Manufacturing method of hastelloy shielding sleeve for intelligent micro-strain welding

Abstract

The invention belongs to the field of mechanical engineering, particularly relates to a manufacturing method of a hastelloy shielding sleeve for intelligent micro-strain welding, and aims to solve the problems of high residual stress, serious element segregation and low yield in traditional welding. According to the method, a multi-physics field coupling digital twinborn model is constructed, a high-temporal-spatial-resolution in-situ sensing system is deployed, and temperature, strain and component data of a molten pool are collected in real time; the laser power, the scanning speed and the cooling parameters are dynamically adjusted through model prediction control, and accurate control over heat input and the cooling rate is achieved; by combining spectral feedback closed-loop compensation element segregation, the component deviation of the welding seam is less than or equal to + / -0.5%; and ultrasonic impact and electromagnetic induction composite treatment is adopted after welding. According to the method, the residual stress is smaller than or equal to 150 MPa, the grain size is smaller than or equal to 10 microns, the intergranular corrosion resistance is improved by more than two times, the single-piece manufacturing period is smaller than 45 minutes, the yield is larger than 98%, the electromagnetic shielding effectiveness is larger than 90 dB, and the strict requirements of high-end equipment such as nuclear power equipment and spaceflight equipment for the high-reliability shielding sleeve are met.

Description

Technical Field

[0001] This invention belongs to the field of mechanical engineering, and specifically relates to a method for manufacturing Hastelloy shielding sleeves using intelligent micro-strain welding. Background Technology

[0002] With the continuous evolution of high-end equipment manufacturing and special material welding processes, Hastelloy, due to its excellent corrosion resistance and high-temperature stability, is widely used in critical shielding structures in nuclear power, aerospace, and chemical industries. Traditional shielding sleeve manufacturing often employs conventional fusion welding or mechanical connections, which struggle to achieve a balance between high density and structural integrity under micro-strain control. This leads to problems such as intergranular corrosion, thermal stress cracking, and electromagnetic shielding effectiveness degradation during service. Especially under complex operating conditions, the non-uniform distribution of welding heat input and the uncontrollable cooling rate further exacerbate microstructural heterogeneity, posing a severe challenge to the long-term reliability of components.

[0003] However, existing welding processes lack the ability to intelligently control the thermal and strain fields, making it difficult to avoid grain coarsening and residual stress concentration in the weld region. Simultaneously, the elemental segregation behavior of Hastelloy in the high-temperature molten pool lacks real-time sensing and dynamic compensation mechanisms, causing the weld chemical composition to deviate from the design window and weakening its resistance to intergranular corrosion. Furthermore, traditional manufacturing processes rely on offline inspection and empirical parameter settings, failing to achieve online defect suppression and closed-loop performance optimization during welding. This results in large fluctuations in yield and poor consistency, making it difficult to meet the stringent requirements of high-end equipment for the service life and safety redundancy of shielding sleeves.

[0004] Therefore, a method for manufacturing Hastelloy shielding sleeves using intelligent micro-strain welding is desired. Summary of the Invention

[0005] The purpose of this invention is to provide a method for manufacturing Hastelloy shielding sleeves using intelligent micro-strain welding, which can effectively solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for manufacturing a Hastelloy shielding sleeve for intelligent micro-strain welding includes the following specific steps: Step (1) Constructing a multi-physics coupled welding digital twin model: Based on the geometric structure, material thermophysical parameters, and service conditions of the Hastelloy shielding sleeve, a multi-physics coupled model including heat conduction, fluid dynamics, and solid mechanics is established, and element diffusion dynamics equations are embedded to predict the flow behavior of the molten pool, the evolution of the microstructure, and the distribution of residual stress; Step (2) Deploying a high spatiotemporal resolution in-situ sensing system: A multimodal sensing array, including an infrared thermal imager, a laser ultrasonic probe, and a spectral analysis unit, is deployed around the welding area to collect the temperature field, strain field, and spectral signal of the molten pool in real time, with a sampling frequency of not less than 10 kHz and a spatial resolution of 0.1 mm; Step (3) Performing dynamic parameter adjustment based on model predictive control: The in-situ sensing data is input into the digital twin model, and the laser power is optimized online through the model predictive control algorithm. The rate, scanning speed, protective gas flow rate and auxiliary cooling intensity are adjusted to control the heat input gradient within the range of 5 to 15 degrees Celsius per millimeter and the cooling rate within the range of 50 to 200 degrees Celsius per second; Step (4) Implement a closed-loop compensation mechanism for elemental segregation: Based on the concentration deviation of nickel, molybdenum and chromium elements fed back by the spectral analysis unit, dynamically adjust the wire feeding composition or laser energy distribution to control the content deviation of each main element in the weld area within ±0.5% and ensure that the chemical composition is within the anti-intergranular corrosion design window; Step (5) Complete the micro-strain control and performance verification after welding: Immediately after welding, start the local electromagnetic induction heating and ultrasonic impact composite treatment, apply ultrasonic vibration with a frequency of 20 kHz and an amplitude of 10 to 30 micrometers to the weld and heat-affected zone, and apply gradient temperature control through the induction coil to convert the residual tensile stress into compressive stress. Finally, verify the compactness, grain size and electromagnetic shielding effectiveness of the shielding sleeve through online eddy current detection and rapid metallographic analysis.

[0007] Preferably, in step (1), the multiphysics coupling model adopts a finite element-lattice Boltzmann hybrid solution strategy, wherein heat conduction and stress field are discretized using eight-node hexahedral elements, and the molten pool flow is simulated using the lattice Boltzmann method. The time step is set to 1 microsecond, and the mesh size is refined to 0.05 mm in the molten pool region to accurately capture the dynamic evolution process of the molten pool boundary.

[0008] Preferably, in step (2), the infrared thermal imager operates in a wavelength range of 3 to 5 micrometers, has a temperature measurement range of 200 to 2000 degrees Celsius, and a temperature measurement accuracy of ±2 degrees Celsius; the laser ultrasonic probe uses pulsed laser excitation and interferometer reception, which can simultaneously acquire surface strain and internal sound velocity changes for inversion of residual stress distribution; the spectral analysis unit is based on the principle of atomic emission spectroscopy, and collects molten pool plasma radiation through an optical fiber probe with a spectral resolution of 0.1 nanometers, which can identify the characteristic spectral line intensities of nickel, molybdenum, chromium, and iron elements in real time.

[0009] Preferably, in step (3), the model prediction control algorithm takes minimizing the width of the heat-affected zone and the peak value of residual stress as the objective function, and the constraints include the molten pool stability criterion and the upper limit of grain growth rate. The control cycle is 1 millisecond, and a closed loop of perception-modeling-optimization-execution is completed in each cycle to ensure that the welding process is always in the micro-strain stable domain.

[0010] Preferably, in step (4), the wire feeding composition compensation is achieved by a multi-channel precision wire feeding mechanism, each channel is loaded with Hastelloy welding wire of different proportions, and the wire feeding rate of each channel is dynamically adjusted according to the element concentration deviation, with an adjustment response delay of less than 5 milliseconds; the laser energy distribution compensation is achieved by a programmable spatial light modulator, which shapes the laser beam into a ring or bi-peak distribution to suppress the volatilization of elements in the central region.

[0011] Preferably, in step (5), the ultrasonic impact head is made of tungsten carbide, the impact frequency is 20 kHz, the impact force is controlled between 50 and 150 Newtons, and the processing path is scanned along the weld direction at a speed of 2 mm per second; the electromagnetic induction heating uses a multi-turn copper coil, and an alternating current with a frequency of 10 to 50 kHz is passed through it, the temperature gradient of the heating area is controlled within 10 degrees Celsius per millimeter, and the processing time lasts for 30 to 120 seconds.

[0012] Preferably, the online eddy current detection adopts a multi-frequency excitation mode with an excitation frequency range of 10 kHz to 1 MHz. Phase analysis is used to distinguish between surface cracks and subsurface pores, and the detection sensitivity can reach 0.1 mm defects. The rapid metallographic analysis uses a micro-sampling device to drill 0.5 mm diameter micropillar samples in non-critical areas. After electrolytic polishing, a portable electron backscatter diffraction system is used to determine the grain orientation and size. The analysis time is less than 5 minutes.

[0013] Preferably, the method is integrated into a fully automated welding workstation equipped with a six-axis collaborative robot with a repeatability accuracy of ±0.02 mm. The robot end effector integrates a laser welding head, a sensor array, and an ultrasonic impact head. Seamless process switching is achieved through a unified timing controller, and the manufacturing cycle of a single shielding sleeve is less than 45 minutes.

[0014] Preferably, the digital twin model automatically loads the thermophysical parameters of the corresponding material batch based on the batch number of the shielding sleeve before each welding, and adaptively corrects the model parameters through historical welding data. The model prediction error is continuously reduced through a Bayesian update mechanism to ensure long-term manufacturing consistency.

[0015] Compared with the prior art, the present invention has the following beneficial effects: Achieving precise control of micro-strain By deeply integrating multi-physics digital twins with high spatiotemporal resolution in-situ sensing, a collaborative closed-loop control of the thermal field, strain field, and composition field was achieved for the first time in Hastelloy welding. This reduced the peak value of welding residual stress to below 150 MPa and controlled the grain size to within 10 micrometers, which is significantly better than the residual stress above 300 MPa and the grain size above 50 micrometers of traditional processes. This fundamentally suppressed the initiation of thermal stress cracks and intergranular corrosion.

[0016] Ensure the stability of the chemical composition of the weld The closed-loop compensation mechanism for elemental segregation based on real-time spectral feedback controls the fluctuation of the main element content in the weld to within ±0.5%, ensuring the stable precipitation of key phases in Hastelloy (such as γ matrix and Mo-rich phase), improving the resistance to intergranular corrosion by more than 2 times, and meeting the requirements of ASTM G28 standard method A. In contrast, traditional processes often lead to excessive corrosion rates due to component segregation.

[0017] Improve manufacturing efficiency and consistency The fully automated workstation integrates the entire process of sensing, welding, control and verification, with a single-piece manufacturing cycle of less than 45 minutes and a yield rate of over 98%. It improves efficiency by 3 times compared to the traditional process that relies on offline testing and manual parameter adjustment, and reduces the standard deviation of performance fluctuation between batches to within 5%, meeting the stringent requirements of high-end equipment for the high reliability of shielding sleeves.

[0018] Achieving online defect suppression and closed-loop performance optimization By combining model predictive control with ultrasonic-induction composite post-processing, not only can the formation of defects such as porosity and cracks be dynamically suppressed during the welding process, but the residual stress state can also be actively regulated after welding. This ensures that the electromagnetic shielding effectiveness of the shielding sleeve is consistently above 90 dB in the 1 to 10 GHz frequency band, and the service life is predicted to be extended to more than 10 years, providing high safety redundancy for key fields such as nuclear power and aerospace. Attached Figure Description

[0019] Figure 1 This is a flowchart of the overall technical solution of the present invention. Detailed Implementation

[0020] Example 1 Please refer to Figure 1 To make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to specific embodiments.

[0021] Currently, the manufacturing process of Hastelloy shielding sleeves faces technical challenges such as excessive residual welding stress, coarse grains, severe chemical segregation, and unstable post-weld performance. These issues lead to thermal stress cracks and intergranular corrosion defects, affecting the service life and safety of Hastelloy shielding sleeves in high-reliability fields like nuclear power and aerospace. To address these problems, this invention proposes an intelligent micro-strain welding method for manufacturing Hastelloy shielding sleeves, and applies it to the entire intelligent manufacturing process of Hastelloy shielding sleeves.

[0022] In the above-mentioned intelligent micro-strain welding method for manufacturing Hastelloy shielding sleeves, step (1) involves constructing a multi-physics coupled welding digital twin model: based on the geometric structure, material thermophysical parameters, and service conditions of the Hastelloy shielding sleeve, a multi-physics coupled model including heat conduction, fluid dynamics, and solid mechanics is established, and element diffusion dynamics equations are embedded to predict the flow behavior of the molten pool, the evolution of the microstructure, and the distribution of residual stress. Specifically, in step (1), the multi-physics coupled model adopts a finite element-lattice Boltzmann hybrid solution strategy, where heat conduction and stress fields are discretized using eight-node hexahedral elements, and the molten pool flow is simulated using the lattice Boltzmann method. The time step is set to 1 microsecond, and the mesh size is refined to 0.05 mm in the molten pool region to accurately capture the dynamic evolution process of the molten pool boundary. The model first obtains the geometric contour of the shielding sleeve through 3D CAD modeling, including the inner diameter, outer diameter, wall thickness, and end structural features. All geometric data are imported into the simulation platform in STL format. Subsequently, based on the specific grade of Hastelloy C-276 or C-22, its thermophysical and mechanical parameters, including density, specific heat capacity, thermal conductivity, coefficient of thermal expansion, elastic modulus, Poisson's ratio, yield strength, and fracture toughness, were retrieved from the material database. These parameter values ​​were derived from the ASTM E145 standard test report, ensuring data accuracy better than ±3%. Service condition information included the operating temperature range (up to 800 degrees Celsius), pressure rating (up to 10 MPa), type of corrosive medium (such as acidic environments containing chloride ions), and electromagnetic shielding frequency requirements (1 to 10 GHz). These conditions were input as boundary conditions into the model. The model space was divided into a non-uniform mesh, with 1 mm cube elements used for the overall region, while localized mesh refinement was applied to the molten pool and its surrounding heat-affected zone, with a minimum mesh size of 0.05 mm and a total mesh count of approximately 120 million. The heat conduction module employed transient thermal analysis equations.

[0023] in For density, For specific heat capacity, For temperature, Thermal conductivity, The laser heat source term is described using a Gaussian distribution function, and its power density decreases exponentially with spatial position. The fluid dynamics section is based on the incompressible Navier-Stokes equations, incorporating surface tension, gravity, buoyancy, and the Marangoni effect. The lattice Boltzmann method is used to solve for the velocity and pressure fields. This method has a natural advantage in handling complex free surface motions and can accurately simulate the rotation, fluctuation, and splashing phenomena of the molten pool. The solid mechanics module uses elastoplastic constitutive relations, considers temperature dependence, and calculates the residual stress field caused by thermal expansion and contraction. The element diffusion kinetics equations introduce Fick's second law.

[0024] in For the first The concentration of elements (such as Ni, Mo, Cr, Fe), The diffusion coefficient is given, which varies with temperature and is determined by the Arrhenius formula. During model initialization, the initial temperature is set to room temperature (25 degrees Celsius), the initial stress is zero, and the initial elemental distribution is assumed to be uniform in the parent material. The solver employs an explicit time integration method with a fixed time step of 1 microsecond. Each iteration completes the coupled update of the thermal field, flow field, stress field, and composition field. The model output includes the molten pool shape, temperature gradient, velocity vector, stress distribution map, and concentration contour maps of each element. All results are stored in VTK format for subsequent real-time control.

[0025] In the above-mentioned intelligent micro-strain welding method for manufacturing Hastelloy shielding sleeves, step (2) involves deploying a high spatiotemporal resolution in-situ sensing system: a multimodal sensing array is deployed around the welding area, including an infrared thermal imager, a laser ultrasonic probe, and a spectral analysis unit, to collect the molten pool temperature field, strain field, and melt composition spectral signals in real time, with a sampling frequency of not less than 10 kHz and a spatial resolution of 0.1 mm. Specifically, in step (2), the infrared thermal imager operates in a wavelength range of 3 to 5 micrometers, with a temperature measurement range of 200 to 2000 degrees Celsius and a temperature measurement accuracy of ±2 degrees Celsius; the laser ultrasonic probe uses pulsed laser excitation and interferometer reception to simultaneously acquire surface strain and internal sound velocity changes for inverting residual stress distribution; the spectral analysis unit, based on the principle of atomic emission spectroscopy, collects molten pool plasma radiation through an optical fiber probe with a spectral resolution of 0.1 nanometers, and can identify the characteristic spectral line intensities of nickel, molybdenum, chromium, and iron elements in real time. The system consists of three independent but synchronously operating sensors, all mounted on the robot's end effector and maintained in a fixed relative position with the laser welding head to ensure that the measurement field of view always covers the weld pool. The infrared thermal imager, model FLIR A655sc, is equipped with a dual-band detector, with a main wavelength of 3 to 5 micrometers, suitable for high-temperature metal radiation measurement. Its image resolution is 640×512 pixels, the frame rate is set to 10 kHz, and the single-frame exposure time is 100 nanoseconds. Blackbody calibration ensures temperature measurement accuracy. The laser ultrasonic probe consists of an Nd:YAG pulsed laser (wavelength 1064 nm, pulse width 10 nanoseconds, repetition frequency 10 kHz) and a Michelson interferometer. The laser beam is focused on the surface of the weld pool, generating thermoelastic stress waves. The interferometer receives the reflected light and demodulates the surface displacement signal, thereby calculating the surface strain rate. Simultaneously, the internal sound velocity of the material is inverted through the sound wave propagation time difference. Combined with the sound velocity-stress relationship model, online inversion of residual stress is achieved. The spectral analysis unit employs ICP-OES (Inductively Coupled Plasma Optical Emission Spectroscopy) technology, using an argon plasma torch as its light source. The temperature reaches 10,000 degrees Celsius, effectively exciting atoms in the molten pool to transition to excited states and radiate light of specific wavelengths. The fiber optic probe, 1 mm in diameter with a protective window at the tip, is inserted 5 mm above the molten pool. The collected light signal is split by a grating and then enters a CCD detector. The spectral resolution is 0.1 nm, and the scanning range covers the 200-800 nm wavelength band, focusing on monitoring the intensity of characteristic spectral lines such as Ni (343.48 nm), Mo (313.29 nm), Cr (357.87 nm), and Fe (371.99 nm). All sensor data is transmitted to the central control computer via gigabit Ethernet. The data packet includes a timestamp, spatial coordinates, raw signal values, and pre-processed characteristic values, with a time synchronization error of less than 1 microsecond, ensuring precise alignment of multi-source data.

[0026] In the above-mentioned intelligent micro-strain welding method for manufacturing Hastelloy shielding sleeves, step (3) involves dynamic parameter adjustment based on model predictive control: in-situ sensing data is input into a digital twin model, and the laser power, scanning speed, protective gas flow rate, and auxiliary cooling intensity are optimized online through a model predictive control algorithm to control the heat input gradient within the range of 5 to 15 degrees Celsius per millimeter and maintain the cooling rate within the range of 50 to 200 degrees Celsius per second. Specifically, in step (3), the model predictive control algorithm takes minimizing the width of the heat-affected zone and the peak value of residual stress as the objective function, and the constraints include the molten pool stability criterion and the upper limit of grain growth rate. The control cycle is 1 millisecond, and a closed loop of sensing-modeling-optimization-execution is completed in each cycle to ensure that the welding process is always in the micro-strain stable domain. This control process is executed by a high-performance industrial computer, the core of which is a model predictive controller (MPC). At the start of each 1-millisecond control cycle, the system acquires the current molten pool temperature field from an infrared thermal imager, surface strain and internal sound velocity data from a laser ultrasonic probe, and elemental concentration feedback from a spectral analysis unit. These data, after filtering and normalization, are input as the current state vector into the digital twin model. Based on the current state, the model predicts the evolution trends of the thermal field, stress field, and composition field within the next 10 milliseconds, generating a set of possible future trajectories. The objective function of the controller is defined as:

[0027] in For the first The predicted width of the heat-affected zone. The desired width (usually set to 0.5 mm). For the first The peak value of residual stress predicted in step step, The safety threshold is 150 MPa. and The weighting coefficients are set to 0.6 and 0.4 respectively to balance the importance of the two objectives. Constraints include a molten pool stability criterion (maximum molten pool amplitude not exceeding 0.2 mm with no spatter) and a maximum grain growth rate (austenite grain growth rate not exceeding 0.1 μm / s to prevent grain coarsening). Under these constraints, the controller solves for the optimal control sequence that minimizes the objective function. This sequence includes commands for changes in laser power, scanning speed, shielding gas flow rate, and auxiliary cooling intensity over the next 10 milliseconds. Since only the first control action is executed, with the remaining commands used as prediction reserves, rolling optimization is achieved. The optimized control commands are sent via the CAN bus to the laser power supply, servo motor driver, gas flow meter, and cooling system for real-time parameter adjustment. The entire closed-loop process is completed within 1 millisecond, ensuring timely response to rapidly changing welding processes.

[0028] In the above-mentioned intelligent micro-strain welding method for manufacturing Hastelloy shielding sleeves, step (4) implements a closed-loop compensation mechanism for elemental segregation: based on the concentration deviations of nickel, molybdenum, and chromium elements fed back by the spectral analysis unit, the wire feeding composition or laser energy distribution is dynamically adjusted to control the content deviation of each main element in the weld area within ±0.5%, ensuring that the chemical composition is within the design window for resistance to intergranular corrosion. Specifically, in step (4), the wire feeding composition compensation is achieved through a multi-channel precision wire feeding mechanism, with each channel loaded with Hastelloy welding wire of different proportions. The wire feeding rate of each channel is dynamically adjusted according to the elemental concentration deviation, with an adjustment response delay of less than 5 milliseconds. Laser energy distribution compensation is achieved through a programmable spatial light modulator, shaping the laser beam into a ring or bi-peak distribution to suppress element volatilization in the central region. This mechanism includes two parallel compensation paths. The first path involves wire feeding composition compensation, employing a three-channel wire feeding system. Channel A loads welding wire with a Ni content of 58%, channel B loads welding wire with a Mo content of 16%, and channel C loads welding wire with a Cr content of 16%. All three types of welding wire are solid flux-cored wires with a diameter of 1.2 mm. A spectral analysis unit monitors the concentrations of Ni, Mo, and Cr elements in the molten pool in real time, comparing them with preset target values ​​(Ni 58%, Mo 16%, Cr 16%) to calculate the concentration deviation. .like If the value is greater than 0.5%, the wire feed rate of the corresponding channel is increased; conversely, it is decreased. The wire feed rate adjustment is driven by a servo motor and implemented through a PID controller, with a response delay of less than 5 milliseconds, ensuring rapid compensation. The second path is laser energy distribution compensation, employing a programmable spatial light modulator (SLM). This device consists of 1024×768 micromirror units, each of which can be independently tilted, thereby changing the phase distribution of the incident laser. Based on spectral feedback, if increased Mo volatilization is detected in the central region, the SLM reshapes the laser beam into a ring distribution, concentrating energy at the edge of the molten pool and preventing overheating in the center. If Ni enrichment is detected, the laser beam is reshaped into a bimodal distribution, forming two energy peaks to promote molten pool stirring and improve compositional uniformity. The SLM control signal is generated by a computer based on real-time feedback, with an update frequency of 10 kHz, ensuring synchronization between the energy distribution and the molten pool state.

[0029] In the above-mentioned intelligent micro-strain welding method for manufacturing Hastelloy shielding sleeves, step (5) completes post-weld micro-strain control and performance verification: immediately after welding, local electromagnetic induction heating and ultrasonic impact composite treatment are initiated, applying ultrasonic vibration with a frequency of 20 kHz and an amplitude of 10 to 30 micrometers to the weld and heat-affected zone, while simultaneously applying gradient temperature control through an induction coil to convert residual tensile stress into compressive stress. Finally, the compactness, grain size, and electromagnetic shielding effectiveness of the shielding sleeve are verified through online eddy current detection and rapid metallographic analysis. Specifically, in step (5), the ultrasonic impact head is made of tungsten carbide, the impact frequency is 20 kHz, the impact force is controlled between 50 and 150 Newtons, and the processing path is scanned along the weld direction at a speed of 2 mm per second; electromagnetic induction heating uses a multi-turn copper coil, with an alternating current of 10 to 50 kHz, the temperature gradient of the heating area is controlled within 10 degrees Celsius per millimeter, and the processing time lasts for 30 to 120 seconds. This composite treatment process is seamlessly connected after welding. First, an ultrasonic impact head carried by the robot's end effector contacts the weld surface. The impact head is made of tungsten carbide cemented carbide with a hardness exceeding HRA 90, exhibiting strong wear resistance. Connected to an ultrasonic generator, the impact head has a fixed frequency of 20 kHz and an amplitude adjustable between 10 and 30 micrometers. The impact force is controlled by a hydraulic system, ranging from 50 to 150 Newtons. The impact head moves uniformly along the weld direction at a speed of 2 millimeters per second, impacting the weld and the heat-affected zones on both sides point by point. Each impact lasts approximately 0.5 milliseconds, forming a dense plastic deformation layer. Simultaneously, a multi-turn copper induction coil surrounds the weld area, passing an alternating current with a frequency of 10 to 50 kHz, generating an alternating magnetic field that induces eddy currents and heats the weld area. By adjusting the current frequency and amplitude, the temperature gradient in the heated area is controlled to within 10 degrees Celsius per millimeter to prevent localized overheating. The processing time is set according to the weld length, typically ranging from 30 to 120 seconds, ensuring that the entire weld area is adequately treated. This composite treatment introduces a large number of dislocations through ultrasonic vibration, disrupting the original grain boundaries. Simultaneously, gradient temperature control promotes dislocation recrystallization and ultimately transforms residual tensile stress into beneficial compressive stress, improving fatigue life. Immediately after treatment, online monitoring is performed. Online eddy current testing employs a multi-frequency excitation mode, with an excitation frequency range of 10 kHz to 1 MHz. Phase analysis distinguishes between surface cracks and subsurface porosity, achieving a detection sensitivity of up to 0.1 mm defects. The detection probe moves along the weld seam, acquiring impedance signals. Fourier transform is used to extract the phase angle at different frequencies; surface cracks exhibit low-frequency phase abrupt changes, while subsurface porosity exhibits mid-frequency phase drift. Rapid metallographic analysis involves drilling 0.5 mm diameter micropillar samples from non-critical areas using a micro-sampling device. After electrolytic polishing, a portable electron backscatter diffraction system is used to determine grain orientation and size, with an analysis time of less than 5 minutes.This system can quickly generate grain orientation maps (IPF) and grain size distribution histograms to confirm whether the grain size is less than 10 micrometers and whether the grain orientation is random, in order to evaluate the welding quality.

[0030] The method is integrated into a fully automated welding workstation equipped with a six-axis collaborative robot with a repeatability of ±0.02 mm. The robot's end effector integrates a laser welding head, a sensor array, and an ultrasonic impact head. Seamless process switching is achieved through a unified timing controller, with a single shielding sleeve manufacturing cycle of less than 45 minutes. The workstation consists of an ABB IRB 1200 collaborative robot with a payload capacity of 12 kg, a working radius of 1.2 m, and a repeatability of ±0.02 mm, meeting high-precision welding requirements. The robot's end effector quick-change interface integrates a laser welding head (1 kW power, 1070 nm wavelength), a multimodal sensor array (infrared thermal imager, laser ultrasonic probe, spectral analysis unit), and an ultrasonic impact head. All equipment moves collaboratively through the robotic arm, achieving integrated operation of welding, sensing, control, and post-processing. The unified timing controller, based on a real-time operating system (RTOS), coordinates the sequence of actions and time synchronization of each subsystem. For example, before welding begins, the controller first commands the robot to move to the starting point, then starts the laser power supply for preheating, then activates all sensors for self-checking, and finally issues the welding command. During the welding process, the controller schedules sensing, modeling, optimization, and execution tasks at 1-millisecond intervals. After welding is completed, the controller automatically switches to post-processing mode, sequentially performing ultrasonic impact and induction heating. The entire process requires no manual intervention, and the manufacturing cycle for a single shielding sleeve from loading to unloading is less than 45 minutes, with a yield rate consistently above 98%.

[0031] The digital twin model automatically loads the thermophysical parameters of the corresponding material batch based on the shielding sleeve batch number before each welding operation, and adaptively corrects the model parameters using historical welding data. The model prediction error is continuously reduced through a Bayesian update mechanism, ensuring long-term manufacturing consistency. This mechanism is activated before each welding task begins. The system reads the batch number of the shielding sleeve to be processed, which is associated with a thermophysical parameter table provided by the material supplier. The parameter table is stored in a database and contains key parameters such as thermal conductivity, specific heat capacity, and elastic modulus for each batch of materials, with an accuracy better than ±2%. The model automatically loads the parameters of the corresponding batch, replacing the default values, ensuring that the model input is consistent with the actual material. In addition, the system maintains a historical welding database, recording the input parameters (laser power, scanning speed, etc.), sensed data (temperature field, strain field, etc.), and final outputs (residual stress, grain size, etc.) for each welding operation. After each welding operation, the system compares the actual result with the model prediction result and calculates the prediction error. This error is used as observation data and input into the Bayesian update framework to update the prior probability distribution of uncertain parameters in the model, such as the uncertainty range of the heat source efficiency factor and the thermal expansion coefficient of the material. By continuously accumulating data, the model's prediction accuracy has gradually improved, and the long-term prediction error standard deviation has decreased from the initial 15% to below 5%, significantly improving the stability and predictability of the manufacturing process.

[0032] Example 2 To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0033] Currently, the manufacturing process of Hastelloy shielding sleeves faces technical challenges such as excessive residual welding stress, coarse grains, severe chemical segregation, and unstable post-weld performance. These issues lead to thermal stress cracks and intergranular corrosion defects, affecting the service life and safety of Hastelloy shielding sleeves in high-reliability fields like nuclear power and aerospace. To address these problems, this invention proposes an intelligent micro-strain welding method for manufacturing Hastelloy shielding sleeves, and applies it to the entire intelligent manufacturing process of Hastelloy shielding sleeves.

[0034] In the above-mentioned intelligent micro-strain welding method for manufacturing Hastelloy shielding sleeves, step (1) involves constructing a multi-physics coupled welding digital twin model: based on the geometric structure, material thermophysical parameters, and service conditions of the Hastelloy shielding sleeve, a multi-physics coupled model including heat conduction, fluid dynamics, and solid mechanics is established, and element diffusion dynamics equations are embedded to predict the flow behavior of the molten pool, the evolution of the microstructure, and the distribution of residual stress. Specifically, in step (1), the multi-physics coupled model adopts a finite element-lattice Boltzmann hybrid solution strategy, where heat conduction and stress fields are discretized using eight-node hexahedral elements, the molten pool flow is simulated using the lattice Boltzmann method, the time step is set to 1 microsecond, and the mesh size is refined to 0.05 mm in the molten pool region to accurately capture the dynamic evolution process of the molten pool boundary. In this embodiment, the solver architecture is further optimized. GPU acceleration is employed, allocating the finite element solution tasks for heat conduction and stress fields to the NVIDIA A100 GPU. Leveraging its massively parallel computing capabilities, the single iteration time is reduced from 10 milliseconds on a traditional CPU to 0.8 milliseconds, significantly improving real-time performance. Simultaneously, a machine learning agent is introduced. A lightweight neural network is trained, taking the current molten pool temperature and velocity fields as input and outputting a stress field prediction for the next time step. This agent is trained offline using a large amount of simulation data, exhibiting extremely fast inference speed. It can be used to quickly estimate stress evolution trends, serving as an auxiliary prediction tool for MPC and reducing the computational burden on the main model.

[0035] In the above-mentioned intelligent micro-strain welding method for manufacturing Hastelloy shielding sleeves, step (2) involves deploying a high spatiotemporal resolution in-situ sensing system: a multimodal sensing array is deployed around the welding area, including an infrared thermal imager, a laser ultrasonic probe, and a spectral analysis unit, to collect the molten pool temperature field, strain field, and melt composition spectral signals in real time, with a sampling frequency of not less than 10 kHz and a spatial resolution of 0.1 mm. Specifically, in step (2), the infrared thermal imager operates in the 3 to 5 micrometer band, has a temperature measurement range of 200 to 2000 degrees Celsius, and a temperature measurement accuracy of ±2 degrees Celsius; the laser ultrasonic probe uses pulsed laser excitation and interferometer reception to simultaneously acquire surface strain and internal sound velocity changes for inverting residual stress distribution; the spectral analysis unit, based on the principle of atomic emission spectroscopy, collects molten pool plasma radiation through an optical fiber probe with a spectral resolution of 0.1 nanometers, and can identify the characteristic spectral line intensities of nickel, molybdenum, chromium, and iron elements in real time. In this embodiment, the spectral analysis unit is upgraded to a dual-channel configuration. The first channel focuses on high-precision measurement of Ni, Mo, and Cr elements, while the second channel is extended to detect trace impurity elements such as Si, Mn, S, and P, with a concentration threshold set at 0.05% to monitor the potential risk of brittle phase precipitation. The dual-channel data is combined using a multiplexer to ensure time synchronization.

[0036] In the above-mentioned intelligent micro-strain welding method for manufacturing Hastelloy shielding sleeves, step (3) involves dynamic parameter adjustment based on model predictive control: in-situ sensing data is input into a digital twin model, and the laser power, scanning speed, protective gas flow rate, and auxiliary cooling intensity are optimized online through the model predictive control algorithm to control the heat input gradient within the range of 5 to 15 degrees Celsius per millimeter and the cooling rate within the range of 50 to 200 degrees Celsius per second. Specifically, in step (3), the model predictive control algorithm takes minimizing the width of the heat-affected zone and the peak value of residual stress as the objective function, and the constraints include the molten pool stability criterion and the upper limit of grain growth rate. The control cycle is 1 millisecond, and a closed loop of sensing-modeling-optimization-execution is completed in each cycle to ensure that the welding process is always in the micro-strain stable domain. In this embodiment, the MPC algorithm introduces a robust optimization framework, considering sensor noise and model uncertainty, and replaces the variables in the objective function with a combination of expected value and variance, i.e. ,in The risk aversion coefficient is set to 0.3, which allows the controller to pursue optimal performance while maintaining stability, thus avoiding drastic parameter fluctuations caused by small disturbances.

[0037] In the above-mentioned intelligent micro-strain welding Hastelloy shielding sleeve manufacturing method, step (4) implements a closed-loop compensation mechanism for elemental segregation: based on the concentration deviations of nickel, molybdenum, and chromium elements fed back by the spectral analysis unit, the wire feeding composition or laser energy distribution is dynamically adjusted to control the content deviation of each main element in the weld area within ±0.5%, ensuring that the chemical composition is within the design window for resistance to intergranular corrosion. Specifically, in step (4), the wire feeding composition compensation is achieved through a multi-channel precision wire feeding mechanism, with each channel loaded with Hastelloy welding wire of different proportions. The wire feeding rate of each channel is dynamically adjusted according to the elemental concentration deviation, and the adjustment response delay is less than 5 milliseconds. Laser energy distribution compensation is achieved through a programmable spatial light modulator, which shapes the laser beam into a ring or bi-peak distribution to suppress the volatilization of elements in the central region. In this embodiment, the wire feeding mechanism is increased to four channels, and the newly added channel D is loaded with welding wire containing Nb elements to suppress σ phase precipitation. When the spectrum detects that the Nb content is lower than the target value, the wire feeding rate of channel D is automatically increased.

[0038] In the above-mentioned intelligent micro-strain welding method for manufacturing Hastelloy shielding sleeves, step (5) completes post-weld micro-strain control and performance verification: immediately after welding, local electromagnetic induction heating and ultrasonic impact combined treatment are started, applying ultrasonic vibration with a frequency of 20 kHz and an amplitude of 10 to 30 micrometers to the weld and heat-affected zone, while applying gradient temperature control through an induction coil to convert residual tensile stress into compressive stress, and finally verifying the compactness, grain size and electromagnetic shielding effectiveness of the shielding sleeve through online eddy current detection and rapid metallographic analysis. Specifically, in step (5), the impact head of the ultrasonic impact treatment is made of tungsten carbide, the impact frequency is 20 kHz, the impact force is controlled at 50 to 150 Newtons, and the treatment path is scanned along the weld direction at a speed of 2 mm per second; the electromagnetic induction heating uses a multi-turn copper coil, and an alternating current with a frequency of 10 to 50 kHz is passed through it, the temperature gradient of the heating area is controlled within 10 degrees Celsius per millimeter, and the treatment time lasts for 30 to 120 seconds. In this embodiment, the ultrasonic impact head is equipped with a rotatable eccentric wheel, which causes it to oscillate slightly laterally during impact, thereby expanding the plastic deformation area and enhancing the stress release effect.

[0039] The method is integrated into a fully automated welding workstation equipped with a six-axis collaborative robot with a repeatability accuracy of ±0.02 mm. The robot's end effector integrates a laser welding head, a sensor array, and an ultrasonic impact head. Seamless process switching is achieved through a unified timing controller, and the manufacturing cycle for a single shielding sleeve is less than 45 minutes. In this embodiment, the workstation adds an online cleaning unit, which is automatically executed between welding and post-processing to remove welding slag and oxides, ensuring the effectiveness of subsequent processing.

[0040] The digital twin model automatically loads the thermophysical parameters of the corresponding material batch based on the shielding sleeve batch number before each welding operation, and adaptively corrects the model parameters using historical welding data. The model prediction error is continuously reduced through a Bayesian update mechanism, ensuring long-term manufacturing consistency. In this embodiment, the Bayesian update mechanism introduces an online learning mode, allowing the model to update parameters in real time during continuous production without downtime for training, further improving adaptability.

[0041] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for manufacturing a Hastelloy shielding sleeve using intelligent micro-strain welding, characterized in that: Includes the following steps: Step (1) Construct a multi-physics coupled welding digital twin model: Based on the geometry, material thermophysical parameters and service conditions of the Hastelloy shielding sleeve, establish a multi-physics coupled model including heat conduction, fluid dynamics and solid mechanics, and embed element diffusion dynamics equations to predict the flow behavior of the molten pool, microstructure evolution and residual stress distribution. Step (2) Deploy a high spatiotemporal resolution in-situ sensing system: Deploy a multimodal sensing array around the welding area, including an infrared thermal imager, a laser ultrasonic probe and a spectral analysis unit, to collect the temperature field, strain field and melt composition spectral signals of the molten pool in real time, with a sampling frequency of not less than 10 kHz and a spatial resolution of 0.1 mm. Step (3) Perform dynamic parameter adjustment based on model predictive control: Input the in-situ sensing data into the digital twin model, and optimize the laser power, scanning speed, protective gas flow rate and auxiliary cooling intensity online through the model predictive control algorithm, so that the heat input gradient is controlled within the range of 5 to 15 degrees Celsius per millimeter and the cooling rate is maintained at 50 to 200 degrees Celsius per second; Step (4) Implement a closed-loop compensation mechanism for elemental segregation: Based on the concentration deviations of nickel, molybdenum, and chromium elements fed back by the spectral analysis unit, dynamically adjust the wire feeding composition or laser energy distribution to control the content deviation of each main element in the weld area within ±0.5%, ensuring that the chemical composition is within the design window for resistance to intergranular corrosion. Step (5) Complete post-weld micro-strain control and performance verification: Immediately after welding, start local electromagnetic induction heating and ultrasonic impact combined treatment, apply ultrasonic vibration with a frequency of 20 kHz and an amplitude of 10 to 30 micrometers to the weld and heat-affected zone, and apply gradient temperature control through induction coil to convert residual tensile stress into compressive stress. Finally, verify the compactness, grain size and electromagnetic shielding effectiveness of the shielding sleeve through online eddy current detection and rapid metallographic analysis.

2. The method for manufacturing a Hastelloy shielding sleeve using intelligent micro-strain welding according to claim 1, characterized in that: In step (1), the multiphysics coupling model adopts a finite element-lattice Boltzmann hybrid solution strategy. The heat conduction and stress field are discretized using eight-node hexahedral elements, and the molten pool flow is simulated using the lattice Boltzmann method. The time step is set to 1 microsecond, and the mesh size is refined to 0.05 mm in the molten pool region to accurately capture the dynamic evolution process of the molten pool boundary. During model initialization, the initial temperature is set to 25 degrees Celsius, the initial stress is zero, and the initial element distribution is uniformly distributed in the parent material. The solver adopts the explicit time integration method, and each iteration completes the coupled update of the thermal field, flow field, stress field and composition field.

3. The method for manufacturing a Hastelloy shielding sleeve using intelligent micro-strain welding according to claim 1, characterized in that: In step (2), the infrared thermal imager operates in the 3 to 5 micrometer band, with a temperature measurement range of 200 to 2000 degrees Celsius and a temperature measurement accuracy of ±2 degrees Celsius. The laser ultrasonic probe uses pulsed laser excitation and interferometer reception to simultaneously acquire surface strain and internal sound velocity changes, which are used to invert the residual stress distribution. The spectral analysis unit is based on the principle of atomic emission spectroscopy and collects molten pool plasma radiation through an optical fiber probe. The spectral resolution is 0.1 nanometers, and it can identify the characteristic spectral line intensities of nickel, molybdenum, chromium, and iron elements in real time. All sensor data is transmitted to the central control computer via gigabit Ethernet, with a time synchronization error of less than 1 microsecond.

4. The method for manufacturing a Hastelloy shielding sleeve using intelligent micro-strain welding according to claim 1, characterized in that: In step (3), the model predictive control algorithm takes minimizing the width of the heat-affected zone and the peak residual stress as its objective function. The constraints include the melt pool stability criterion and the upper limit of the grain growth rate. The control cycle is 1 millisecond, and a closed loop of perception-modeling-optimization-execution is completed within each cycle. The objective function is defined as: in Set to 0.5 mm. Set to 150 MPa and The values ​​are set to 0.6 and 0.4 respectively; the molten pool stability criterion requires that the maximum amplitude of the molten pool does not exceed 0.2 mm and that no splashing occurs.

5. The method for manufacturing a Hastelloy shielding sleeve using intelligent micro-strain welding according to claim 1, characterized in that: In step (4), the wire feeding composition compensation is achieved through a multi-channel precision wire feeding mechanism. Each channel is loaded with Hastelloy welding wire of different proportions. The wire feeding rate of each channel is dynamically adjusted according to the element concentration deviation, and the adjustment response delay is less than 5 milliseconds. The laser energy distribution compensation is achieved through a programmable spatial light modulator, which shapes the laser beam into a ring or bi-peak distribution to suppress the volatilization of elements in the central region. The wire feeding mechanism adopts a three-channel configuration, which is loaded with solid flux-cored welding wires with Ni content of 58%, Mo content of 16%, and Cr content of 16%, respectively, with a diameter of 1.2 mm.

6. The method for manufacturing a Hastelloy shielding sleeve using intelligent micro-strain welding according to claim 5, characterized in that: The wire feeding mechanism is expanded to a four-channel configuration, with the new channel loading welding wire containing Nb to suppress σ phase precipitation; when the spectrum detects that the Nb content is lower than the target value, the wire feeding rate of this channel is automatically increased; the programmable spatial light modulator is composed of 1024×768 micromirror units, with an update frequency of 10 kHz, and dynamically adjusts the energy distribution of the laser beam according to real-time feedback.

7. The method for manufacturing a Hastelloy shielding sleeve using intelligent micro-strain welding according to claim 1, characterized in that: In step (5), the ultrasonic impact head is made of tungsten carbide, the impact frequency is 20 kHz, the impact force is controlled between 50 and 150 Newtons, and the processing path is scanned along the weld direction at a speed of 2 mm per second; the electromagnetic induction heating uses a multi-turn copper coil, and an alternating current with a frequency of 10 to 50 kHz is passed through it. The temperature gradient of the heating area is controlled within 10 degrees Celsius per millimeter, and the processing time lasts for 30 to 120 seconds; the ultrasonic impact head is equipped with a rotatable eccentric wheel structure, which makes it oscillate laterally during impact, thereby expanding the plastic deformation area.

8. The method for manufacturing a Hastelloy shielding sleeve using intelligent micro-strain welding according to claim 1, characterized in that: The online eddy current detection adopts a multi-frequency excitation mode with an excitation frequency range of 10 kHz to 1 MHz. It distinguishes between surface cracks and subsurface pores through phase analysis, and the detection sensitivity reaches 0.1 mm defects. The rapid metallographic analysis uses a micro-sampling device to drill 0.5 mm diameter micropillar samples in non-critical areas. After electrolytic polishing, the grain orientation and size are determined using a portable electron backscatter diffraction system. The analysis time is less than 5 minutes, confirming that the grain size is less than 10 micrometers and the grain orientation is random.

9. The method for manufacturing a Hastelloy shielding sleeve using intelligent micro-strain welding according to claim 1, characterized in that: The method is integrated into a fully automated welding workstation equipped with a six-axis collaborative robot with a repeatability accuracy of ±0.02 mm. The robot's end effector integrates a laser welding head, a sensor array, and an ultrasonic impact head. Seamless process switching is achieved through a unified timing controller. The workstation also includes an online cleaning unit that automatically removes welding slag and oxides between welding and post-processing. The manufacturing cycle for a single shielding sleeve is less than 45 minutes, and the yield rate is consistently above 98%.

10. The method for manufacturing a Hastelloy shielding sleeve using intelligent micro-strain welding according to claim 1, characterized in that: The digital twin model automatically loads the thermophysical parameters of the corresponding material batch based on the batch number of the shielding sleeve before each welding, and adaptively corrects the model parameters through historical welding data; the model prediction error is continuously reduced through the Bayesian update mechanism, and the long-term prediction error standard deviation is reduced from the initial 15% to below 5%; The Bayesian update mechanism supports online learning mode, allowing the model to update parameters in real time during continuous production without downtime training, ensuring long-term manufacturing consistency.

Images