Laser Cladding Leveling Control System and Method

Abstract

This invention relates to the field of laser cladding additive manufacturing technology, and provides a laser cladding smoothing control system, including a laser, a laser cladding processing head, a robot, a worktable, a positioner, a powder feeder, a protective gas device, a vision acquisition unit, a laser cladding system controller, and a computer system. The computer system performs image processing and feature extraction on real-time molten pool images, inputs the data to the MPC controller, and outputs process parameter adjustment values. The laser cladding system controller sends corresponding process parameter adjustment commands to the laser, robot, and powder feeder according to the process parameter adjustment values, adjusting the laser power, laser scanning speed, and powder feeding rate in real time. This invention allows for online monitoring of various molten pool characteristics and real-time feedback based on a predictive control model. By adjusting various process parameters in real time, it compensates for geometric errors during processing, resulting in a uniformly smooth surface on the formed parts, improving the surface quality and dimensional accuracy of the laser-clad workpiece.

Description

Technical Field

[0001] This invention relates to the field of metal laser cladding additive manufacturing technology, and in particular to process control technology for cladding processes, specifically to a laser cladding planarization control system and method. Background Technology

[0002] Metal laser cladding forms metal structural parts by depositing molten metal powder layer by layer from the bottom up. It can directly generate complex and dense structural parts from computer model data, and has broad application prospects in industries such as intelligent manufacturing, aerospace, and petrochemicals. However, in the complex laser-powder-gas coupling process, geometric deviations and cumulative errors caused by thermal effects can seriously affect the geometric accuracy and flatness of the parts. In particular, the deposition process of complex melt channels is prone to producing uneven surfaces and inconsistent widths, which reduces the repeatability and quality of metal laser cladding products.

[0003] To achieve flatness in machined parts, existing technologies have attempted to improve it in various ways, such as by using molten pool vibration to homogenize the molten pool and thus improve flatness, by using post-processing processes such as milling to improve flatness, by using laser remelting technology and by modifying process parameters at specific locations in the program for specific parts, and by improving the geometric characteristics of the produced parts through closed-loop control to improve flatness.

[0004] In the prior art, patent application CN204224703U discloses a three-dimensional measuring device for laser cladding. It features a laser cladding mechanism, a laser milling mechanism, and a contour measuring instrument arranged in a dispersed manner above a movable work platform. By detecting the three-dimensional shape information of the workpiece, it achieves online shaping of the workpiece. However, the equipment is costly and has large measurement errors. Patent application CN104807410A discloses a laser cladding rapid prototyping layer height measuring device and closed-loop control method. It uses three laser 2D displacement sensors to convert image signals into cladding layer height data, enabling real-time control of the single-layer lifting amount of the cladding head. However, its complex structure is unsuitable for additive manufacturing of complex structural parts, and using the average cladding layer height as the layer height value results in uneven surfaces being compensated by the same parameter, failing to meet the required effect. For example, patent application CN108247059A discloses a layer height control system and method for a coaxial powder-feeding laser melting forming equipment. This system controls the layer height during processing by comparing the laser spot diameter with a process database and calculating the resulting value to determine the lifting height of the coaxial laser nozzle. However, the system relies solely on the laser spot diameter for its side-axis camera mounting, resulting in insufficient accuracy and reliability.

[0005] Existing technologies are often time-consuming, labor-intensive, costly, and unstable. They may also suffer from limited information acquisition, susceptibility to external environmental interference, inability to guarantee measurement accuracy, and poor robustness. If the control index is singular, it may improve the quality of part forming to some extent, but it is far from meeting the process requirements for smoothing the surface of the part. Summary of the Invention

[0006] The purpose of this invention is to provide a laser cladding leveling control system and method, which can monitor various characteristic information of the molten pool online and make real-time feedback based on a predictive control model. By adjusting various process parameters in real time, it can compensate for geometric errors in the processing, so as to make the surface of the formed parts uniformly level and improve the surface quality and dimensional accuracy of the laser cladding workpiece.

[0007] A first aspect of the present invention provides a laser cladding planarization control method, comprising:

[0008] Step 1: Obtain the molten pool image stream information of metal laser cladding additive manufacturing;

[0009] Step 2: Select the region of interest from multiple frames of molten pool images based on the molten pool image stream information;

[0010] Step 3: Based on the region of interest, use the surface fitting method to fit the ROI melt pool image, and obtain the fitting formula coefficients and goodness-of-fit statistics through fitting.

[0011] Step 4: Perform dimensionality reduction on the obtained coefficients of multiple fitting formulas and goodness-of-fit statistics, and determine the n key features that affect the molten pool as flattening feature vectors.

[0012] Step 5: Using the flattening feature vector and the laser cladding process parameter vector as the dataset, and based on the MATLAB system identification toolbox, perform multi-input multi-output system identification to obtain the time constant, steady-state gain, time delay, and corresponding state-space model of the relationship between the flatness feature parameters and the corresponding process parameter data; wherein the time constant, steady-state gain, and time delay correspond to a set of state coefficients of the state-space model; wherein the process parameters include laser power, laser scanning speed, and powder feeding rate;

[0013] Step 6: Using the state-space model as the structure of the model predictive controller, establish the model predictive controller, whose input is the flatness characteristic parameter and whose output is the adjustment amount of the laser cladding process parameter, and configure the control parameters of the model predictive controller.

[0014] Step 7: During the metal laser cladding additive manufacturing process, real-time acquisition of molten pool image data is performed, planarization feature vectors are extracted, and the extracted planarization feature vectors are used as input to the model predictive controller. The model predictive controller outputs process parameter adjustment amounts and sends them to the laser cladding system controller; and

[0015] Step 8: The laser cladding system controller sends the corresponding process parameter adjustment instructions to the laser, robot and powder feeder respectively according to the process parameter adjustment amount, and adjusts the laser power, laser scanning speed and powder feeding rate in real time.

[0016] A second aspect of the present invention provides a laser cladding leveling control system, comprising a laser, a laser cladding processing head, a robot, a worktable, a positioner, a powder feeder, a protective gas device, a vision acquisition unit, a laser cladding system controller, and a computer system.

[0017] The powder feeder is used to feed powder onto the surface of the workbench;

[0018] The protective gas device is configured to deliver protective gas to the surface of the worktable through the laser cladding head;

[0019] The laser is used to emit a laser beam, which is shaped by the optical system in the laser cladding head to form a laser spot on the surface of the worktable, and then melts and deposits the powder on the surface of the worktable.

[0020] The worktable is mounted on the positioner and is configured to move with the positioner.

[0021] The vision acquisition unit includes at least one image acquisition device for acquiring images of the molten pool located on the surface of the worktable.

[0022] The laser cladding head is mounted on the robot and can be driven by the robot to move and change position.

[0023] The laser cladding system controller is used to control the operation of the laser, laser cladding processing head, robot, worktable, positioner, powder feeder, and protective gas device.

[0024] The visual acquisition unit is connected to the computer system and sends the acquired molten pool image to the computer system;

[0025] The computer system is provided with at least one processor and at least one memory. The at least one memory stores operable instructions, which, when executed by the one or more processors, cause the one or more processors to perform operations, including the aforementioned laser cladding leveling control method.

[0026] As an optional implementation, the at least one image acquisition device is a CCD camera, the optical axis of which is mounted perpendicular to the central axis of the laser cladding head, and the CCD camera acquires the image of the molten pool through a 45° reflector set inside the laser cladding head.

[0027] As an optional implementation, the laser cladding system controller is connected to the laser, robot, powder feeder, and protective gas device via I / O, and is configured to control the laser power, powder feeding rate, laser scanning speed, and protective gas flow rate.

[0028] Therefore, according to the laser cladding leveling control system of the above embodiments of the present invention, the control method based on the system can monitor the characteristic information of the molten pool online and make real-time feedback based on the predictive control model. By adjusting multiple process parameters in real time, it can compensate for the geometric error at each point in the processing, so as to make the surface of the formed part uniformly level and improve the surface quality and dimensional accuracy of the laser cladding workpiece.

[0029] It should be understood that all combinations of the foregoing concepts and the additional concepts described in more detail below may be considered part of the inventive subject matter of this disclosure, provided that such concepts do not contradict each other. Furthermore, all combinations of the claimed subject matter are considered part of the inventive subject matter of this disclosure.

[0030] The foregoing and other aspects, embodiments, and features of the teachings of the present invention will be more fully understood from the following description in conjunction with the accompanying drawings. Other additional aspects of the invention, such as features and / or beneficial effects of exemplary embodiments, will become apparent from the following description or may be learned through practice of specific embodiments according to the teachings of the present invention. Attached Figure Description

[0031] The accompanying drawings are not intended to be drawn to scale. In the drawings, each identical or nearly identical component shown in the various figures may be denoted by the same reference numeral. For clarity, not every component is labeled in each figure. Embodiments of various aspects of the invention will now be described by way of example and with reference to the accompanying drawings.

[0032] Figure 1 This is a schematic diagram of the laser cladding leveling control system according to an embodiment of the present invention.

[0033] Figure 2 This is a flowchart illustrating the laser cladding leveling control method according to an embodiment of the present invention.

[0034] Figure 3 This is a schematic diagram of a molten pool image obtained by surface fitting according to an embodiment of the present invention.

[0035] Figure 4This is a schematic diagram illustrating how an embodiment of the present invention uses experiments to determine the correlation between characteristic parameters and changes in molten pool characteristics.

[0036] Figure 5 This is a schematic diagram illustrating the cumulative impact of the PCA-based method on the characteristics of the molten pool in an embodiment of the present invention.

[0037] Figure 6 This is a schematic diagram of the configuration model predictive controller control parameters according to an embodiment of the present invention.

[0038] The meanings of the various reference numerals in the attached figures are defined as follows:

[0039] 10-Laser; 11-Powder feeder; 12-Protective gas device; 13-Laser cladding head; 14-Robot; 15-Reflector; 16-CCD camera;

[0040] 20 - Laser cladding system controller; 30 - Computer system;

[0041] 100 - Workbench; 110 - Positioner. Detailed Implementation

[0042] To better understand the technical content of the present invention, specific embodiments are described below in conjunction with the accompanying drawings.

[0043] Various aspects of the invention are described in this disclosure with reference to the accompanying drawings, which illustrate numerous illustrative embodiments. The embodiments of this disclosure are not necessarily intended to encompass all aspects of the invention. It should be understood that the various concepts and embodiments described above, as well as those described in more detail below, can be implemented in any of many ways, because the concepts and embodiments disclosed herein are not limited to any particular implementation. Furthermore, some aspects of the invention disclosed may be used alone or in any suitable combination with other aspects of the invention disclosed.

[0044] Combination Figure 1 The laser cladding leveling control system of the embodiment shown is characterized by including a laser 10, a powder feeder 11, a protective gas device 12, a laser cladding processing head 13, a robot 14, a vision acquisition unit, a laser cladding system controller 20, a computer system 30, and a workbench 100 and a positioner 110.

[0045] Powder feeder 11 is used to feed powder to the surface of worktable 100.

[0046] The protective gas device 12 is configured to deliver protective gas to the surface of the worktable 100 via a protective gas nozzle, which is integrated into the laser cladding head. The protective gas is, in particular, argon.

[0047] Laser 10 is used to emit a laser beam. After beam shaping by the optical system in the laser cladding head, a laser spot is formed on the surface of the worktable to melt and deposit the powder on the surface of the worktable 100.

[0048] The worktable 100 is mounted on the positioner 110 and is configured to move with the positioner.

[0049] The vision acquisition unit includes at least one image acquisition device, such as a CCD camera 16, for acquiring images of the molten pool on the surface of the worktable.

[0050] The laser cladding head 13 is mounted on the robot 14 and can be driven by the robot 14 to move and change positions, so that the robot 14 can move according to a predetermined trajectory and scanning speed under the control of the laser cladding system controller 20.

[0051] In an optional embodiment, robot 14 may be a commercially available multi-shutdown industrial robot.

[0052] In an embodiment of the present invention, the laser cladding system controller 20 may be an industrial-grade PLC control system used to control the operation of the laser 10, powder feeder 11, protective gas device 12, laser cladding processing head 13, robot 14 and positioner 110.

[0053] like Figure 1 As shown, the visual acquisition unit is connected to the computer system 30 and sends the acquired molten pool image to the computer system 30.

[0054] The computer system 30 is equipped with at least one processor and at least one memory. The at least one memory stores operable instructions. When the instructions are executed by one or more processors, they cause one or more processors to perform operations, such as image processing, feature extraction, and model prediction. The controller outputs the process parameter adjustment amount of laser cladding and sends it to the laser cladding system controller 20.

[0055] The laser cladding system controller 20 sends the corresponding process parameter adjustment instructions to the laser 10, robot 14 and powder feeder 11 respectively according to the process parameter adjustment amount, and adjusts the laser power, laser scanning speed and powder feeding rate in real time.

[0056] As an optional implementation, the aforementioned image acquisition device is a CCD camera. The optical axis of the CCD camera 16 is installed in a direction perpendicular to the central axis of the laser cladding head. The CCD camera 16 acquires the image of the molten pool through a 45° reflector 15 disposed inside the laser cladding head.

[0057] As an optional implementation, the CCD camera 16 is configured to connect to a computer system via a GigE interface to achieve real-time transmission of image information.

[0058] The laser cladding system controller 20 is connected to the laser 10, powder feeder 11, protective gas device 12, and robot 14 via I / O, and can control the laser power, powder feeding rate, laser scanning speed, and protective gas flow rate.

[0059] Combined with the present invention Figure 2 As shown, the process of the laser cladding planarization control method according to an embodiment of the present invention includes:

[0060] Step 1: Obtain the molten pool image stream information of metal laser cladding additive manufacturing;

[0061] Step 2: Select the region of interest from multiple frames of molten pool images based on the molten pool image stream information;

[0062] Step 3: Based on the region of interest, use the surface fitting method to fit the ROI melt pool image, and obtain the fitting formula coefficients and goodness-of-fit statistics through fitting.

[0063] Step 4: Perform dimensionality reduction on the obtained coefficients of multiple fitting formulas and goodness-of-fit statistics, and determine the n key features that affect the molten pool as flattening feature vectors.

[0064] Step 5: Using the flattening feature vector and the laser cladding process parameter vector as the dataset, and based on the MATLAB system identification toolbox, perform multi-input multi-output system identification to obtain the time constant, steady-state gain, time delay, and corresponding state-space model of the relationship between the flatness feature parameters and the corresponding process parameter data; wherein the time constant, steady-state gain, and time delay correspond to a set of state coefficients of the state-space model; wherein the process parameters include laser power, laser scanning speed, and powder feeding rate;

[0065] Step 6: Using the state-space model as the structure of the model predictive controller, establish a model predictive controller (MPC), whose input is the flatness characteristic parameter and whose output is the adjustment amount of the laser cladding process parameter, and configure the control parameters of the model predictive controller.

[0066] Step 7: During the metal laser cladding additive manufacturing process, real-time acquisition of molten pool image data is performed, planarization feature vectors are extracted, and the extracted planarization feature vectors are used as input to the model predictive controller. The model predictive controller outputs process parameter adjustment amounts and sends them to the laser cladding system controller; and

[0067] Step 8: The laser cladding system controller sends the corresponding process parameter adjustment instructions to the laser, robot and powder feeder respectively according to the process parameter adjustment amount, and adjusts the laser power, laser scanning speed and powder feeding rate in real time.

[0068] As an optional approach, in step 2, a surface fitting method is used to fit the ROI melt pool image. The fitting formula used is a high-order polynomial, expressed as:

[0069] z = p 00 +p 10 *x+p 01 *y+p 20 *x 2 +…+p 21 *x 2 *y+p 12 *x*y 2 +p 03 *y 3

[0070] Where (x, y) represents the pixel coordinates of a pixel, z represents the grayscale value of the pixel at coordinates (x, y), and p 00 p 10 p 01 p 20 p 11 p 02 p 21 p 12 p 03 These are the polynomial coefficients.

[0071] Figure 3 An exemplary illustration shows a molten pool image obtained through the aforementioned surface fitting. By fitting the molten pool surface, 14 characteristic parameters are obtained, including the aforementioned 9 polynomial coefficients and 5 goodness-of-fit statistics. The 5 goodness-of-fit statistics are sum-variance (SSE), coefficient of determination (R-square), root mean square error (MSE), corrected coefficient of determination (adj R-square), and root mean square error (RMSE).

[0072] As an optional implementation, after selecting the region of interest (ROI), filtering is also included to eliminate noise and preserve feature information in the image.

[0073] As an optional approach, in step 3, the n feature parameters that have the greatest impact on part flattening are selected as the flattening feature vector by experimentally comparing the correlation between the 14 feature parameters and the changes in the geometric features of the molten pool. For example... Figure 4 The example shown is an experiment that compares the correlation between feature parameters and changes in the geometric features of the molten pool to select a smoothing feature vector.

[0074] As an optional approach, combine Figure 5 The illustration of PCA analysis shows that in step 3, based on principal component analysis, the 14 feature parameters are ranked by importance, and n feature parameters whose cumulative influence on the molten pool characteristics reaches or exceeds a predetermined threshold are selected as the smoothed feature vector. The smoothed feature vector includes at least one polynomial coefficient and at least one goodness-of-fit statistic.

[0075] As an optional approach, in step 3, for the 14 feature parameters obtained, a random forest classifier is used to sort the feature parameters that affect the flatness of the parts by importance, and the n feature parameters that have the greatest impact on the flatness of the parts are used as the flatness feature vector.

[0076] Alternatively, in step 4, the state equations of the state-space model are expressed as follows:

[0077] A(γ)y(k)=B(γ)u(k)+e(k)

[0078] In the formula, y(k) is the smoothness feature vector at time k, u(k) is the process parameter vector corresponding to time k, A(γ) and B(γ) are the state coefficients respectively, and e(k) is white noise.

[0079] Alternatively, in step 4, the control parameters of the model prediction controller are configured, including:

[0080] Configure the sampling period, prediction range, and control range of the model predictive controller, as well as configure the input and / or output constraints and weights. For example... Figure 6 The diagram illustrates an example of setting the sampling period, prediction range, and control range. The configuration of input and / or output constraints and weights can be adjusted and configured through pre-printing and experimentation according to the foregoing embodiments of the present invention.

[0081] It should be understood that in the process of laser cladding additive manufacturing, by continuously acquiring images of the molten pool and extracting features, the MPC model predictive controller can obtain the molten pool flatness deviation vector based on real-time image information, thereby predicting the adjustment amount of laser cladding process parameters. Based on this, the process parameters, namely laser power, laser scanning speed and powder feeding rate, can be adjusted in real time to compensate for geometric errors in the processing, so as to make the surface of the formed parts uniform and flat, and improve the surface quality and dimensional accuracy of the laser cladding workpiece.

[0082] Below, we will use a specific example of a leveling control process to further illustrate the implementation process of the aforementioned embodiments.

[0083] (1) Information acquisition: The CCD camera acquires image information of the molten pool and transmits the image stream to the computer in real time.

[0084] (2) Feature parameter extraction: In the MATLAB environment, the region of interest (ROI) is selected and filtered based on image processing algorithms, and the ROI melt pool image is fitted using a surface fitting method. The fitting formula based on polynomials is as follows:

[0085] z = p 00 +p 10 *x+p 01 *y+p 20 *x 2 +…+p 21 *x 2 *y+p 12 *x*y 2 +p 03 *y 3

[0086] In the formula, (x, y) are the pixel coordinates, z is the gray value of the pixel, and p 00 p 10 p 01 p 20 p 11 p 02 p 21 p 12 p 03 These are polynomial coefficients. After fitting the molten pool surface, 14 characteristic parameters can be obtained, including 9 polynomial coefficients and 5 goodness-of-fit statistics: sum of variances (SSE), coefficient of determination (R-square), root mean square error (MSE), corrected coefficient of determination (adj R-square), and root mean square error (RMSE).

[0087] (3) Dimensionality reduction of feature parameters: In order to reduce the amount of data while taking into account the main features of the melt pool, the dimension reduction of feature parameters in step (2) can be achieved in a variety of ways. This invention uses PCA analysis as an example for illustration.

[0088] Based on principal component analysis (PCA), the 14 feature parameters were ranked by importance, and selected based on their cumulative influence on the molten pool characteristics exceeding 98%. Finally, as an example, we selected 7 of these features as the smoothing feature vector:

[0089] T1 = [p 01 p 11 p 21 p 12 SSE, R-squared, RMSE.

[0090] (4) System identification: using the flatness feature parameter vector T1 = [p 01 p 11 p 21 p12 The dataset consists of [SSE, R-squared, RMSE] and the corresponding process parameter vector T2 = [P, θ, R]. Multiple-input multiple-output (MIMO) system identification is performed using the MATLAB system identification toolbox to obtain the time constant, steady-state gain, time delay, and corresponding state-space model of the relationship between the smoothness feature parameters and the corresponding process parameter data. The state equation of this state-space model can be expressed as:

[0091] A(γ)y(k)=B(γ)u(k)+e(k)

[0092] In the formula, y(k) is the flatness feature parameter vector at time k, u(k) is the process parameter vector at time k, A(γ) and B(γ) are state coefficients, and e(k) is white noise.

[0093] The aforementioned P, θ, and R represent laser power, laser scanning speed, and powder feeding rate, respectively.

[0094] (5) Design MPC (Model Predictive Controller): Use the state space model obtained in step (4) as the structure of MPC, configure the sampling period, prediction range and control range of the controller, define the input and output constraints and weights, and complete the MPC design.

[0095] (6) During the laser cladding process, the CCD camera is used to acquire real-time image information of the molten pool, extract the feature vector from step (3), and process the obtained smoothed feature vector T1 = [p 01 p 11 p 21 p 12 [SSE, R-squared, RMSE] are used as inputs to the MPC. The MPC outputs a process parameter adjustment vector T3 = [ΔP, Δθ, ΔR] and sends it to the laser cladding system controller.

[0096] Where ΔP, Δθ, and ΔR represent the laser power adjustment value, scanning speed adjustment value, and powder feeding rate adjustment value, respectively.

[0097] (7) The laser cladding system controller sends process parameter adjustment instructions to the laser, robot and powder feeder respectively, and adjusts the laser power, scanning speed and powder feeding rate in real time.

[0098] (8) Repeat steps (6) and (7) until the entire part is printed.

[0099] While the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the invention. Those skilled in the art can make various modifications and refinements without departing from the spirit and scope of the invention. Therefore, the scope of protection of the present invention shall be determined by the claims.

Claims

1. A method for controlling the leveling of laser cladding, characterized in that, include: Step 1: Obtain the molten pool image stream information of metal laser cladding additive manufacturing; Step 2: Select the region of interest from multiple frames of molten pool images based on the molten pool image stream information; Step 3: Based on the region of interest, use the surface fitting method to fit the ROI melt pool image, and obtain the fitting formula coefficients and goodness-of-fit statistics through fitting. Step 4: Perform dimensionality reduction on the obtained coefficients of multiple fitting formulas and goodness-of-fit statistics, and determine the n key features that affect the molten pool as flattening feature vectors. Step 5: Using the flattening feature vector and the laser cladding process parameter vector as the dataset, and based on the MATLAB system identification toolbox, perform multi-input multi-output system identification to obtain the time constant, steady-state gain, time delay, and corresponding state-space model of the relationship between the flatness feature parameters and the corresponding process parameter data; wherein the time constant, steady-state gain, and time delay correspond to a set of state coefficients of the state-space model; wherein the process parameters include laser power, laser scanning speed, and powder feeding rate; Step 6: Using the state-space model as the structure of the model predictive controller, establish the model predictive controller, whose input is the flatness characteristic parameter and whose output is the adjustment amount of the laser cladding process parameter, and configure the control parameters of the model predictive controller. Step 7: During the metal laser cladding additive manufacturing process, real-time image data of the molten pool is acquired, and a flattening feature vector is extracted. This extracted flattening feature vector is used as input to the model predictive controller. The model predictive controller obtains the molten pool flatness deviation vector based on the real-time image information, outputs the process parameter adjustment amount, and sends it to the laser cladding system controller. Step 8: The laser cladding system controller sends the corresponding process parameter adjustment instructions to the laser, robot and powder feeder respectively according to the process parameter adjustment amount, and adjusts the laser power, laser scanning speed and powder feeding rate in real time to compensate for geometric errors in the processing process and make the surface of the formed parts uniform and flat. In step 2, the surface fitting method is used to fit the ROI melt pool image. The fitting formula used is a high-order polynomial, expressed as: z=p 00 +p 10 *x+p 01 *y+p 20 *x 2 +…+p 21 *x 2 *y+p 12 *x*y 2 +p 03 *y 3 Where (x,y) represents the pixel coordinates of a pixel, z represents the grayscale value of the pixel at coordinates (x,y), and p 00 ,p 10 p 01 ,p 20 p 11 ,p 02 p 21 p 12 p 03 These are the polynomial coefficients; By fitting the surface of the molten pool, 14 characteristic parameters are obtained, including the aforementioned 9 polynomial coefficients and 5 goodness-of-fit statistics. The 5 goodness-of-fit statistics are sum-variance (SSE), coefficient of determination (R-square), root mean square (MSE), corrected coefficient of determination (adj R-square), and root mean square (RMSE). In step 4, the state equation of the state-space model is expressed as: A(γ)y(k)=B(γ)u(k)+e(k) In the formula, y(k) is the smoothness feature vector at time k, u(k) is the process parameter vector corresponding to time k, A(γ) and B(γ) are the state coefficients respectively, and e(k) is white noise.

2. The laser cladding planarization control method according to claim 1, characterized in that, In step 3, based on principal component analysis, the 14 feature parameters are ranked by importance, and n feature parameters whose cumulative influence on the molten pool characteristics reaches or exceeds a predetermined threshold are selected as the smoothing feature vector. The smoothing feature vector includes at least one polynomial coefficient and at least one goodness-of-fit statistic.

3. The laser cladding planarization control method according to claim 1, characterized in that, In step 3, for the 14 feature parameters obtained, a random forest classifier is used to sort the feature parameters that affect the flatness of the parts by importance, and the n feature parameters that have the greatest impact on the flatness of the parts are used as the flatness feature vector.

4. The laser cladding planarization control method according to claim 1, characterized in that, In step 3, the n feature parameters that have the greatest impact on the flattening of the part are selected as the flattening feature vector by comparing the correlation between the 14 feature parameters and the changes in the geometric features of the molten pool through experiments.

5. The laser cladding planarization control method according to claim 1, characterized in that, In step 4, configuring the control parameters of the model prediction controller includes: Configure the sampling period, prediction range, and control range of the model predictive controller, as well as configure the constraints and weights of the inputs and / or outputs.

6. A laser cladding leveling control system, characterized in that, It includes lasers, laser cladding processing heads, robots, worktables, positioners, powder feeders, protective gas devices, vision acquisition units, laser cladding system controllers, and computer systems; The powder feeder is used to feed powder onto the surface of the workbench; The protective gas device is configured to deliver protective gas to the surface of the worktable through the laser cladding head; The laser is used to emit a laser beam, which is shaped by the optical system in the laser cladding head to form a laser spot on the surface of the worktable, and then melts and deposits the powder on the surface of the worktable. The worktable is mounted on the positioner and is configured to move with the positioner. The vision acquisition unit includes at least one image acquisition device for acquiring images of the molten pool located on the surface of the worktable. The laser cladding head is mounted on the robot and can be driven by the robot to move and change position. The laser cladding system controller is used to control the operation of the laser, laser cladding processing head, robot, worktable, positioner, powder feeder, and protective gas device. The visual acquisition unit is connected to the computer system and sends the acquired molten pool image to the computer system; The computer system is provided with at least one processor and at least one memory, the at least one memory storing operable instructions, which, when executed by the one or more processors, cause the one or more processors to perform operations, including the process of the laser cladding leveling control method as described in any one of claims 1-5.

7. The laser cladding leveling control system according to claim 6, characterized in that, The at least one image acquisition device is a CCD camera, the optical axis of which is mounted perpendicular to the central axis of the laser cladding head, and the CCD camera acquires the image of the molten pool through a 45° reflector set inside the laser cladding head.

8. The laser cladding leveling control system according to claim 6, characterized in that, The laser cladding system controller is connected to the laser, robot, powder feeder, and protective gas device via I / O, and is configured to control the laser power, powder feeding rate, laser scanning speed, and protective gas flow rate.

Images