Arc Additive Manufacturing Method for Martensitic Stainless Steel Impeller Blades

Abstract

The present invention discloses an arc additive method for martensitic stainless steel impeller blades, which mainly includes pre-processing of the substrate, including substrate fixing, substrate cleaning and substrate preheating; additive manufacturing of the impeller blades, during additive manufacturing, controlling the interlayer temperature, monitoring the molten pool state and using vibration method to eliminate thermal stress; post-processing of the impeller blades obtained by additive manufacturing, including heat treatment, testing and machining. By controlling the interlayer temperature, controlling the process parameters of the additive process, segmented vibration aging and overall vibration aging, the accumulation of thermal stress in the additive process is effectively eliminated. The subsequent heat treatment, on the one hand, regulates the microstructure and mechanical properties of the martensitic stainless steel impeller blades, and on the other hand, further releases the residual thermal stress.

Description

Technical Field

[0001] The present invention relates to the field of arc material addition technology, and in particular to an arc material addition process method for martensitic stainless steel impeller blades. Background Art

[0002] As an important mechanical component, impellers are widely used in industrial and automotive fields such as pumps, fans, automobiles and ships. Unlike applications in other fields, impellers in the marine field are generally large in size due to the large power required. In addition, in order to meet the requirements of fluid dynamics, the impeller has a complex shape. The traditional production and manufacturing method of marine impellers is through billet subtractive manufacturing, that is, the alloy material is used as a blank by casting or forging, and then semi-finished or finished by CNC machine tools. In special cases, manual polishing may be required to form the final finishing. This subtractive manufacturing production process is complex and will generate a large amount of processing waste, resulting in serious material waste. In addition, for special large impeller blades, casting requires re-molding, which is costly.

[0003] In contrast, additive manufacturing can directly obtain the structural outline of a component, allowing subsequent subtractive machining to achieve the desired dimensions, even without further machining. This significantly shortens production cycles and maximizes material utilization. Compared to other additive manufacturing methods, arc additive manufacturing offers advantages such as simple equipment, high production efficiency, and the ability to manufacture parts without the constraints of equipment space. This makes it particularly suitable for the additive manufacturing of large components such as marine impellers.

[0004] When additive manufacturing parameters are inappropriate, components are prone to defects such as porosity, lack of fusion, and microcracks, reducing material performance and service life. In the actual additive manufacturing process for large workpieces, especially those with complex shapes, fluctuations in external conditions such as ambient temperature and humidity, as well as equipment factors, can cause certain additive manufacturing parameters to become unstable throughout the entire process. This necessitates timely adjustment of these parameters. Furthermore, when manufacturing large impeller blades, the complex thermal field can easily generate thermal stresses in the component, leading to deformation and even macrocracks, resulting in scrap. Traditional methods for removing these thermal stresses include post-weld heat treatment and vibration aging. However, these post-treatments can pose significant challenges. Due to the significant accumulation of thermal stress in large additive manufacturing components, stress relief during thermal annealing can lead to localized stress concentrations, severely impacting the overall morphology and dimensions of the component. Furthermore, vibration aging can only achieve limited thermal stress removal, making it difficult to meet product requirements. Maintaining component microstructural stability and reducing thermal stress during the additive manufacturing process are key to achieving high-quality arc-assisted martensitic stainless steel impeller blades.

[0005] Therefore, research innovation is needed to solve the above problems. Summary of the Invention

[0006] The purpose of the invention is to provide an arc additive method for martensitic stainless steel impeller blades to solve the above-mentioned problems existing in the prior art.

[0007] Technical solution, according to one aspect of the present application, an arc additive method for martensitic stainless steel impeller blades comprises the following steps:

[0008] Step S1, pre-treating the substrate, including substrate fixing, substrate cleaning and substrate preheating;

[0009] Step S2: performing additive manufacturing on the impeller blades. During the additive manufacturing process, the interlayer temperature is controlled, the molten pool state is monitored, and a vibration method is used to eliminate thermal stress.

[0010] Step S3: performing post-processing on the impeller blades obtained by additive manufacturing, including heat treatment, testing and machining.

[0011] According to one aspect of the present application, step S1 further comprises:

[0012] Step S11: Based on the size of the impeller blades, a material of a predetermined size is selected as an additive manufacturing substrate, and the substrate is fastened to the carrier platform with bolts to prevent deformation caused by excessive heat accumulation in the substrate during the additive process;

[0013] Step S12: cleaning the substrate surface to remove oil stains, water stains and oxide film on the surface;

[0014] Step S13: placing a predetermined number of ceramic heating plates under the substrate to heat the substrate and the additive impeller blades; and wrapping the ceramic heating plates with thermal insulation asbestos to reduce heat dissipation.

[0015] According to one aspect of the present application, step S2 further comprises:

[0016] Step S21: During additive manufacturing, continuously heating the ceramic sheet to maintain a predetermined interlayer temperature;

[0017] Step S22: using the arc welding molten pool visual perception module to collect a molten pool image of a predetermined area, extracting molten pool defects in the additive process from the molten pool image, and calling a preconfigured control parameter set to adjust additive process parameters based on the monitoring result;

[0018] Step S23 : For each predetermined height of the added material, the vibration spectrum in the multi-dimensional spectrum vibration module is called and applied to the impeller blades to remove the thermal stress of the added impeller blades.

[0019] According to one aspect of the present application, step S3 is further:

[0020] Step S31, normalizing the obtained impeller blades at 900-1200° C. and tempering them at 500-650° C.;

[0021] Step S32: perform dimensional inspection, non-destructive inspection, chemical composition inspection and mechanical property inspection; if the inspection is qualified, perform machining and surface treatment.

[0022] According to one aspect of the present application, in step S2, the process of calling a preconfigured control parameter set to adjust the additive process parameters is further as follows:

[0023] Step S221: Using a numerical simulation method to establish a three-dimensional finite element model of the impeller blade, simulating the temperature field, stress field, and deformation field during the additive process based on the initial additive process parameters and boundary conditions, dividing the impeller blade into a predetermined number of slices based on the three-dimensional finite element model, and outputting characteristic indicators of each slice, including maximum principal stress, maximum principal strain, and maximum displacement;

[0024] Step S222: Using a machine learning method to establish a relationship model between additive process parameters and the number of defects, the model is trained and optimized based on the training data set and the validation data set, and a predicted value of the number of defects corresponding to each set of additive process parameters is output;

[0025] Step S223: Optimizing additive process parameters using a genetic algorithm, searching for an optimal solution based on an objective function and constraints, and outputting the optimal additive process parameters for each slice; the objective function is to minimize the number of defects, and the constraints include ensuring that stress, strain, and displacement do not exceed limit values;

[0026] Step S224: configure the obtained additive process parameters in the arc additive equipment, and call the process parameters corresponding to each slice in sequence during operation.

[0027] According to one aspect of the present application, step S23 further includes:

[0028] Step S231: Perform spectrum analysis on the material-reinforced impeller blade using a spectrum analysis method to obtain the resonant frequency and multi-dimensional vibration mode of the impeller blade, and record the amplitude and phase of each resonant frequency;

[0029] Step S232: Use a multi-dimensional spectrum vibration aging method to screen out five multi-dimensional resonance peaks that can produce maximum dynamic stress and optimal stress superposition effect, and perform multi-dimensional vibration aging treatment on the reinforced impeller blade to eliminate and equalize residual stress.

[0030] Step S233: forming a multi-dimensional spectrum vibration module and configuring it in the arc additive equipment. During operation, the vibration spectrum in the multi-dimensional spectrum vibration module is called and acts on the impeller blades.

[0031] According to one aspect of the present application, the process of establishing the temperature field, stress field, and deformation field in step S221 further includes:

[0032] Step S221a: call a three-dimensional finite element model, including a substrate, a deposited layer, and an arc heat source, and use simulation software to perform simulation calculations;

[0033] Step S221b: Construct and use a double ellipsoid heat source model to describe the heat input characteristics of the arc heat source, and determine the heat source parameters based on the experimentally measured current, voltage, and speed; use the birth-death unit method to simulate the formation and disappearance of the deposited layer during the additive process, and determine the position and thickness of the deposited layer based on the experimentally set number of additive layers and passes;

[0034] Step S221c: using a coupled temperature-displacement-phase change analysis method to solve the temperature field, stress field, and phase change field during the additive manufacturing process, taking into account the nonlinear constitutive relationship and phase change latent heat effect of stainless steel;

[0035] Step S221d: Output the temperature field distribution diagram and node thermal cycle curve for different numbers of additive layers, and compare them with the experimentally measured temperature data and metallographic structure to verify the accuracy and reliability of the model.

[0036] According to one aspect of the present application, in step S221, the process of dividing the impeller blade into a predetermined number of slices according to the three-dimensional finite element model is further as follows:

[0037] Step S221i: Segment and slice the three-dimensional model of the impeller blade according to the arc additive manufacturing process requirements and basic parameters of each part of the impeller blade to generate deposition paths and process parameters; the basic parameters include thickness, shape, and position;

[0038] Step S221ii: Determine the adverse effects of thermal stress on each slice based on its shape and printing order, evaluate the thermal stress based on the acquired image and temperature, and optimize the printing process parameters based on the evaluation results;

[0039] Step S221iii: Using image processing software and a temperature sensor, the pre-printing process of each slice is monitored in real time, and parameters such as the image and temperature of each slice are obtained. These parameters are compared with preset target values to determine whether each slice achieves the expected effect.

[0040] Step S221iv: If the expected effect is achieved, the slice parameters are determined; if the expected effect is not achieved, the slice parameters are modified again until a slice set is obtained.

[0041] In another embodiment of the present application, slices corresponding to key areas can be extracted, and then these areas can be printed with emphasis and precision, thereby improving overall molding efficiency. In addition, slicing the model also facilitates the subsequent use of a spatiotemporal sequence model to generate high-precision process parameters, enabling precise control of process parameters.

[0042] In another embodiment of the present application, the process of identifying key slices is as follows:

[0043] The three-dimensional finite element model is retrieved and the curvature, normal vector, thickness and other parameters of each triangular face in the model are calculated; thresholds or ranges are set according to the printing quality standards, and areas exceeding or below these thresholds or ranges are marked to form slices of the key areas to be selected; the slices of the key areas to be selected are modified or optimized by adjusting the resolution, smoothness, topology, etc. to obtain a set of key area slices; printing parameters such as layer thickness and printing speed are set for different areas of each key area slice, and the optimal slice parameters are adjusted and obtained based on the printing requirements and characteristics, and then configured to the printing device.

[0044] According to one aspect of the present application, in step S22, the process of collecting a molten pool image of a predetermined area by using an arc welding molten pool visual perception module, and extracting molten pool defects in the additive process by using the molten pool image further includes:

[0045] Step S22a, collecting a molten pool image of a predetermined area through an arc welding molten pool visual perception module;

[0046] Step S22b: Construct and use a convolutional neural network and an attention mechanism to extract and classify features from the image, identify different types of defects, and output the number and location of defects, including holes, cracks, and pores.

[0047] Step S22c: reconstruct and generate the image using a variational autoencoder, estimate the missing or blocked parts of the image, and output a complete defect image;

[0048] Step S22d: Use a deep residual network to perform regression analysis on the image, predict the temperature value corresponding to each pixel in the image, and output a temperature field distribution map; use a conditional generative adversarial network to transform and synthesize the image, generate the tissue structure corresponding to each pixel in the image based on the temperature field and phase transition law in the image, and output the tissue structure map;

[0049] Step S22e: Find the location of the defect image from the organizational structure diagram, and output the defect image including data including the temperature field and phase change parameters.

[0050] According to one aspect of the present application, step S222 further includes:

[0051] Step S222a: Using a multivariate spatiotemporal sequence model, model and predict the core variables in the additive process, taking into account the interdependence and temporal dynamics of the core variables, and outputting the future value of each variable; the core variables include current, voltage, speed, temperature, and stress;

[0052] Step S222b: Build and use a graph neural network model to learn and represent the relationship between core variables in the additive process, automatically extract the hidden spatial dependencies between the core variables, and output the relationship strength between each pair of core variables.

[0053] Beneficial effects: By controlling interlayer temperature, process parameters, and both segmented and overall VSA treatments, thermal stress accumulation during the additive process is effectively eliminated. Subsequent heat treatment not only modulates the microstructure and mechanical properties of the martensitic stainless steel impeller blades, but also further releases residual thermal stresses. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is a flow chart of the present invention.

[0055] Figure 2 It is a flow chart of step S1 of the present invention.

[0056] Figure 3 It is a flow chart of step S2 of the present invention.

[0057] Figure 4 It is a flow chart of step S3 of the present invention.

[0058] Figure 5 It is a simulation digital-analog schematic diagram of the present invention. DETAILED DESCRIPTION

[0059] like Figure 1 As shown, according to one aspect of the present application, an arc additive method for martensitic stainless steel impeller blades includes the following steps:

[0060] Step S1, pre-treating the substrate, including substrate fixing, substrate cleaning and substrate preheating;

[0061] Step S2: performing additive manufacturing on the impeller blades. During the additive manufacturing process, the interlayer temperature is controlled, the molten pool state is monitored, and a vibration method is used to eliminate thermal stress.

[0062] Step S3: performing post-processing on the impeller blades obtained by additive manufacturing, including heat treatment, testing and machining.

[0063] In this embodiment, by controlling the interlayer temperature, segmented vibration aging and overall vibration aging effectively eliminate the thermal stress accumulation in the additive process. The subsequent heat treatment, on the one hand, regulates the microstructure and mechanical properties of the martensitic stainless steel impeller blades, and on the other hand, further releases the residual thermal stress. Thanks to the stress elimination before and during the additive process, the stress released by the heat treatment is relatively small, and will not cause changes in the macroscopic morphology and size of the additive workpiece, thus avoiding problems such as increased process steps and extended production cycles due to dimensional correction, and even the scrapping of workpieces due to failure to meet dimensional requirements. With the help of the arc welding pool visual perception image system to monitor and control the microscopic defects of additive components, product quality control is effectively achieved, the yield of workpieces is improved, and costs are reduced.

[0064] According to one aspect of the present application, step S1 further comprises:

[0065] Step S11: Based on the size of the impeller blades, a material of a predetermined size is selected as an additive manufacturing substrate, and the substrate is fastened to the carrier platform with bolts to prevent deformation caused by excessive heat accumulation in the substrate during the additive process;

[0066] Based on the size and shape of the impeller blades, select an appropriate support platform. The support platform should have sufficient strength and rigidity to withstand the thermal stress and deformation during the additive process. Place the substrate on the support platform, aligning the centerlines of the substrate and the support platform so that the substrate is parallel to the support platform. Drill holes and install bolts at the four corners of the substrate to lock the substrate to the support platform and prevent it from moving or falling off during the additive process. Securing the substrate with bolts ensures stability and reliability during the additive process, preventing the substrate from moving or falling off, and improving additive quality and efficiency.

[0067] Step S12: Clean the substrate surface to remove oil, water stains, and oxide films. Use a brush or cloth to remove dust, oil, water stains, and other contaminants from the substrate surface. Use alcohol or other solvents to clean the substrate surface to remove residual oil, water stains, and other substances. Use sandpaper or other abrasives to polish the substrate surface to remove oxide films, rust, and other substances, and to improve surface roughness. This removes surface contaminants and oxides, improves surface cleanliness and activity, ensures metallurgical bonding and structural properties during the additive process, and avoids defects such as oxidation, pores, and cracks.

[0068] Step S13: A predetermined number of ceramic heating plates are placed beneath the baseplate to heat the baseplate and the additive impeller blades. The ceramic heating plates are wrapped with insulating asbestos to reduce heat dissipation. To address the high sensitivity of martensitic stainless steel to cold cracking, the ceramic heating plates are used to uniformly heat the baseplate to above 200°C before additive manufacturing, effectively preventing the occurrence of cold cracks.

[0069] According to one aspect of the present application, step S2 further comprises:

[0070] Step S21: During additive manufacturing, continuously heating the ceramic sheet to maintain a predetermined interlayer temperature;

[0071] Ceramic heating plates are used as auxiliary heat sources to uniformly heat the substrate and additive impeller blades to ensure that the temperature of the entire workpiece is above 200°C. According to the additive height of the workpiece, the number of ceramic heating plates is increased in stages to ensure heating efficiency and uniformity of the workpiece temperature field. Using equipment such as temperature sensors or infrared thermal imagers, the temperature changes in the deposition area and surrounding areas are measured in real time. Based on the temperature data, parameters such as the cooling rate and temperature gradient are calculated. According to the preset interlayer temperature target value, the power and time of the ceramic plate heating are automatically adjusted, or the auxiliary cooling device is started to control the interlayer temperature within an appropriate range. By controlling the interlayer temperature, the cooling rate and temperature gradient in the additive process can be effectively optimized to obtain a fine and uniform martensitic structure with high strength and toughness, and avoid structural defects caused by overheating or overcooling.

[0072] Step S22: using the arc welding molten pool visual perception module to collect a molten pool image of a predetermined area, extracting molten pool defects in the additive process from the molten pool image, and calling a preconfigured control parameter set to adjust additive process parameters based on the monitoring result;

[0073] The arc welding pool visual perception module collects images of the molten pool in a predetermined area, and uses image processing software to analyze the molten pool images to extract characteristic parameters such as the shape, size, temperature, and surface tension of the molten pool. Based on the molten pool characteristic parameters, it is determined whether the molten pool has defects such as pores, cracks, spatter, etc., and the number, location, and degree of the defects are calculated. Depending on the situation of the defect, the pre-configured control parameter set is called, and appropriate control parameters such as current, voltage, speed, shielding gas, etc. are selected. The welding parameters are adjusted in time according to the feedback signal to eliminate or reduce the occurrence of defects. By timely adjusting the additive process parameters, the characteristic parameters such as the shape, size, temperature, and surface tension of the molten pool can be effectively controlled to ensure the stability and quality of the molten pool during the additive process and avoid defects.

[0074] Step S23 : For each predetermined height of the added material, the vibration spectrum in the multi-dimensional spectrum vibration module is called and applied to the impeller blades to remove the thermal stress of the added impeller blades.

[0075] According to the residual stress distribution of the impeller blades, the target area and direction where stress needs to be eliminated are determined. According to the amplitude and phase of each resonant frequency, the stress wave amplitude and phase generated by each resonant frequency in the target area and direction are calculated, and the total stress wave is obtained according to the superposition principle. For each predetermined height of the additive, the vibration spectrum in the multi-dimensional spectrum vibration module is called, and a vibration wave with an opposite phase to the total stress wave is applied to the impeller blades through the exciter to achieve destructive interference between the stress waves, thereby removing the thermal stress of the additively added impeller blades. The multi-dimensional spectrum vibration method can effectively remove the thermal stress of the additively added impeller blades, improve the defects such as internal stress, residual stress and surface roughness of the workpiece after additive, and improve the mechanical properties of the workpiece such as strength, toughness, corrosion resistance and wear resistance.

[0076] In a further embodiment, the process of optimizing the multi-dimensional resonance peak is as follows:

[0077] Fix the workpiece on the vibration platform, select appropriate excitation points and pickup points, and connect the exciter and sensor.

[0078] Start the vibrator to make the workpiece vibrate periodically and collect the sensor signal at the same time.

[0079] Use Fourier transform or other mathematical methods to perform spectrum analysis on the signal and obtain the spectrum diagram of the artifact.

[0080] Identify the resonant frequency and corresponding vibration mode of the workpiece from the spectrum diagram, and record the amplitude and phase of each resonant frequency.

[0081] Based on the residual stress distribution in the workpiece, the target areas and directions for stress relief are determined. Based on the amplitude and phase of each resonant frequency, the magnitude and direction of the dynamic stress generated by each resonant frequency in the target area and direction are calculated. From all resonant frequencies, frequencies that result in the same or similar directions as the residual stress are selected as candidate frequencies. From the candidate frequencies, five frequencies that result in the same or similar amplitudes as the residual stress are selected as the optimal frequencies. If there are fewer than five candidate frequencies, additional frequencies are selected from other resonant frequencies. Each optimal frequency is verified and adjusted to ensure it effectively eliminates and equalizes residual stress.

[0082] According to one aspect of the present application, step S3 is further:

[0083] Step S31, normalizing the impeller blades obtained at 900-1200°C and tempering them at 500-650°C; placing the impeller blades obtained by additive manufacturing in a heat treatment furnace, heating them to 900-1200°C, keeping them warm for a certain period of time, then cooling them to room temperature and performing normalizing treatment. The purpose of normalizing treatment is to eliminate the residual stress generated during the additive process, improve the microstructure, and increase the strength and toughness. The impeller blades after normalizing treatment are placed in a heat treatment furnace again, heating them to 500-650°C, keeping them warm for a certain period of time, then cooling them to room temperature and performing tempering treatment. The purpose of tempering treatment is to stabilize the martensitic structure, reduce hardness and brittleness, and improve corrosion resistance and wear resistance. Through heat treatment, the microstructure and mechanical properties of the impeller blades after additive manufacturing can be effectively improved, so that they have higher strength, toughness, corrosion resistance and wear resistance, meeting marine requirements.

[0084] Step S32: Perform dimensional inspection, non-destructive inspection, chemical composition inspection and mechanical property inspection; if the inspection is qualified, perform machining and surface treatment. In a certain embodiment, the following scheme can be adopted: perform dimensional inspection on the impeller blades after heat treatment, use a three-coordinate measuring machine or other precision instruments to measure the outer dimensions, form and position tolerances, surface roughness and other parameters of the impeller blades, compare them with the design requirements, and determine whether they are qualified. Perform non-destructive inspection on the impeller blades after heat treatment, use an ultrasonic flaw detector or other non-destructive inspection equipment to detect whether there are cracks, pores, inclusions and other defects inside the impeller blades, and evaluate them according to national standards or industry standards to determine whether they are qualified. Perform chemical composition inspection on the impeller blades after heat treatment, use a spectrometer or other chemical analysis equipment to analyze the content and proportion of various elements in the impeller blades, compare them with the design requirements, and determine whether they are qualified. The mechanical properties of the impeller blades after heat treatment are tested. A tensile testing machine or other mechanical testing equipment is used to perform tensile tests on the impeller blades at different temperatures to measure parameters such as tensile strength, yield strength, and elongation. Parameters such as elastic modulus and Poisson's ratio are calculated and compared with the design requirements to determine whether they are qualified. If all the above tests are qualified, the impeller blades are machined and surface treated. Machining includes removing burrs, spatter, excess material, etc. generated during the additive process, as well as finishing the impeller blades to achieve the size and shape required by the design. Surface treatment includes polishing, sandblasting, electroplating, etc. on the impeller blades to improve their surface quality and appearance. Through testing and processing, it can be effectively ensured that the impeller blades after additive manufacturing meet the design requirements, meet marine requirements, and improve the performance and quality of the product.

[0085] According to one aspect of the present application, in step S2, the process of calling a preconfigured control parameter set to adjust the additive process parameters is further as follows:

[0086] Step S221: Using a numerical simulation method to establish a three-dimensional finite element model of the impeller blade, simulating the temperature field, stress field, and deformation field during the additive process based on the initial additive process parameters and boundary conditions, dividing the impeller blade into a predetermined number of slices based on the three-dimensional finite element model, and outputting characteristic indicators of each slice, including maximum principal stress, maximum principal strain, and maximum displacement;

[0087] Using pre-configured software, draw a three-dimensional geometric model of the impeller blades according to the design drawings of the impeller blades and export it to a file in a predetermined format. Using finite element analysis software, import the file in a predetermined format, mesh the impeller blades, generate a three-dimensional finite element model, and set the appropriate unit type and unit size. Using finite element analysis software, simulate the temperature field, stress field, and deformation field during the additive process based on the initial additive process parameters and boundary conditions, and output the characteristic indicators of each slice, including the maximum principal stress, maximum principal strain, and maximum displacement. By establishing a three-dimensional finite element model of the impeller blades, the temperature field, stress field, and deformation field during the additive process can be effectively simulated, and the characteristic indicators of each slice can be obtained, providing basic data for subsequent process parameter optimization.

[0088] Step S222: Utilize machine learning methods to establish a relationship model between additive process parameters and the number of defects. The model is trained and optimized based on the training and validation datasets, and a predicted number of defects corresponding to each set of additive process parameters is output. In one embodiment, arc additive manufacturing can be performed on the same material using different combinations of additive process parameters (e.g., current, voltage, speed, shielding gas, etc.). Defect detection is then performed on the post-additive manufacturing samples, and the number of defects corresponding to each set of additive process parameters is counted. Utilize machine learning methods to select an appropriate algorithm (e.g., linear regression, neural network, support vector machine, etc.). Based on the experimental dataset, a relationship model between additive process parameters and the number of defects is established, and the model is trained and optimized. Utilize machine learning methods to test and evaluate the model based on the validation dataset, and a predicted number of defects corresponding to each set of additive process parameters is output. By establishing a relationship model between additive process parameters and the number of defects, the number of defects corresponding to different combinations of additive process parameters can be effectively predicted, providing a reference for subsequent process parameter optimization.

[0089] Step S223: Optimizing additive process parameters using a genetic algorithm, searching for an optimal solution based on an objective function and constraints, and outputting the optimal additive process parameters for each slice; the objective function is to minimize the number of defects, and the constraints include ensuring that stress, strain, and displacement do not exceed limit values;

[0090] Using a genetic algorithm, a set of feasible solutions, i.e., a set of additive process parameters that satisfy the constraints, is initialized based on the objective function and constraints. The genetic algorithm is used to calculate the objective function value for each feasible solution, i.e., predict the corresponding number of defects based on the relational model, and sort and select the feasible solutions based on the number of defects. The genetic algorithm is used to perform crossover and mutation operations on the selected feasible solutions to generate new feasible solutions, and the previous steps are repeated until the termination conditions are met, such as reaching the maximum number of iterations or the optimal solution remains stable. The genetic algorithm is used to output the optimal solution, i.e., the optimal additive process parameters for each slice. By optimizing the additive process parameters, the optimal process parameters can be effectively selected based on the characteristic indicators and predicted number of defects of each slice, ensuring the stability and quality of the melt pool during the additive process and minimizing the number of defects.

[0091] Step S224: configure the obtained additive process parameters in the arc additive equipment, and call the process parameters corresponding to each slice in sequence during operation.

[0092] In a certain embodiment, the optimal additive process parameters for each slice are stored in a preconfigured control parameter set, which includes the current, voltage, speed, shielding gas and other parameters corresponding to each slice. The preconfigured control parameter set is configured in the arc additive equipment so that it can automatically call the corresponding process parameters according to the order and number of the slices. During operation, the process parameters corresponding to each slice are called in sequence to perform arc additive manufacturing, and the molten pool state and interlayer temperature are monitored in real time, and fine-tuning is performed as needed. By calling the preconfigured control parameter set to adjust the additive process parameters, the optimal process parameters can be effectively selected based on the characteristic indicators of each slice and the predicted value of the number of defects, to ensure the stability and quality of the molten pool during the additive process and avoid defects.

[0093] According to one aspect of the present application, step S23 further includes:

[0094] Step S231: Perform spectrum analysis on the material-reinforced impeller blade using a spectrum analysis method to obtain the resonant frequency and multi-dimensional vibration mode of the impeller blade, and record the amplitude and phase of each resonant frequency;

[0095] Using vibration sensors or accelerometers, vibration tests are performed on the reinforced impeller blades. The vibration signals of the impeller blades at different frequencies are collected and converted into digital signals. Using spectrum analysis software, a Fourier transform is performed on the digital signals to obtain a spectrum diagram of the impeller blades. The resonant frequencies and multidimensional vibration modes of the impeller blades are then identified. The amplitude and phase of each resonant frequency are recorded using the spectrum analysis software and stored in a multidimensional spectrum vibration module for subsequent vibration aging treatment. Spectral analysis effectively determines the resonant frequencies and multidimensional vibration modes of the impeller blades, providing basic data for subsequent vibration aging treatment.

[0096] Step S232: Use a multi-dimensional spectrum vibration aging method to screen out five multi-dimensional resonance peaks that can produce maximum dynamic stress and optimal stress superposition effect, and perform multi-dimensional vibration aging treatment on the reinforced impeller blade to eliminate and equalize residual stress.

[0097] Step S233: forming a multi-dimensional spectrum vibration module and configuring it in the arc additive equipment. During operation, the vibration spectrum in the multi-dimensional spectrum vibration module is called and acts on the impeller blades.

[0098] Using a multidimensional spectrum vibration aging method, the five multidimensional resonance peaks that produce the maximum dynamic stress and optimal stress superposition effect are screened based on the amplitude and phase of each resonant frequency, and their corresponding vibration waveforms and parameters are determined. This multidimensional spectrum vibration aging method can be used to perform multidimensional vibration aging on impeller blades that have been reinforced. Vibration waves from five multidimensional resonance peaks with opposite phases to the residual stress are applied to the impeller blades via an exciter, achieving destructive interference between the stress waves, thereby eliminating and equalizing residual stress. A multidimensional spectrum vibration module is created and deployed within the arc additive manufacturing equipment. During operation, the vibration spectrum within the module is applied to the impeller blades. This multidimensional vibration aging treatment effectively eliminates and equalizes residual stress in the impeller blades that have been reinforced, improving the performance and lifespan of the workpiece.

[0099] According to one aspect of the present application, the process of establishing the temperature field, stress field, and deformation field in step S221 further includes:

[0100] Step S221a: call a three-dimensional finite element model, including the substrate, the deposited layer, and the arc heat source, and use simulation software to perform simulation calculations to reflect the thermomechanical coupling effect in the additive process, taking into account the nonlinear constitutive relationship of the material and the phase change latent heat effect.

[0101] Step S221b: Construct and use a double ellipsoid heat source model to describe the heat input characteristics of the arc heat source, and determine the heat source parameters based on the experimentally measured current, voltage, and speed; use the birth and death unit method to simulate the generation and disappearance of the deposited layer in the additive process, and determine the position and thickness of the deposited layer based on the experimentally set number of additive layers and passes; in order to accurately describe the heating effect of the arc heat source on the substrate and the deposited layer, the movement and direction change of the arc heat source are taken into account.

[0102] Step S221c: Use the coupled temperature-displacement-phase change analysis method to solve the temperature field, stress field, and phase change field in the additive process, taking into account the nonlinear constitutive relationship and latent heat effect of phase change of stainless steel; consider the mutual influence between temperature, displacement, and phase change in the additive process to obtain the distribution law of temperature, stress, and phase change.

[0103] Step S221d: Output the temperature field distribution diagram and nodal thermal cycle curve for different numbers of additive layers and compare them with the experimentally measured temperature data and metallographic structure to verify the accuracy and reliability of the model. This evaluates the model's applicability and effectiveness, providing a basis for subsequent process parameter optimization.

[0104] By establishing temperature fields, stress fields, and deformation fields, we can effectively conduct thermal analysis of the additive process, obtain the distribution patterns of temperature, stress, and phase change in the additive process, and provide basic data for subsequent process parameter optimization.

[0105] In summary, a double-ellipsoid heat source model is used to describe the heat input characteristics of the arc heat source. Heat source parameters are determined based on experimentally measured parameters such as current, voltage, and speed, accurately simulating the heating effect of the arc heat source on the deposited layer. The birth-and-death unit method is used to simulate the formation and disappearance of the deposited layer during the additive process. The position and thickness of the deposited layer are determined based on the experimentally set number of additive layers and passes, effectively simulating the geometric changes during the additive process. Thermal stresses are eliminated using a vibration method, and the molten pool state is monitored using image processing software and temperature sensors. The welding parameters are adjusted promptly based on feedback signals, effectively controlling the surface quality, defects, and microstructure of the deposited layer, thereby improving the forming quality. By optimizing process parameters, the degree of oxidation is reduced and the surface quality is improved. Furthermore, numerical simulations are used to optimize the additive process parameters and paths, avoiding blind experimentation, wasting resources, and saving costs.

[0106] According to one aspect of the present application, in step S221, the process of dividing the impeller blade into a predetermined number of slices according to the three-dimensional finite element model is further as follows:

[0107] Step S221i: Segment and slice the three-dimensional model of the impeller blade according to the arc additive manufacturing process requirements and basic parameters of each part of the impeller blade to generate deposition paths and process parameters; the basic parameters include thickness, shape, and position;

[0108] Slice thickness: Slice thickness affects deposition efficiency and quality. Excessively large slice thickness will result in loose connections between deposited layers, resulting in defects; excessively small slice thickness will result in excessively long deposition times, increasing costs. A slice thickness of 0.5-2 mm is generally selected. Slice shape: Slice shape affects the thermal stress and deformation of the deposited layer. Complex curves or curved surface shapes will result in uneven cooling of the deposited layer, resulting in greater thermal stress and deformation; simple straight lines or plane shapes will result in uniform cooling of the deposited layer, resulting in less thermal stress and deformation. Simple and continuous curves or surfaces are generally selected as slice shapes. Slice position and printing order: The printing order affects the temperature gradient between deposited layers and the interlayer temperature. A reasonable printing order can keep the temperature gradient between deposited layers within a certain range, avoiding excessively high or low interlayer temperatures. The printing order is generally selected to print layer by layer from the inside out or from the outside in.

[0109] Step S221ii: Determine the adverse effects of thermal stress on each slice based on its shape and printing order, evaluate the thermal stress based on the acquired image and temperature, and optimize the process parameters during printing based on the evaluation results; use finite element analysis software to evaluate the thermal stress of each slice, calculate the maximum principal stress, maximum principal strain, maximum displacement and other indicators generated by each slice during the printing process, and compare them with the strength, plasticity, fracture and other properties of the material to determine whether each slice exceeds the limit value or is close to the limit value.

[0110] In one embodiment, based on the thermal stress assessment results, printing parameters are adjusted for each slice to reduce thermal stress and deformation. If a slice has a simple shape and is less susceptible to thermal stress, the printing speed can be increased, while current and voltage can be reduced to improve deposition efficiency and reduce costs. If a slice has a complex shape and is more susceptible to thermal stress, the printing speed can be reduced, while current and voltage can be increased to ensure deposition quality and minimize deformation.

[0111] Step S221iii: Using image processing software and a temperature sensor, the pre-printing process of each slice is monitored in real time, and parameters such as the image and temperature of each slice are obtained. These parameters are compared with preset target values to determine whether each slice achieves the expected effect.

[0112] Based on the monitoring results of parameters such as images and temperature, feedback and corrections are provided to the printing parameters of each slice to achieve adaptive control of the printing process. If the image of a slice shows problems such as uneven deposits or obvious defects, the printing speed can be reduced, and parameters such as current and voltage can be increased to improve the surface quality and integrity of the deposited layer. If the temperature of a slice shows that the deposited layer is overheating or undercooling, the printing speed, current, voltage and other parameters can be increased or decreased to adjust the temperature gradient and interlayer temperature of the deposited layer.

[0113] Step S221iv: If the expected effect is achieved, the slice parameters are determined; if the expected effect is not achieved, the slice parameters are modified again until a slice set is obtained.

[0114] According to one aspect of the present application, in step S22, the process of collecting a molten pool image of a predetermined area by using an arc welding molten pool visual perception module, and extracting molten pool defects in the additive process by using the molten pool image further includes:

[0115] Step S22a, collecting a molten pool image of a predetermined area through an arc welding molten pool visual perception module;

[0116] High-speed cameras or other image acquisition devices are used to capture the morphology and dynamic changes of the melt pool during the arc additive process in real time and convert them into digital images. By acquiring melt pool images, information on the morphology and dynamic changes of the melt pool during the arc additive process can be effectively obtained, providing basic data for subsequent defect extraction and analysis.

[0117] Step S22b: Construct and use a convolutional neural network and an attention mechanism to extract and classify features from the image, identify different types of defects, and output the number and location of defects, including holes, cracks, and pores; use a convolutional neural network to extract features from the image, use an attention mechanism to divide the image into regions and assign weights, use a classifier to predict labels for the image, identify different types of defects, and output the number and location of defects.

[0118] Step S22c: reconstruct and generate the image using a variational autoencoder, estimate the missing or occluded parts in the image, and output a complete defective image; encode and decode the image using a variational autoencoder, sample the latent variables using the reparameterization technique, reconstruct and generate the image using a generator, estimate the missing or occluded parts in the image, and output a complete defective image.

[0119] Step S22d: Use a deep residual network to perform regression analysis on the image, predict the temperature value corresponding to each pixel in the image, and output a temperature field distribution map; use a conditional generative adversarial network to transform and synthesize the image, generate the tissue structure corresponding to each pixel in the image based on the temperature field and phase transition law in the image, and output a tissue structure map; use a deep residual network to perform regression analysis on the image, use the residual module to enhance feature expression capabilities, use a regressor to predict the temperature value corresponding to each pixel in the image, and output a temperature field distribution map. Use a conditional generative adversarial network to transform and synthesize the image, use a discriminator to determine the authenticity of the generated image, use a generator to generate the tissue structure based on the temperature field and phase transition law, and output a tissue structure map.

[0120] Step S22e: Find the location of the defect image in the organizational structure diagram and output the defect image, including data on the temperature field and phase transformation parameters. In the organizational structure diagram, different phase transformation regions, such as austenite, martensite, and ferrite, are distinguished based on different colors or grayscale values. Based on the location of the defect image, the corresponding phase transformation region is found and the defect image, including data on the temperature field and phase transformation parameters, is output.

[0121] By extracting melt pool defects, different types of defects can be effectively identified, and the number and location of defects can be output, providing a reference for subsequent process parameter adjustments. Furthermore, missing or obscured portions of the image can be effectively estimated, and a complete defect image can be output, providing complete data for subsequent defect analysis. The temperature value and microstructure corresponding to each pixel in the image can also be effectively predicted, and temperature field distribution maps and microstructure diagrams can be output, providing basic data for subsequent temperature field and phase change field analysis.

[0122] According to one aspect of the present application, step S222 further includes:

[0123] Step S222a: Using a multivariate spatiotemporal sequence model, model and predict the core variables in the additive process, taking into account the interdependence and temporal dynamics of the core variables, and outputting the future value of each variable; the core variables include current, voltage, speed, temperature, and stress;

[0124] Using a multivariate spatiotemporal sequence model, we model and predict core variables in the additive process, utilizing deep learning or statistical learning methods such as long-short-term memory networks and vector autoregressive models. Based on historical datasets, we build a multivariate spatiotemporal sequence model, train it, and optimize it. This model predicts core variables in the additive process, outputs the future value of each core variable, and adjusts and controls process parameters based on the predicted values. Core variables include current, voltage, speed, temperature, and stress.

[0125] By establishing a multivariate spatiotemporal sequence model, the core variables in the additive process can be effectively modeled and predicted, taking into account the interdependence and temporal dynamics among the core variables, and providing a basis for subsequent process parameter adjustments.

[0126] Step S222b: Build and use a graph neural network model to learn and represent the relationship between core variables in the additive process, automatically extract the hidden spatial dependencies between the core variables, and output the relationship strength between each pair of core variables.

[0127] The parameters of each slice are predicted using a multivariate spatiotemporal sequence model, which improves the precision of simulation and prediction.

[0128] A graph neural network model is constructed and utilized to learn and represent the relationships between core variables in the additive process. Using graph structures and neural network methods, such as graph convolutional networks and graph attention networks, the model is built based on historical datasets and trained and optimized. The graph neural network model represents the relationships between core variables in the additive process. The model outputs the strength of the relationship between each pair of core variables, and process parameters are adjusted and controlled based on this relationship strength. Core variables include current, voltage, speed, temperature, and stress.

[0129] By establishing a graph neural network model, we can effectively learn and represent the relationship between the core variables in the additive process, automatically extract the hidden spatial dependencies between the core variables, and provide a basis for subsequent process parameter adjustments.

[0130] Another embodiment of the present application further includes a laser unit and / or ultrasonic unit for assisting arc additive manufacturing, wherein a laser beam is added in front of the arc heat source as an auxiliary heat source. An ultrasonic generator is installed above the workpiece to apply periodic ultrasonic radiation to the workpiece. This can utilize the cavitation effect and micro-jet effect of ultrasound to improve the flow of the molten pool and dendrite growth on a microscale, thereby obtaining a fine and uniform martensitic structure. The high energy density and high speed characteristics of the laser are utilized to achieve precise temperature control and stress release in a short time.

[0131] Before and after each additive pass, periodic ultrasonic radiation is applied to the workpiece with a radiation frequency range of 20-40kHz and a radiation time of 10-20 s.

[0132] After each layer is deposited, the deposited layer is locally laser heated at a temperature of 500-600°C for 1-2 s.

[0133] In another embodiment of the present application, it also includes using equipment such as temperature sensors or infrared thermal imagers to measure the temperature changes in the deposition area and the surrounding area in real time, and calculating parameters such as the cooling rate and temperature gradient based on the temperature data. According to the preset target values, the welding parameters or auxiliary cooling devices are automatically adjusted to control the tissue performance.

[0134] In one embodiment, the parameters are as follows:

[0135] like Figure 1 As shown in the figure, based on the 3D modeling of the workpiece, the impeller blades are 1400 mm long and 1100 mm high. Based on the characteristics of the arc additive manufacturing process, the digital model was optimized, with a 4 mm margin added to the impeller blades on each side and a 10 mm margin added to the top. Based on this, a Q235 low-carbon steel plate with dimensions of 1500 × 1500 mm and a thickness of 100 mm was selected as the baseplate. To prevent deformation caused by excessive heat accumulation on the baseplate during the additive process, the baseplate was secured to the support platform with bolts. The baseplate surface was cleaned to ensure it was free of oil, water stains, oxide films, and other factors that could affect additive manufacturing quality. Ceramic heating and insulation wrapping were used to maintain the overall workpiece temperature above 200°C. Furthermore, the number of ceramic heaters was increased in stages during the printing process. At 0 mm, two ceramic heaters were used to continuously heat the baseplate. Two more ceramic heaters were added at a printing height of 500 mm, and two more at a printing height of 1000 mm. All ceramic heaters were tightly wrapped with asbestos blankets to minimize heat dissipation.

[0136] The arc additive manufacturing process for a workpiece includes interlayer temperature control, melt pool monitoring, segmented vibration aging, and overall vibration aging, as follows:

[0137] A 0Cr16Ni5Mo additive wire with a diameter of 1.2mm was used, and a shielding gas mixture of 98% Ar and 2% CO2 was used. The additive parameters were: additive voltage of 15-25V, additive current of 150-250A, wire feed speed of 5-10m / min, and welding speed of 5-15mm / s. Interpass temperature control: Before each additive pass, the surface temperature of the component being added is measured. The next additive pass can only be performed when the temperature is at least 200°C. Melt pool monitoring: During the additive process, real-time visual perception of the melt pool is performed online, achieving a dimensional accuracy of ±1.5mm. Based on the visual perception of the arc weld pool, defects in the additive process, such as porosity and lack of fusion, are detected, with an accuracy rate of 70-90%. When the defect results extracted by the melt pool monitoring system are a≤K≤b, the additive manufacturing process parameters are fine-tuned, adjusting the heat input to reduce the number of defects. When K≥b, the number of defects is excessive and has significantly impacted material properties, and the workpiece is scrapped. Segmented vibration aging: To prevent excessive thermal stress accumulation, a multi-dimensional spectrum vibration aging process is used to remove thermal stress in the workpiece after each addition of a certain height h (h is set to 100mm). Specific parameters include a vibration frequency range of 50-120Hz and a duration of 30-60min. After the workpiece is added, multi-dimensional spectrum vibration aging is performed again to remove thermal stress. Specific parameters include a vibration frequency range of 50-120Hz and a duration of 120-180min. The workpiece heat treatment includes normalizing and tempering. The specific parameters are as follows: In a heat treatment furnace, normalizing is performed at 900-1200°C, followed by tempering at 500-650°C.

[0138] The unqualified requirements for mechanical properties are: tensile strength less than 800Mpa, yield strength less than 600Mpa, elongation less than 18%, cross-sectional shrinkage less than 35%, and average impact energy of Charpy V-notch impact test less than 35J.

[0139] The acceptable chemical composition percentages for chemical composition testing are: C ≤ 0.060%; Si ≤ 1.50%; Mn ≤ 2.00%; S ≤ 0.030%; P ≤ 0.035%; 15.00% ≤ Cr ≤ 17.50%; 3.50% ≤ Ni ≤ 6.00%; and Mo ≤ 1.50%. Ultrasonic testing must comply with NB / T47013.3-2015, "Nondestructive Testing of Pressure Equipment - Part 3: Ultrasonic Testing." The internal quality must not be lower than weld quality level 2. The test area must be polished flat and smooth before testing. Radiographic testing must comply with NB / T47013.2-2015, "Nondestructive Testing of Pressure Equipment - Part 2: Radiographic Testing." The internal quality must not be lower than weld quality level 2.

[0140] In summary, real-time melt pool detection and dimensional inspection and correction during the additive process control the quality of additive workpieces at both the micro and macro levels, effectively improving the success rate of workpiece qualification and subsequent rework cycles. Compared to traditional casting methods, this application has the advantage of eliminating the need for mold opening, shortening the manufacturing process, and simplifying machining, significantly reducing manufacturing costs and shortening product delivery cycles.

[0141] The preferred embodiments of the present invention are described in detail above. However, the present invention is not limited to the specific details in the above embodiments. Within the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention. These equivalent transformations all fall within the protection scope of the present invention. In different scenarios, in order to achieve corresponding effects, the parameters and may be different between the embodiments.

Claims

1. Arc additive manufacturing method for martensitic stainless steel impeller blades, characterized in that: The steps include: Step S1, pre-treating the substrate, including substrate fixing, substrate cleaning and substrate preheating; Step S2: performing additive manufacturing on the impeller blades. During the additive manufacturing process, the interlayer temperature is controlled, the molten pool state is monitored, and a vibration method is used to eliminate thermal stress. Step S3, post-processing the impeller blades obtained by additive manufacturing, including heat treatment, testing and machining; The step S2 is further as follows: Step S21: During additive manufacturing, continuously heating the ceramic sheet to maintain a predetermined interlayer temperature; Step S22: using the arc welding molten pool visual perception module to collect a molten pool image of a predetermined area, extracting molten pool defects in the additive process from the molten pool image, and calling a preconfigured control parameter set to adjust additive process parameters based on the monitoring result; Step S23: For each predetermined height of the added material, the vibration spectrum in the multi-dimensional spectrum vibration module is called and applied to the impeller blades to remove thermal stress of the added impeller blades; In step S2, the process of calling the pre-configured control parameter set to adjust the additive process parameters is further as follows: Step S221: Using a numerical simulation method to establish a three-dimensional finite element model of the impeller blade, simulating the temperature field, stress field, and deformation field during the additive process based on the initial additive process parameters and boundary conditions, dividing the impeller blade into a predetermined number of slices based on the three-dimensional finite element model, and outputting characteristic indicators of each slice, including maximum principal stress, maximum principal strain, and maximum displacement; Step S222: Using a machine learning method to establish a relationship model between additive process parameters and the number of defects, the model is trained and optimized based on the training data set and the validation data set, and a predicted value of the number of defects corresponding to each set of additive process parameters is output; Step S223: Optimizing additive process parameters using a genetic algorithm, searching for an optimal solution based on an objective function and constraints, and outputting the optimal additive process parameters for each slice; the objective function is to minimize the number of defects, and the constraints include ensuring that stress, strain, and displacement do not exceed limit values; Step S224: configure the obtained additive process parameters in the arc additive equipment, and call the process parameters corresponding to each slice in sequence during operation.

2. The arc material addition method for martensitic stainless steel impeller blades according to claim 1, characterized in that: The step S1 is further as follows: Step S11: Based on the size of the impeller blades, a material of a predetermined size is selected as an additive manufacturing substrate, and the substrate is fastened to the carrier platform with bolts to prevent deformation caused by excessive heat accumulation in the substrate during the additive process; Step S12: cleaning the substrate surface to remove oil stains, water stains and oxide film on the surface; Step S13: placing a predetermined number of ceramic heating plates under the substrate to heat the substrate and the additive impeller blades; and wrapping the ceramic heating plates with thermal insulation asbestos to reduce heat dissipation.

3. The arc material addition method for martensitic stainless steel impeller blades according to claim 1, characterized in that: The step S3 is further as follows: Step S31, normalizing the obtained impeller blades at 900-1200° C. and tempering them at 500-650° C.; Step S32: perform dimensional inspection, non-destructive inspection, chemical composition inspection and mechanical property inspection; if the inspection is qualified, perform machining and surface treatment.

4. The arc material addition method for martensitic stainless steel impeller blades according to claim 3, characterized in that: The step S23 further includes: Step S231: Perform spectrum analysis on the material-reinforced impeller blade using a spectrum analysis method to obtain the resonant frequency and multi-dimensional vibration mode of the impeller blade, and record the amplitude and phase of each resonant frequency; Step S232: Using a multi-dimensional spectrum vibration aging method, five multi-dimensional resonance peaks that can produce maximum dynamic stress and optimal stress superposition effect are screened out, and multi-dimensional vibration aging treatment can be performed on the reinforced impeller blade to eliminate and equalize residual stress; Step S233: forming a multi-dimensional spectrum vibration module and configuring it in the arc additive equipment. During operation, the vibration spectrum in the multi-dimensional spectrum vibration module is called and acts on the impeller blades.

5. The arc material addition method for martensitic stainless steel impeller blades according to claim 1, characterized in that: The process of simulating the temperature field, stress field and deformation field in the additive process in step S221 further includes: Step S221a: call a three-dimensional finite element model, including a substrate, a deposited layer, and an arc heat source, and use simulation software to perform simulation calculations; Step S221b: Construct and use a double ellipsoid heat source model to describe the heat input characteristics of the arc heat source, and determine the heat source parameters based on the experimentally measured current, voltage, and speed; use the birth-death unit method to simulate the formation and disappearance of the deposited layer during the additive process, and determine the position and thickness of the deposited layer based on the experimentally set number of additive layers and passes; Step S221c: using a coupled temperature-displacement-phase change analysis method to solve the temperature field, stress field, and phase change field during the additive manufacturing process, taking into account the nonlinear constitutive relationship and phase change latent heat effect of stainless steel; Step S221d: Output the temperature field distribution diagram and node thermal cycle curve for different numbers of additive layers, and compare them with the experimentally measured temperature data and metallographic structure to verify the accuracy and reliability of the model.

6. The arc material addition method for martensitic stainless steel impeller blades according to claim 1, characterized in that: In step S221, the process of dividing the impeller blade into a predetermined number of slices according to the three-dimensional finite element model is further as follows: Step S221i: Segment and slice the three-dimensional model of the impeller blade according to the arc additive manufacturing process requirements and basic parameters of each part of the impeller blade to generate deposition paths and process parameters; the basic parameters include thickness, shape, and position; Step S221ii: Determine the adverse effects of thermal stress on each slice based on its shape and printing order, evaluate the thermal stress based on the acquired image and temperature, and optimize the printing process parameters based on the evaluation results; Step S221iii: Using image processing software and a temperature sensor, the pre-printing process of each slice is monitored in real time, and parameters such as the image and temperature of each slice are obtained. These parameters are compared with preset target values to determine whether each slice achieves the expected effect. Step S221iv: If the expected effect is achieved, the slice parameters are determined; if the expected effect is not achieved, the slice parameters are modified again until a slice set is obtained.

7. The arc material addition method for martensitic stainless steel impeller blades according to claim 1, characterized in that: In step S22, the arc welding molten pool visual perception module collects a molten pool image of a predetermined area, and the process of extracting molten pool defects in the additive process through the molten pool image further includes: Step S22a, collecting a molten pool image of a predetermined area through an arc welding molten pool visual perception module; Step S22b: Construct and use a convolutional neural network and an attention mechanism to extract and classify features from the image, identify different types of defects, and output the number and location of defects, including holes, cracks, and pores. Step S22c: reconstruct and generate the image using a variational autoencoder, estimate the missing or blocked parts of the image, and output a complete defect image; Step S22d: Use a deep residual network to perform regression analysis on the image, predict the temperature value corresponding to each pixel in the image, and output a temperature field distribution map; use a conditional generative adversarial network to transform and synthesize the image, generate the tissue structure corresponding to each pixel in the image based on the temperature field and phase transition law in the image, and output the tissue structure map; Step S22e: Find the location of the defect image from the organizational structure diagram, and output the defect image including data including the temperature field and phase change parameters.

8. The arc additive manufacturing method for martensitic stainless steel impeller blades according to claim 1, wherein: The step S222 further includes: Step S222a: Using a multivariate spatiotemporal sequence model, model and predict the core variables in the additive process, taking into account the interdependence and temporal dynamics of the core variables, and outputting the future value of each variable; the core variables include current, voltage, speed, temperature, and stress; Step S222b: Build and use a graph neural network model to learn and represent the relationship between core variables in the additive process, automatically extract the hidden spatial dependencies between the core variables, and output the relationship strength between each pair of core variables.

Images