A real-time control method and system for welding deformation of a large coil box of a fusion reactor

Abstract

The application discloses a kind of real-time control method and system of fusion reactor large coil box welding deformation, comprising the following steps: establishing welding deformation prediction model;Arrange multi-channel sensor array, and collect monitoring data in welding process;Monitoring data is input into welding deformation prediction model, and the predicted welding deformation amount is calculated;The predicted welding deformation amount is compared with preset deformation threshold, and active control instruction is triggered;Start with welding active control device, and dynamic mechanical action is applied to high temperature plastic zone behind welding pool;At the same time, according to the deformation parameter of weld area, the action parameter of welding active control device is closed-loop adjusted;After completing a weld, the actual welding process parameters and measured deformation data of this weld are fed back to welding deformation prediction model, and the deformation prediction parameter of subsequent weld is corrected;Therefore, the application realizes the real-time closed-loop control of thick-wall structure welding deformation, and is suitable for the high-precision manufacturing requirement of fusion reactor superconducting magnet large coil box.

Description

Technical Field

[0001] This invention relates to the field of welding technology, and in particular to a real-time control method and system for welding deformation of a large coil box in a fusion reactor. Background Technology

[0002] The superconducting magnet system in fusion reactors (such as the China Experimental Fusion Reactor (CFETR) and the Compact Fusion Experimental Device (BEST)) is a core component for confining plasma. The large coil box, serving as the structural support for the superconducting magnet, is characterized by its large size (meters to tens of meters), thick walls (50 mm to 150 mm), dense welds, and the use of 316LN austenitic stainless steel. Furthermore, the welding and manufacturing of this coil box requires extremely high dimensional accuracy and extremely low residual stress, as it must maintain structural stability at an extremely low temperature of 4.2K (-269℃).

[0003] Currently, in the actual manufacturing of large coil boxes for fusion reactors, the following engineering problems are faced in controlling welding deformation.

[0004] First, the coil box is a thick plate, multi-layer, multi-pass welded structure. The thermal cycles and stress accumulation of each pass are coupled with each other, resulting in complex deformation evolution. Existing methods mostly rely on empirical estimation, which deviates significantly from reality. This leads to the setting of the anti-deformation amount being based on experience, resulting in low deformation prediction accuracy and a high rework rate.

[0005] Second, the existing process adopts an open-loop mode of "testing after welding and correcting if the deviation exceeds the tolerance". Once the deformation exceeds the tolerance, subsequent correction is difficult and time-consuming. Although there are technical solutions such as welding hammering, they are mainly applied to thin plates or general structures. There is a lack of special process methods for the thick-walled structure of large coil boxes in fusion reactors. Moreover, most of them are open-loop control, which cannot be dynamically adjusted according to real-time deformation feedback and lacks real-time monitoring and control means.

[0006] Third, existing deformation control only targets the assembly accuracy at room temperature, without including the dimensional shrinkage requirements at extremely low temperatures of 4.2 K in the control objectives, and without considering the constraints of low-temperature service, which poses the risk of "passing at room temperature but failing at low temperature".

[0007] Therefore, developing a welding deformation control method and system suitable for thick-walled structures of large coil boxes in fusion reactors, capable of establishing a quantitative mapping relationship between room-temperature welding deformation and 4.2K cryogenic service size requirements, and integrating real-time sensing and active control, is an urgent problem to be solved in this field. Summary of the Invention

[0008] The purpose of this invention is to solve the problems of low deformation prediction accuracy, lack of real-time monitoring and control methods, and failure to consider low-temperature service constraints in the prior art.

[0009] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0010] A method for real-time control of welding deformation of a large coil box in a fusion reactor includes the following steps:

[0011] Establish a welding deformation prediction model;

[0012] A multi-channel sensor array is arranged in the weld area and heat-affected zone of the coil box to collect monitoring data in real time during the welding process; the monitoring data includes welding heat input parameters, interpass temperature parameters, and weld area deformation parameters.

[0013] The real-time collected monitoring data is input into the welding deformation prediction model to calculate the predicted welding deformation under the current welding state in real time.

[0014] The predicted welding deformation is compared with a preset deformation threshold. If the predicted welding deformation exceeds the preset deformation threshold, an active control command is triggered.

[0015] According to the active control command, the welding active control device is activated to apply dynamic mechanical action to the high-temperature plastic zone behind the weld pool; at the same time, according to the deformation parameters of the weld area collected in real time, the action parameters of the welding active control device are adjusted in a closed loop until the predicted welding deformation amount falls back to within the preset deformation threshold.

[0016] After completing a weld, the actual welding process parameters and measured deformation data of this weld are fed back to the welding deformation prediction model to correct the deformation prediction parameters of subsequent welds.

[0017] Furthermore, the preset deformation threshold is calculated in reverse using a room temperature and low temperature dimensional chain mapping model, based on the allowable dimensional shrinkage of the fusion reactor coil box in a 4.2K low-temperature service environment and the assembly gap requirements.

[0018] Furthermore, the active control device during welding includes a welding hammer device or a welding temperature difference stretching device.

[0019] Furthermore, the welding hammer device includes:

[0020] At least one hammer, a follower bracket mounted behind the welding head, and an adjustment component for controlling the distance between the hammer and the weld surface;

[0021] The hammer head end is provided as a carbide hammer tip or a copper buffer head.

[0022] Furthermore, the welding temperature difference stretching device includes:

[0023] At least one set of induction heating coils arranged in front of the weld pool, and a liquid nitrogen or water-cooled composite cooling plate arranged behind the weld pool and on the back of the coil box;

[0024] By controlling the heating rate of the heating zone and the cooling rate of the cooling zone, a controllable temperature difference tensile stress field is established in the weld zone.

[0025] Furthermore, the multi-channel sensor array includes:

[0026] Welding current sensor, welding voltage sensor, temperature sensor, displacement sensor;

[0027] The welding current sensor and welding voltage sensor are installed at the output end of the welding power source and are used to calculate the instantaneous heat input and cumulative heat input of the welding device.

[0028] The temperature sensor includes a thermocouple and an infrared thermometer; the thermocouple is installed on the bevel sidewall by spot welding; the infrared thermometer is arranged in a manner aligned with the back of the molten pool; the temperature sensor is used to monitor the interlayer temperature.

[0029] The displacement sensor includes a laser displacement sensor and / or a strain gauge, used to measure the transverse shrinkage and angular deformation of the weld in real time; the laser displacement sensor is arranged on both sides of the welding torch, and the strain gauge is attached to the base material area on the back of the weld.

[0030] Furthermore, the method for establishing the room temperature and low temperature size chain mapping model includes:

[0031] The coefficient of linear expansion of the coil box material was measured from room temperature to 4.2K, and a database of material thermal shrinkage was established.

[0032] The allowable assembly clearance at 4.2K is reverse-mapped to the target manufacturing tolerance at room temperature, serving as a constraint for controlling welding deformation.

[0033] Furthermore, the welding deformation prediction model is established based on the three-dimensional geometric model of the coil box, material physical property parameters, and preset welding process parameters;

[0034] The physical properties of the material include the temperature-dependent curves of density, specific heat capacity, thermal conductivity, elastic modulus, yield strength, and coefficient of linear expansion.

[0035] The preset welding process parameters include welding method, welding current, arc voltage, welding speed, heat input, weld sequence, and interpass temperature control requirements.

[0036] Furthermore, the real-time calculation of the predicted welding deformation under the current welding state is achieved based on the thermo-elastic-plastic finite element theory and the inherent strain method.

[0037] A real-time control system for welding deformation of a large coil box in a fusion reactor includes:

[0038] The deformation prediction module, equipped with a thermo-coupled finite element solver and an inherent strain database, is used to receive real-time sensor data and output the predicted deformation.

[0039] The multi-source sensing module includes a welding current sensor, a welding voltage sensor, a temperature sensor array, a displacement sensor array, and a data acquisition card, which are used to collect heat input, interpass temperature, and weld deformation parameters in real time during the welding process.

[0040] The active execution module, including a welding hammer device or a welding temperature difference stretching device, is used to receive control commands and apply dynamic mechanical action to the weld zone.

[0041] The central controller is electrically connected to the deformation prediction module, the multi-source sensing module, and the active execution module, and outputs control commands to achieve closed-loop feedback control.

[0042] The beneficial effects of this invention are:

[0043] First, the real-time control method for welding deformation of large coil boxes in fusion reactors provided by this invention, through a closed-loop active control technology solution of "real-time acquisition - deformation prediction - threshold judgment - active control - closed-loop adjustment - step-by-step correction", has played a role in upgrading welding deformation control from "post-correction" to "process active control", avoiding deformation accumulation to an irreversible degree, and significantly improving the deformation control accuracy of multi-layer and multi-pass welding of thick-walled structures. It solves the problem in the existing technology of "measuring after welding and correcting after exceeding the tolerance" which leads to the inability to detect deformation in real time during the welding process and the difficulty in remedying deformation after exceeding the tolerance.

[0044] Secondly, the real-time control method for welding deformation of large coil boxes in fusion reactors provided by this invention establishes a material thermal shrinkage database by measuring the linear expansion coefficient of the coil box material from room temperature to 4.2K temperature range, and reverse-maps the allowable assembly gap at 4.2K low temperature to the target manufacturing tolerance at room temperature. This solves the problem in the prior art where welding deformation control only targets the assembly accuracy at room temperature without considering the dimensional shrinkage constraints under extremely low temperature conditions, leading to "passing at room temperature but failing at low temperature". This method ensures that the coil box meets the assembly accuracy requirements at extremely low temperatures. At the same time, it realizes the quantitative determination of the welding deformation control threshold under the differentiated low temperature shrinkage constraints of each region of the thick-walled, multi-layer, multi-pass welded structure, and the real-time control of the welding process based on this threshold. Attached Figure Description

[0045] Figure 1 A flowchart illustrating a real-time control method for welding deformation of a large coil box in a fusion reactor, provided by the present invention.

[0046] Figure 2 This is a schematic diagram of the welding hammer device used in one embodiment of a real-time control method for welding deformation of a large coil box in a fusion reactor provided by the present invention.

[0047] Figure 3 Provided by the present invention Figure 2 A magnified view of a portion of the hammer assembly;

[0048] Figure 4 A schematic diagram of the architecture of a real-time control method for welding deformation of a large coil box in a fusion reactor provided by the present invention;

[0049] Figure 5 This is a schematic diagram of the process for correcting the welding deformation of a large coil box in a fusion reactor in real time, as provided by the present invention.

[0050] Attached diagrams: 1. Locking caster wheel; 2. Welding machine base; 3. I-beam block; 4. Horizontal Y-axis slide column; 5. Horizontal X-axis slide column; 6. Vertical Z-axis slide column; 7. Adjustable corner welding machine assembly; 8. Hammer head assembly; 9. Welding part;

[0051] 71. Welding machine planar corner joint; 72. Welding machine vertical corner joint;

[0052] 81. Adjustable pressure cylinder; 82. Hammer head. Detailed Implementation

[0053] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0054] Example 1:

[0055] Reference Figure 1 The control flow of this invention is described in detail below:

[0056] Step S100: First, establish a welding deformation prediction model. This model is based on the three-dimensional geometric model of the coil box, material physical property parameters, and preset welding process parameters.

[0057] The three-dimensional geometric model of the coil box reflects the geometry and dimensions of the structure to be welded, including plate thickness, weld location, and bevel type. For large coil boxes in fusion reactors, the plate thickness ranges from 50mm to 150mm, and the weld types include butt welds, fillet welds, and deep, narrow-gap, multi-layer, multi-pass welds. The three-dimensional geometric model can be created using computer-aided design software and imported into the subsequent analysis system in standard geometric data formats (such as STEP, IGES, etc.).

[0058] Material physical properties include the temperature-dependent curves of density, specific heat capacity, thermal conductivity, elastic modulus, yield strength, and coefficient of linear expansion. These parameters significantly influence the calculation of the welding temperature and stress fields. Taking 316LN austenitic stainless steel as an example, its thermal conductivity and specific heat capacity exhibit non-linear changes from room temperature to high temperatures, while its elastic modulus and yield strength decrease significantly with increasing temperature. These material parameters can be obtained from material handbooks or through actual measurements. In one embodiment of this invention, the temperature range for collecting material physical properties covers the entire temperature range from room temperature to the material's melting point, ensuring that corresponding material property data are available for interpolation at each temperature point during the welding thermal cycle.

[0059] The preset welding process parameters include welding method, welding current, arc voltage, welding speed, heat input, weld sequence, and interpass temperature control requirements. The welding method can be narrow-gap tungsten inert gas (TIG) welding or narrow-gap submerged arc welding. Welding current, arc voltage, and welding speed together determine the magnitude of the welding heat input. Weld sequence refers to the order in which the weld passes are arranged in a multi-layer, multi-pass welding process; different weld sequences lead to different thermal cycling histories and stress accumulation paths. Interpass temperature control requirements refer to the target temperature range to which the workpiece needs to be cooled before each weld pass; the level of the interpass temperature directly affects the heat input conditions and deformation development of subsequent weld passes.

[0060] The finite element method can be used to establish a welding deformation prediction model.

[0061] Specifically, in this embodiment, the welding deformation prediction model is constructed as follows:

[0062] (1) Geometric model: A finite element mesh model is established based on the actual three-dimensional CAD model of the coil box. The mesh is refined in the weld zone and a transition mesh is used in the base material zone. For multi-layer and multi-pass welding, the "birth and death element method" is used to simulate the layer-by-layer filling of weld metal.

[0063] (2) Material Model: The material parameters (elastic modulus, yield strength, coefficient of linear expansion, thermal conductivity, specific heat capacity, etc.) of 316LN austenitic stainless steel are set as temperature-dependent functions, covering the complete temperature range from 293K to the material melting point. The material parameters are obtained through high-temperature tensile tests and thermophysical property tests, or by referencing data from the 316LN material handbook.

[0064] (3) Heat source model: A double ellipsoidal heat source model is used to simulate the welding heat input. The parameters of the heat source model are determined by the actual welding process parameters (welding current, voltage, speed) through heat source model calibration tests.

[0065] (4) Boundary conditions: including thermal boundary conditions (convective heat transfer coefficient, radiative heat dissipation coefficient) and mechanical boundary conditions (fixture constraint position, constraint direction), which are determined according to the actual welding fixture and welding process.

[0066] (5) Solution method: The sequential coupling method is adopted. First, nonlinear transient heat conduction analysis is performed to obtain the temperature field, and then the temperature field is applied as a heat load to perform thermo-elastic-plastic stress-strain analysis.

[0067] Specifically, a finite element mesh model is first established based on the three-dimensional geometric model of the coil box. For thick-walled structures of large coil boxes, the mesh needs to be appropriately refined in the weld zone and heat-affected zone to ensure the accuracy of temperature and stress field calculations. In regions far from the weld, a relatively coarse mesh can be used to improve computational efficiency. Then, material physical property parameters are assigned to different regions of the finite element model, and heat source models and boundary conditions are set according to preset welding process parameters. The heat source model is used to simulate the heating effect of the welding arc on the workpiece and can use a moving heat source model (such as a double ellipsoidal heat source model or a Gaussian distributed heat source model). Boundary conditions include convective and radiative heat transfer conditions between the workpiece and the surrounding environment.

[0068] After the finite element model is established, the temperature field distribution and stress-strain field distribution under specific welding process conditions can be obtained through numerical simulation of the welding process. The temperature field is calculated based on the nonlinear transient heat conduction equation, taking into account the nonlinear changes of material thermophysical parameters with temperature and the influence of latent heat of phase transformation. The stress-strain field is calculated based on the thermo-elastic-plastic constitutive model, taking into account the changes of material yield strength, elastic modulus and other mechanical parameters with temperature and the plastic flow law.

[0069] In one embodiment of the present invention, the welding deformation prediction model includes not only the finite element model itself, but also a library of inherent strain parameters calibrated based on field measurement data. The inherent strain parameter library is stored and categorized by plate thickness, weld bead position, and heat input range. Inherent strain refers to the residual plastic strain generated in the weld and heat-affected zone during welding, which is the fundamental cause of welding deformation. By performing thermo-elastic-plastic finite element analysis or experimental measurements on typical joints beforehand, inherent strain data under different plate thicknesses, weld bead positions, and heat input conditions can be obtained. Categorizing and storing this data forms the inherent strain parameter library. In actual deformation prediction, the corresponding inherent strain data can be quickly retrieved from the inherent strain parameter library based on the current plate thickness, weld bead position, and heat input conditions. Then, the welding deformation distribution of the overall structure is obtained through a single elastic finite element calculation. This prediction method based on the inherent strain method has high computational efficiency and is suitable for real-time deformation prediction during the welding process.

[0070] Step S200, Arrangement and data acquisition of the multi-channel sensor array:

[0071] A multi-channel sensor array is deployed in the weld zone and heat-affected zone of the coil box to collect monitoring data in real time during the welding process. The monitoring data includes welding heat input parameters, interpass temperature parameters, and weld zone deformation parameters.

[0072] Specifically, the multi-channel sensor array includes a welding current sensor, a welding voltage sensor, a temperature sensor, and a displacement sensor.

[0073] The welding current sensor and welding voltage sensor are installed at the output end of the welding power source. The welding current sensor measures the welding current in real time, and the welding voltage sensor measures the arc voltage in real time. Based on the real-time measurements of the welding current and welding voltage, the instantaneous heat input and cumulative heat input of the welding apparatus can be calculated. The instantaneous heat input reflects the heating power of the welding arc on the workpiece at the current moment, while the cumulative heat input reflects the total heating energy from the start of welding to the current moment. The magnitude of the heat input directly affects the distribution of the welding temperature field and the degree of welding deformation.

[0074] Temperature sensors include thermocouples and infrared thermometers. Thermocouples are spot-welded to the bevel sidewall. Since the bevel sidewall is one of the areas with the most drastic temperature changes during welding, spot-welding thermocouples to the sidewall allows for accurate monitoring of temperature changes in this area. The infrared thermometer is positioned so that it is aligned with the back of the molten pool. The infrared thermometer uses the principle of thermal radiation to achieve non-contact temperature measurement; aligning it with the back of the molten pool allows for measurement of the temperature distribution in the heat-affected zone behind the molten pool. Temperature sensors are used to monitor interpass temperature. Interpass temperature refers to the temperature of the previous weld and its surrounding area in multi-pass, multi-layer welding, before the next pass is applied. The level of interpass temperature directly affects the heat input conditions and cooling rate of subsequent weld passes, thus affecting the accumulation of welding deformation.

[0075] The displacement sensors include laser displacement sensors and / or strain gauges, used to measure the transverse shrinkage and angular deformation of the weld in real time. As one optional option in this embodiment, the laser displacement sensor is arranged on both sides of the welding torch. Utilizing the principle of laser triangulation, the laser displacement sensor can non-contactly measure the displacement change of the weld surface relative to the sensor. By arranging the laser displacement sensor on both sides of the welding torch, the displacement changes on both sides of the weld can be measured separately, thereby calculating the transverse shrinkage and angular deformation of the weld. As another optional option in this embodiment, the strain gauge is attached to the base material area on the back of the weld and arranged in a triaxial strain rosette pattern. The strain gauge utilizes the resistance strain effect to measure the dynamic strain changes in the base material area during the welding process. Specifically, the triaxial strain rosette can adopt two standard layout schemes: 0°, 45°, 90° (right-angle strain rosette) or 0°, 60°, 120° (equiangular strain rosette). Both schemes can be used to calculate the magnitude and direction of the principal stresses under plane stress when the principal stress direction is unknown; the specific selection depends on the site layout conditions and measurement habits.

[0076] All sensor signals are synchronously acquired via a data acquisition card and transmitted to the host computer. The data acquisition card features multi-channel synchronous sampling to ensure that all sensor signals are synchronized in time. The sampling frequency is determined based on the sensor type and system configuration; generally, welding current and voltage signals are sampled at higher frequencies (e.g., 1kHz–10kHz), while temperature and displacement signals can be sampled at relatively lower frequencies (e.g., 100Hz–1kHz). Furthermore, the synchronously acquired data from each channel carries a unified timestamp, facilitating subsequent data fusion and analysis.

[0077] Step S300 involves inputting the real-time collected monitoring data into the welding deformation prediction model to calculate the predicted welding deformation under the current welding state in real time. The real-time calculation of the predicted welding deformation under the current welding state is based on the thermo-elastic-plastic finite element theory and the inherent strain method.

[0078] Specifically, during the welding process, the temperature of different areas of the workpiece continuously changes as the welding heat source moves. Real-time collected monitoring data (including welding heat input parameters, interpass temperature parameters, and weld deformation parameters) reflects the actual conditions of the current welding state. This invention inputs these measured data into a pre-established welding deformation prediction model, which can correct the model's calculation parameters, making deformation prediction more accurate.

[0079] Specifically, real-time acquired welding heat input parameters (including real-time values ​​of welding current and arc voltage) are used to determine the heat source intensity at the current moment. In the welding deformation prediction model, the parameters of the heat source model (such as heat source power and heat source distribution shape) are adjusted in real time based on the measured welding current and voltage values. Real-time acquired interpass temperature parameters are used to determine the initial temperature conditions before the current welding pass. In multi-layer, multi-pass welding, the interpass temperature before each pass directly affects the temperature field distribution and stress-strain development of that pass. The real-time acquired interpass temperature values ​​serve as the initial conditions input to the model for thermal analysis. Real-time acquired weld zone deformation parameters (including transverse shrinkage and angular deformation) are used to verify and correct the model's prediction results. Furthermore, by comparing the measured deformation with the model's predicted deformation, the model's prediction accuracy can be determined, and the model parameters can be corrected online.

[0080] Based on the thermo-elastic-plastic finite element theory, the stress-strain relationship during welding follows the thermo-elastic-plastic constitutive equation. Within each time step, the temperature field distribution is first calculated according to the heat conduction equation, and then the stress and strain field distributions are calculated according to the thermo-elastic-plastic constitutive equation. During the calculation, the nonlinear changes in material physical properties with temperature, as well as the loading and unloading conditions of plastic flow, need to be considered.

[0081] Based on the inherent strain method, within each time step, the corresponding inherent strain data can be retrieved from the inherent strain parameter library according to the current welding conditions (plate thickness, weld position, heat input, etc.). The inherent strain is applied to the weld area of ​​the structure in the form of an equivalent load, and then the distribution of welding deformation can be obtained through a single elastic finite element calculation. This method has high computational efficiency and is suitable for real-time deformation prediction in each sampling period during the welding process.

[0082] In one embodiment of the present invention, a deformation prediction calculation is performed once per sampling period (e.g., every 100ms). After the measured data of the current sampling period is input into the model, the model completes the calculation and outputs the predicted welding deformation within a set time (e.g., less than the sampling period). In this way, the system designed in this invention can continuously monitor the development trend of deformation during the welding process.

[0083] Step S400: Compare the predicted welding deformation with a preset deformation threshold. If the predicted welding deformation exceeds the preset deformation threshold, trigger an active control command.

[0084] At this point, the 4.2K cryogenic service constraint of the large coil box of the fusion reactor is not a simple problem of "uniform dimensional shrinkage after temperature reduction". Its special and complex nature lies in the fact that the coil box is a thick-walled (50mm~150mm) multi-layer, multi-pass welded structure. The welding thermal cycle history and residual stress distribution of different weld positions and plate thicknesses are significantly different. When cooling from room temperature to 4.2K, the integral shrinkage rate of each region is not uniform, and the welding residual stress will be further redistributed at low temperature, resulting in different low-temperature dimensional shrinkage in each region. Existing technology only uses room temperature assembly accuracy as the control target, without establishing a quantitative mapping relationship between "room temperature welding deformation - low temperature dimensional shrinkage - 4.2K final assembly accuracy", nor providing a method to calculate the control threshold of differentiated welding deformation in each region based on this mapping relationship. Therefore, even if the existing technology uses closed-loop control to suppress welding deformation at room temperature, it cannot ensure that the final size after cooling to 4.2K meets the assembly gap requirements of the superconducting magnet.

[0085] For thick-walled austenitic stainless steel structures such as the large coil box of a fusion reactor, the service condition is an extremely low temperature environment of 4.2K. Unlike ordinary welded structural components, which only need to control dimensional accuracy at room temperature, the welding deformation control of this structure must also take into account the thermal shrinkage effect during cooling from 293K to 4.2K. Therefore, the welding deformation control threshold of this invention is not arbitrarily set based on the assembly accuracy at room temperature, but is derived by reverse derivation based on the allowable assembly gap at the low temperature of 4.2K.

[0086] This is because the fusion reactor coil box needs to maintain structural stability at an extremely low temperature of 4.2K. When the material is cooled from room temperature (approximately 293K) to 4.2K, its dimensions shrink due to thermal contraction. In this embodiment, the integral shrinkage rate of 316LN austenitic stainless steel from 293K to 4.2K is approximately 0.3%. That is, a component with dimensions of 1000mm at room temperature will shrink by approximately 3mm after cooling to 4.2K. This shrinkage is not negligible for the high-precision assembly of superconducting magnets.

[0087] Therefore, when setting the welding deformation control threshold, the assembly accuracy at room temperature should not be the sole objective; the impact of low-temperature shrinkage must also be considered. If only room-temperature accuracy is the target, there is a risk that the dimensions may be acceptable at room temperature but out of tolerance after cooling to 4.2K, i.e., "acceptable at room temperature, but fails at low temperature."

[0088] The room temperature and low temperature size chain mapping model in this invention is used to solve this problem. The method for establishing this model includes the following steps.

[0089] First, the coefficient of linear expansion of the coil box material was measured from room temperature to 4.2K to establish a database of material thermal shrinkage. The coefficient of linear expansion is a crucial parameter of a material's thermophysical properties, defined as the relative change in material dimensions for every 1°C change in temperature. For 316LN austenitic stainless steel, the coefficient of linear expansion decreases with decreasing temperature, tending to a relatively small value in the extremely low temperature range. Through actual measurements, curves showing the coefficient of linear expansion as a function of temperature across the entire temperature range from 293K to 4.2K were obtained. Integrating the coefficient of linear expansion from 293K to 4.2K with temperature yields the total shrinkage rate from 293K to 4.2K. The core formula is:

[0090] ;

[0091] in, These are the dimensions of the material at a low temperature of 4.2K; These are the dimensions of the material at room temperature (293K). The integral shrinkage rate of the material from 293K to 4.2K (obtained from the material thermal shrinkage database).

[0092] This is because when the material is cooled from 293K to 4.2K, the dimensions shrink according to the integral shrinkage rate. Contraction occurs. This is based on the definition of the coefficient of linear expansion. ,in, The change in length The original length, The change in temperature is the total shrinkage rate, which can be obtained by integrating the temperature.

[0093] Then, the allowable assembly clearance at 4.2K is inversely mapped to the target manufacturing tolerance at room temperature, serving as a constraint for controlling welding deformation. Specifically, given the allowable deviation range of a certain mating dimension at 4.2K and the shrinkage rate of the material from 293K to 4.2K, the tolerance range that should be controlled for that dimension at room temperature can be derived. The core formula for this inverse mapping is:

[0094] ;

[0095] in, This is the target room temperature size of the material derived from the reverse calculation; The required dimensions of the material at 4.2K are determined by the size chain equation.

[0096] It should also be noted that for multi-layer, multi-pass welded structures, the materials at different weld positions experience different thermal cycling histories during the welding process, resulting in differences in the microstructure and residual stress levels of each region, which in turn affects their low-temperature shrinkage behavior when cooled from 293K to 4.2K.

[0097] Let the equivalent peak temperature of the i-th weld zone be... Then the integral contraction rate of this region Calculate using the following formula:

[0098]

[0099] in, This is the equivalent linear expansion coefficient, and its value depends on the current temperature. and the highest temperature experienced in the region :

[0100]

[0101] In the formula, The standard linear expansion coefficient of the base material (obtained by actual measurement using a differential thermal expansion meter); The additional thermal shrinkage correction caused by the peak temperature is obtained by conducting thermal expansion tests on the sample after simulated thermal cycling in advance. This is the relaxation constant, with a value ranging from 300 to 500K.

[0102] By substituting the aforementioned differentiated integral shrinkage rate into the size chain mapping model, the differentiated room temperature control thresholds for each weld zone can be obtained.

[0103] Furthermore, through this reverse mapping, the present invention can transform the dimensional requirements under low-temperature service conditions into control targets in the room-temperature manufacturing process.

[0104] In one embodiment of the present invention, taking a key mating dimension of a coil box as an example, the allowable deviation at a low temperature of 4.2K is ±0.5mm, and the integral shrinkage rate of 316LN from 293K to 4.2K is 0.003 (i.e. 0.3%). Therefore, the target manufacturing tolerance of this dimension at room temperature should be ±0.5mm divided by (1-0.003), which is approximately ±0.5015mm.

[0105] At this point, the predicted welding deformation is compared with the preset deformation threshold. If the predicted welding deformation exceeds the preset deformation threshold, it indicates that under the current welding condition, without intervention, the final welding deformation will exceed the allowable range under cryogenic service conditions. The system immediately triggers an active control command.

[0106] To verify the technical effects of this invention, the following experimental scheme can be used to obtain verification data: Table 1: Comparison of Implementation Methods for Different Control Methods

[0107] control group No active control (conventional welding) — 3 items Assembly clearance deviation at 4.2K Experimental group 1 Welding hammering Hammering frequency 70Hz, hammering force 600N 3 items Assembly clearance deviation at 4.2K Experimental group 2 Temperature difference stretching Heating temperature 500℃, cooling temperature -50℃ 3 items Assembly clearance deviation at 4.2K Experimental group 3 Welding hammering + thermal stretching Combined use 3 items Assembly clearance deviation at 4.2K

[0108] The measurement method was as follows: After the coil boxes of each experimental group were welded, the key assembly dimensions were measured at room temperature; then, they were placed in a low temperature environment of 4.2K, and the dimensions at the same position were remeasured using a laser tracker; the low temperature dimensional deviation of each group was calculated, and the effects of different control methods were compared.

[0109] In this embodiment, the expected assembly gap deviations at 4.2K for each experimental group are as follows: Table 2: Comparison of Expected Effects of Different Control Methods

[0110] control group No active control Without active intervention during welding, the assembly gap deviation is greatest at low temperatures. Experimental group 1 Welding hammering By subjecting the weld metal to plastic elongation through periodic impact, the low-temperature assembly gap deviation was reduced compared to the control group. Experimental group 2 Temperature difference stretching By creating a longitudinal tensile stress field through a temperature gradient, shrinkage deformation is suppressed, resulting in a smaller assembly gap deviation at low temperatures compared to the control group. Experimental group 3 Welding hammering + thermal stretching By combining two control methods, welding deformation can be synergistically suppressed, resulting in optimal assembly clearance deviation at low temperatures.

[0111] It is expected that the assembly gap deviation of experimental groups 1, 2 and 3 at a low temperature of 4.2K will be smaller than that of the control group, among which experimental group 3 (used in combination with welding hammer and thermal stretching) will have the most significant effect.

[0112] In step S500, the system activates the welding hammer device to apply dynamic mechanical action to the high-temperature plastic zone behind the weld pool according to the active control command.

[0113] The welding hammer device includes at least one hammer head, a follower bracket mounted behind the welding head, and an adjusting element for controlling the distance between the hammer head and the weld surface. The hammer head end is equipped with a carbide tip or a copper buffer head that matches the hardness of the coil box steel. In this embodiment, the adjusting element is configured as an adjustable pressure cylinder.

[0114] Reference Figure 2 and 3 The working principle of the welding hammer device is as follows:

[0115] During welding, the workpiece is placed on the welding machine base; the adjustable corner welding machine assembly is controlled by the horizontal Y-axis sliding column installed on the welding machine base by the I-shaped pad block, the horizontal X-axis sliding column slidably installed on the horizontal Y-axis sliding column, and the vertical Z-axis sliding column slidably installed on the horizontal Y-axis sliding column to weld the workpiece.

[0116] During the welding process, the welding angle can also be adjusted by the welding machine plane angle joint and the welding machine vertical angle joint in the adjustable corner welding machine assembly;

[0117] During welding, a high-temperature plastic zone exists behind the weld pool. This zone contains materials with high temperatures and low yield strength, existing in a plastic state. When the hammer, following the adjustable corner welding machine assembly, applies impact force to this high-temperature plastic zone, compressive plastic deformation occurs. This compressive plastic deformation partially offsets the tensile plastic strain generated during welding cooling, thereby reducing the final weld deformation. It should also be noted that the effectiveness of the hammering depends on factors such as the hammering frequency, hammering force, hammering position, and the distance between the hammer and the weld surface.

[0118] In this embodiment, the hammer head reciprocates, and the hammering frequency and force can be controlled as needed. A follower bracket is installed behind the welding head, allowing the hammer head to move with the welding head, maintaining a relatively constant relative position between the hammer head and the molten pool. An adjustment component is used to adjust the distance between the hammer head and the weld surface to accommodate variations in plate thickness and bevel shape. In this embodiment, the cemented carbide hammer tip has high hardness and wear resistance, making it suitable for high-frequency hammering operations. Alternatively, a copper buffer head with good toughness and thermal conductivity can be selected to buffer the impact during hammering, reducing damage to the workpiece surface.

[0119] It should be noted that the adjusting component adjusts the distance between the hammer head and the weld surface to accommodate variations in plate thickness and bevel type.

[0120] (1) Effect of plate thickness: As the plate thickness increases (50mm→150mm), the vertical distance between the weld surface and the welding torch increases, and the hammer needs to be adjusted downward accordingly to maintain an effective contact distance with the weld surface. The adjustment amount is automatically calculated by the control system according to the change of plate thickness.

[0121] (2) Influence of bevel type: Different bevel types (V-shaped, U-shaped, narrow gap) result in different weld surface width and depth. For V-shaped bevel, the hammer can be offset from the weld center by a certain distance; for narrow gap bevel, the hammer needs to be aligned with the weld center.

[0122] (3) Adjustment method: The adjustment component receives instructions from the central controller and drives the hammer head to move in the vertical direction. The specific adjustment amount is automatically calculated by the control system or manually set through the human-machine interface based on the combination parameters of plate thickness and bevel form.

[0123] After the welding hammer device is activated, the system adjusts its operating parameters in a closed loop based on real-time collected deformation parameters of the weld area. Specifically, laser displacement sensors and / or strain gauges measure the lateral shrinkage and angular deformation of the weld in real time, and these measured deformation data reflect the current control effect of the welding hammer. If the measured deformation is still large, it indicates that the current control effect is insufficient, and the system needs to increase the hammering frequency or hammering force. If the deformation has been effectively controlled, the system maintains the current control parameters or appropriately reduces the hammering intensity. The closed-loop feedback adjustment of this system continues until the predicted welding deformation falls back to within the preset deformation threshold.

[0124] It should be noted that the core of closed-loop control lies in using the measured deformation as a feedback signal, comparing it with a preset deformation threshold, and automatically adjusting the control parameters based on the magnitude and direction of the deviation. Through this feedback control mechanism, the present invention enables the control action to adaptively follow changes in the welding state, avoiding the drawbacks of fixed parameters in open-loop control.

[0125] Specifically, during hammering, the real-time feedback signal from the strain sensor is used to adjust the hammering parameters in a closed loop.

[0126] In this embodiment, the central controller employs a proportional-integral-derivative (PID) control algorithm to achieve closed-loop adjustment of the hammering parameters. Using the lateral contraction amount as the control target, the specific adjustment logic is as follows:

[0127] Let the current measured lateral shrinkage be... The preset deformation threshold is Define the deviation amount .

[0128] when When the value is greater than 0 (i.e., the deformation exceeds the threshold), the controller calculates the adjustment amount of the hammering frequency according to the PID algorithm.

[0129] Specific adjustment examples are as follows: Table 3: Comparison Table of Deviation and Frequency Adjustment Amount

[0130] e≤0.05 0 50 (reference frequency) 0.05 <e≤0.10 +5 55 0.10 <e≤0.20 +10 60 e>0.20 +20 70

[0131] When the measured deformation falls below the threshold (e < 0), the controller maintains the current parameters and gradually reduces the hammering frequency by 5Hz each time until the deformation stabilizes, until it returns to the reference frequency of 50Hz.

[0132] The frequency adjustment values ​​mentioned above are example values. In actual applications, they can be adjusted according to factors such as plate thickness, welding speed, and heat input. The reference frequency of 50Hz can be set within the range of 30Hz to 110Hz.

[0133] Step S600: After completing a weld, the actual welding process parameters and measured deformation data of this weld are fed back to the welding deformation prediction model to correct the deformation prediction parameters of subsequent welds.

[0134] Because the walls of large coil boxes in fusion reactors are quite thick (50mm–150mm), their welding typically requires a multi-layer, multi-pass welding process. Specifically, multi-layer, multi-pass welding refers to the process of sequentially welding multiple layers along the thickness direction of a single weld joint, with each layer consisting of multiple passes. In multi-layer, multi-pass welding, the thermal cycles and stress accumulation of each pass are coupled, resulting in complex deformation evolution. The welding conditions for each pass (such as interpass temperature and constraint conditions) differ, leading to different deformation patterns in each pass.

[0135] Existing methods typically use uniform process parameters and reverse deformation settings for all weld passes. Since the differences in deformation patterns between each pass are not considered, this uniform setting often leads to insufficient deformation control in some passes and excessive deformation control in others, making it difficult to guarantee the overall deformation accuracy.

[0136] Unlike existing technologies, this invention employs a pass-by-pass correction strategy. After each weld pass is completed, the system acquires the actual welding process parameters (including actual welding current, voltage, speed, etc.) and measured deformation data (including post-weld transverse shrinkage, angular deformation, etc.) for that pass. This actual data is fed back to the welding deformation prediction model, which recalibrates the inherent strain parameters and deformation prediction parameters based on the actual conditions of that weld pass. Then, the corrected model is used to predict the deformation development of subsequent weld passes, and the welding sequence, heat input, and reverse deformation amount of subsequent weld passes are adjusted accordingly.

[0137] Reference Figure 5 Specifically, the process for making corrections one by one is as follows:

[0138] First, before welding the first weld pass, the welding parameters for the first pass are set based on the initial welding process parameters and deformation prediction model. After the first weld pass is completed, the actual post-weld deformation is measured and compared with the model prediction value. Based on the comparison results, the parameters in the model related to the first pass (such as the actual inherent strain value of this pass, heat input efficiency coefficient, etc.) are corrected.

[0139] Then, the corrected model is used to predict the deformation of the second weld. Based on the corrected prediction results, the welding parameters for the second weld are adjusted (such as adjusting the heat input, changing the weld sequence, etc.). After the second weld is completed, the actual deformation is measured again and the model is corrected.

[0140] Finally, this process is repeated for each weld pass until all weld passes are completed.

[0141] Furthermore, this invention, through this step-by-step correction strategy, ensures that the welding parameters of each weld are optimized based on the actual data of the previous weld, making the deformation prediction of subsequent welds more accurate and the control more precise.

[0142] Example 2:

[0143] Unlike Embodiment 1, in this embodiment, in step S500, the system activates the welding temperature difference stretching device according to the active control command to apply dynamic mechanical action to the high-temperature plastic zone behind the weld pool.

[0144] The welding temperature difference stretching device includes at least one set of induction heating coils arranged in front of the weld pool, and a liquid nitrogen or water-cooled composite cooling plate arranged behind the weld pool and on the back of the coil box. This invention establishes a controllable temperature gradient in the weld zone by controlling the heating rate of the heating zone and the cooling rate of the cooling zone, thereby forming a longitudinal tensile stress field.

[0145] It should be noted that the working principle of the welding temperature difference stretching device is as follows:

[0146] First, an induction heating coil is placed in front of the weld pool to preheat the base material before it enters the welding area. A cooling device (liquid nitrogen or water-cooled composite cooling plate) is placed behind the weld pool and on the back of the coil box to rapidly cool the welded area. This creates a temperature difference field in the weld zone, with a high temperature in front and a low temperature behind. Due to the thermal expansion and contraction effect of materials, the high-temperature area tends to expand, and the low-temperature area tends to contract, thus generating a tensile stress field in the weld zone. This tensile stress field cancels out the compressive stress field generated during welding and cooling, thereby reducing welding deformation. By controlling the heating power of the induction heating coil and the cooling medium flow rate of the cooling plate, the heating rate of the heating zone and the cooling rate of the cooling zone can be adjusted, thereby controlling the magnitude and distribution of the tensile stress field due to the temperature difference.

[0147] Next, after the welding temperature difference stretching device is activated, the system adjusts the device's parameters in a closed loop based on the real-time collected deformation parameters of the weld area. Specifically, laser displacement sensors and / or strain gauges measure the lateral shrinkage and angular deformation of the weld in real time. These measured deformation data reflect the current control effect of the welding temperature difference stretching. If the deformation is still large, it indicates that the current control effect is insufficient, and the system increases the heating power or cooling rate to enhance the temperature difference stretching effect. If the deformation has been effectively controlled, the system maintains the current control parameters or appropriately reduces the control intensity. This closed-loop feedback adjustment continues until the predicted welding deformation falls back to within the preset deformation threshold.

[0148] In one embodiment of the present invention, for welds with high appearance quality requirements, the temperature difference stretching device is preferred for control, because temperature difference stretching is a non-contact control method and will not leave hammer marks on the workpiece surface.

[0149] Example 3

[0150] Reference Figure 4In this embodiment, the present invention provides a real-time control system for welding deformation of a large coil box in a fusion reactor. The control system includes a deformation prediction module, a multi-source sensing module, an active execution module, and a central controller.

[0151] Specifically, the deformation prediction module is equipped with a thermo-coupled finite element solver and an inherent strain database, used to receive real-time sensor data and output predicted deformation. The thermo-coupled finite element solver performs calculations for heat conduction analysis and thermo-elastic-plastic stress-strain analysis. The inherent strain database stores inherent strain data categorized by plate thickness, weld bead location, and heat input range.

[0152] It should be noted that after receiving real-time data from the multi-source sensing module, the deformation prediction module calls the thermo-coupled finite element solver and the inherent strain database to perform rapid calculations and output the predicted welding deformation. When the predicted deformation exceeds a preset threshold, the deformation prediction module generates a control command and transmits it to the active execution module.

[0153] Additionally, it should be noted that the inherent strain database is constructed using the following method:

[0154] (1) Data source: The inherent strain data under different conditions are obtained by calculating typical welded joints using the thermo-elastic-plastic finite element method or by measuring actual welding experiments.

[0155] (2) Influencing factors: The main influencing factors of inherent strain include plate thickness, groove type, welding heat input, material grade and joint type. The database is classified and stored according to the above factors.

[0156] (3) Database structure: Each data record includes plate thickness, bevel type, heat input, material grade, joint type, longitudinal inherent strain, transverse inherent strain and thickness direction inherent strain;

[0157] (4) Data expansion: For working conditions that are not directly obtained through calculation or experiment, interpolation methods (linear interpolation or spline interpolation) can be used to obtain approximate inherent strain data;

[0158] (5) Database update: During the actual welding process, the measured deformation data is fed back to the database to correct and supplement the inherent strain data, so as to achieve continuous optimization of the database.

[0159] Specifically, the multi-source sensing module includes a welding current sensor, a welding voltage sensor, a temperature sensor array, a displacement sensor array, and a data acquisition card, used to acquire heat input, interpass temperature, and weld deformation parameters in real time during the welding process. The welding current sensor and welding voltage sensor are installed at the output end of the welding power source. The temperature sensor array includes thermocouples and an infrared thermometer. The displacement sensor array includes a laser displacement sensor and / or strain gauges. The data acquisition card synchronously acquires and converts the signals from each sensor to analog-to-digital data, and then transmits the digitized data to the central controller and deformation prediction module.

[0160] Specifically, the active execution module includes either a welding hammer device or a welding temperature difference stretching device, used to receive control commands and apply dynamic mechanical action to the weld zone. The welding hammer device includes a hammer head, a follow-up support, and an adjusting component. The welding temperature difference stretching device includes an induction heating coil and a liquid nitrogen or water-cooled composite cooling plate. After receiving control commands from the central controller, the active execution module activates the corresponding execution device and adjusts the action parameters in real time according to the adjustment commands from the central controller.

[0161] The central controller connects the deformation prediction module, the multi-source sensing module, and the active execution module to achieve closed-loop feedback control. The central controller receives the predicted deformation and control commands from the deformation prediction module and transmits the control commands to the active execution module. Simultaneously, the central controller receives real-time deformation monitoring data from the multi-source sensing module, calculates the control deviation based on the measured deformation, generates adjustment commands, and transmits them to the active execution module, thus achieving closed-loop feedback regulation. The central controller also coordinates the data and control flows between the modules to ensure stable system operation.

[0162] The dynamic mechanical action described in this invention refers to the time-varying mechanical load applied to the weld and heat-affected zone metal, which are still at a high temperature during the welding process. This mechanical load includes, but is not limited to: the periodic impact force generated by the welding hammer, causing plastic elongation of the weld metal; and the thermal stress generated by the temperature difference stretching device, which establishes a controllable temperature gradient in the weld zone through heating and cooling, forming a longitudinal tensile stress field. Both of these methods can be used independently or in combination to counteract shrinkage deformation during the welding process and reduce residual stress.

[0163] Example 4

[0164] The following example of a specific welding process illustrates the overall processing flow of this invention:

[0165] During the manufacturing process of a large coil box for a fusion reactor, a butt weld with a length of 2m and a plate thickness of 80mm is required. The coil box is made of 316LN austenitic stainless steel, and the welding method is narrow-gap tungsten inert gas welding (TIG), employing a multi-layer, multi-pass welding process, requiring a total of 20 welding passes.

[0166] First, a welding deformation prediction model needs to be established. A finite element mesh model is established based on the three-dimensional geometric model of the coil box. The material physical properties of 316LN (including density, specific heat capacity, thermal conductivity, elastic modulus, yield strength, and linear expansion coefficient as a function of temperature) are input, and preset welding process parameters are set (welding current 300A, arc voltage 28V, welding speed 150mm / min, heat input of approximately 3.36kJ / mm, welding sequence from bottom to top layer by layer, and interpass temperature controlled not higher than 150℃).

[0167] Next, a multi-channel sensor array is deployed. Welding current and welding voltage sensors are installed at the welding power source output. Thermocouples are spot-welded to the bevel sidewalls of the weld to monitor interpass temperature. Laser displacement sensors are positioned on both sides of the welding torch to measure the lateral shrinkage and angular deformation of the weld in real time. Strain gauges are attached to the base material area on the back of the weld. All sensors are connected to the central controller and deformation prediction module via a data acquisition card.

[0168] The first weld seam is then welded. During the welding process, sensors collect and monitor data in real time and transmit it to the deformation prediction module. Based on the measured data, the deformation prediction module uses thermo-elastic-plastic finite element theory and the inherent strain method to calculate and predict the welding deformation in real time. In this embodiment, when the welding reaches approximately 300 mm in length, the predicted welding deformation (such as lateral shrinkage) reaches a preset deformation threshold (calculated inversely using a room-temperature / low-temperature dimensional chain mapping model based on the 4.2 K low-temperature service requirements; in this embodiment, it is 0.5 mm). At this point, the central controller triggers an active control command.

[0169] Specifically, this invention employs the thermo-elastic-plastic finite element method to calculate the inherent strain of a typical welded joint. The specific implementation steps of this method are as follows:

[0170] (1) Establish a three-dimensional finite element model of a typical welded joint (such as a butt joint or a T-joint), refine the mesh in the weld zone, and use a transition mesh in the base material zone.

[0171] (2) Thermal analysis stage: The moving heat source model (double ellipsoidal heat source model or Gaussian heat source model) is used to simulate the welding heat input, and the temperature field distribution at each moment during the welding process is obtained through nonlinear transient heat conduction calculation.

[0172] (3) Mechanical analysis stage: The temperature field at each time point obtained from the thermal analysis is used as a thermal load and applied to the structural model one by one. The material parameters (elastic modulus, yield strength, coefficient of linear expansion, etc.) are set as temperature-dependent functions and solved using a thermo-elastic-plastic constitutive model to obtain the distribution of welding residual stress and deformation.

[0173] It should be noted that the specific parameter settings (such as mesh size, time step, etc.) of the specific modeling and solution process of the thermo-elastic-plastic finite element method are determined through conventional experiments based on actual working conditions.

[0174] The inherent strain method is implemented in this patent according to the following steps:

[0175] The first step is to use finite element software to calculate the deformation results of a typical welded joint (for a specific plate thickness, bevel type, and welding parameters), and then extract the residual plastic strain data of the weld area from the results.

[0176] The second step is to classify and store the residual plastic strain data obtained under different conditions into a database according to plate thickness, bevel type, and welding parameters.

[0177] The third step is to apply the strain data corresponding to the conditions in the database directly to the weld area of ​​the coil box model when calculating the overall deformation of the large coil box. This will allow for a quick elastic calculation and obtain the overall deformation result.

[0178] This method is computationally much more efficient than simulating the entire welding process. The database is queried and retrieved based on plate thickness, bevel type, and thermal input. If no perfectly matching database record is found, it can be obtained through interpolation.

[0179] The active execution module initiates the welding hammering device. The hammer head of this device impacts the high-temperature plastic zone behind the weld pool at a set initial frequency (e.g., 20Hz) and initial hammering force. During hammering, a laser displacement sensor continuously measures the transverse shrinkage of the weld. The central controller calculates the deviation from a preset threshold based on the real-time measured transverse shrinkage. If the shrinkage continues to increase, the hammering frequency is increased to 30Hz or the hammering force is increased. Once the transverse shrinkage falls below 0.5mm and stabilizes, the current hammering parameters are maintained. This closed-loop adjustment continues until the current weld pass is completed.

[0180] After the first weld pass is completed, the actual post-weld transverse shrinkage and angular deformation are measured. These measured data, along with the actual welding process parameters for the first pass, are fed back to the welding deformation prediction model. The model corrects the inherent strain parameters and deformation prediction parameters based on the actual data from the first pass. Then, the corrected model is used to predict the deformation development of the second weld pass, and the welding parameters for the second pass are adjusted based on the prediction results (e.g., slightly reducing the heat input to reduce the deformation trend). The second weld pass is then welded according to the adjusted parameters, repeating the above process of real-time monitoring, deformation prediction, threshold judgment, active control, and closed-loop adjustment.

[0181] This process was repeated step by step, with the model corrected and parameters adjusted for each subsequent weld pass based on measured data. After 20 weld passes, all welds were completed. Because each weld pass was monitored and actively controlled in real time, and the parameters for subsequent weld passes were optimized based on the measured data from the previous pass, the overall welding deformation was effectively controlled within the preset deformation threshold, meeting the dimensional accuracy requirements under 4.2K low-temperature service conditions.

[0182] After all welding is completed, a laser tracker can be used to measure the dimensional accuracy and flatness of key parts of the coil box, and X-ray diffraction or blind hole method can be used to detect residual stress in the weld zone. Dimensional re-measurement is then performed at 4.2K on a cryogenic platform to confirm that the fit clearance meets the design requirements.

[0183] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A real-time control method for welding deformation of a large coil box in a fusion reactor, characterized in that, Includes the following steps: Establish a welding deformation prediction model; A multi-channel sensor array is arranged in the weld area and heat-affected zone of the coil box to collect monitoring data in real time during the welding process; the monitoring data includes welding heat input parameters, interpass temperature parameters, and weld area deformation parameters. The real-time collected monitoring data is input into the welding deformation prediction model to calculate the predicted welding deformation under the current welding state in real time. The predicted welding deformation is compared with a preset deformation threshold. If the predicted welding deformation exceeds the preset deformation threshold, an active control command is triggered. According to the active control command, the welding active control device is activated to apply dynamic mechanical action to the high-temperature plastic zone behind the weld pool; at the same time, according to the deformation parameters of the weld area collected in real time, the action parameters of the welding active control device are adjusted in a closed loop until the predicted welding deformation amount falls back to within the preset deformation threshold. After completing a weld, the actual welding process parameters and measured deformation data of this weld are fed back to the welding deformation prediction model to correct the deformation prediction parameters of subsequent welds.

2. The method according to claim 1, characterized in that, The preset deformation threshold is calculated in reverse using a room temperature and low temperature dimensional chain mapping model, based on the allowable dimensional shrinkage of the fusion reactor coil box in a 4.2K low-temperature service environment and the assembly clearance requirements.

3. The method according to claim 1, characterized in that, The active control device for welding includes a welding hammer device or a welding temperature difference stretching device.

4. The method according to claim 3, characterized in that, The welding hammer device includes: At least one hammer, a follower bracket mounted behind the welding head, and an adjustment component for controlling the distance between the hammer and the weld surface; The hammer head end is provided as a carbide hammer tip or a copper buffer head.

5. The method according to claim 3, characterized in that, The temperature difference stretching device includes: At least one set of induction heating coils arranged in front of the weld pool, and a liquid nitrogen or water-cooled composite cooling plate arranged behind the weld pool and on the back of the coil box; By controlling the heating rate of the heating zone and the cooling rate of the cooling zone, a controllable temperature difference tensile stress field is established in the weld zone.

6. The method according to claim 1, characterized in that, The multi-channel sensor array includes: Welding current sensor, welding voltage sensor, temperature sensor, displacement sensor; The welding current sensor and welding voltage sensor are installed at the output end of the welding power source and are used to calculate the instantaneous heat input and cumulative heat input of the welding device. The temperature sensor includes a thermocouple and an infrared thermometer; the thermocouple is installed on the bevel sidewall by spot welding; the infrared thermometer is arranged in a manner aligned with the back of the molten pool; the temperature sensor is used to monitor the interlayer temperature. The displacement sensor includes a laser displacement sensor and / or a strain gauge, used to measure the transverse shrinkage and angular deformation of the weld in real time; the laser displacement sensor is arranged on both sides of the welding torch, and the strain gauge is attached to the base material area on the back of the weld.

7. The method according to claim 2, characterized in that, The method for establishing the room temperature and low temperature size chain mapping model includes: The coefficient of linear expansion of the coil box material was measured from room temperature to 4.2K, and a database of material thermal shrinkage was established. The allowable assembly clearance at 4.2K low temperature is reverse-mapped to the target manufacturing tolerance at room temperature, and used as a constraint condition for welding deformation control.

8. The method according to claim 1, characterized in that, The welding deformation prediction model is established based on the three-dimensional geometric model of the coil box, material physical property parameters, and preset welding process parameters. The physical properties of the material include the temperature-dependent curves of density, specific heat capacity, thermal conductivity, elastic modulus, yield strength, and coefficient of linear expansion. The preset welding process parameters include welding method, welding current, arc voltage, welding speed, heat input, weld sequence, and interpass temperature control requirements.

9. The method according to claim 1, characterized in that, The real-time calculation of the predicted welding deformation under the current welding state is achieved based on the thermo-elastic-plastic finite element theory and the inherent strain method.

10. A real-time control system for welding deformation of a large coil box in a fusion reactor, characterized in that, include: The deformation prediction module, equipped with a thermo-coupled finite element solver and an inherent strain database, is used to receive real-time sensor data and output the predicted deformation. The multi-source sensing module includes a welding current sensor, a welding voltage sensor, a temperature sensor array, a displacement sensor array, and a data acquisition card, which are used to collect heat input, interpass temperature, and weld deformation parameters in real time during the welding process. The active execution module, including a welding hammer device or a welding temperature difference stretching device, is used to receive control commands and apply dynamic mechanical action to the weld zone. The central controller is electrically connected to the deformation prediction module, the multi-source sensing module, and the active execution module, and outputs control commands to achieve closed-loop feedback control.

Images