A blade repair method and system based on laser cladding simulation model

Abstract

The present invention discloses a blade repair method and system based on a laser cladding simulation model, which relates to the field of blade repair technology. A laser cladding simulation model is established, high-speed laser cladding process parameters are set as model variables to input into the laser cladding simulation model, the laser cladding process under different parameter combinations is simulated, and simulation results of the cladding layer are obtained. Cladding samples under different process parameter combinations are prepared and feature analysis is performed, and relevant analysis data is recorded. Based on the relevant analysis data, a performance prediction model with a corresponding mapping relationship between the process parameter combination and the blade coating performance is established to predict the coating performance of the blade under different process parameter combinations. In combination with the specific service environment conditions of the blade, performance constraints are set, the process parameter combination is optimized, the optimal preparation process parameter combination is determined, the actual damaged blade is laser cladding repaired, the laser cladding process is monitored in real time, and it is decided whether to intervene. Quality inspection is performed after the repair is completed.

Description

Technical Field

[0001] The present invention relates to the technical field of blade repair, and in particular to a blade repair method and system based on a laser cladding simulation model. Background Art

[0002] In aerospace, energy, and power generation, blades, as critical components, are subject to long-term exposure to harsh environments such as high temperature, high pressure, and high-speed erosion. These blades are susceptible to damage such as wear and corrosion, which can lead to performance degradation or even failure. Traditional blade repair methods, such as welding and thermal spraying, suffer from low repair accuracy, large heat-affected zones, and difficulty maintaining mechanical properties after repair. Laser cladding technology, with its advantages of low dilution rate, small heat-affected zone, and the ability to perform in-situ repairs, shows great potential for application in blade repair.

[0003] However, the laser cladding process involves complex physical and chemical changes, and numerous and mutually coupled process parameters, such as laser power, laser scanning speed, and powder feed speed, significantly affect the microstructure, coating-substrate bonding strength, corrosion resistance, and wear resistance of the cladding coating. Currently, in the actual repair process, the selection of process parameters mostly relies on experience and lacks scientific and effective theoretical guidance, resulting in uneven quality of repaired blades and difficulty in meeting the growing demand for high performance. Therefore, how to deeply analyze the impact of process parameters on blade repair quality based on the laser cladding simulation model and optimize the process parameters to obtain the best preparation process has become an urgent problem to be solved. Summary of the Invention

[0004] In order to solve the above problems, the purpose of the present invention is to provide a blade repair method and system based on a laser cladding simulation model.

[0005] The purpose of the present invention can be achieved by the following technical solution: A blade repair method based on a laser cladding simulation model comprises the following steps:

[0006] Step S1: Establishing a laser cladding simulation model and setting model parameters. Setting high-speed laser cladding process parameters as model variables and inputting them into the laser cladding simulation model. Simulating the laser cladding process under different parameter combinations and obtaining simulation results of the cladding layer.

[0007] Step S2: Based on the simulation results, prepare cladding samples under different process parameter combinations, perform feature analysis on the cladding samples, and record relevant analysis data;

[0008] Step S3: Based on the correlation analysis data, a performance prediction model is established to map the corresponding relationship between the process parameter combination and the blade coating performance. The performance prediction model is operated based on the test data to predict the blade coating performance under different process parameter combinations.

[0009] Step S4: Based on the specific service environment conditions of the blade, performance constraints are set, and with the goal of achieving the best comprehensive coating performance, the process parameter combination is optimized to determine the best preparation process parameter combination;

[0010] Step S5: Using the optimal combination of preparation process parameters, perform laser cladding repair on the actual damaged blade, monitor the key working parameters of the laser cladding process in real time, and decide whether to intervene. After the repair is completed, perform quality inspection on the blade.

[0011] Furthermore, the process of establishing a laser cladding simulation model and setting model parameters includes:

[0012] Obtaining the geometric information of the blade and constructing a three-dimensional solid model of the blade based on the geometric information;

[0013] Obtain all process steps in the entire laser cladding process, establish a step operation model corresponding to each process step, and construct a laser cladding process model based on all step operation models;

[0014] An initial convolutional neural network framework model is constructed using deep learning technology. The blade 3D solid model and the laser cladding process model are synchronously mapped into the convolutional neural network framework model, and then the convolutional neural network framework model is constructed into a laser cladding simulation model.

[0015] Set the model parameters corresponding to the laser cladding simulation model, which include blade material parameters, cladding material parameters, and environmental parameters.

[0016] Furthermore, the high-speed laser cladding process parameters are set as model variables and input into the laser cladding simulation model, the laser cladding process under different parameter combinations is simulated, and the simulation results of the cladding layer are obtained. The process includes:

[0017] Set the high-speed laser cladding process parameters corresponding to the laser cladding simulation model. The high-speed laser cladding process parameters include laser power, laser scanning speed, and powder feeding speed, which are denoted as W and V respectively. 扫描 and V 送粉 ;

[0018] Set the power safety range and record it as Ω[w];

[0019] Set the effective speed ranges corresponding to the laser scanning speed and the powder feeding speed, and record them as Ω[v 扫描 ] and Ω[v 送粉 ];

[0020] Keep W Total laser power in Ω[W], excluding W The total laser power of Ω[w] generates the laser power set;

[0021] Keep V 扫描 Ω[v 扫描 ] total laser scanning speed, excluding V 扫描 Ω[v 扫描 ] to generate a laser scanning speed set;

[0022] Keep V 送粉 Ω[v 送粉 ] total powder feeding speed, excluding V 送粉 Ω[v 送粉 ] to generate a powder feeding speed set;

[0023] The laser power set, laser scanning speed set and powder feeding speed set are integrated as model variables and input into the laser cladding simulation model. A laser power, laser scanning speed and powder feeding speed are randomly selected from the laser power set, laser scanning speed set and powder feeding speed set respectively and integrated as parameter combinations to generate several different parameter combinations. The laser cladding simulation model simulates the laser cladding process of the blade under different parameter combinations to obtain simulation results of different cladding layers on the blade.

[0024] Furthermore, the process of preparing cladding samples under different process parameter combinations, performing characteristic analysis on the cladding samples, and recording relevant analysis data includes:

[0025] Construct a sample information library, which is used to store cladding sample ratio information corresponding to different simulation results. Input any simulation result obtained by the laser cladding simulation model into the sample information library to obtain cladding sample ratio information that matches the current simulation result. Prepare cladding samples with the working parameter combination corresponding to the current simulation result based on the cladding sample ratio information.

[0026] Repeat the above operation and input all simulation results into the sample information library to obtain the cladding sample ratio information corresponding to each simulation result, and prepare the cladding samples under different process parameter combinations; perform feature analysis on all cladding samples to obtain relevant analysis data of the cladding samples, including geometric feature parameters, microscopic feature parameters, mechanical property parameters, and functional performance parameters.

[0027] Furthermore, based on the relevant analysis data, a performance prediction model is established to map the corresponding relationship between the process parameter combination and the blade coating performance. The performance prediction model is operated based on the test data. The process of predicting the blade coating performance under different process parameter combinations includes:

[0028] Perform initial model selection based on the amount of relevant analysis data, then build the model architecture corresponding to the initial performance prediction model, and define the input features and output targets corresponding to the initial performance prediction model;

[0029] The input features are different combinations of process parameters; the output targets are different coating performance indicators corresponding to the blade, including blade hardness, residual stress and defect rate;

[0030] Construct a functional mapping relationship between a process parameter combination and a coating performance indicator. Set a partition ratio to divide the relevant analysis data into a training set and a validation set. Input the training set into the performance prediction model under the initial model architecture for model training. Use the validation set to obtain the corresponding model evaluation index. When the model evaluation index meets the requirements, construct the final performance prediction model based on the functional mapping relationship between the process parameter combination and the blade coating performance.

[0031] The test data is input into the performance prediction model to predict the coating performance of the blade under different process parameter combinations. The coating performance of the blade includes four performance levels: excellent, good, qualified and unqualified.

[0032] Furthermore, based on the specific service environment conditions of the blade, performance constraints are set, and the process parameter combination is optimized with the goal of achieving the best overall coating performance. The process of determining the best preparation process parameter combination includes:

[0033] The specific service environment conditions of the blade include temperature conditions, mechanical load conditions, and corrosion environment conditions; performance constraints are set, including the threshold intervals or endpoint thresholds corresponding to blade hardness, residual stress, defect rate, laser power, laser scanning speed, and powder feeding speed;

[0034] The coating performance of the blade is set at an excellent performance level as the goal of optimizing the overall coating performance. A multi-objective optimization model is constructed based on the goal of optimizing the overall coating performance, the specific service environment conditions of the blade, and the performance constraints.

[0035] Different process parameter combinations are used as debugging data for the multi-objective optimization model. The coating index coefficient is obtained in real time by the multi-objective optimization model, and an excellent quantitative value is set. When the coating index coefficient ≥ the excellent quantitative value, the current process parameter combination is selected as the candidate data, and all the candidate data are integrated to generate a candidate data set, and the optimization conditions of the candidate data set are set;

[0036] When the coating index coefficient is less than the excellent quantitative value, the corresponding process parameter combination is eliminated;

[0037] The optimization condition is: select the process parameter combination with the highest coating index coefficient in the candidate data set as the optimal preparation process parameter combination.

[0038] Furthermore, the optimal combination of preparation process parameters is used to perform laser cladding repair on actual damaged blades. The key working parameters of the laser cladding process are monitored in real time to determine whether to intervene. After the repair is completed, the blade quality inspection process includes the following:

[0039] Use the optimal preparation process parameters to operate laser cladding related equipment, perform laser cladding repair on actual damaged blades, and monitor the laser cladding process in real time to obtain all monitoring objects and monitoring parameters of each monitoring object during the laser cladding repair process;

[0040] The monitoring parameters of each monitoring object are integrated as the key working parameters of the laser cladding process;

[0041] Monitoring targets include melt pool temperature, melt pool morphology, powder delivery, and laser power;

[0042] Decide whether to intervene in the laser cladding repair process based on key working parameters;

[0043] If any of the melt pool temperature, melt pool morphology, powder delivery and laser power does not meet the standards, it is decided to adjust the laser cladding related equipment and intervene in the laser cladding repair process. Otherwise, it is decided not to intervene.

[0044] After the blade repair is completed, the blade is subjected to quality inspection, which includes surface defect inspection and internal defect inspection, and a quality inspection report of the blade is generated after the quality inspection is completed.

[0045] Furthermore, a blade repair system based on the laser cladding simulation model includes:

[0046] Laser cladding simulation module, used to establish a laser cladding simulation model and set model parameters. It sets high-speed laser cladding process parameters as model variables and inputs them into the laser cladding simulation model to simulate the laser cladding process under different parameter combinations and obtain simulation results of the cladding layer.

[0047] The sample characteristic analysis module is used to prepare cladding samples under different process parameter combinations based on the simulation results, perform characteristic analysis on the cladding samples, and record relevant analysis data;

[0048] The coating performance prediction module establishes a performance prediction model that maps the corresponding relationship between process parameter combinations and blade coating performance based on relevant analysis data. The performance prediction model is operated based on test data to predict the coating performance of the blade under different process parameter combinations.

[0049] The process parameter optimization module is used to set performance constraints based on the specific service environment conditions of the blade, optimize the process parameter combination with the goal of achieving the best comprehensive coating performance, and determine the best preparation process parameter combination;

[0050] The blade repair quality inspection module uses the optimal combination of preparation process parameters to perform laser cladding repair on actual damaged blades, monitors the key working parameters of the laser cladding process in real time, and decides whether to intervene. After the repair is completed, the blade is quality inspected.

[0051] Compared with the prior art, the beneficial effects of the present invention are as follows: by establishing a laser cladding simulation model, the laser cladding process under different process parameter combinations can be simulated without conducting actual experiments, thereby quickly screening out parameter combinations with potential optimization space, reducing the experimental cost, and providing effective theoretical guidance; by preparing cladding specimens under different process parameter combinations, and establishing a mapping relationship between process parameters and coating performance after characteristic analysis, the coating performance prediction is made more accurate, which helps to adjust the process parameters in a targeted manner and improve the comprehensive performance of the blade coating; in combination with the specific service environment conditions of the blade, performance constraints are set, and process parameters are optimized with the goal of achieving the best comprehensive coating performance, ensuring that the blade coating has the best performance under specific working conditions; using the best preparation process parameter combination for laser cladding repair, real-time monitoring of key working parameters, ensuring that the repair process is stable and controllable, and to a certain extent improving the repair quality of the blade. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 This is a flow chart of a blade repair method based on a laser cladding simulation model according to an embodiment of the present application. DETAILED DESCRIPTION

[0053] like Figure 1 As shown, a blade repair method based on a laser cladding simulation model includes the following steps:

[0054] Step S1: Establishing a laser cladding simulation model and setting model parameters. Setting high-speed laser cladding process parameters as model variables and inputting them into the laser cladding simulation model. Simulating the laser cladding process under different parameter combinations and obtaining simulation results of the cladding layer.

[0055] Step S2: Based on the simulation results, prepare cladding samples under different process parameter combinations, perform feature analysis on the cladding samples, and record relevant analysis data;

[0056] Step S3: Based on the correlation analysis data, a performance prediction model is established to map the corresponding relationship between the process parameter combination and the blade coating performance. The performance prediction model is operated based on the test data to predict the blade coating performance under different process parameter combinations.

[0057] Step S4: Based on the specific service environment conditions of the blade, performance constraints are set, and with the goal of achieving the best comprehensive coating performance, the process parameter combination is optimized to determine the best preparation process parameter combination;

[0058] Step S5: Using the optimal combination of preparation process parameters, perform laser cladding repair on the actual damaged blade, monitor the key working parameters of the laser cladding process in real time, and decide whether to intervene. After the repair is completed, perform quality inspection on the blade.

[0059] It should be further explained that, in the specific implementation process, the process of establishing a laser cladding simulation model and setting model parameters includes:

[0060] Obtaining the geometric information of the blade and constructing a three-dimensional solid model of the blade based on the geometric information;

[0061] The geometric information of the blade includes blade shape, blade length, blade circumference, blade thickness, blade cross-section information, blade large-scale drawing information, blade main view information, and blade exploded view information; the geometric information of the blade is used to reflect the physical attribute characteristics of the blade at the visual level;

[0062] Obtain all process steps in the entire laser cladding process, establish the corresponding step operation model for each process step, and follow the step execution order of the process steps, taking the first step operation model as the fusion reference body and the other step operation models as the individuals to be fused. Each individual model to be fused is fused to the fusion reference body in turn, thereby constructing the laser cladding process model;

[0063] An initial convolutional neural network framework model is constructed using deep learning technology. The blade 3D solid model and the laser cladding process model are synchronously mapped into the convolutional neural network framework model, and then the convolutional neural network framework model is constructed into a laser cladding simulation model.

[0064] Set the model parameters corresponding to the laser cladding simulation model. The model parameters include blade material parameters, cladding material parameters, and environmental parameters. The blade material parameters are used to characterize the material properties corresponding to the blade that needs to be repaired by cladding. The cladding material parameters are used to characterize all material properties corresponding to the material used for cladding repair.

[0065] The blade material parameters specifically include thermophysical parameters, blade mechanical performance parameters, blade optical characteristic parameters and historical state parameters; among which, the thermophysical parameters include thermal conductivity, specific heat capacity, density, thermal expansion coefficient and melting point; the blade mechanical performance parameters include elastic modulus, Poisson's ratio, yield strength and creep parameters; the blade optical characteristic parameters are expressed as laser absorptivity; the historical state parameters include initial residual stress and microstructure;

[0066] Cladding material parameters specifically include cladding powder characteristic parameters, cladding layer mechanical properties parameters, solid-liquid phase transition parameters, and powder optical properties parameters. Cladding powder characteristic parameters include particle size distribution and powder flow rate; cladding layer mechanical properties parameters also include elastic modulus, Poisson's ratio, and yield strength; solid-liquid phase transition parameters include latent heat of fusion, latent heat of solidification, surface tension, and temperature coefficient; powder optical properties parameters include powder absorptivity and cladding layer reflectivity.

[0067] The environmental parameters specifically include parameters related to heat dissipation conditions, parameters related to protective gas, and parameters related to substrate fixing conditions; among them, parameters related to heat dissipation conditions include convective heat transfer coefficient, radiation heat transfer coefficient, and ambient temperature; parameters related to protective gas include gas type and gas flow rate; parameters related to substrate fixing conditions include constraint method and initial preload force.

[0068] It should be further explained that, in the specific implementation process, the high-speed laser cladding process parameters are set as model variables and input into the laser cladding simulation model, the laser cladding process under different parameter combinations is simulated, and the simulation results of the cladding layer are obtained. The process includes:

[0069] Set the high-speed laser cladding process parameters corresponding to the laser cladding simulation model. The high-speed laser cladding process parameters include laser power, laser scanning speed, and powder feeding speed. The laser power, laser scanning speed, and powder feeding speed are denoted as W and V, respectively. 扫描 and V 送粉 ;

[0070] Set the power safety interval corresponding to the laser power and record it as Ω[w];

[0071] Set the effective speed range corresponding to the laser scanning speed and record it as Ω[v 扫描 ];

[0072] Set the effective speed range corresponding to the powder feeding speed and record it as Ω[v 送粉 ];

[0073] Among them, the power safety interval Ω[w] is used to limit the upper and lower limits of the laser power when it is in safe operation. Ω[w] = [W-min, W-max]. The speed effective ranges of the laser scanning speed and the powder feeding speed are used to limit the speed value range during laser scanning or powder feeding. Only the laser scanning speed and powder feeding speed within the speed effective range are considered to be effective operations of laser cladding, which can better;

[0074] Keep W Total laser power in Ω[W], excluding W The total laser power of Ω[w] generates the laser power set;

[0075] Keep V 扫描 Ω[v 扫描 ] total laser scanning speed, excluding V 扫描 Ω[v 扫描 ] to generate a laser scanning speed set;

[0076] Keep V 送粉 Ω[v 送粉 ] total powder feeding speed, excluding V 送粉 Ω[v 送粉 ] to generate a powder feeding speed set;

[0077] Integrate the laser power set, laser scanning speed set, and powder feeding speed set as model variables, and input the model variables into the laser cladding simulation model. The laser cladding simulation model randomly selects a laser power, laser scanning speed, and powder feeding speed from the laser power set, laser scanning speed set, and powder feeding speed set, and integrates them as a parameter combination, thereby generating several different parameter combinations.

[0078] The laser cladding simulation model simulates the laser cladding process of the blade under different parameter combinations based on the obtained several different parameter combinations, and then obtains the simulation results of the blade corresponding to different cladding layers.

[0079] It should be further explained that, in the specific implementation process, based on the simulation results, the cladding samples under different process parameter combinations are prepared, the characteristics of the cladding samples are analyzed, and the relevant analysis data are recorded. The process includes:

[0080] Constructing a sample information database, which is used to store the cladding sample ratio information corresponding to different simulation results. The cladding sample ratio information includes the ratio information between different chemical components of the cladding material, the ratio information between the substrate material and the cladding material, and the ratio parameters of the powder physical properties;

[0081] Input any simulation result obtained by the laser cladding simulation model into the sample information library, match the cladding sample ratio information that matches the current simulation result from the sample information library, prepare the cladding sample with the working parameter combination corresponding to the current simulation result according to the cladding sample ratio information, repeat the above operation, input all simulation results into the sample information library, and then match the cladding sample ratio information corresponding to each simulation result from the sample information library, thereby completing the preparation of cladding samples under different process parameter combinations;

[0082] Perform feature analysis on all cladding samples, and obtain relevant analysis data of the cladding samples including geometric feature parameters, microscopic feature parameters, mechanical property parameters and functional performance parameters through feature analysis records;

[0083] Among them, the geometric characteristic parameters include the size of the cladding layer corresponding to the cladding sample, such as: weld width, weld height and dilution rate, and the surface morphology of the cladding sample, such as: roughness, cracks and pores. The size of the cladding layer is obtained by measuring with an optical microscope or a 3D profilometer. The microscopic characteristic parameters include the microstructural characteristics and interface bonding characteristics of the cladding sample. The microstructural characteristics include grain size, dendrite spacing and phase composition, which are obtained by metallographic microscope; the interface bonding characteristics include the continuity of the interface between the substrate material and the cladding layer corresponding to the cladding sample and the element diffusion intensity, which are obtained by line scanning.

[0084] Among them, the mechanical performance parameters include the hardness distribution, residual stress and bonding strength of the cladding sample. Multi-point tests are performed along the depth direction of the corresponding cladding layer of the cladding sample to obtain the hardness distribution; the surface and internal stresses are measured by X-ray diffraction or neutron diffraction to obtain the residual stress of the cladding sample, and the bonding strength of the cladding sample is obtained through tensile and shear tests; the functional performance parameters include the wear resistance, corrosion resistance and high temperature resistance of the cladding sample.

[0085] It should be further explained that, in the specific implementation process, based on the relevant analysis data, a performance prediction model is established to map the corresponding relationship between the process parameter combination and the blade coating performance. The performance prediction model is operated based on the test data. The process of predicting the blade coating performance under different process parameter combinations includes:

[0086] Initial model selection based on the amount of relevant analysis data;

[0087] The amount of relevant analysis data is denoted as D 相关 , which is a natural number greater than 0, unit: group;

[0088] When D 相关 When <100, the linear regression equation is used to construct the model architecture corresponding to the initial performance prediction model;

[0089] When 100≤D 相关 When <1000, a shallow neural network is used to construct the model architecture corresponding to the initial performance prediction model;

[0090] When D 相关 When the number is greater than 10,000, deep learning is used to construct the model architecture corresponding to the initial performance prediction model;

[0091] Define the input features and output targets corresponding to the initial performance prediction model;

[0092] The input feature is a cladding simulation model corresponding to different process parameter combinations, which are identified as a set X; different process parameter combinations are composed of any value of laser power, any value of laser scanning speed, and any value of powder feeding speed, thereby constructing a parameter set including different laser powers, laser scanning speeds, and powder feeding speeds;

[0093] Then we have the set X = <any value of laser power, any value of laser scanning speed, any value of powder feeding speed>;

[0094] The output target is the different coating performance indicators corresponding to the blade, marked as Y;

[0095] Then the value of Y includes blade hardness, residual stress and defect rate;

[0096] The function mapping relationship between the laser power, laser scanning speed, and powder feeding speed in the process parameter combination and a certain coating performance index is constructed. The function mapping relationship is expressed as follows:

[0097] Y = a × laser power + b × laser scanning speed + c × powder feeding speed + d;

[0098] Among them, a, b, and c are the linear fitting coefficients of laser power, laser scanning speed, and powder feeding speed, respectively, and d is the tuning coefficient of the corresponding function mapping relationship between the process parameter combination and the coating performance index;

[0099] Set the division ratio and divide the relevant analysis data into training set and validation set according to the division ratio;

[0100] The training set is input into the performance prediction model under the initial model architecture to perform corresponding model training. The corresponding model evaluation indicators are obtained through the validation set. The model evaluation indicators include normalization rate, generalization rate and mean square error.

[0101] When the normalization rate, generalization rate, and mean square error are all within their respective preset standard ranges, the final performance prediction model is constructed based on the functional mapping relationship between the process parameter combination and the blade coating performance.

[0102] Acquire test data, which is a combination of process parameters for blade cladding repair, and input the test data into a performance prediction model. The performance prediction model then predicts the coating performance of the blade under different process parameter combinations.

[0103] The coating performance of the blade is characterized by blade hardness, residual stress and defect rate;

[0104] The coating performance of the blade includes four performance levels: excellent, good, qualified and unqualified.

[0105] It should be further explained that, in the specific implementation process, the performance constraints are set in combination with the specific service environment conditions of the blade, and the comprehensive optimization of coating performance is taken as the goal. The process of determining the best preparation process parameter combination includes:

[0106] Obtain the specific service environment conditions of the blade, which include temperature conditions, mechanical load conditions, and corrosion environment conditions. The temperature conditions are used to characterize the high-temperature environment in which the blade is located. The mechanical load conditions are used to characterize the extreme values of centrifugal force and airflow impact that the blade can withstand, as well as the vibration fatigue that the blade can withstand.

[0107] The corrosion environment condition is used to characterize the real-time concentration of salt spray, sulfide and other chemical corrosion gases in the blade's current service environment;

[0108] Set performance constraints. The performance constraints consist of threshold intervals or endpoint thresholds corresponding to blade hardness, residual stress, defect rate, laser power, laser scanning speed, and powder feeding speed. Examples of the performance constraints are as follows:

[0109] ;

[0110] Among them, the constraint types of residual stress ≤ -300MPa, defect rate ≤ 1% and blade hardness ≥ 450HV0.5 are all endpoint thresholds, while W∈[1000,3000], V 扫描 ∈[40, 100] mm / s and V 送粉 The constraint types of ∈[8,20]mm / s are all threshold intervals;

[0111] The coating performance of the blade is set at the excellent performance level as the goal of optimizing the overall coating performance. A multi-objective optimization model is constructed based on the goal of optimizing the overall coating performance, the specific service environment conditions of the blade, and the performance constraints.

[0112] The construction process of the multi-objective optimization model is as follows:

[0113] Construct an objective optimization function between the specific service environment conditions of the blade and the coating performance of the blade, and record the current objective optimization function as f1;

[0114] Constructing a target optimization function between the performance constraint conditions and the coating performance of the blade, and recording the current target optimization function as f2;

[0115] An initial multi-objective optimization model is constructed using deep learning technology, and the objective optimization functions f1 and f2 are associated and mapped to the initial multi-objective optimization model. The optimal comprehensive performance of the coating is used as the model iteration stopping condition of the initial multi-objective optimization model. The initial multi-objective optimization model is continuously iterated until the model prediction accuracy of the multi-objective optimization model meets the preset expected accuracy and the optimal comprehensive performance of the coating is met. The current iteration is stopped and the final multi-objective optimization model is constructed.

[0116] Different process parameter combinations are used as debugging data for the multi-objective optimization model. The coating index coefficient is obtained in real time by the multi-objective optimization model, and an excellent quantitative value is set. When the coating index coefficient ≥ the excellent quantitative value, the current process parameter combination is selected as the candidate data, and all the candidate data are integrated to generate a candidate data set, and the optimization conditions corresponding to the candidate data set are set;

[0117] The coating index coefficient is recorded as μ, and the expression formula of the coating index coefficient is as follows:

[0118] μ=H -n ×w1+L -n ×w2+D -n ×w3+P -n ×w4+K -n ×w5+Q -n ×w6;

[0119] Among them, H -n Indicates the normal hardness after numerical normalization, L -n represents the normal residual stress after numerical normalization, D -n It represents the normal defect rate after numerical normalization, P -n It represents the normal laser power after numerical normalization, K -n It represents the normal laser scanning speed after numerical normalization, Q -n Indicates the normal powder feeding speed after numerical normalization;

[0120] w1, w2, w3, w4, w5, and w6 are the weighted values of normal hardness, normal residual stress, normal defect rate, normal laser power, normal laser scanning speed, and normal powder feeding speed, respectively. Among them, w1 + w2 + w3 + w4 + w5 + w6 = 1. The values of w1-w6 are set as needed. w1-w6 are all real numbers greater than 0. The value range of the coating index coefficient μ is (0, 1);

[0121] H -n 、L -n 、D -n 、P -n , K -n , and Q-n The respective calculation formulas are as follows:

[0122] H -n =(H-H -min ) / (H -max -H -min );

[0123] L -n =(L-L -min ) / (L -max -L -min );

[0124] D -n =1-[(D-D -min ) / (D -max -D -min )];

[0125] P -n =(P-P -min ) / (P -max -P -min );

[0126] K -n =(K-K -min ) / (K -max -K -min );

[0127] Q -n =(Q-Q -min ) / (Q -max -Q -min );

[0128] Among them, the above H -n 、L -n 、D -n 、P -n , K -n , and Q -n The relevant parameters in each calculation formula are described as follows:

[0129] H is the normal hardness of any value, H -min is the minimum value of normal hardness, H -max is the maximum value of normal hardness;

[0130] L is the normal residual stress of any value, L -min is the minimum value of normal residual stress, L -max is the maximum value of normal residual stress;

[0131] D is the normal defect rate of any value, D -min is the minimum value of the normal defect rate, D -max is the maximum value of normal defect rate;

[0132] P is the normal laser power of any value, P -min is the minimum value of normal laser power, P -max is the maximum value of normal laser power;

[0133] K is the normal laser scanning speed of any value, K -min is the minimum value of normal laser scanning speed, K -max is the maximum value of normal laser scanning speed;

[0134] Q is the normal powder feeding speed of any value, Q -min is the minimum value of normal powder feeding speed, Q -max It is the maximum value of normal powder feeding speed;

[0135] The value of the excellent quantization value is set to 0.8, which can be adjusted according to actual conditions in the future;

[0136] When the coating index coefficient is less than the excellent quantitative value, the corresponding process parameter combination is eliminated;

[0137] The optimization condition is: select the process parameter combination with the highest coating index coefficient in the candidate data set, and use the process parameter combination as the optimal preparation process parameter combination.

[0138] It should be further explained that in the specific implementation process, the optimal preparation process parameter combination is used to perform laser cladding repair on the actual damaged blade. The key working parameters of the laser cladding process are monitored in real time to determine whether to intervene. After the repair is completed, the blade quality inspection process includes the following:

[0139] Select the actual damaged blade that needs to be repaired, operate the laser cladding related equipment with the optimal preparation process parameters, and then perform laser cladding repair on the actual damaged blade. The laser cladding process is monitored in real time. Through real-time monitoring, all monitoring objects corresponding to the laser cladding repair process and the monitoring parameters of each monitoring object are obtained;

[0140] The monitoring parameters of each monitoring object are integrated as the key working parameters of the laser cladding process;

[0141] The monitoring objects include molten pool temperature, molten pool morphology, powder delivery and laser power;

[0142] The monitoring parameters of the molten pool temperature include peak temperature and trough temperature;

[0143] The monitoring parameters of the molten pool morphology include the molten pool width and height;

[0144] Monitoring parameters of powder delivery include powder delivery rate and powder delivery utilization;

[0145] Laser power monitoring parameters include real-time power and peak power;

[0146] Decide whether to intervene in the laser cladding repair process based on key working parameters;

[0147] If any of the melt pool temperature, melt pool morphology, powder delivery and laser power does not meet the standards, it is decided to adjust the laser cladding related equipment and intervene in the laser cladding repair process. Otherwise, it is decided not to intervene.

[0148] Among them, the contents of judging whether the molten pool temperature, molten pool morphology, powder delivery and laser power meet the standards are as follows:

[0149] For the molten pool temperature, when the peak temperature is higher than the preset upper temperature limit, or the trough temperature is lower than the preset lower temperature limit, it is judged that the current molten pool temperature does not meet the standard; otherwise, it is judged that the molten pool temperature meets the standard;

[0150] For the molten pool morphology, when the molten width deviation corresponding to the molten width is greater than 10%, or the molten height deviation corresponding to the molten height is greater than 5%, the current molten pool morphology is judged to be unqualified; otherwise, the molten pool morphology is judged to be qualified;

[0151] For powder delivery, when the powder delivery rate fluctuation is greater than 5%, or the powder delivery utilization rate is lower than the preset utilization rate lower limit threshold, it is judged that the current powder delivery does not meet the standard; otherwise, it is judged that the powder delivery meets the standard;

[0152] For laser power, when the power deviation of the real-time power is greater than 15%, or the peak power is greater than the preset power upper limit, it is judged that the current laser power does not meet the standard; otherwise, it is judged that the laser power meets the standard;

[0153] After the blade repair is completed, the blade is subjected to quality inspection, which includes surface defect inspection and internal defect inspection, and then the corresponding surface detail information data and internal detail information data after the blade repair is completed are generated respectively, and integrated as the blade quality inspection report.

[0154] The present invention also provides a blade repair system based on a laser cladding simulation model, the system comprising:

[0155] Laser cladding simulation module, used to establish a laser cladding simulation model and set model parameters. It sets high-speed laser cladding process parameters as model variables and inputs them into the laser cladding simulation model to simulate the laser cladding process under different parameter combinations and obtain simulation results of the cladding layer.

[0156] The sample characteristic analysis module is used to prepare cladding samples under different process parameter combinations based on the simulation results, perform characteristic analysis on the cladding samples, and record relevant analysis data;

[0157] The coating performance prediction module establishes a performance prediction model that maps the corresponding relationship between process parameter combinations and blade coating performance based on relevant analysis data. The performance prediction model is operated based on test data to predict the coating performance of the blade under different process parameter combinations.

[0158] The process parameter optimization module is used to set performance constraints based on the specific service environment conditions of the blade, optimize the process parameter combination with the goal of achieving the best comprehensive coating performance, and determine the best preparation process parameter combination;

[0159] The blade repair quality inspection module uses the optimal combination of preparation process parameters to perform laser cladding repair on actual damaged blades, monitors the key working parameters of the laser cladding process in real time, and decides whether to intervene. After the repair is completed, the blade is quality inspected.

[0160] The above embodiments are only used to illustrate the technical method of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A blade repair method based on a laser cladding simulation model, characterized in that: The following steps are involved: Step S1: Establishing a laser cladding simulation model and setting model parameters. Setting high-speed laser cladding process parameters as model variables and inputting them into the laser cladding simulation model. Simulating the laser cladding process under different parameter combinations and obtaining simulation results of the cladding layer. Step S2: Based on the simulation results, prepare cladding samples under different process parameter combinations, perform feature analysis on the cladding samples, and record relevant analysis data; Step S3: Based on the correlation analysis data, a performance prediction model is established to map the corresponding relationship between the process parameter combination and the blade coating performance. The performance prediction model is operated based on the test data to predict the blade coating performance under different process parameter combinations. Step S4: In combination with the specific service environment conditions of the blade, performance constraints are set, and with the comprehensive optimization of coating performance as the goal, the process parameter combination is optimized to determine the optimal preparation process parameter combination, including: the specific service environment conditions of the blade include temperature conditions, mechanical load conditions and corrosion environment conditions; performance constraints are set, including threshold intervals or endpoint thresholds corresponding to blade hardness, residual stress, defect rate, laser power, laser scanning speed and powder feeding speed; the coating performance of the blade is set to an excellent performance level as the goal of comprehensive optimization of coating performance, and a multi-objective optimization model is constructed based on the goal of optimal comprehensive coating performance, the specific service environment conditions of the blade and the performance constraints; different process parameter combinations are used as debugging data of the multi-objective optimization model, and the coating index coefficient is obtained in real time by the multi-objective optimization model, and an excellent quantitative value is set. When the coating index coefficient is ≥ the excellent quantitative value, the current process parameter combination is selected as candidate data, all candidate data are integrated to generate a candidate data set, and optimization conditions for the candidate data set are set; when the coating index coefficient is < the excellent quantitative value, the corresponding process parameter combination is eliminated; Step S5: Use the best preparation process parameter combination to perform laser cladding repair on the actual damaged blade, monitor the key working parameters of the laser cladding process in real time, and decide whether to intervene. After the repair is completed, perform quality inspection on the blade, including: use the best preparation process parameters to operate laser cladding related equipment, perform laser cladding repair on the actual damaged blade, and monitor the laser cladding process in real time to obtain all monitoring objects of the laser cladding repair process and the monitoring parameters of each monitoring object; the monitoring parameters of each monitoring object are integrated as the key working parameters of the laser cladding process; the monitoring objects include molten pool temperature, molten pool morphology, powder delivery and laser power; decide whether to intervene in the laser cladding repair process based on the key working parameters; if any of the molten pool temperature, molten pool morphology, powder delivery and laser power does not meet the standards, decide to adjust the laser cladding related equipment and intervene in the laser cladding repair process, otherwise, decide not to intervene; when the blade repair is completed, perform quality inspection on the blade, the quality inspection includes surface defect inspection and internal defect inspection, and generate a blade quality inspection report after the quality inspection is completed.

2. The blade repair method based on the laser cladding simulation model according to claim 1 is characterized in that: The process of establishing a laser cladding simulation model and setting model parameters includes: Obtaining the geometric information of the blade and constructing a three-dimensional solid model of the blade based on the geometric information; Obtain all process steps in the entire laser cladding process, establish a step operation model corresponding to each process step, and construct a laser cladding process model based on all step operation models; An initial convolutional neural network framework model is constructed using deep learning technology. The blade 3D solid model and the laser cladding process model are synchronously mapped into the convolutional neural network framework model, and then the convolutional neural network framework model is constructed into a laser cladding simulation model. Set the model parameters corresponding to the laser cladding simulation model, which include blade material parameters, cladding material parameters, and environmental parameters.

3. The blade repair method based on the laser cladding simulation model according to claim 2 is characterized in that: The process of setting high-speed laser cladding process parameters as model variables and inputting them into the laser cladding simulation model, simulating the laser cladding process under different parameter combinations, and obtaining the simulation results of the cladding layer includes: Set the high-speed laser cladding process parameters corresponding to the laser cladding simulation model. The high-speed laser cladding process parameters include laser power, laser scanning speed, and powder feeding speed, which are denoted as W and V respectively. 扫描 and V 送粉 ; Set the power safety range and record it as Ω[w]; Set the effective speed ranges corresponding to the laser scanning speed and the powder feeding speed, and record them as Ω[v 扫描 ] and Ω[v 送粉 ]; Keep W Total laser power in Ω[W], excluding W The total laser power of Ω[w] generates the laser power set; Keep V 扫描 Ω[v 扫描 ] total laser scanning speed, excluding V 扫描 Ω[v 扫描 ] to generate a laser scanning speed set; Keep V 送粉 Ω[v 送粉 ] total powder feeding speed, excluding V 送粉 Ω[v 送粉 ] to generate a powder feeding speed set; The laser power set, laser scanning speed set and powder feeding speed set are integrated as model variables and input into the laser cladding simulation model. A laser power, laser scanning speed and powder feeding speed are randomly selected from the laser power set, laser scanning speed set and powder feeding speed set respectively and integrated as parameter combinations to generate several different parameter combinations. The laser cladding simulation model simulates the laser cladding process of the blade under different parameter combinations to obtain simulation results of different cladding layers on the blade.

4. The blade repair method based on the laser cladding simulation model according to claim 3 is characterized in that: Based on the simulation results, the process of preparing cladding samples under different process parameter combinations, performing feature analysis on the cladding samples, and recording relevant analysis data includes: Construct a sample information library, which is used to store cladding sample ratio information corresponding to different simulation results. Input any simulation result obtained by the laser cladding simulation model into the sample information library to obtain cladding sample ratio information that matches the current simulation result. Prepare cladding samples with the working parameter combination corresponding to the current simulation result based on the cladding sample ratio information. Repeat the above operation and input all simulation results into the sample information library to obtain the cladding sample ratio information corresponding to each simulation result, and prepare the cladding samples under different process parameter combinations; perform feature analysis on all cladding samples to obtain relevant analysis data of the cladding samples, including geometric feature parameters, microscopic feature parameters, mechanical property parameters, and functional performance parameters.

5. The blade repair method based on the laser cladding simulation model according to claim 4 is characterized in that: Based on the correlation analysis data, a performance prediction model is established to map the corresponding relationship between process parameter combinations and blade coating performance. The performance prediction model is operated based on the test data. The process of predicting the blade coating performance under different process parameter combinations includes: Perform initial model selection based on the amount of relevant analysis data, then build the model architecture corresponding to the initial performance prediction model, and define the input features and output targets corresponding to the initial performance prediction model; The input features are different combinations of process parameters; the output targets are different coating performance indicators corresponding to the blade, including blade hardness, residual stress and defect rate; Construct a functional mapping relationship between a process parameter combination and a coating performance indicator. Set a partition ratio to divide the relevant analysis data into a training set and a validation set. Input the training set into the performance prediction model under the initial model architecture for model training. Use the validation set to obtain the corresponding model evaluation index. When the model evaluation index meets the requirements, construct the final performance prediction model based on the functional mapping relationship between the process parameter combination and the blade coating performance. The test data is input into the performance prediction model to predict the coating performance of the blade under different process parameter combinations. The coating performance of the blade includes four performance levels: excellent, good, qualified and unqualified.

6. A blade repair system based on a laser cladding simulation model, used to implement the blade repair method according to any one of claims 1 to 5, characterized in that: The system includes: Laser cladding simulation module, used to establish a laser cladding simulation model and set model parameters. It sets high-speed laser cladding process parameters as model variables and inputs them into the laser cladding simulation model to simulate the laser cladding process under different parameter combinations and obtain simulation results of the cladding layer. The sample characteristic analysis module is used to prepare cladding samples under different process parameter combinations based on the simulation results, perform characteristic analysis on the cladding samples, and record relevant analysis data; The coating performance prediction module establishes a performance prediction model that maps the corresponding relationship between process parameter combinations and blade coating performance based on relevant analysis data. The performance prediction model is operated based on test data to predict the coating performance of the blade under different process parameter combinations. The process parameter optimization module is used to set performance constraints based on the specific service environment conditions of the blade, optimize the process parameter combination with the goal of achieving the best comprehensive coating performance, and determine the best preparation process parameter combination; The blade repair quality inspection module uses the optimal combination of preparation process parameters to perform laser cladding repair on actual damaged blades, monitors the key working parameters of the laser cladding process in real time, and decides whether to intervene. After the repair is completed, the blade is quality inspected.

Images