentz force within the weld pool, driving the solid-liquid interface to generate mechanical micro-oscillations at the same frequency as the characteristic frequency. A laser excitation module then excites ultrasonic waves to propagate downwards. When the longitudinal ultrasonic wave reflects at the micro-oscillating solid-liquid interface, the transient displacement of the interface modulates the acoustic phase of the ultrasonic echo, generating an ultrasonic echo carrying the characteristic frequency.

[0013] The interference receiving module converts the mechanical displacement of the substrate surface into the original interference voltage signal. Then, the phase-locked extraction module uses the same frequency reference reference signal to perform orthogonal mixing and low-pass filtering, shifting the target signal in the high-frequency modulation band to the baseband and filtering out broadband noise, thereby extracting the ultrasonic time of flight.

[0014] The main control module reconstructs the sound velocity distribution model of the nonlinear temperature-varying medium to compensate for acoustic delay, obtains actual melt depth data, and feeds back the phase and amplitude of the alternating magnetic field based on the state assessment results to generate an upward electromagnetic lifting force in the molten pool, thereby achieving closed-loop control of the welding penetration state.

[0015] Preferably, the quadrature electromagnetic coil array consists of four independently wound electromagnetic excitation coils, which are uniformly and symmetrically distributed in a ring and arranged perpendicularly to each other in pairs. The excitation current generator includes a signal generation unit and a power amplification unit. The signal generation unit is used to receive operating parameter commands and generate a low-voltage reference waveform signal, and the power amplification unit is used to amplify the low-voltage reference waveform signal and output an alternating excitation current signal to the quadrature electromagnetic coil array.

[0016] Preferably, the alternating magnetic field couples with the welding current inside the weld pool region to generate a volume Lorentz force; the volume Lorentz force drives the solid-liquid interface at the bottom of the weld pool region to generate mechanical micro-oscillations with corresponding characteristic frequencies.

[0017] Preferably, the laser excitation module includes a nanosecond-level Q-switched pulsed solid-state laser and an optical collimating and focusing lens group. The pulsed laser beam excites broadband ultrasound. The broadband ultrasound is reflected at the solid-liquid interface in a state of mechanical micro-oscillation, generating an ultrasound echo carrying the phase information of the corresponding mechanical micro-oscillation law.

[0018] Preferably, the interference receiving module includes a continuous wave probe laser, an optical beam splitter interferometer group, and a high-frequency photodetector; the continuous wave probe laser outputs a probe laser to acquire reflected light carrying an optical phase shift; the reflected light enters the optical beam splitter interferometer group and interferes with the reference beam, and is converted into a raw interference voltage signal by the high-frequency photodetector.

[0019] Preferably, the original interference voltage signal consists of a modulated ultrasonic echo signal component and a broadband interference component; the broadband interference component includes a high-intensity acoustic-optical coupling interference component, a thermo-elastic background noise component, and a system thermal noise component; the modulated ultrasonic echo signal component carries the frequency domain modulation characteristics of the corresponding characteristic frequency.

[0020] Preferably, the phase-locked extraction module includes a waveform conversion and precision phase-shifting circuit and two dual-balanced analog multipliers; the waveform conversion and precision phase-shifting circuit converts the reference reference signal into an in-phase reference signal and a quadrature reference signal; the two dual-balanced analog multipliers multiply the original interference voltage signal with the in-phase reference signal and the quadrature reference signal respectively, and output the in-phase mixing signal and the quadrature mixing signal.

[0021] Preferably, the phase-locked extraction module further includes a low-pass filter circuit and a digital signal processor; the in-phase mixer signal and the quadrature mixer signal are processed by the low-pass filter circuit into an in-phase baseband signal and a quadrature baseband signal; the digital signal processor performs an envelope amplitude calculation algorithm on the in-phase baseband signal and the quadrature baseband signal and extracts the transient echo envelope amplitude, and locks the time-of-flight data corresponding to the peak amplitude of the transient echo envelope amplitude through peak optimization logic.

[0022] Preferably, the main control module runs a transient heat conduction calculation unit, which is used to reconstruct a one-dimensional spatial temperature distribution profile and establish a sound velocity distribution model; the main control module constructs an integral equation based on the time-of-flight data and the sound velocity distribution model, executes a numerical discretization approximation and iterative solution algorithm, and calculates the actual melting depth.

[0023] Preferably, the main control module calculates the remaining unfused physical thickness and the transient heat accumulation index, and uses the transient heat accumulation index to correct the basic safe thickness threshold to obtain the dynamic critical threshold; when the remaining unfused physical thickness is less than the dynamic critical threshold, it sends a control parameter adjustment command to the alternating magnetic field module to adjust the magnetic flux density amplitude and magnetic field phase angle of the alternating magnetic field, and generates a time-averaged Lorentz force in the weld pool area as an upward electromagnetic lifting force.

[0024] This invention provides an intelligent welding control system for precast box girder frames. It has the following advantages:

[0025] 1. This invention uses the volumetric Lorentz force generated by the coupling of an alternating magnetic field and welding current to drive the solid-liquid interface of the molten pool to generate mechanical micro-oscillations with corresponding characteristic frequencies. This causes the reflected ultrasonic echo to carry the characteristic frequency in phase. Combined with the phase-locked extraction module, the same-frequency reference reference signal is used to perform orthogonal phase-locked demodulation processing, shifting the target signal to the baseband and filtering out broadband interference. This allows for the extraction of ultrasonic time-of-flight data in a strong arc acoustic-optical noise environment, improving the anti-interference capability of the system detection.

[0026] 2. This invention utilizes the transient heat conduction calculation unit of the main control module to reconstruct the one-dimensional spatial temperature distribution profile inside the base material, establishes a sound velocity distribution model, and combines it with the extracted time-of-flight data to construct an integral equation for iterative solution. This eliminates the interference of nonlinear changes in ultrasonic sound velocity caused by non-uniform temperature gradient fields on ranging, implements sound velocity compensation for variable temperature media, and improves the accuracy of actual melting depth data.

[0027] 3. This invention uses the main control module to calculate the transient heat accumulation index to dynamically correct the basic safety thickness threshold. Combined with the remaining unfused physical thickness, it assesses the pressure-bearing state of the underlying base material. When a burn-through risk is determined, the system actively adjusts the magnetic flux density amplitude and magnetic field phase angle of the alternating magnetic field to generate an upward electromagnetic lifting force inside the molten pool to offset part of the downward load of the liquid metal. This achieves closed-loop physical intervention on the depth of the molten pool, preventing incomplete penetration and burn-through defects. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the system functional module architecture of the present invention;

[0029] Figure 2 This is a schematic diagram of the system operation method of the present invention;

[0030] Figure 3 This is a schematic diagram comparing the original signal with the phase-locked loop extraction envelope of the present invention;

[0031] Figure 4 This is a schematic diagram comparing the dynamic actual melting depth and electromagnetic intervention response of the present invention;

[0032] Figure 5 This is a schematic diagram showing the final product yield comparison of the present invention. Detailed Implementation

[0033] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] Please see the appendix Figure 1 This invention provides an intelligent welding control system for a precast box girder frame, comprising: an alternating magnetic field module, a laser excitation module, an interference receiving module, a phase-locked extraction module, and a main control module.

[0035] An alternating magnetic field module is deployed in the peripheral area of ​​the welding torch. The alternating magnetic field module includes an array of orthogonal electromagnetic coils and an excitation current generator connected to the array of orthogonal electromagnetic coils.

[0036] The laser excitation module is integrated at the front end of the welding torch, and the optical path of the laser excitation module is aligned with the solidification zone of the base material surface behind the weld pool.

[0037] The interference receiving module and the laser excitation module are arranged adjacent to each other at the front end of the welding torch, with the optical receiving end of the interference receiving module aligned with the ultrasonic excitation area of ​​the laser excitation module.

[0038] The phase-locked extraction module establishes electrical connections with the interference receiving module and the alternating magnetic field module, respectively, and the main control module establishes bidirectional data communication connections with the alternating magnetic field module, the laser excitation module, and the phase-locked extraction module, respectively.

[0039] See attached document Figure 2 This invention provides an intelligent welding control method for precast box girder frames, comprising the following steps:

[0040] The main control module acquires the bevel size and base material properties of the node to be welded and sets the target weld depth data; the main control module sends an initialization control command to the alternating magnetic field module;

[0041] The excitation current generator of the alternating magnetic field module generates an alternating excitation current signal with a specific characteristic frequency. The alternating excitation current signal drives the orthogonal electromagnetic coil array to generate an alternating magnetic field in the welding pool area. The alternating magnetic field module synchronously sends the characteristic frequency to the phase-locked extraction module as a reference signal.

[0042] The alternating magnetic field and the welding current inside the molten pool are coupled to generate a volume Lorentz force. The alternating volume Lorentz force establishes a force field in the liquid metal, driving the solid-liquid interface at the bottom of the molten pool to generate mechanical micro-oscillations with the characteristic frequency output by the alternating magnetic field module.

[0043] The laser excitation module emits a pulsed laser beam onto the surface of the solidification zone at a set repetition frequency. The light energy of the pulsed laser beam excites broadband ultrasound on the surface of the base material through the thermoelastic effect. The broadband ultrasound propagates along the thickness direction of the base material and is reflected at the solid-liquid interface, which is in a state of mechanical micro-oscillation. According to the acoustic Doppler effect, the reflected ultrasound echo is physically modulated in acoustic phase by the mechanical micro-oscillation law of the solid-liquid interface.

[0044] The interference receiving module receives the transient surface displacement of the base material caused by the ultrasonic echo. The interference receiving module converts the transient surface displacement into a raw interference voltage signal containing frequency domain modulation characteristics and transmits the raw interference voltage signal to the phase-locked extraction module.

[0045] The phase-locked extraction module uses a reference signal to perform quadrature phase-locked demodulation and low-pass filtering on the original interference voltage signal. The phase-locked extraction module extracts the peak amplitude of the ultrasonic echo envelope from the processed signal and locks the flight time data corresponding to the peak amplitude.

[0046] The main control module receives flight time data and calculates the current actual melt depth by combining it with the system's built-in parameters. The main control module compares the actual melt depth with the target melt depth data to obtain the deviation value. Based on the status of the deviation value, the main control module outputs control parameter adjustment instructions to the alternating magnetic field module. The alternating magnetic field module receives the control parameter adjustment instructions and changes the amplitude and phase parameters of the alternating excitation current signal in real time.

[0047] The alternating magnetic field module has the physical ability to convert electrical signals into a spatial dynamic magnetic field. The upper-level functional features of the alternating magnetic field module are realized through lower-level physical components. Specifically, it includes a signal generation unit, a power amplification unit, and an orthogonal electromagnetic coil array. The signal generation unit and the power amplification unit together constitute the specific implementation structure of the excitation current generator.

[0048] The orthogonal electromagnetic coil array is fixedly installed on the outside of the protective gas cover around the welding torch. The orthogonal electromagnetic coil array consists of four independently wound electromagnetic excitation coils. The four electromagnetic excitation coils are evenly and symmetrically distributed in a ring on a horizontal projection plane perpendicular to the central axis of the welding torch. The first coil and the second coil are arranged opposite each other along the X-axis of the horizontal plane, and the third coil and the fourth coil are arranged opposite each other along the Y-axis of the horizontal plane, forming a spatially orthogonal physical geometric topology. The four electromagnetic excitation coils are wrapped with high-temperature resistant aluminum silicate ceramic heat insulation material to block the heat radiation of the welding arc to the copper windings of the coils.

[0049] The main control module stores a process parameter table for the precast box girder nodes. Based on the process parameter table, the main control module sets the working parameter instructions for the alternating magnetic field module. The working parameter instructions specifically include current peak amplitude data, characteristic frequency data, and initial phase data. The signal generation unit of the alternating magnetic field module receives the working parameter instructions and uses the internal direct digital frequency synthesizer to generate a low-voltage reference waveform signal with the corresponding parameters.

[0050] For the specific circuit design of the power amplifier unit, those skilled in the art can use a full-bridge inverter circuit or an H-bridge push-pull amplifier circuit to amplify the low-voltage reference waveform signal. The selection of switching transistors and the setting of driving logic are well-known technologies in this field and will not be elaborated here.

[0051] The power amplifier unit outputs the amplified alternating excitation current signal to both ends of the winding of the quadrature electromagnetic coil array. The alternating excitation current signal is a continuous analog electrical signal that changes periodically with time. The instantaneous current magnitude of the alternating excitation current signal at any time point is equal to the peak current amplitude multiplied by the sine of the overall phase angle. This overall phase angle is formed by adding the product of the constant twice pi, the characteristic frequency, and the time variable to the initial phase.

[0052] While generating a low-voltage reference waveform signal, the signal generation unit sends the square wave pulse signal corresponding to the characteristic frequency to the phase-locked extraction module through a hardware splitter via a synchronous communication bus. This square wave pulse signal serves as a reference signal to establish frequency locking synchronization between the magnetic field transmitter and the acoustic receiver.

[0053] After receiving the alternating excitation current signal, the orthogonal electromagnetic coil array generates an external alternating magnetic field that changes periodically in the physical space around the tip of the welding torch. The magnetic lines of force of the external alternating magnetic field penetrate the protective gas area and cover the entire molten metal pool and the heat-affected zone of the base material around it.

[0054] The welding arc acts directly on the bevel base material of the precast box girder node within the working area. The arc plasma introduces a high-density welding current into the base material. The introduced welding current forms a welding current density field that radiates from the surface to the bottom within the three-dimensional physical space of the molten metal pool. The external alternating magnetic field covering the molten pool area couples with the welding current density field inside the molten metal, continuously exciting and generating volumetric Lorentz force within the entire volume of the molten metal.

[0055] Based on the fundamental principles of electrohydrodynamics, the magnitude and direction of the force vector at any physical coordinate point within the liquid metal are equal to the cross product of the transient welding current density vector at that coordinate point and the transient external alternating magnetic field vector passing through that coordinate point. Since the alternating excitation current signal is in a continuous periodic alternating state, the external alternating magnetic field continuously changes its polarity and intensity according to its characteristic frequency, thus causing the volumetric Lorentz force inside the liquid metal to exhibit an alternating force field with the same characteristic frequency.

[0056] The alternating force field directly intervenes in and changes the fluid dynamics state of the liquid metal. The main control module sends control parameter adjustment commands containing new parameters to the alternating magnetic field module, thereby changing the initial phase and amplitude parameters of the alternating excitation current signal. Under the action of the control commands, the spatial distribution and time-averaged force direction of the alternating force field change accordingly.

[0057] When the system detects that the liquid metal is prone to collapse or burn-through due to the heat accumulation inside the box girder, the alternating magnetic field module adjusts the output phase of the excitation current generator. The adjusted phase parameter causes the transient external alternating magnetic field vector and the transient welding current density vector to be cross-multiplied. The resulting volumetric Lorentz force forms an upward-pointing time-averaged resultant force in the direction perpendicular to the surface of the base material. This upward resultant force forms an electromagnetic lifting effect inside the liquid metal, which is used to directly offset part of the downward load of the liquid metal under the gravitational field.

[0058] When the system detects a tendency for incomplete penetration at the root of the groove, the alternating magnetic field module synchronously adjusts the amplitude and phase parameters of the alternating excitation current signal. The increase in amplitude increases the magnitude of the transient external alternating magnetic field vector, and the change in phase strengthens the tangential component of the volumetric Lorentz force inside the molten pool. The increase in tangential force drives the liquid metal to overcome its own viscous resistance and generate a strong downward convection circulation. The fluid convection forces the ultra-high temperature liquid metal below the arc to the narrow groove root area, changing the energy transfer path that relies solely on heat conduction. The forced convection heat transfer mode increases the melting volume of the bottom base material.

[0059] For the finite element analysis and numerical simulation of the transient temperature field distribution and viscous fluid dynamic properties inside the weld pool, those skilled in the art can use existing multiphysics coupling calculation software to build the model. The specific three-dimensional mesh generation and thermodynamic boundary condition setting are well-known technologies in this field and will not be elaborated here.

[0060] The alternating magnetic field output by the alternating magnetic field module continuously excites the alternating volume Lorentz force inside the molten liquid metal pool, which directly acts on the solid-liquid interface at the bottom of the pool. The solid-liquid interface is the transition layer between the molten high-temperature liquid metal and the underlying solid substrate.

[0061] Under the periodic compression and stretching action of alternating volume Lorentz force, liquid metal generates periodic hydrodynamic pressure fluctuations in the boundary layer near the solid-liquid interface. These hydrodynamic pressure fluctuations overcome the viscous resistance of the liquid metal and the surface tension of the phase interface, forcing the entire solid-liquid interface to generate forced vibrations on the reference surface at the macroscopic melting depth position, establishing mechanical micro-oscillations at the physical level. The oscillation frequency of the mechanical micro-oscillations strictly follows the characteristic frequency of the alternating excitation current signal.

[0062] A mathematical model of micro-oscillation displacement at the solid-liquid interface is established to quantify this process. The model describes the transient position change at the solid-liquid interface as a superposition of depth evolution and microscopic high-frequency displacement. The specific formula for calculating the transient depth position at the solid-liquid interface is as follows:

[0063] ;

[0064] In the formula, Indicates at any time Transient depth position of the solid-liquid interface; It represents the macroscopic actual depth of the molten pool at the current moment, and characterizes the average physical depth of the solid-liquid interface. Its value shows a low-frequency and slow change with the macroscopic accumulation of welding heat input. This represents the amplitude of the mechanical micro-oscillations driven by the alternating volume Lorentz force; This indicates the characteristic frequency set and output by the alternating magnetic field module. This represents the phase difference in mechanical displacement response caused by the viscosity of the liquid metal and the damping effect of fluid dynamics; Indicates at any time The transient microscopic oscillation displacement of the solid-liquid interface relative to its actual macroscopic melting depth.

[0065] In the mathematical model of micro-oscillation displacement, the amplitude parameter of mechanical micro-oscillation is directly related to the amplitude of alternating excitation current signal. The physical magnitude of the amplitude is limited by the fluid damping coefficient of the liquid metal itself and the attenuation of dynamic mass. The phase difference parameter of mechanical displacement response reflects the time lag effect generated when the alternating physical force field acts on the viscous fluid interface. The main control module maintains the characteristic frequency constant through commands throughout the welding cycle, ensuring that alternating dynamic energy of the corresponding frequency is continuously injected into the liquid metal. This ensures that the mechanical micro-oscillation can exist stably and continuously in the strong arc noise environment, enabling the solid-liquid interface to have the physical conditions to act as an acoustic modulation source.

[0066] For the determination of the viscous damping coefficient of precast box girder base material of different materials in high temperature liquid environment, and the specific calculation method of metal surface tension coefficient, those skilled in the art can consult the relevant thermal property standard manual of metal materials for matching. The acquisition of its basic parameters and conventional algebraic calculation methods are well known technologies in this field, and will not be elaborated here.

[0067] The laser excitation module contains a nanosecond-level Q-switched pulsed solid-state laser and an optical collimating and focusing lens group. The main control module sends a trigger signal to the laser excitation module according to the timing requirements of the welding cycle. After receiving the trigger signal, the laser excitation module emits a nanosecond-level short pulse laser beam at a set repetition frequency.

[0068] The pulsed laser beam is transmitted and shaped by the optical collimating and focusing lens group. Its optical path is precisely directed to the surface of the base material in the solidification zone behind the tail of the weld pool. The center of the focused spot of the emitting end optical path maintains a set spatial physical distance from the edge of the weld pool. This spatial physical distance ensures that the base material in the laser irradiation area has completed the physical phase transition process from liquid to solid, while avoiding severe optical scattering interference of the welding arc plasma on the incident pulsed laser beam.

[0069] When the solidified substrate surface is irradiated by a pulsed laser beam, it rapidly absorbs light energy. Within an extremely short nanosecond timescale, the skin layer region of the substrate surface converts the absorbed light energy into heat energy, resulting in a large transient temperature gradient within the local material volume. The rapid increase in the transient temperature gradient triggers rapid physical thermal expansion of the surface material. The laser excitation module adjusts the single-pulse output energy and the focused spot area to strictly limit the laser energy density in the irradiated area below the damage and ablation threshold of the substrate, so that the surface energy conversion process is kept entirely within the non-destructive thermoelastic physical range.

[0070] The localized transient thermal expansion of the base material surface is constrained by the mechanical and physical constraints of the surrounding relatively cold base material region, exciting a transient thermoelastic stress field within the base material surface structure. As the energy of a single pulse laser rapidly decays and disappears, the accumulated thermoelastic stress field is released instantaneously. The stress field release process is transformed into outward-radiating mechanoelastic mechanical waves within the base material medium, thus completing the thermoelastic excitation physical mechanism of broadband ultrasound.

[0071] The excited broadband ultrasound includes Rayleigh surface waves propagating along the surface of the base material and acoustic volume waves radiating deep into the interior of the base material. Among them, the longitudinal wave component, which propagates downward perpendicular to the surface of the base material, serves as the main detection signal. It penetrates into the interior of the material along the thickness direction. Influenced by the heat input conduction process during welding, a nonlinear spatial thermodynamic temperature gradient field is distributed from the surface to the bottom of the base material. When the longitudinal wave passes through this variable temperature medium, its sound velocity and acoustic impedance characteristics change dynamically with the temperature distribution at the physical coordinate point. The longitudinal wave continues to penetrate downwards at a dynamic sound velocity until it reaches the solid-liquid interface below the molten metal pool, which is in a state of mechanical micro-oscillation.

[0072] For the calibration of laser energy absorption rate on different metal surfaces, and the specific modulation circuit topology of the Q-switching switch in the laser control module, those skilled in the art can use conventional laser ultrasonic non-destructive testing standards and specifications for parameter matching and selection. The combination of optical components and the basic driving principle are well-known technologies in this field and will not be elaborated here.

[0073] The longitudinal wave component of broadband ultrasound propagates continuously downward along the thickness direction of the base material and reaches the solid-liquid interface region at the bottom of the base material. There is a significant difference in acoustic impedance between the liquid metal and the solid base material in terms of physical density and acoustic propagation speed. When the longitudinal wave component passes through this interface with a sudden change in acoustic impedance, it undergoes physical reflection, generating an ultrasonic reflection echo that is folded back towards the upper surface of the base material.

[0074] At the moment when the longitudinal wave component of the ultrasound arrives and is reflected, the solid-liquid interface is in a state of mechanical micro-oscillation driven by alternating Lorentz force. Since the reflection boundary of the sound wave has a dynamic motion velocity and periodic physical displacement along the longitudinal wave propagation direction, the ultrasound reflected echo is directly physically interfered with by the Doppler effect during the generation process.

[0075] When an incident ultrasonic wave comes into contact with a dynamically moving solid-liquid interface, the instantaneous motion of the interface causes changes in the acoustic frequency and phase characteristics of the reflected echo. According to the classical Doppler effect, the instantaneous frequency shift of the reflected wave is directly proportional to the instantaneous velocity of the interface, specifically, twice the instantaneous velocity of the interface divided by the local medium sound velocity and then multiplied by the original frequency of the incident wave. The instantaneous velocity of the solid-liquid interface corresponds to the first derivative of its mechanical micro-oscillation displacement with respect to time.

[0076] The transient displacement of the interface strictly follows the forced oscillation law based on the characteristic frequency set by the alternating magnetic field module. The actual acoustic physical path length experienced by the ultrasonic wave when traveling to and from the interface changes periodically at the same frequency with the small displacement of the interface. The micro-dynamic change of the acoustic propagation path length is transformed into the dynamic phase shift of the ultrasonic reflected echo in the time domain.

[0077] After physical reflection through the mechanical micro-oscillation interface, the ultrasonic reflected echo is superimposed with the temporal pattern of micro-oscillation displacement in the phase dimension on the basis of its original broadband acoustic signal. The reflected echo derives frequency-shifted sideband characteristics corresponding to the characteristic frequency of the alternating magnetic field in the frequency domain. The applied alternating magnetic field mechanically modulates the bottom solid-liquid interface through the magnetohydrodynamic mechanism, injecting the characteristic frequency as a physical reference into the ultrasonic echo signal, thus completing the physical modulation process of the acoustic phase.

[0078] The ultrasonic reflected echo with frequency domain modulation characteristics detaches from the solid-liquid interface, reverses along the original propagation path and crosses the parent material medium with a nonlinear temperature gradient field. The ultrasonic reflected echo eventually reaches the upper surface region of the solidified parent material and induces mechanical vibration of microparticles with corresponding acoustic frequencies and phases in the surface material.

[0079] For the derivation of the conventional physical equations for the frequency and phase shift of sound waves caused by the Doppler effect, as well as the basic algebraic calculation of the reflection and transmission coefficients on the cross section of acoustic impedance difference, those skilled in the art can refer to the basic acoustic physics theory. The derivation and calculation methods are well-known technologies in this field and will not be elaborated here.

[0080] The interference receiving module specifically includes a continuous wave probe laser, an optical beam splitter interferometer group, and a high-frequency photodetector to achieve non-contact optical measurement of minute displacements on the surface of the substrate.

[0081] A continuous-wave probe laser outputs a stable single-frequency probe laser beam. This probe laser, after passing through an optical beam splitter interferometer array, is focused onto the physical region where the ultrasonic reflection echo from the substrate surface arrives. The ultrasonic reflection echo, with frequency-domain modulation characteristics, induces transient mechanical vibrations along the thickness direction in the surface particles of the substrate upon reaching the upper surface. These transient mechanical vibrations cause transient surface displacements in the microscopic geometric positions of the substrate surface over time.

[0082] When the probe laser illuminates the surface of the base material undergoing transient surface displacement, it generates reflected light. The optical phase of the reflected light dynamically shifts in sync with the transient surface displacement. The reflected light, carrying this optical phase shift, returns along its original path to the optical beam-splitting interferometer assembly, where it physically interferes with the reference beam retained within the system. This interference converts the optical phase shift, which is difficult to extract directly, into a dynamic change in light intensity. A high-frequency photodetector receives the interfered beam and converts the time-series change in light intensity into the original interference voltage signal. .

[0083] The welding work area is in an extreme physical environment, and the original interference voltage signal output by the high-frequency photodetector is coupled with complex broadband interference components. The system establishes a component model of the original interference voltage signal for quantitative analysis, and the specific mathematical combination formula for the signal is as follows:

[0084] ;

[0085] In the formula, This represents the original interference voltage signal output by the interference receiving module to the downstream circuit. This represents a useful modulated ultrasonic echo signal component that carries a characteristic frequency injected by an alternating magnetic field in the acoustic phase, and whose signal envelope contains actual ultrasonic time-of-flight information, with an amplitude on the order of extremely weak levels. This represents the high-intensity acoustic-optical coupling interference component generated by the welding arc plasma. This component includes broadband acoustic emission signals caused by the high-frequency mechanical impact of the arc force on the molten pool region, and photoelectric shot noise caused by the penetration of optical devices by the intense arc light that is not completely filtered out; This represents the low-frequency thermoelastic background noise component caused by the thermal expansion of the microcrystalline lattice inside the base material under the action of the high-temperature thermal field gradient during welding. This represents the inherent broadband system thermal noise component of the photoelectric conversion components and front-end amplifier circuit inside the interference receiving module during operation.

[0086] The interference components cover an extremely wide frequency band, and their overall power spectral density is much higher than that of the modulated ultrasonic echo signal components in the ultrasonic operating frequency band. The interference components and the modulated ultrasonic echo signal components severely overlap in both the time and frequency domains. Under extremely low signal-to-noise ratio conditions, conventional bandpass filtering techniques cannot separate the modulated ultrasonic echo signal components from the strong background noise without destroying the echo envelope characteristics. This complex physical interference condition constitutes the prerequisite input condition for the downstream phase-locked extraction module to execute the orthogonal phase-locked demodulation algorithm.

[0087] For the frequency stabilization control mechanism of continuous wave probe lasers, the physical matching method of optical path difference between reference light and measurement light in interference optical path, and the selection rules of response frequency band of avalanche photodiode in high frequency photodetectors, those skilled in the art can refer to the standard design manual of existing non-contact laser vibrometers to build the system. Its basic photoelectric conversion principle and internal optical lens arrangement structure are well known technologies in this field and will not be elaborated here.

[0088] The front-end hardware circuit of the phase-locked loop extraction module receives the raw interference voltage signal transmitted by the interference receiving module and simultaneously receives the reference signal sent by the alternating magnetic field module. The reference signal is a square wave pulse with a characteristic frequency.

[0089] The phase-locked loop extraction module is equipped with waveform conversion and precision phase shifting circuits. These circuits convert the square wave pulse state reference signal into a sine wave and generate two independent reference signals. Both reference signals have a physical frequency equal to the characteristic frequency and maintain a 90-degree orthogonal relationship in the phase dimension. The reference signal with zero phase offset is defined as the in-phase reference signal, and the reference signal with a 90-degree phase offset is defined as the quadrature reference signal.

[0090] After the original interference voltage signal enters the phase-locked loop (PLL) extraction module, it is divided into two branch signals with identical physical characteristics by a passive power divider. The PLL extraction module integrates two parallel dual-balanced analog multipliers. The first dual-balanced analog multiplier receives the first branch signal and the in-phase reference signal, performs continuous analog voltage multiplication on the two in the time domain, and outputs an in-phase mixed signal. The second dual-balanced analog multiplier receives the second branch signal and the quadrature reference signal, performs the same multiplication operation in the time domain, and outputs a quadrature mixed signal.

[0091] The system establishes a mathematical expression for the quadrature mixing mechanism to describe the signal processing at the hardware level. The specific calculation formulas for in-phase and quadrature mixed signals are as follows:

[0092] ;

[0093] ;

[0094] In the formula, This represents the transient in-phase mixer signal output from the first dual-balanced analog multiplier. This represents the transient quadrature mixing signal output from the second dual-balanced analog multiplier; This represents the fixed voltage amplitude of the sinusoidal reference signal output by the waveform conversion and precision phase-shifting circuit. This represents the in-phase reference signal output by the waveform conversion and precision phase-shifting circuit. This represents the quadrature reference signal output by the waveform conversion and precision phase shift circuit.

[0095] The analog signal multiplication operation in the time domain corresponds to the spectral convolution and translation in the frequency domain in terms of physical mechanism. The useful modulated ultrasonic echo signal component carrying the characteristic frequency in the original interference voltage signal, after being multiplied with the reference signal of the same frequency, generates a high-frequency component with twice the characteristic frequency and a DC low-frequency component with a frequency difference of zero, according to the physical law of trigonometric function product sum and difference.

[0096] This physical operation directly shifts the high-frequency modulation band of the target signal to the zero-frequency baseband region. Various random broadband interference components coupled to the original interference voltage signal, lacking fixed phase consistency and frequency correlation with the reference signal, are randomly scattered to different high-frequency bands after mixing, failing to form a stable DC component accumulation in the zero-frequency region. This hardware-level quadrature mixing operation achieves effective separation of the target signal from strong background noise in the frequency domain through the physical path.

[0097] For the transistor-level circuit topology of the Gilbert unit inside the dual-balanced analog multiplier, and the calculation method of the microstrip line impedance matching of the passive power divider, those skilled in the art can refer to the basic design specifications of microwave RF circuits for hardware routing. The basic working principle of the circuit and the selection of components are well-known technologies in this field, and will not be elaborated here.

[0098] The phase-locked loop extraction module is internally configured with two parallel matched low-pass filter circuits. The in-phase mixing signal output from the first dual-balanced analog multiplier is input to the first low-pass filter, and the quadrature mixing signal output from the second dual-balanced analog multiplier is input to the second low-pass filter.

[0099] The cutoff frequency of the low-pass filter is set to be much lower than the characteristic frequency of the alternating magnetic field. The low-pass filter performs physical filtering on the input mixing signal, filtering out high-frequency components at twice the characteristic frequency and random broadband noise that is scattered by spectrum shifting, and retaining only the zero-frequency baseband slow-changing signal generated by Doppler phase modulation and demodulation. This physical process outputs the filtered in-phase baseband signal and quadrature baseband signal.

[0100] The integrated digital signal processor (DSP) within the phase-locked loop (PLL) extraction module performs high-speed analog-to-digital conversion and mathematical operations on the filtered in-phase and quadrature baseband signals. The DSP executes an envelope amplitude calculation algorithm, completely eliminating the physical influence of the initial random phase of the ultrasonic echo on signal amplitude extraction through the square root operation of the sum of squares of the orthogonal vectors. The specific calculation formula for the transient echo envelope amplitude extracted by the DSP is as follows:

[0101] ;

[0102] In the formula, This represents the transient echo envelope amplitude extracted by the phase-locked loop extraction module at any time tt; It represents the square of the in-phase baseband signal at any time t, and in a physical sense, it characterizes the transient power level of the in-phase component; It represents the square of the orthogonal baseband signal at any time t, and in a physical sense, it characterizes the transient power level of the orthogonal components.

[0103] As the time variable advances linearly, the digital signal processor continuously records the specific magnitude of the transient echo envelope amplitude in the time domain, forming a time-series envelope curve that includes the ultrasonic echo energy intensity.

[0104] The digital signal processor has a peak optimization logic running inside. Within the set effective detection time window, it compares the envelope amplitude of the current continuous sampling points with the historical maximum value in the register in real time. When it detects that the current amplitude shows a local extreme value evolution characteristic of first rising and then falling, and the voltage amplitude corresponding to the local extreme value exceeds the set system physical noise background threshold, the digital signal processor locks the physical extreme value point as the peak value of the real ultrasonic reflected echo.

[0105] At the instant the main control module sends a trigger electrical signal to the laser excitation module, the timer inside the synchronous phase-locked extraction module marks the emission time of the pulsed laser beam as the zero point. The digital signal processor uses the internal high-frequency hardware clock counter to calculate the time difference between the zero point and the moment when the amplitude peak appears. This time difference is the flight time data of the ultrasonic wave completing the round-trip propagation inside the parent material.

[0106] Time-of-flight data directly characterizes, at the physical level, the actual absolute time taken for an ultrasonic wave to complete a single spatial physical path traversal from the surface of the substrate to the solid-liquid interface and back to the surface of the substrate under the current dynamic temperature field medium. Obtaining this time parameter enables the system to accurately extract effective acoustic features characterizing the actual melting depth from a complex, high-noise environment.

[0107] For the active Butterworth analog circuit topology design of low-pass filters and the sampling timing configuration method of high-speed analog-to-digital converters inside digital signal processors, those skilled in the art can refer to the digital signal processing hardware standard manual to build the system. The basic analog filtering principle and digital sampling discretization rule are well-known technologies in this field and will not be elaborated here.

[0108] After the phase-locked extraction module obtains the flight time data of the ultrasonic wave to complete the round-trip propagation, the system enters the dynamic actual penetration depth calculation stage. During the welding process, due to the high-density arc heat input, there is a nonlinear heat conduction phenomenon from the upper surface to the bottom solid-liquid phase interface inside the base material, which establishes a non-uniform spatial temperature gradient field in the physical thickness direction.

[0109] The physical propagation speed of ultrasonic longitudinal waves within a metallic medium is highly sensitive to the local absolute temperature of the medium itself. Increased temperature leads to thermal expansion of the crystal lattice and physical attenuation of the elastic modulus within the metallic material. Consequently, the local physical sound velocity of the ultrasonic longitudinal wave decreases nonlinearly with increasing temperature. The time-of-flight data is coupled with the nonlinear acoustic delay effect accumulated when the ultrasonic wave passes through this variable-temperature medium. Directly applying fixed sound velocity parameters at room temperature for simple algebraic ranging will introduce significant physical errors. Variable-temperature medium sound velocity compensation must be performed to restore the true physical depth spatial position.

[0110] The main control module contains a transient heat conduction calculation unit. By combining macroscopic heat input parameters such as current welding current, arc voltage, and welding speed, it reconstructs a one-dimensional spatial temperature distribution profile along the thickness of the base material in real time. This profile quantitatively describes the local transient physical temperature at any depth. Simultaneously, it retrieves pre-stored sound velocity-temperature coupling curve functions for the corresponding metal material to establish a sound velocity distribution model in a spatial physical coordinate system.

[0111] Based on the acquired flight time data and sound speed distribution model, the main control module constructs an integral equation for dynamically calculating the actual melting depth. This integral equation mathematically maps the acoustic propagation time to the physical spatial path; the specific integral calculation formula is as follows:

[0112] ;

[0113] In the formula, This represents the actual flight time data of the ultrasonic reflected echo calculated and output by the phase-locked loop extraction module. This represents the result obtained by reconstructing the physical coordinates using a transient heat conduction calculation unit. The transient physical temperature at that location; Indicates the transient physical temperature The transient acoustic propagation speed of ultrasonic longitudinal waves in this local microscopic physical medium under the condition; It represents the definite integral interval from the initial coordinate zero point of the upper surface of the parent material to the macroscopic actual melt depth coordinate of the solid-liquid interface in the spatial physical coordinate system. Its physical meaning is the spatial continuous accumulation process over the entire unidirectional propagation path. This represents a tiny physical space thickness differential unit divided along the thickness direction of the base material; It represents the unidirectional physical cumulative time taken for the ultrasonic longitudinal wave to propagate from the upper surface of the parent material to the solid-liquid interface.

[0114] Since the target variable, the actual macroscopic melting depth, exists as the upper limit of integration in the nonlinear integral equation, the microprocessor of the main control module obtains its numerical solution by executing a numerical discretization approximation and iterative solution algorithm. The microprocessor sets an initial depth iteration value and accumulates the time differential components stepwise along the spatial physical coordinate variables. The iteration process terminates when the accumulated total computation time and half of the flight time data reach the set approximation tolerance requirement.

[0115] The accumulated spatial coordinates at this point are locked by the system as the dynamic actual melt depth after compensation for the sound velocity in the variable-temperature medium. This value eliminates the dynamic interference of the non-uniform thermodynamic physical field on the sound wave propagation speed, accurately restoring the absolute physical coordinates of the interface between the high-temperature liquid metal and the underlying solid substrate. The main control module uses this dynamic actual melt depth as a feedback variable to participate in the closed-loop control algorithm calculation of the alternating magnetic field module, realizing physical intervention in the welding quality of the precast box girder nodes.

[0116] For the acoustic physical calibration experimental method of the elastic modulus of different metallic materials deteriorating with temperature, and the basic code architecture of the Newton-Raphson nonlinear numerical iterative approximation algorithm executed by the microprocessor, those skilled in the art can refer to the materials physics handbook and conventional numerical calculation and analysis theory for adaptation. The acquisition of basic constants and the setting of iterative convergence conditions are well-known technologies in this field and will not be elaborated here.

[0117] The main control module continuously receives dynamic actual weld depth data calculated and output by the phase-locked extraction module. In the continuous welding operation of precast box girders, the system needs to monitor the physical boundary extending from the bottom of the molten pool to the bottom surface of the base material in real time to prevent structural defects such as incomplete penetration or excessive burn-through. The main control module stores the initial design physical thickness parameters of the base material to be welded.

[0118] The microprocessor inside the main control module extracts the current dynamic actual melt depth and performs a geometric difference calculation with the initial design physical thickness of the base material to obtain the current remaining unfused physical thickness. This value, at a physical level, characterizes the absolute safety margin between the solid-liquid interface and the bottom of the base material. The specific geometric difference calculation formula is as follows:

[0119] ;

[0120] In the formula, Indicates at any time The calculated remaining unfused physical thickness; This indicates the pre-set initial design physical thickness of the base material; This represents the dynamic actual melting depth obtained at the current moment after sound speed compensation calculation.

[0121] In the continuous welding process of large thick plate structures, relying solely on the geometric thickness allowance cannot accurately determine the actual penetration risk. The welding heat input generates a strong and complex heat accumulation effect inside the material. The solid base material at the bottom of the molten pool is subjected to the heat conduction effect of the high-temperature liquid metal above for a long time, and its local absolute temperature continues to rise. The high temperature state causes the mechanical yield strength of the bottom base material to physically decrease, and its ability to bear the gravity of the liquid metal above and the mechanical impact force of the electric arc is significantly reduced.

[0122] To quantify the above physical process, the main control module establishes a heat accumulation characteristic evaluation model, calculates the effective heat input accumulation within a specific time range, and the microprocessor acquires real-time welding electrical parameters and performs numerical integration within a set sliding time window. The specific calculation formula for the transient heat accumulation index is as follows:

[0123] ;

[0124] In the formula, Indicates at any time The resulting transient heat accumulation index characterizes the cumulative heat load of a local area; In the time dimension, from the past From time to any time The interval of definite integrals; This represents the arc thermal efficiency coefficient under the current welding process conditions; This represents the transient arc voltage at the moment corresponding to the integral variable; This represents the transient welding current at the moment corresponding to the integral variable; This represents a tiny unit of time integration.

[0125] After obtaining the remaining unfused physical thickness and transient heat accumulation index, the logic judgment unit configured inside the microprocessor executes a multi-level state evaluation algorithm. The microprocessor uses the transient heat accumulation index to dynamically compensate the preset basic safety thickness threshold, generating a dynamic critical threshold that changes in real time with the degree of heat accumulation. As the local accumulated heat load increases, the bearing capacity of the underlying base material weakens, and the system automatically increases the allowable minimum safety thickness margin.

[0126] The logic judgment unit compares the real-time remaining unfused physical thickness with the dynamic critical threshold. When the remaining unfused physical thickness is greater than the dynamic critical threshold, the system determines that it is currently in a safe fusion state and the solid-liquid interface has sufficient physical support strength. When the remaining unfused physical thickness is equal to or slightly lower than the dynamic critical threshold, the system determines that it has entered a critical full-penetration state and the underlying base material is close to the edge of thermodynamic instability. If the remaining unfused physical thickness continues to decrease sharply and exceeds the set absolute tolerance lower limit, the system determines that there is a risk of burn-through and the physical intervention mechanism must be triggered immediately.

[0127] The dynamic evaluation results of the above penetration state are stored in the internal register of the main control module in the form of status flag bits, serving as the direct logical prerequisite for the subsequent closed-loop control module to adjust the welding energy output.

[0128] For the material mechanics calibration method for the nonlinear decay of the high-temperature yield strength of the base material, and the register data shifting and accumulation algorithm for the sliding window numerical integration of the microprocessor, those skilled in the art can refer to the material engineering specifications and the basic logic of digital signal processing to build it. The correspondence of its physical parameters and the basic software architecture are well-known technologies in this field and will not be elaborated here.

[0129] The main control module initiates closed-loop control logic based on the melt penetration state assessment results stored in the register. When the system determines that it is currently in a critical full melt penetration state or a burn-through risk state, the microprocessor sends a dynamic adjustment command to the alternating magnetic field module. After receiving the command, the alternating magnetic field module generates a directional electromagnetic force inside the molten pool by adjusting the physical parameters of the applied alternating magnetic field, thereby physically intervening in the high-temperature liquid metal.

[0130] The alternating magnetic field and the welding current flowing downwards into the molten pool are physically orthogonal in space, generating a Lorentz force within the liquid metal according to electromagnetic principles. This Lorentz force acts directly on the liquid metal particles, altering the fluid dynamics distribution within the molten pool. The main control module dynamically adjusts the amplitude and phase of the output excitation current by controlling the inverter drive circuit within the alternating magnetic field module, thereby changing the amplitude of the magnetic flux density of the applied magnetic field and the magnetic field phase angle between it and the welding current.

[0131] The system establishes a mathematical model for the time-averaged Lorentz force density, quantifying the physical influence of magnetic field phase angle and amplitude on macroscopic electromagnetic forces. The specific calculation formulas are as follows:

[0132] ;

[0133] In the formula, This represents the time-averaged Lorentz force volume density experienced by the liquid metal during the alternating cycle. This indicates the amplitude of the welding current density flowing into the molten pool region; This represents the amplitude of the applied magnetic flux density from the alternating magnetic field module to the molten pool region. The cosine value representing the phase angle difference of the magnetic field, in terms of physical mechanism, characterizes the degree of synchronous coupling between the alternating magnetic field and the welding current in the time phase. Its magnitude directly determines the proportion of the effective component of the time-averaged Lorentz force in a specific direction.

[0134] The main control module executes a dynamic optimization adjustment algorithm. The microprocessor uses the real-time acquired remaining unfused physical thickness as feedback input and calculates the deviation from the set dynamic critical threshold. Based on the magnitude of the deviation, the microprocessor prioritizes adjusting the magnetic field phase angle difference. System control Approaching zero degrees or 180 degrees maximizes the absolute value of the cosine function, thus aligning the direction of the Lorentz force vertically upwards. This vertically upward time-averaged Lorentz force acts as an electromagnetic lifting force, physically counteracting the gravity of the liquid metal and the downward mechanical impact force generated by the arc plasma, thereby mitigating the downward pressure and thermal penetration of the high-temperature liquid metal onto the underlying solid substrate.

[0135] Based on phase angle optimization, if the rate of deviation reduction does not reach the set convergence criterion, the microprocessor synchronously increases the output voltage of the inverter drive circuit and increases the amplitude of the applied magnetic flux density. The amplitude of the electromagnetic lifting force is linearly amplified, further strengthening the mechanical support effect on the bottom of the molten pool. As the lifting force increases, the downward physical displacement of the liquid metal is restricted, and the expansion process of the dynamic actual melting depth is forcibly slowed down.

[0136] When the main control module detects that the remaining unfused physical thickness has rebounded and stabilized above the dynamic critical threshold, the microprocessor determines that the molten pool has escaped the risk of burn-through. The system then reverses this process by reducing the magnetic flux density amplitude and restoring the initial magnetic field phase angle setting, keeping the penetration depth within a safe fluctuation range, thus completing the closed-loop physical control of the entire process.

[0137] For the pulse width modulation switching timing control method of the insulated gate bipolar transistor in the inverter drive circuit, and the hardware compensation circuit for the current response delay of the excitation coil inductance, those skilled in the art can refer to the design specifications of high-power power electronic converters. The basic inverter topology and duty cycle calculation principles are well-known technologies in this field and will not be elaborated here.

[0138] Specific application examples:

[0139] Example 1: Intelligent welding operation for full penetration of thick plates in precast box girders of a high-speed railway super bridge;

[0140] Scenario Deployment:

[0141] This embodiment is applied to a large bridge steel structure manufacturing plant, focusing on full penetration butt welding of the load-bearing nodes of precast box girders for a cross-river high-speed railway bridge. The base material to be welded is Q345qD bridge-specific steel, with an initial design physical thickness of 20mm. The target actual penetration depth is controlled at around 17mm, with a safety margin of approximately 3mm to prevent burn-through.

[0142] Hardware configuration:

[0143] Welding station: 1 heavy-duty gantry-type automatic MAG welding robot.

[0144] Alternating magnetic field module: An array of orthogonal electromagnetic coils deployed outside the welding torch protective cover, connected to a full-bridge inverter excitation current generator.

[0145] Laser excitation and interference receiving module: integrated at the front end of the welding torch, it includes a 1064nm nanosecond-level Q-switched pulsed solid-state laser and a high-frequency continuous wave optical interference receiver.

[0146] Computing hub: The main control module equipped with a digital signal processor and a multiphysics transient heat conduction calculation unit.

[0147] Production flow and data interaction process:

[0148] Step 1: Ultrasonic excitation and phase-locked demodulation locking;

[0149] Action: Welding initiated (current 350A). The alternating magnetic field module applies a 40kHz alternating magnetic field to the molten pool, driving the solid-liquid interface to generate synchronous mechanical micro-oscillations. The system establishes a mathematical model of the micro-oscillation displacement at the solid-liquid interface: Simultaneously, the laser excitation module emits pulsed laser light behind the molten pool to excite broadband ultrasonic waves.

[0150] Sensing: The interferometric receiving module captures transient surface displacements and outputs a raw interferometric voltage signal mixed with strong arc-induced acousto-optic noise. The composition model of this signal is as follows: .

[0151] Demodulation and Lock-in: The phase-locked loop (PLL) extraction module utilizes a 40kHz reference signal, performing quadrature mixing and low-pass filtering via a double-balanced analog multiplier. The in-phase and quadrature mixed signals are calculated as follows: , .

[0152] Judgment: The system successfully extracted the ultrasonic echo envelope from strong background noise, and extracted the transient echo envelope amplitude by taking the square root of the sum of squares of orthogonal vectors. The system then uses optimization logic to pinpoint the peak amplitude, accurately acquiring ultrasonic time-of-flight data.

[0153] Step 2: Sound velocity compensation and dynamic melt depth calculation;

[0154] Mapping: The main control module retrieves the thermophysical data of the base material, reconstructs a one-dimensional temperature distribution profile based on the current welding electrical parameters, and establishes a corresponding sound velocity distribution model.

[0155] Calculation: The system performs numerical iterations based on time-of-flight data and a nonlinear sound velocity distribution model to construct an integral equation for dynamically calculating the actual melting depth, thus eliminating the sound velocity attenuation error caused by high temperature. .

[0156] Output: The current high-precision dynamic actual weld depth is calculated to be 16.5mm, and the remaining unfused physical thickness is obtained by performing geometric difference calculation. The formula is then used to... The calculated remaining unfused physical thickness is: .

[0157] Step 3: Assessment of thermal accumulation and attribution of electromagnetic intervention;

[0158] State evolution: As welding continues, the temperature of the base material at the bottom rises sharply, and the yield strength decreases. The main control module performs numerical integration through a sliding window to calculate the transient heat accumulation index: Based on this index, the system dynamically adjusts the basic safety thickness threshold from 2.5mm to 3.2mm (dynamic critical threshold).

[0159] Attribution and Action: When the system detects that the remaining unfused physical thickness has dropped to 3.0 mm (less than the dynamic critical threshold of 3.2 mm), it determines that the underlying base material is at risk of thermodynamic instability and collapse (burn-through).

[0160] Closed-loop control: The main control module immediately sends a command to the alternating magnetic field module to adjust the magnetic field phase angle difference to approach 0 degrees and increase the excitation current amplitude. The alternating magnetic field and welding current are orthogonally coupled, and the mathematical model of their time-averaged Lorentz force density is as follows: .when When it approaches 0, By maximizing the instantaneous generation of a vertically upward time-averaged Lorentz force (electromagnetic lifting force) within the molten pool, the gravity of the liquid metal is forcibly counteracted, and the expansion of the melt depth is successfully contained and stabilized at the 17mm safety line.

[0161] Experimental verification and effect comparison:

[0162] To verify the actual effectiveness of the system, a comparative experiment was conducted in the aforementioned steel structure workshop over a period of 30 days, involving a total of 1,200 thick plate main welds.

[0163] Control group: Traditional MAG automated welding equipment was used, without alternating magnetic field interference or ultrasonic in-situ detection feedback. Welding parameter adjustments relied entirely on the experience-based presets of the process engineer, and heat accumulation depended entirely on natural heat dissipation.

[0164] Experimental group: The intelligent welding control system of the present invention was activated, and orthogonal phase-locked extraction feedback and electromagnetic lifting closed-loop control were enabled.

[0165] Experimental data presentation:

[0166] See attached document Figure 3 Chart description: The horizontal axis represents the time window (microseconds), and the vertical axis represents the voltage / envelope amplitude.

[0167] Data Interpretation: The light gray, cluttered waveform in the figure represents the raw interference voltage signal captured by the interferometric receiving module. Its signal-to-noise ratio is low, and the target echo is completely submerged in strong arc noise. The smooth black solid line represents the echo envelope after demodulation by the experimental group's quadrature phase-locked loop extraction module. The black markers with circles represent the peak values ​​(time-of-flight data) automatically locked by the system.

[0168] In contrast, traditional ultrasonic testing is completely ineffective when the electric arc is activated (blind zone); while this system achieves effective feature extraction in extreme noise environments through Doppler frequency shift feature modulation and demodulation mechanism.

[0169] See attached document Figure 4 Chart description: The horizontal axis represents welding advancement time (seconds), the left vertical axis represents physical space depth (mm), and the right vertical axis represents electromagnetic lifting force (N).

[0170] Data Interpretation:

[0171] Black dotted line (control group): As heat accumulation intensifies, the molten pool descends uncontrollably, breaking through the 20mm baseline at 60 seconds, resulting in a severe burn-through defect.

[0172] Black solid line (experimental group): When the melting depth reaches around 17mm (reaching the dynamic critical threshold boundary), the dark gray solid line (representing the electromagnetic lifting force) rises rapidly. Under the physical intervention of the lifting force, the trajectory of the black solid line tends to stabilize at the optimal target depth of 17mm.

[0173] In contrast, the experimental group directly intervened in the fluid dynamics state through electromagnetic mechanics, solving the industry pain point that the later stage of continuous welding of thick plates is prone to collapse and burn-through.

[0174] See attached document Figure 5 Chart description: The horizontal axis represents the number of experimental days (day 1 to day 30), and the vertical axis represents the final product yield.

[0175] Data Interpretation: Black solid line with asterisk (experimental group): The yield rate reached over 98% on the first day of use. With the adaptive fine-tuning of system control parameters, the yield rate stabilized at an extremely high level of 99.5%.

[0176] Gray area with dotted squares (control group): The yield rate is affected by ambient temperature, slight differences between batches of boards and human operation conditions, and fluctuates greatly between 85% and 92%.

[0177] Table 1. Summary of Effect Comparison

[0178] Conclusion: Experimental results show that this invention successfully extracts accurate weld depth data from extremely harsh arc noise through laser-ultrasound combined with orthogonal phase-locked loop technology; the problem of data mapping distortion is solved based on heat accumulation assessment and sound velocity compensation. Most importantly, the directional Lorentz force generated by the alternating magnetic field precisely intervenes in the molten pool fluid. The closed-loop control system constructed based on this can quickly and intelligently mitigate the risk of burn-through, significantly improving the intelligence level and overall economic benefits of welding and manufacturing extra-thick precast box girders.

Claims

1. An intelligent welding control system for precast box girder frames, characterized in that, include: An alternating magnetic field module, deployed in the peripheral area of ​​the welding torch, includes an array of orthogonal electromagnetic coils and an excitation current generator connected to the array of orthogonal electromagnetic coils, used to generate an alternating magnetic field in the weld pool area and output a reference signal with a corresponding characteristic frequency. The laser excitation module is integrated at the front end of the welding torch. The optical path of the emitting end is aligned with the surface of the base material in the solidification zone to emit a pulsed laser beam and generate broadband ultrasonic waves. An interference receiving module is arranged adjacent to the laser excitation module, with its optical receiving end aligned with the ultrasonic excitation region of the laser excitation module. It is used to convert the transient surface displacement caused by the ultrasonic echo generated after the broadband ultrasonic wave is reflected into the original interference voltage signal. The phase-locked extraction module receives the reference reference signal and uses the reference reference signal to perform orthogonal phase-locked demodulation processing on the original interference voltage signal, thereby locking and extracting the corresponding time-of-flight data. The phase-locked extraction module includes waveform conversion and phase-shifting circuits and two dual-balanced analog multipliers. The waveform conversion and phase shifting circuit converts the reference reference signal into an in-phase reference signal and a quadrature reference signal; The two-channel dual-balanced analog multipliers multiply the original interference voltage signal with the in-phase reference signal and the quadrature reference signal respectively, and output the in-phase mixer signal and the quadrature mixer signal. The main control module calculates the current actual melting depth based on the flight time data and built-in parameters, compares the actual melting depth with the target melting depth data to obtain the deviation value, and outputs control parameter adjustment instructions to the alternating magnetic field module. The built-in parameters include the thermophysical data of the base material; The main control module runs a transient heat conduction calculation unit, which reconstructs a one-dimensional spatial temperature distribution profile and establishes a sound velocity distribution model based on the thermophysical data of the parent material. The main control module constructs an integral equation based on the flight time data and the sound speed distribution model, and executes a numerical discretization approximation and iterative solution algorithm to calculate the actual melting depth. The built-in parameters include the total thickness of the base material and the basic safety thickness threshold. The main control module calculates the remaining unfused physical thickness based on the difference between the total thickness of the base material and the actual melt depth, and calculates the transient heat accumulation index. The transient heat accumulation index is then used to correct the basic safety thickness threshold to obtain the dynamic critical threshold. When the remaining unfused physical thickness is less than the dynamic critical threshold, the main control module sends the control parameter adjustment command to the alternating magnetic field module based on the deviation value, so as to adjust the magnetic flux density amplitude and magnetic field phase angle of the alternating magnetic field, and generate a time-averaged Lorentz force as an upward electromagnetic lifting force within the weld pool region.

2. The intelligent welding control system for a precast box girder frame according to claim 1, characterized in that, The orthogonal electromagnetic coil array consists of four independently wound electromagnetic excitation coils, which are uniformly and symmetrically distributed in a ring and arranged perpendicularly to each other in pairs. The excitation current generator includes a signal generation unit and a power amplification unit. The signal generation unit is used to receive operating parameter commands, generate a low-voltage reference waveform signal, and synchronously output the reference reference signal. The power amplification unit is used to amplify the low-voltage reference waveform signal and output an alternating excitation current signal to the quadrature electromagnetic coil array.

3. The intelligent welding control system for a precast box girder frame according to claim 1, characterized in that, The alternating magnetic field couples with the welding current inside the weld pool region to generate a volumetric Lorentz force. The volumetric Lorentz force drives the solid-liquid interface at the bottom of the weld pool region to generate mechanical micro-oscillations corresponding to the characteristic frequency.

4. The intelligent welding control system for a precast box girder frame according to claim 3, characterized in that, The laser excitation module includes a nanosecond-level Q-switched pulsed solid-state laser and an optical collimating and focusing lens group, wherein the pulsed laser beam excites the broadband ultrasound. The broadband ultrasonic wave is reflected at the solid-liquid interface in the state of mechanical micro-oscillation, generating an ultrasonic echo carrying phase information corresponding to the mechanical micro-oscillation pattern.

5. The intelligent welding control system for a precast box girder frame according to claim 4, characterized in that, The interference receiving module includes a continuous wave detection laser, an optical beam splitter interferometer group, and a high-frequency photodetector. The continuous wave probe laser outputs a probe laser to irradiate the surface of the base material that generates the transient surface displacement, and obtains reflected light carrying the optical phase shift corresponding to the transient surface displacement; The reflected light enters the optical beam splitter interferometer group and interferes with the reference beam, and is converted into the original interference voltage signal by the high-frequency photodetector.

6. The intelligent welding control system for a precast box girder frame according to claim 5, characterized in that, The original interference voltage signal is composed of modulated ultrasonic echo signal components and broadband interference components; The broadband interference components include high-intensity acoustic-optical coupling interference components, thermo-elastic background noise components, and system thermal noise components. The modulated ultrasonic echo signal component carries frequency domain modulation features corresponding to the characteristic frequency.

7. The intelligent welding control system for a precast box girder frame according to claim 1, characterized in that, The phase-locked extraction module also includes a low-pass filter circuit and a digital signal processor; The in-phase mixer signal and the quadrature mixer signal are processed by the low-pass filter circuit into an in-phase baseband signal and a quadrature baseband signal; The digital signal processor performs an envelope amplitude calculation algorithm on the in-phase baseband signal and the quadrature baseband signal and extracts the transient echo envelope amplitude. It then uses peak optimization logic to lock the time-of-flight data corresponding to the peak amplitude of the transient echo envelope.

Images