A method for repairing a thin-walled titanium alloy moving component coating based on laser cladding

Abstract

The present application relates to the technical field of laser cladding, and particularly provides a kind of based on laser cladding's thin-walled titanium alloy moving component coating repair method, comprising the following steps: cleaning the surface of the workpiece to be repaired to obtain the substrate surface that meets the cladding requirement;Adjust the cladding parameter, make laser irradiation to the area to be repaired and melt the substrate to form a molten pool, synchronously send the cladding powder to the molten pool, prepare wear-resistant and corrosion-resistant coating on the substrate surface;Synchronously obtain the temperature field near the molten pool, by controlling the temperature of peripheral inert gas, the workpiece is controlled slow cooling under the protection atmosphere to the cladding workpiece;Based on the thermal-mechanical coupling reduced-order model, the current temperature field and the deformation trend of the workpiece under the cladding parameter are simulated synchronously, and the cladding parameter and the temperature of inert gas are adjusted in real time based on the simulation result;After finishing and polishing treatment to the cladding position, annealing treatment is carried out again, and the workpiece repair is completed;The present application can effectively reduce the deformation amount of workpiece.

Description

Technical Field

[0001] This invention relates to the field of laser cladding technology, and in particular to a method for repairing coatings on thin-walled titanium alloy moving components based on laser cladding. Background Technology

[0002] Given the consistently high humidity, temperature, salt spray, and ultraviolet radiation conditions in tropical open ocean areas, ship flight deck temperatures can reach 70°C, hangar temperatures are generally above 30°C, and relative humidity is typically above 80%. Specialized equipment operates under harsh marine environments characterized by high humidity, salt spray, and intense sunlight. Affected by sea fog, tides, and seawater splashes, critical moving components such as landing gear suffer high-speed wear from prolonged impacts. Furthermore, the high humidity and salinity of the marine environment exacerbate atmospheric corrosion, making them highly susceptible to both corrosion and operational conditions, leading to performance degradation and even service failure. Therefore, a rapid-response repair coating is necessary as a corrosion- and wear-resistant protection strategy.

[0003] TA15 (Ti-6.5Al-2Zr-1Mo-1V) is a titanium alloy with low density, good corrosion resistance and excellent high-temperature mechanical properties. Therefore, it is often used as a key moving component in the aerospace field, such as the retraction and extension mechanism / guide rail sliding mechanism on carrier-based aircraft.

[0004] Existing surface repair technologies for titanium alloys mainly include electroplating, electroless plating, thermal spraying, and conventional laser cladding. Electroplating and electroless plating suffer from defects such as weak coating adhesion and severe environmental pollution; thermal spraying coatings have high porosity, poor density, and insufficient corrosion resistance and fatigue resistance. When conventional laser cladding technology is applied to thin-walled titanium alloy components, it suffers from prominent problems such as difficulty in controlling heat input, easy deformation of the thin-walled substrate, numerous pore and crack defects in the cladding layer, low repair efficiency, and high cost, making it difficult to meet the rapid repair needs of high-precision thin-walled moving parts. Therefore, this application proposes a method for repairing coatings on thin-walled titanium alloy moving parts based on laser cladding. Summary of the Invention

[0005] The purpose of this invention is to provide a method for repairing coatings on thin-walled titanium alloy moving components based on laser cladding, so as to solve the problem that conventional laser cladding technology is prone to workpiece deformation when applied to thin-walled titanium alloy components.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A method for repairing coatings on thin-walled titanium alloy moving components based on laser cladding, the method comprising the following steps: Clean the surface of the workpiece to be repaired to obtain a substrate surface that meets the cladding requirements; Adjust the cladding parameters so that the laser irradiates the area to be repaired and melts the substrate to form a molten pool. Simultaneously, cladding powder is sent to the molten pool to prepare a wear-resistant and corrosion-resistant coating on the surface of the substrate. The temperature field near the molten pool is acquired synchronously by an infrared thermal imager, and the temperature of the surrounding inert gas is controlled to allow the workpiece to be slowly cooled in a controlled manner under a protective atmosphere. The thermo-mechanical coupling reduced-order model constructed based on intrinsic orthogonal decomposition and Galerkin projection method synchronously simulates the deformation trend of the workpiece under the current temperature field and cladding parameters, and adjusts the cladding parameters and the temperature of the inert gas in real time based on the simulation results. After finishing and polishing the cladding area, annealing is performed to complete the workpiece repair.

[0007] Furthermore, the cladding powder is a mixture of TC4 titanium alloy and WC tungsten carbide powder.

[0008] Furthermore, the mass ratio of the TC4 titanium alloy to the WC tungsten carbide is 7:3.

[0009] Furthermore, the nozzle used in the powder feeding step adopts a double-layer protective gas structure. The protective gas injection end of the nozzle adopts two concentric annular nozzles. The inner protective gas protects the molten pool from oxidation, and the temperature of the outer protective gas is lower than that of the inner protective gas.

[0010] Furthermore, while simulating deformation trends, the shape of the current workpiece is also acquired simultaneously. When the difference between the workpiece deformation and the predicted value exceeds a threshold, the simulation parameters, cladding parameters, and inert gas temperature are adjusted in real time based on the difference.

[0011] Furthermore, after the cladding work is completed and before finishing, the outer contour of the workpiece is scanned, and the workpiece is repaired by grinding and deformation correction during finishing.

[0012] Furthermore, the method for real-time adjustment of cladding parameters and inert gas temperature based on digital twin technology includes the following steps: A thermo-mechanical coupled reduced-order model reflecting the temperature field and deformation trend is constructed by using the intrinsic orthogonal decomposition method and the Galerkin projection method. An ensemble Kalman filter algorithm is used to correct the temperature state variables of the reduced-order model in real time, using real-time acquired temperature field data as observations. Using the corrected state as the initial state, the model predictive controller aims to minimize the maximum deformation in the prediction time domain, while constraining the cooling rate within a preset window. It simultaneously solves for and outputs the optimal adjustment values ​​for laser power, scanning speed, and ambient inert gas temperature.

[0013] Furthermore, the steps for constructing a reduced-order model include: Parameter combinations are generated in the parameter space of laser power, scanning speed, cooling gas temperature, spot diameter, and powder feeding rate by Latin hypercube sampling. Transient thermo-mechanical coupling simulations were performed on each set of parameters using a full-order finite element model to generate snapshot datasets of temperature and deformation fields. The snapshot dataset is subjected to mean reduction and singular value decomposition. The first r left singular vectors whose cumulative energy proportion exceeds a preset threshold are extracted as basis functions of the reduced-order model.

[0014] In summary, the present invention has the following advantages compared with the prior art: This invention uses a digital twin-based thermo-mechanical coupled finite element model to simulate and predict the future deformation trend of a workpiece under current process parameters in real time. Based on this, the laser cladding parameters and the temperature of the controllable slow-cooling peripheral inert gas are adjusted online and synchronously. This closed-loop system, which integrates "prediction" and "control", can actively and dynamically regulate the balance of heat input and heat output during the cladding process, thereby controlling the macroscopic thermal deformation of thin-walled workpieces to an extremely low level from the source and reducing the amount of workpiece deformation. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating the steps of a method for repairing coatings on thin-walled titanium alloy moving components based on laser cladding, as disclosed in an embodiment of the present invention.

[0016] Figure 2 This is a schematic diagram of the coaxial laser cladding head used in the method for repairing coatings on thin-walled titanium alloy moving components based on laser cladding disclosed in an embodiment of the present invention.

[0017] Figure 3 for Figure 2 Sectional view of AA.

[0018] Figure 4 This is a flowchart illustrating the steps of model building and closed-loop control in the method for repairing thin-walled titanium alloy moving component coatings based on laser cladding, as disclosed in an embodiment of the present invention.

[0019] Figure label: 100. Core housing; 101. Laser channel; 102. Mounting flange; 103. Connection interface; 200. Powder feeding housing; 201. Powder feeding chamber; 202. Powder feeding interface; 300. Protective gas housing; 301. Protective gas chamber; 302. Protective gas interface; 303. Diverter hole; 400. Cooling gas housing; 401. Cooling chamber; 402. Spiral cooling channel; 500. Interface housing; 501. Cooling interface. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0021] like Figure 1 As shown, one embodiment of the present invention provides a repair method for thin-walled titanium alloy moving component coatings based on laser cladding, used to repair TA15 workpieces, especially thin-walled parts of the workpieces, to reduce deformation during the repair process.

[0022] In this embodiment, the hardware system used for the repair work includes: a fiber laser (laser power 1.5kW), a coaxial laser cladding head, a powder feeder, a five-axis linkage worktable, an infrared thermal imager (frame rate 200Hz, band 3-5μm, temperature measurement range 200-2500℃, pixel resolution 640×512), a refrigerant circulation temperature control unit, a laser profilometer (Z-axis measurement accuracy ±3μm), and an industrial control computer. The industrial control computer is equipped with a digital twin model, an integrated Kalman filter data assimilation module, a model predictive controller, and a multi-sensor data fusion program. The hardware is connected according to the cladding process, and system initialization and calibration are completed. The fiber laser is connected to the coaxial laser cladding head via optical fiber. The end of the coaxial laser cladding head is equipped with an optical structure to shape the laser beam transmitted to the head into a linear beam. The powder feeder is connected to the coaxial laser cladding head via a powder feeding tube to uniformly deliver cladding powder to the cladding area. The powder feeder uses a carrier gas powder feeding method. An infrared thermal imager is installed... The infrared thermal imager, located beside the coaxial laser cladding head, collects real-time temperature field distribution data of the molten pool and heat-affected zone. It is connected to the industrial control computer via a gigabit Ethernet interface, transmitting the collected temperature field data to a multi-sensor data fusion program in real time. A laser profilometer, mounted on one side of the coaxial laser cladding head, measures the shape of the workpiece's cladding area and nearby easily deformable areas in real time. The laser profilometer is also connected to the industrial control computer via a gigabit Ethernet interface. A refrigerant circulation temperature control unit is connected to the coaxial laser cladding head and to the industrial control computer via a gigabit Ethernet interface. The control structures for the laser and powder feeder are also connected to the industrial control computer via a gigabit Ethernet interface.

[0023] The repair method includes the following steps: Step S1: Clean the surface of the workpiece to be repaired to obtain a substrate surface that meets the cladding requirements; This step is a pre-repair cleaning process used to obtain a substrate surface that meets the cladding requirements. Specifically, the surface of the TA15 titanium alloy workpiece to be repaired undergoes degreasing cleaning, oxide layer removal, and surface roughening treatment in sequence.

[0024] The specific operation is as follows: use an organic solvent (such as acetone) to clean and remove surface oil stains, then wipe with anhydrous ethanol and dry thoroughly; select 70-mesh alumina sand particles, and uniformly sandblast the repair area and surrounding area under a sandblasting pressure of 0.5MPa to roughen the surface. Control the surface roughness Ra of the substrate after sandblasting to be within the range of 3.2 to 6.3μm. Finally, confirm that the substrate has no defects such as cracks or delamination through non-destructive testing (ultrasonic non-destructive testing) to obtain a clean roughened surface that meets the requirements of laser cladding.

[0025] Alternatively, use 400-grit, 800-grit, 1200-grit, and 2000-grit sandpaper to pre-treat the surface of the moving component, initially leveling it and gradually refining it to form a base surface. Then, immerse the component in an acidic mixed solution of HF and HNO3 for 2 minutes to remove the natural oxide film on the surface of the component, rinse it with deionized water, and dry it as soon as possible.

[0026] Step S2: Adjust the cladding parameters so that the laser irradiates the area to be repaired and melts the substrate to form a molten pool. Simultaneously, cladding powder is sent to the molten pool to prepare a wear-resistant and corrosion-resistant coating on the surface of the substrate. Specifically, in this step, the pre-treated workpiece is placed on a workbench for positioning. The workbench has a fixture corresponding to the workpiece, which is used to clamp the workpiece in a preset position.

[0027] The cladding powder is a composite powder of 70% TC4 (Ti-6Al-4V) and 30% spherical WC, with a particle size range of 53-150μm (corresponding to mesh sizes of -200 and +325). The WC powder is selected from casting crushed particles that have undergone spheroidization treatment. Before use, the composite powder is vacuum dried at 120℃ for 2 hours and sieved through a vibrating screen (200 mesh + 325 mesh double-layer screen) to ensure uniform powder composition and good flowability (Hall flow meter measurement value ≤18s / 50g).

[0028] The initial cladding process parameters were set as follows: laser power 1.5kW, scanning speed 1.5m / min, powder feeding rate 8.5g / min, laser spot diameter 3mm, single-pass transverse shift 1.5mm, overlap rate 50%, powder-carrying gas flow rate 5L / min, and high-purity argon gas with a purity ≥99.99% was used as the protective gas. A central protective gas flow of 25L / min was provided through a coaxial central nozzle to form an inert gas protective atmosphere around the molten pool to prevent high-temperature oxidation.

[0029] Under the above parameters, the laser irradiates the area to be repaired and melts the surface of the substrate to form a molten pool. At the same time, the composite powder is precisely fed into the molten pool through a coaxial powder feeding method. As the laser beam moves, the molten pool solidifies to form a continuous and dense wear-resistant and corrosion-resistant cladding layer.

[0030] It should be noted that, in the absence of a protective gas channel in the coaxial laser cladding head, a cladding scheme can also be used in an argon chamber environment, with an argon filling time of 5-10 minutes.

[0031] For areas where right-angle regions inside the groove of a workpiece are difficult to clad, a custom-made lateral nozzle can be inserted into the workpiece, and the laser power and spot diameter can be adjusted to prepare a gradient transition coating.

[0032] Step S3: Simultaneously acquire the temperature field near the molten pool using an infrared thermal imager, and control the temperature of the surrounding inert gas to allow the workpiece to be slowly cooled in a controlled manner under a protective atmosphere.

[0033] In this step, during the cladding process, the external inert gas temperature control structure is activated simultaneously to actively intervene in the cooling rate of the cladding layer.

[0034] Specifically, this embodiment adopts the following... Figure 2 The coaxial laser cladding head shown includes a core housing 100, a powder feeding housing 200, a protective gas housing 300, a cooling gas housing 400, and an interface housing 500. The core housing 100 has a stepped shaft structure, and a laser channel 101 is provided at the central axis. The laser channel 101 is a through hole that passes through the core housing 100 along the axis. One end of the core housing 100 is provided with a connection interface 103 for connecting a laser or optical structure, so that the laser beam input into the laser channel 101 can be transmitted to the cladding area along the laser channel 101. The core housing 100 is also provided with a mounting flange 102 near the connection interface 103. The mounting flange 102 is used to connect an external robotic arm to drive the entire cladding head to move in space.

[0035] The powder feeding housing 200 is a stepped housing with a stepped structure on its inner side. The core housing 100 is fixedly connected to the powder feeding housing 200 by a threaded structure. The connection interface 103 is located at the end of the core housing 100 away from the powder feeding housing 200. The inner side of the end of the powder feeding housing 200 away from the connection interface 103 is a conical inner wall. The end of the core housing 100 away from the connection interface 103 is a conical structure. There is a gap between the conical inner wall of the powder feeding housing 200 and the conical structure of the core housing 100, which forms an annular powder feeding channel. A powder feeding interface 202 is provided on the side wall of the powder feeding housing 200. The powder feeding interface 202 is used to connect an external powder feeder. The powder feeding interface 202 communicates with the annular powder feeding channel. There is an annular gap between the large end of the conical structure on the powder feeding interface 202 and the large end of the conical inner wall of the powder feeding housing 200. This gap forms a powder feeding cavity 201. The powder feeding interface 202 is provided in multiple ways and is evenly distributed along the circumference to ensure the uniformity of powder delivery.

[0036] The protective gas housing 300 is a stepped housing fixed to the end of the powder feeding housing 200. The outer wall of the end of the powder feeding housing 200 away from the connection interface 103 is tapered. The protective gas housing 300 is a trumpet-shaped housing. The large end of the protective gas housing 300 is fixed to the powder feeding housing 200 by a threaded structure. An annular protective gas cavity 301 is formed between the inner wall of the protective gas housing 300 and the tapered outer wall of the powder feeding housing 200. The protective gas cavity 301 is used to transport protective gas. A protective gas interface 302 is provided on the side wall of the protective gas housing 300. The protective gas interface 302 is used to connect to an external gas source. The protective gas interface 302 penetrates the inner and outer sides of the protective gas housing 300. Multiple protective gas interfaces 302 are provided and are evenly distributed circumferentially.

[0037] The cooling gas shell 400 has a conical shell structure. The large end of the cooling gas shell 400 is fixedly connected to the protective gas shell 300. An annular cooling cavity 401 is formed between the inner wall of the cooling gas shell 400 and the conical outer wall of the protective gas shell 300. The cooling cavity 401 is used to transport cooling gas. The protective gas shell 300 is provided with a number of diversion holes 303 that communicate with the interior of the protective gas cavity 301. The diversion holes 303 are evenly distributed circumferentially, so that the protective gas entering the protective gas cavity 301 can... The gas enters the cooling chamber 401 through the diversion hole 303. A spiral cooling channel 402 is provided on the outside of the cooling gas shell 400. An interface shell 500 is fixed to the outside of the cooling gas shell 400 via a threaded structure. The interface shell 500 covers the outside of the spiral cooling channel 402, forming a closed channel. A cooling interface 501 is provided on the side wall of the interface shell 500, used to connect to an external cooling gas source. The cooling interface 501 communicates with the spiral cooling channel 402. The cooling gas shell 400 is made of a material with good thermal conductivity, such as copper alloy or stainless steel, ensuring that heat can be quickly transferred to the cooling gas in the spiral cooling channel 402. When the protective gas entering the cooling chamber 401 flows through the inner wall of the cooling gas shell 400, it can transfer heat to the cooling gas shell 400, which is then carried away by the cooling gas in the spiral cooling channel 402, achieving cooling of the outer protective gas.

[0038] Preferably, the inner wall of the cooling gas shell 400 is machined with spherical or cylindrical protrusions to increase the heat exchange area and improve the cooling efficiency.

[0039] In this embodiment, when the coaxial laser cladding head is in use, the fiber laser or optical structure is connected to the connection interface 103, the laser is transmitted to the molten pool through the laser channel 101, the powder feeding interface 202 is connected to the powder feeder, the protective gas interface 302 is connected to the protective gas source, and the cooling interface 501 is connected to the refrigerant circulation temperature control unit.

[0040] Preferably, the inner wall of the protective gas shell 300 is coated with a zirconium oxide heat insulation layer to reduce temperature transfer and reduce the impact of heat on the cooling gas.

[0041] The powder feeding cavity 201, the protective gas cavity 301, and the cooling cavity 401 are provided with annular openings at the ends near the laser channel 101. From the inside to the outside, the openings are the opening of the powder feeding cavity 201, the opening of the protective gas cavity 301, and the opening of the cooling cavity 401, forming a three-layer coaxial annular structure.

[0042] During operation, the protective gas (high-purity argon gas with a purity ≥ 99.99%) enters the coaxial laser cladding head at a total flow rate of 30L / min. Of this, 25L / min is ejected from the opening of the protective gas chamber 301 to form a central protective gas flow that directly covers the molten pool, 5L / min is sent into the cooling chamber 401 through the diversion hole 303, and 5L / min of the peripheral protective gas is ejected from the opening of the cooling chamber 401 and sprayed onto the workpiece.

[0043] During the simultaneous laser cladding and controlled slow cooling process, a coaxial infrared thermal imager is used to acquire real-time thermal images of the temperature distribution of the molten pool and a surrounding 30mm×30mm area at a frame rate of 200Hz. The data is transmitted to an industrial control computer via gigabit Ethernet. The image processing software in the industrial control computer has been pre-calibrated at multiple points using a blackbody furnace (calibration points cover 200℃ to 2000℃, with 100℃ intervals), establishing a calibration curve between pixel grayscale values ​​and absolute temperature. The software calculates each frame of thermal image into an absolute temperature distribution cloud map in real time, and extracts the temperature value of a 10mm long line segment 5mm behind the molten pool along the current laser scanning direction. The temperature change sequence of each point on this line segment over time is recorded. By performing backward difference calculation on this temperature-time sequence, the instantaneous cooling rate (in ℃ / s) at this location is obtained in real time, serving as a key feedback variable for the closed-loop control system. At the same time, the software also calculates the cooling rate distribution cloud map of each pixel in the entire monitoring area, providing richer spatial distribution observation data for the digital twin model.

[0044] Step S4: Based on the intrinsic orthogonal decomposition and Galerkin projection method, the reduced-order model is used to simultaneously simulate the deformation trend of the workpiece under the current temperature field and cladding parameters, and the cladding parameters and the temperature of the inert gas are adjusted in real time based on the simulation results.

[0045] As a specific embodiment of the present invention, in this step, a digital twin system is deployed in the industrial control computer to run in parallel with the physical cladding process. This system forms a real-time closed-loop control operating at a frequency of 10-20Hz through five core stages: model building, data assimilation, trend prediction, collaborative optimization, and instruction execution. Figure 4 As shown, the specific technical solution is as follows: Step S4.1: Construction of the offline training dataset.

[0046] To meet the stringent requirements of real-time control for model computation speed (a single temperature field prediction must be completed within 10ms), this method uses intrinsic orthogonal decomposition (POD) combined with the Galerkin projection method to construct a digital twin reduced-order model (ROM), reducing the degrees of freedom of the high-dimensional finite element model from approximately 500,000 to 25th order.

[0047] A three-dimensional thermo-structural sequential coupled finite element model of a TA15 workpiece laser cladding was established using the ANSYS Workbench platform. The model employed Solid70 hexahedral thermal elements, with the mesh locally refined to 0.2 mm in the cladding region and gradually transitioning to 2 mm in the far field. A Goldak double ellipsoidal moving heat source model was used to simulate the laser thermal input, and the heat source parameters were determined experimentally. The heat source parameters were determined by comparing them with the metallographic morphology of a single cladding section, such as by taking cross-sectional metallographic photographs of the cladding sample and measuring the melt depth, melt width, and heat-affected zone range of the molten pool. The degree of overlap between the finite element calculation of the molten pool morphology and the measured values ​​was used as the optimization objective, and the Nelder-Mead simplex algorithm was used for iterative calibration.

[0048] Thermal boundary conditions include: forced convection on the surface of the cladding layer (determined by the cooling air curtain parameters), natural convection and radiation on the remaining surfaces, and nonlinear changes in material thermal properties (thermal conductivity, specific heat capacity, density) with temperature. The data are from the TA15 alloy property handbook.

[0049] The training dataset was generated by uniformly sampling in the design parameter space using Latin hypercube sampling. The sampling parameters had five dimensions: laser power P∈[1.0,2.0]kW, scanning speed v∈[0.5,2.5]m / min, cooling gas temperature Tgas∈[5,35]℃, spot diameter d∈[2.5,3.5]mm, and powder feeding rate f∈[6.0,10.0]g / min. A single-pass cladding transient simulation was performed under each parameter combination (simulation duration 5s, time step 0.01s). A total of 200 parameter combinations were sampled, generating 200×500=100,000 temperature field snapshots. Each simulation also recorded the Z-axis deformation field of the workpiece after final cooling, which served as training labels for ROM deformation prediction.

[0050] Step S4.2: Construct a reduced-order model based on POD-Galerkin projection, and use the Kriging surrogate model to construct a fast prediction channel from modal coefficients to the maximum deformation of the workpiece.

[0051] The above 100,000 temperature field snapshot matrices are transformed into snapshot matrices. Where N≈500,000 (node ​​degrees of freedom of the FOM), S=100,000 (total number of snapshots), after subtracting the mean from X, the POD basis functions are extracted using singular value decomposition: Take the first r (e.g., r=25) left singular vectors (i.e., the first r columns of U) to form the POD basis function matrix. Among them, the cumulative energy proportion of the first r-order POD modes reaches the threshold (e.g., 99.7%). This is the snapshot matrix after subtracting the mean. It is a left singular vector matrix. It is a singular value diagonal matrix. Let be the POD basis function matrix, and r be the order of the reduced-order model. Let be the i-th order POD basis vector.

[0052] The temperature field T(x,t) is approximated as a linear combination of POD basis functions: ,in, For the temperature field, Given a mean temperature field and a(t) as a reduced-order state variable (modal coefficient), we substitute the above approximation into the transient heat conduction control equation and utilize the Galerkin projection (multiplying both sides of the equation by the left). ), thus obtaining the r-dimensional reduced-order state-space equation: ,in, The time derivative of the modal coefficients. For a reduced-order system matrix, K is the thermal stiffness matrix of the FOM. For the reduced-order control input matrix, To control the input vector (including laser power and cooling gas temperature, etc.). As a constant bias term, the reduced-order state equation takes about 0.5ms to advance in one step (time step 0.01s) on the industrial control computer, which is far below the requirement of a 50ms control cycle.

[0053] In addition to the temperature field, the ROM also needs to predict the Z-axis deformation of the workpiece after cooling. In this embodiment, during the offline stage, 200 sets of simulation results from the training dataset are used to establish the terminal value of the modal coefficient a(t) (i.e., the value at the moment of cooling to room temperature). The mapping relationship from the Z-axis to the Z-axis deformation field is established. Specifically, a deformation prediction surrogate model is constructed using Kriging. This surrogate model employs Gaussian process regression with a Matérn 5 / 2 kernel function. The length-scale hyperparameter θ of the kernel function is determined through maximum likelihood estimation (MLE). The specific method involves using the standardized terminal modal coefficients... Using the maximum Z-axis displacement of the deformation field and the Z-axis deviation of 15 preset measurement points as output labels, and based on 200 sets of training data, the L-BFGS-B algorithm is used to maximize the logarithmic marginal likelihood function to automatically learn the length scale hyperparameters for each dimension. In the training results of this embodiment, the optimal length scale θ values ​​for each input dimension are between 0.3 and 1.8, and the R-values ​​of the training set cross-validation are satisfactory. 2 The value reaches 0.96 or higher.

[0054] Step S4.3: Acquire the real-time temperature field based on the infrared thermal imager, assimilate the data using ensemble Kalman filtering, and solve the cladding parameters based on model prediction.

[0055] During the simultaneous implementation of laser cladding and cooling, real-time temperature field data acquired by an infrared thermal imager and interlayer profile data from a paraxial laser profilometer are used as observation information. The state variables of the ROM are continuously corrected through ensemble Kalman filtering, and the online optimization of process parameters is driven by a model predictive control framework, as detailed below: The infrared thermal imager continuously acquires two-dimensional infrared radiation images (640×512 pixels) of the cladding area at a frame rate of 200Hz. Each frame is first processed by non-uniformity correction (NUC) and defect replacement, and then converted into two-dimensional temperature field data according to the radiance-temperature conversion curve calibrated by the infrared thermal imager at the factory. Temperature field data are mapped onto the finite element mesh nodes of the ROM through spatial bilinear interpolation, forming observation vectors. (m is the number of FOM grid nodes corresponding to the field of view of the infrared thermal imager). Let k be the observation vector at time k. Let be an m-dimensional real vector space.

[0056] The EnKF set size Ne = 30 members is set, and each member corresponds to a perturbation implementation of the ROM mode coefficient a. Assimilation is performed with a period of 50 ms (consistent with the control period). In each assimilation period k, the ROM state equation is first used to predict each member i in the set: ,in, Let represent the predicted state (prior estimate) of the i-th set member at time k. Let Δt be the analysis state (posterior estimate) of the i-th member at the previous time step, and let Δt be the time step, which is 0.05s (50ms). The noise is a process noise that follows a Gaussian distribution. , Let be the process noise covariance matrix.

[0057] After the forecast is completed, the ensemble forecast mean is calculated. And ensemble forecast covariance Then, using the observation data from the infrared thermal imager at the current moment... The update is performed, where the Kalman gain is... In the formula, Here is the Kalman gain matrix. Let R be the observation operator (mapping the state to the observation space), R be the observation error covariance matrix, and the state update equation be... , Let i be the analysis state of the i-th member at time k. To observe the disturbance, The set mean obtained from the analysis This is the optimal estimate of the ROM state at the current moment, which serves as the initial state for MPC control.

[0058] In this embodiment, the observation operator H is a selection matrix with dimensions *m×N, where m is the number of effective observed pixels within the infrared thermal imager's field of view, and N is the total number of mesh nodes in the full-order finite element model (N≈500,000). H is constructed as follows: first, the pixel coordinates of the infrared thermal imager are mapped to the FOM mesh node coordinates using spatial bilinear interpolation, determining the nearest neighbor node number for each infrared pixel in the FOM mesh; then, an m×n selection matrix H is constructed, where... The condition is met if and only if the i-th infrared observation node corresponds to the j-th node coordinate of the FOM mesh, and all other elements are 0, meaning each row has only one 1 and the rest are all 0. Within the ROM framework, the observation operator further maps the FOM node temperatures to the ROM modal coefficient space: , is the POD basis function matrix.

[0059] It should be noted that, in order to prevent the set from collapsing after multiple assimilations, covariance inflation (inflation factor δ=0.02) is applied after each analysis step, and distance-based localization is applied at the same time to eliminate the influence of long-distance spurious correlations in the high-dimensional state space.

[0060] The MPC operates with a control cycle of 50ms. In each cycle k, the controller solves the following finite-time optimal control problem. Control variable vector: ,in, Laser power adjustment (W). This represents the scanning speed adjustment amount (mm / s). Set the cooling gas temperature (°C). All three variables are subject to amplitude and rate constraints: ; ; Prediction time domain Np = 6 steps (i.e., 300ms), control time domain Nc = 3 steps, optimize the objective function: In the formula, The cost function (objective function) is used. To predict the maximum Z-axis deformation of the workpiece at the end of the time domain, This is the control increment vector for the (k+i)th future step. , The weighting coefficients are w1 and w2. The initial value of w1 is 100 and the initial value of w2 is 2. When the control command fluctuates too much, w2 is increased; when the deformation suppression effect is found to be insufficient, w1 is increased.

[0061] The following hard constraints are set: Cooling rate constraint: The average cooling rate of the cladding layer in the temperature range of 800-500℃ shall not exceed 410℃ / s, which is the critical cooling rate of TA15 titanium alloy. Exceeding this rate will result in the formation of a large amount of acicular martensite, leading to material embrittlement; Melt pool peak temperature constraint: 2400℃≤ ≤3000℃ to ensure that the powder melts completely and the substrate is not excessively burned; the overlap rate is kept under control: the deviation of the actual overlap rate from the expected value of 50% shall not exceed ±5%.

[0062] The optimization problem employs an interior-point solver (IPOPTv3.14 open-source library, compiled and optimized in C++, and linked with the MA57 linear solver). A warm-start strategy is used. The time taken for a single solution is in the range of 15–25 ms, stably meeting the real-time requirement of a 50 ms control cycle.

[0063] After each control cycle is completed, the MPC immediately sends the optimized control commands to the laser control board and the refrigerant circulation temperature control unit via the real-time fieldbus. The laser power and scanning speed adjustment commands are sent via the EtherCAT bus, with a command execution delay of less than 5ms. The inert gas temperature setpoint command is sent via the RS485 bus, with a temperature adjustment response time of about 3 to 5 seconds. The end-to-end delay from the completion of the MPC optimization solution to the start of the execution of the laser power / scanning speed command is controlled within 15ms.

[0064] Step S4.4: Generate closed-loop compensation based on interlayer geometric deviation of laser profilometer.

[0065] After each single-pass or single-layer cladding is completed, a laser profilometer scans along five preset measurement paths on the workpiece surface to obtain the current actual contour data of the workpiece. The measured contour is compared with the target contour (i.e., the original design contour of the part) to obtain the deviation field e(x,y). If the maximum deviation field is greater than the threshold (i.e., there is a non-negligible geometric deviation), the deviation field is used as the Z-axis geometric correction amount and directly superimposed on the offline planning path of the next cladding layer for Z-axis compensation. At the same time, the gradient information of the deviation field along the scanning direction is used... Adjust the overlap ratio of each area on the next lower level according to the following rules: If If e > 0 and e > 0 (local thickening of the protrusion), then the overlap rate of this area is reduced by 5%; if If e > 0 and e < 0 (local depression thinning), then the overlap rate of this area is increased by 5%; if | If | < threshold (take 10% of the maximum gradient value), then maintain the original overlap rate.

[0066] It should be noted that e(x,y) is the Z-axis deviation between the measured profile and the target profile at coordinates (x,y); Let x be the partial derivative of the deviation along the scanning direction x.

[0067] Step S5: After finishing and polishing the cladding area, perform annealing to complete the workpiece repair.

[0068] Specifically, in this step, after all the cladding layers are prepared, the workpiece, along with the non-contact air-floating support platform, is moved into a dedicated slow-cooling chamber. The slow-cooling chamber is pre-filled with high-purity argon gas with a purity of ≥99.99% to maintain an inert protective atmosphere (oxygen content monitored ≤50ppm). The temperature control system inside the chamber cools the workpiece uniformly and slowly from approximately 200℃ (the residual temperature of the workpiece after cladding) to room temperature (25±5℃) at a rate of 8℃ / min to prevent thermal stress cracks caused by rapid cooling.

[0069] After slow cooling, the cladding area is subjected to grinding and polishing processes in sequence: a diamond grinding wheel (120# grit) is used for rough grinding with a single feed rate of 0.02mm; then silicon carbide sandpaper is used for fine grinding in stages; finally, diamond polishing paste (W2.5 grit) is used in conjunction with a cloth polishing pad for polishing to make the surface roughness Ra of the repaired area ≤ 0.8μm.

[0070] Finally, the finished workpiece undergoes vacuum annealing to completely eliminate residual stress: the workpiece is placed in a vacuum annealing furnace and a vacuum of 1×10⁻⁶ is applied. -3 Below Pa, heat to 820°C at a heating rate of 10°C / min, hold for 1.5 hours, and then cool to room temperature with the furnace.

[0071] After the above complete repair process, the repaired part is subjected to comprehensive performance testing: The microhardness of the cladding layer was measured using a Vickers hardness tester with a load of 0.8 kgf and a holding time of 10 seconds. Ten points along the depth direction of the coating section were measured and the average value was taken. The result was 1150 HV0.2, which is about 230% higher than the hardness of the TA15 substrate (about 350 HV).

[0072] The wear resistance of the coating was tested using a pin-disc friction and wear tester. The grinding material was GCr15 steel ball, the load was 10N, the sliding speed was 0.1m / s, and the sliding distance was 500m. The specific wear rate of the repair coating was 2.1×10⁻⁶. -6 mm 3 / (N·m), compared to the specific wear rate of the unrepaired TA15 substrate (3.5×10). -6 mm 3 / (N·m)) decreased by 40%.

[0073] The corrosion resistance of the coating was tested under neutral salt spray according to GB / T10125 standard. The test conditions were: temperature inside the spray chamber 35±2℃, sodium chloride solution concentration 50±10g / L, pH value 6.5~7.2, and continuous spraying for 720 hours. After the test, no corrosion defects such as pitting, blistering, or cracking were observed on the coating surface, and the corrosion resistance was significantly better than that of the TA15 substrate.

[0074] The overall deformation of the workpiece was detected using a coordinate measuring machine. Fifteen evenly distributed measurement points were selected on the workpiece surface to measure the Z-axis coordinate deviation before and after repair. The results showed that the maximum deformation was 0.04 mm, which is within the preset control target range of ±0.05 mm, meeting the high-precision dimensional retention requirements for aerospace moving components.

[0075] The method disclosed in this invention deeply integrates a digital twin reduced-order model, ensemble Kalman filter data assimilation, model predictive control multivariate collaborative optimization, and an inert gas temperature-controlled slow cooling structure to form a dual-channel real-time closed-loop control system for heat input and heat output. This achieves, for the first time, high-precision active suppression of laser cladding deformation in 2mm ultrathin-walled titanium alloy moving parts. All algorithm construction processes, hardware structure design basis, key parameter selection criteria, and system timing architecture involved in the method have been clearly and completely explained. Those skilled in the art can fully reproduce the technical solution of this invention and achieve the stated technical effects based on the above disclosure.

[0076] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0077] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The singular forms “a,” “the,” and “the” used in this invention and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0078] It should be understood that although the terms first, second, third, etc., may be used in this invention to describe various information, this information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first information may also be referred to as second information without departing from the scope of this invention, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to a determination."

[0079] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for repairing coatings on thin-walled titanium alloy moving components based on laser cladding, characterized in that, The repair method includes the following steps: Clean the surface of the workpiece to be repaired to obtain a substrate surface that meets the cladding requirements; Adjust the cladding parameters so that the laser irradiates the area to be repaired and melts the substrate to form a molten pool. Simultaneously, cladding powder is sent to the molten pool to prepare a wear-resistant and corrosion-resistant coating on the surface of the substrate. The temperature field near the molten pool is acquired synchronously by an infrared thermal imager, and the temperature of the surrounding inert gas is controlled to allow the workpiece to be slowly cooled in a controlled manner under a protective atmosphere. The thermo-mechanical coupling reduced-order model constructed based on intrinsic orthogonal decomposition and Galerkin projection method synchronously simulates the deformation trend of the workpiece under the current temperature field and cladding parameters, and adjusts the cladding parameters and the temperature of the inert gas in real time based on the simulation results. After finishing and polishing the cladding area, annealing is performed to complete the workpiece repair.

2. The method for repairing coatings on thin-walled titanium alloy moving components based on laser cladding according to claim 1, characterized in that, The cladding powder is a mixture of TC4 titanium alloy and WC tungsten carbide powder.

3. The method for repairing coatings on thin-walled titanium alloy moving components based on laser cladding according to claim 2, characterized in that, The mass ratio of the TC4 titanium alloy to the WC tungsten carbide is 7:

3.

4. The method for repairing the coating of thin-walled titanium alloy moving components based on laser cladding according to claim 1, characterized in that, The nozzle used in the powder feeding step adopts a double-layer protective gas structure. The protective gas injection end of the nozzle adopts two concentric annular nozzles. The inner protective gas protects the molten pool from oxidation, and the temperature of the outer protective gas is lower than that of the inner protective gas.

5. The method for repairing coatings on thin-walled titanium alloy moving components based on laser cladding according to claim 1, characterized in that, When simulating deformation trends, the shape of the current workpiece is also acquired simultaneously. When the difference between the workpiece deformation and the predicted value exceeds a threshold, the simulation parameters, cladding parameters, and inert gas temperature are adjusted in real time based on the difference.

6. The method for repairing coatings on thin-walled titanium alloy moving components based on laser cladding according to claim 1, characterized in that, After the cladding work is completed and before finishing, the outer contour of the workpiece is scanned. During finishing, the workpiece is repaired by grinding and correction of deformation.

7. The method for repairing coatings on thin-walled titanium alloy moving components based on laser cladding according to any one of claims 1-6, characterized in that, The method for real-time adjustment of cladding parameters and inert gas temperature based on digital twin technology includes the following steps: A thermo-mechanical coupled reduced-order model reflecting the temperature field and deformation trend is constructed by using the intrinsic orthogonal decomposition method and the Galerkin projection method. An ensemble Kalman filter algorithm is used to correct the temperature state variables of the reduced-order model in real time, using real-time acquired temperature field data as observations. Using the corrected state as the initial state, the model predictive controller aims to minimize the maximum deformation in the prediction time domain, while constraining the cooling rate within a preset window. It simultaneously solves for and outputs the optimal adjustment values ​​for laser power, scanning speed, and ambient inert gas temperature.

8. The method for repairing coatings on thin-walled titanium alloy moving components based on laser cladding according to claim 7, characterized in that, The steps to construct a reduced-order model include Parameter combinations are generated in the parameter space of laser power, scanning speed, cooling gas temperature, spot diameter, and powder feeding rate by Latin hypercube sampling. Transient thermo-mechanical coupling simulations were performed on each set of parameters using a full-order finite element model to generate snapshot datasets of temperature and deformation fields. The snapshot dataset is subjected to mean reduction and singular value decomposition. The first r left singular vectors whose cumulative energy proportion exceeds a preset threshold are extracted as basis functions of the reduced-order model.

Images