A method for determining the linear energy of welding joints in upper and lower sections of a contact fluidized bed

Abstract

The present invention provides a method for determining the linear energy of a welding joint of an upper and lower section of a contact fluidized bed, which belongs to the technical field of fluidized bed welding, and comprises: establishing a temperature distribution model inside a fluidized bed and a thermal stress change model of a joint during welding; taking temperature distribution and thermal stress as two objective functions, and adopting a game theory method to find the optimal combination of welding parameters to establish a game model, so that the temperature distribution is uniform and the thermal stress change is minimized; solving the game model to obtain an optimal solution set; calculating the process parameters corresponding to each solution in the optimal solution set; according to the process parameters corresponding to each solution, obtaining the actual temperature distribution and thermal stress change of the welding joint during an actual welding test, and comparing the actual test results with the calculation results of the game model, evaluating the prediction accuracy of the game model for optimization, and finally obtaining the optimized linear energy result of the welding joint and outputting it to an operator.

Description

Technical Field

[0001] The present invention belongs to the technical field of fluidized bed welding, and in particular, relates to a method for determining the linear energy of a welding joint of an upper and lower section pair on a contact fluidized bed. Background Art

[0002] Fluidized bed technology is an important equipment technology widely used in chemical, metallurgical, energy and other fields. There are complex flow and temperature field distributions inside the fluidized bed, and the changes in these physical fields have an important impact on the performance indicators of the fluidized bed. Among them, the temperature distribution inside the fluidized bed is a key factor, which not only affects the heat transfer process of the gas-solid two-phase inside the fluidized bed, but also has an important impact on the process performance, product quality and equipment safety of the fluidized bed.

[0003] In fluidized bed equipment, contact fluidized bed is an important structural type. Contact fluidized bed improves heat transfer efficiency by directly transferring heat to the inside of the fluidized bed. It is widely used in processes such as coal gasification, catalytic reaction, and drying. In a contact fluidized bed, the fluid and solid materials are in direct contact, and the heat exchange is more sufficient, so the temperature distribution is more complex. If the temperature distribution is uneven, it will cause local overheating or cooling, affecting the reaction efficiency and product quality. Therefore, the measurement, modeling, and optimization control of the temperature field inside the contact fluidized bed have always been the focus and difficulty of fluidized bed technology research.

[0004] Most existing studies use a combination of experimental measurements, numerical simulations or empirical formulas to analyze and predict the temperature distribution inside a contact fluidized bed. For example, some scholars use sensors such as thermocouples to measure the temperature at different locations inside the fluidized bed and establish a two-dimensional or three-dimensional temperature distribution model. Other studies have established a CFD model inside the fluidized bed based on fluid mechanics and heat and mass transfer theory, and predicted the temperature field through numerical simulation. In addition, some scholars have proposed a temperature distribution prediction model based on empirical formulas, taking into account factors such as the geometric dimensions of the fluidized bed and the fluidization velocity.

[0005] However, the above method has some limitations: 1) Experimental measurement is affected by many factors and it is difficult to fully reflect the complex change law of the temperature field; 2) The numerical simulation process is cumbersome and complex, and requires high accuracy of model parameters; 3) The empirical formula is difficult to cover all process conditions, and the prediction accuracy is limited.

[0006] In addition, process optimization of contact fluidized bed welding is also an important issue. During the welding process, the input of heat will cause the temperature of the joint to rise and the thermal stress to change, which not only affects the quality of the weld, but may also cause local deformation or even crack defects. Therefore, it is necessary to accurately analyze and optimize the temperature field and stress field of the welded joint to ensure the welding quality. However, the existing technology often relies on the experience of the welder to determine the linear energy of the welded joint, resulting in low welding quality of the upper and lower sections of the contact fluidized bed, affecting the performance of the fluidized bed. Summary of the invention

[0007] In view of this, the present invention provides a method for determining the linear energy of welding joints of upper and lower sections of a contact fluidized bed, which can solve the technical problem that the prior art often relies on the experience of welders to determine the linear energy of welding joints, resulting in low welding quality of the upper and lower sections of the contact fluidized bed.

[0008] The present invention is achieved in that:

[0009] A first aspect of the present invention provides a method for determining the linear energy of a welding joint of an upper and lower section pair of a contact fluidized bed, wherein the method comprises the following steps:

[0010] S10. Using finite element simulation or experimental measurement, establish a temperature distribution model inside the fluidized bed to describe the temperature variation in space and time;

[0011] S20, combining material performance parameters and welding process parameters, establishing a thermal stress change model of the joint during welding;

[0012] S30, taking temperature distribution and thermal stress as two objective functions, using game theory to find the optimal combination of welding parameters and establish a game model so that the temperature distribution is uniform and the thermal stress change is minimal;

[0013] S40, determining the constraint conditions of the game model, using the vector consisting of the temperature change and thermal stress change caused by the welding joint line energy as the mediation factor, and solving the game model using a multi-objective optimization algorithm to obtain an optimal solution set;

[0014] S50, calculating the process parameters corresponding to each solution in the optimal solution set, wherein the process parameters at least include welding current, welding speed, and welding gun angle;

[0015] S60, according to the process parameters corresponding to each solution, obtaining the actual temperature distribution and thermal stress change of the welding joint part of the process parameters during the actual welding test, and comparing the actual test results with the calculation results of the game model to evaluate the prediction accuracy of the game model;

[0016] S70, if the prediction accuracy of the game model is less than a preset threshold, the thermal stress change model and the parameters of the game model are adjusted and optimized, and steps S10 to S70 are iteratively performed to guide the prediction accuracy of the game model to be no less than the preset threshold, and the process parameters corresponding to the optimal solution set at this time are the pending process parameters;

[0017] S80, inputting the undetermined process parameters into a pre-trained process parameter-welding joint linear energy model, obtaining an optimized welding joint linear energy result and outputting it to an operator.

[0018] On the basis of the above technical solution, the method for determining the energy input of the welding joint of the upper and lower sections of a contact fluidized bed of the present invention can also be improved as follows:

[0019] The specific steps of S10 are as follows:

[0020] Step 1: Establish a three-dimensional transient heat conduction finite element model inside the fluidized bed using the finite element analysis method, discretize the fluidized bed geometry model into grid units, and set the temperature degree of freedom at each grid node; input the geometry size, material properties, temperature boundary conditions at the top and bottom of the fluidized bed, and internal heat source distribution into the model; obtain the temperature distribution of each point inside the fluidized bed that changes with time through finite element analysis;

[0021] Step 2: Arrange temperature sensors at different locations inside the fluidized bed to measure and record the temperature changes over time; analyze and fit the experimental measurement data to establish a mathematical model that describes the temperature distribution inside the fluidized bed;

[0022] Step 3: Comprehensively analyze the temperature distribution data obtained from finite element analysis and experimental measurement to establish a temperature distribution model that describes the spatial and temporal variation of the temperature inside the fluidized bed.

[0023] The specific steps of S20 are as follows:

[0024] Step 1: Combine welding process parameters, including welding current, welding speed, welding gun angle, and material performance parameters, including thermal conductivity, heat capacity, and thermal expansion coefficient, to establish a finite element model of thermal stress changes at the joint during welding;

[0025] Step 2: Input the temperature field calculated by the temperature distribution model established in step S10 as a thermal load into the thermal-stress coupling analysis model, and solve the stress-strain field distribution of the joint part that changes with time by combining the material constitutive relationship and the geometric boundary conditions to obtain a thermal stress change model of the welding joint part that changes with time.

[0026] Among them, the algorithm used to solve the game model is a multi-objective genetic algorithm.

[0027]

[0028] in, is the average temperature, n is the number of discrete points in the temperature distribution, T i is the temperature of the ith discrete point in the temperature distribution.

[0029] The objective function of the thermal stress in the game model is:

[0030] Among them, σ j is the jth stress component, and m is the number of stress components.

[0031] The constraints of the game model include: temperature change constraints, thermal stress change constraints and process parameter constraints.

[0032] Among them, the preset threshold corresponding to the prediction accuracy of the game model is 0.9.

[0033] The process parameter-welding joint line energy prediction model is a neural network model, and its training data comes from:

[0034] The process parameters and joint line energy data recorded during the actual welding test and the corresponding relationship data between the process parameters and joint line energy calculated by finite element simulation or empirical formula.

[0035] Wherein, the game model is a multi-objective game model, and the objective function of the game model also includes:

[0036] Weld seam quality objective function:

[0037] Among them, h k is the weld height at the kth measuring point, h0 is the ideal weld height, and p is the number of weld measuring points;

[0038] Heat affected zone size objective function:

[0039] Among them, d l is the size of the heat affected zone at the lth measuring point, d0 is the ideal heat affected zone size, and q is the internal heat source distribution.

[0040] Compared with the prior art, the method for determining the linear energy of the welded joints of the upper and lower sections of a contact fluidized bed provided by the present invention has the following beneficial effects: 1. A finite element model of the temperature distribution inside the fluidized bed is established. The model not only takes into account the geometric dimensions, material properties and other parameters of the fluidized bed, but also introduces the temperature boundary conditions at the top and bottom of the fluidized bed and the internal heat source distribution, which can more accurately describe the complex three-dimensional transient temperature field inside the fluidized bed. Compared with the existing temperature distribution model based only on two-dimensional or empirical formulas, this method can more comprehensively reflect the changing law of the temperature field.

[0041] 2. A coupled analysis model for thermal stress changes in welding joints was established. The model uses the established temperature field as a thermal load, combines the material constitutive relationship and geometric boundary conditions, and solves the three-dimensional stress-strain distribution of the joint during welding. This provides key thermal stress change data for subsequent process parameter optimization.

[0042] 3. A multi-objective optimization method for welding process parameters based on game theory is proposed. This method takes temperature distribution uniformity and thermal stress variation minimization as two contradictory optimization objectives and establishes a game model. By determining appropriate constraints, a multi-objective optimization algorithm is used to solve the game model and obtain the optimal process parameter solution set that meets the requirements. Compared with the existing single-objective optimization, this multi-objective optimization method can better balance the two objectives of temperature distribution and thermal stress and improve welding quality.

[0043] 4. A prediction model of process parameter-welding joint energy input was established. The model is based on a machine learning algorithm, which inputs the optimized process parameters and outputs the corresponding welding joint energy input. This provides fast and reliable decision support for subsequent production practice.

[0044] In summary, the comprehensive method proposed in the present invention makes full use of technologies such as finite element analysis, game theory optimization and machine learning to systematically solve the problem of determining the linear energy of contact fluidized bed welding joints. It solves the technical problem that the prior art often relies on the experience of welders to determine the linear energy of welding joints, resulting in low welding quality of the upper and lower sections of the contact fluidized bed. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0046] Figure 1 A flow chart of the method provided by the present invention. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0048] like Figure 1 As shown, it is a flow chart of a method for determining the line energy of a welding joint of an upper and lower section pair of a contact fluidized bed provided by the present invention. The method comprises the following steps:

[0049] S10. Using finite element simulation or experimental measurement, establish a temperature distribution model inside the fluidized bed to describe the temperature variation in space and time;

[0050] S20, combining material performance parameters and welding process parameters, establishing a thermal stress change model of the joint during welding;

[0051] S30, taking temperature distribution and thermal stress as two objective functions, using game theory to find the optimal combination of welding parameters and establish a game model so that the temperature distribution is uniform and the thermal stress change is minimal;

[0052] S40, determining the constraint conditions of the game model, taking the vector consisting of the temperature change and thermal stress change caused by the welding joint line energy as the mediation factor, solving the game model by using a multi-objective optimization algorithm, and obtaining the optimal solution set;

[0053] S50, calculating the process parameters corresponding to each solution in the optimal solution set, where the process parameters at least include welding current, welding speed, and welding gun angle;

[0054] S60, according to the process parameters corresponding to each solution, obtaining the actual temperature distribution and thermal stress change of the welding joint during the actual welding test, and comparing the actual test results with the calculation results of the game model to evaluate the prediction accuracy of the game model;

[0055] S70, if the prediction accuracy of the game model is less than a preset threshold, the parameters of the thermal stress change model and the game model are adjusted and optimized, and steps S10 to S70 are iteratively performed to guide the prediction accuracy of the game model to be no less than the preset threshold, and the process parameters corresponding to the optimal solution set at this time are the pending process parameters;

[0056] S80, input the undetermined process parameters into a pre-trained process parameter-welding joint line energy model, obtain an optimized welding joint line energy result and output it to the operator.

[0057] The specific implementation methods of the above steps are described in detail below:

[0058] The specific implementation of step S10 is as follows:

[0059] Modeling and solving of temperature distribution inside fluidized bed

[0060] First, the finite element analysis method is used to establish a three-dimensional transient heat conduction model inside the fluidized bed. The geometric model of the fluidized bed is discretized into a large number of three-dimensional grid elements, and the temperature degree of freedom is set at each element node. In this model, the following parameters need to be input:

[0061] Geometric dimensions of the fluidized bed: L, W, H

[0062] Thermophysical parameters of fluidized bed materials, such as thermal conductivity k, heat capacity c p , density ρ, etc.

[0063] The temperature boundary conditions at the top and bottom of the fluidized bed can be set to a constant temperature T0

[0064] Heat source distribution inside the fluidized bed q(x, y, z, t)

[0065] According to the three-dimensional transient heat conduction equation:

[0066]

[0067] By using the finite element method for numerical solution, we can obtain the temperature distribution T(x, y, z, t) of each point inside the fluidized bed that changes with time.

[0068] In addition, the temperature distribution data inside the fluidized bed can also be obtained through actual experimental measurements. Temperature sensors can be arranged at different locations inside the fluidized bed to measure and record the temperature changes over time. By analyzing and fitting the experimental measurement data, a mathematical model describing the temperature distribution inside the fluidized bed can also be established, expressed as:

[0069] T(x,y,z,t)=f(x,y,z,t;θ)

[0070] Among them, θ is the model parameter, which can be determined by the least square method and other methods.

[0071] By integrating the results of finite element analysis and experimental measurements, a temperature distribution model is established to describe the spatial and temporal changes in temperature inside the fluidized bed, providing basic data for subsequent welding process analysis.

[0072] The specific implementation of step S20 is as follows:

[0073] Modeling of thermal stress changes in welded joints

[0074] Combined with welding process parameters, such as welding current I, welding speed v, welding gun angle θ, etc., and material performance parameters, such as thermal conductivity k, heat capacity c p, thermal expansion coefficient α, etc., and establish a thermal-stress coupling finite element analysis model for the joint during the welding process.

[0075] First, according to the temperature distribution model established in step S10, the temperature change T(x, y, z, t) of the joint during the welding process is calculated. This temperature field is used as a heat load and input into the stress-strain analysis model. Combining the material constitutive relationship and geometric boundary conditions, the stress-strain field distribution of the joint over time can be solved:

[0076] Strain-displacement relationship:

[0077]

[0078] Stress-strain relationship (linear elasticity):

[0079]

[0080] in, is the thermal strain,

[0081] C ijkl is the elastic constant tensor.

[0082] By coupling the analysis of temperature field and stress field, the three-dimensional thermal stress distribution σ of the welded joint changing with time can be obtained. ij (x, y, z, t).

[0083] The purpose of this step is to provide key thermal stress change data for subsequent game theory optimization. By establishing this model, the thermal stress changes in the joint under different welding process parameters can be analyzed, providing a basis for finding the optimal parameter combination.

[0084] The specific implementation of step S30 is as follows:

[0085] Optimization of welding process parameters based on game theory

[0086] On the basis of the temperature distribution model and thermal stress variation model established in steps S10 and S20, the game theory method is used to find the optimal combination of welding parameters. Specifically, the temperature distribution uniformity and the minimization of thermal stress variation can be used as two objective functions to construct the following game model:

[0087] Assume that the two game players are the modulator and the modulated party. The goal of the modulator is to make the temperature distribution as uniform as possible, that is, to minimize the temperature gradient; the goal of the modulated party is to minimize the change of thermal stress. The two objective functions can be expressed as:

[0088] Modulator objective function:

[0089]

[0090] in, is the average temperature.

[0091] The objective function of the modulated party:

[0092]

[0093] Among them, σ j is the jth stress component.

[0094] Obviously, there is a certain contradiction between the two objective functions. The modulator pursues uniform temperature distribution, while the modulated party pursues minimum thermal stress. Through the game theory method, the optimal strategy of the modulator and the modulated party is found, that is, the welding parameter combination with uniform temperature distribution and minimum thermal stress change is found.

[0095] To this end, it is necessary to determine the corresponding constraints:

[0096] 1. Temperature change constraints:

[0097] T max -T min ≤ΔT max

[0098] Among them, T max and T min are the highest and lowest temperatures of the joint, ΔT max is the maximum allowable temperature change.

[0099] 2. Thermal stress change constraints:

[0100] max{σ1,σ2,σ3}≤σ yield

[0101] Among them, σ1, σ2, σ3 are the three principal stress components, σ yield is the yield strength of the material.

[0102] 3. Process parameter constraints:

[0103] I min ≤I≤I max

[0104] v min ≤v≤v max

[0105] θ min ≤θ≤θ max

[0106] They correspond to the value ranges of welding current, speed and angle respectively.

[0107] The above constraints are combined into a constraint vector g and input into the game model as a mediation factor. Then a multi-objective optimization algorithm, such as a genetic algorithm or a particle swarm optimization algorithm, is used to solve the optimal solution set of the game model. Each solution Corresponding to a set of optimal welding process parameters.

[0108] The purpose of this step is to provide the optimal solution set for the subsequent process parameter determination, satisfying the constraints of temperature change and thermal stress change.

[0109] The specific implementation of step S40 is as follows:

[0110] Establishment of process parameter-welding joint line energy model

[0111] The optimal solution set obtained by solving step S30 Calculate each solution x * The corresponding specific welding process parameters include welding current I, welding speed v and welding gun angle θ.

[0112] In order to quickly output the corresponding optimal process parameter combination according to the target requirements of temperature distribution and thermal stress change, a parameter-response value mapping relationship model can be established:

[0113] x=f(T,σ)

[0114] Where T = [T1, T2, ..., Tx] is the temperature distribution vector, σ = [σ1, σ2, ..., σ m ] is the thermal stress vector, and x=[I, v, θ] is the process parameter vector.

[0115] This model can be trained and established using artificial intelligence algorithms, such as neural networks or support vector machines. The training data comes from the optimal solution set obtained in step S30.

[0116] Through training, a parameter-response value mapping model f(·) can be obtained. Using this model, the corresponding optimal process parameter combination can be quickly predicted by simply inputting the target requirements of temperature distribution and thermal stress change, providing decision support for the subsequent actual welding process.

[0117] The specific implementation of step S60 is as follows:

[0118] Comparison of actual welding test results with the prediction results of the game model

[0119] According to the optimal process parameters determined in step S40, during the actual welding test, measuring equipment such as temperature sensors and strain gauges are used to record the actual temperature distribution T of the welding joint in real time. 实测 and thermal stress σ 实测.

[0120] The actual test results are compared and analyzed with the calculation results of the game model established in step S40 to evaluate the prediction accuracy of the game model. Specifically, the relative error between the measured value and the predicted value can be calculated as an indicator of the model prediction accuracy:

[0121] Temperature distribution prediction accuracy index:

[0122]

[0123] Thermal stress prediction accuracy index:

[0124]

[0125] Among them, T i 预测 and T i 实测 are the predicted temperature and measured temperature of the i-th point, respectively. and are the predicted and measured values ​​of the jth stress component, respectively.

[0126] Through comparative analysis, we can evaluate whether the overall prediction accuracy of the game model meets the requirements. If the accuracy is insufficient, the model parameters need to be further optimized.

[0127] The specific implementation of step S70 is as follows:

[0128] Iterative optimization of game models

[0129] If the prediction accuracy of the game model evaluated in step S60 is less than the preset threshold and The model needs to be optimized and adjusted. Specifically, the following measures can be taken:

[0130] 1. Optimize the thermal stress change model. You can try to introduce other influencing factors such as residual stress and phase change to improve the model's description ability. Or adjust material parameters such as α, E, v, etc. to more accurately reflect the thermal stress changes in the actual welding process.

[0131] 2. Adjust the parameter settings of the game model. You can try to change the weight coefficients ω1 and ω2 of the objective function, or adjust the threshold ΔT of the constraint condition. max , σ yield , in order to find a better solution.

[0132] 3. Increase the number of training samples and enrich the sample types. More temperature distribution and thermal stress change data can be obtained through more finite element simulations or experimental tests to enhance the model's fitting and prediction capabilities.

[0133] 4. Use more complex optimization algorithms. If the basic multi-objective optimization algorithm cannot meet the requirements, you can try more advanced optimization methods such as evolutionary algorithms and reinforcement learning.

[0134] Through iterative optimization of the above measures, until the prediction accuracy of the game model reaches the preset threshold, the final optimal welding process parameter combination can be determined. The purpose of this step is to continuously improve the game model and improve its prediction accuracy.

[0135] The specific implementation of step S80 is as follows:

[0136] Training of process parameter-weld joint energy input prediction model

[0137] Based on the optimal welding process parameter combination determined in the previous steps, a prediction model of process parameter-welding joint line energy is established. This model can be trained and established using machine learning methods.

[0138] Sources of training data can include:

[0139] 1. The process parameters and joint energy input data recorded during the aforementioned actual welding test;

[0140] 2. The corresponding relationship data between process parameters and joint line energy obtained through finite element simulation or empirical formula calculation.

[0141] The process parameters (such as welding current I, speed v, angle θ, etc.) are used as the input features x of the model, and the welding joint line energy E line As the output target y, a prediction model y = f(x) is trained. This model can be trained using machine learning algorithms such as neural networks and support vector machines. During the training process, care should be taken to avoid overfitting, and methods such as cross-validation can be used to evaluate the generalization performance of the model.

[0142] Finally, the optimal process parameter x determined in step S70 is * Input it into the trained prediction model f(·) to get the optimized welding joint line energy result Provide decision-making basis for operators.

[0143]

[0144] The specific variables involved are explained in the following table:

[0145]

[0146]

[0147] Specifically, the principle of the present invention is:

[0148] First, the finite element analysis method is used to establish a three-dimensional transient temperature distribution model inside the fluidized bed. This model takes into account the geometric dimensions, material properties and other parameters of the fluidized bed, as well as the temperature boundary conditions at the top and bottom of the fluidized bed and the internal heat source distribution, and can more accurately describe the change law of the complex three-dimensional temperature field. Through this model, detailed temperature data of each point inside the fluidized bed that changes with time can be obtained, providing a basis for subsequent welding process analysis.

[0149] Secondly, based on the established temperature distribution model, combined with welding process parameters and material performance parameters, a thermal-stress coupling analysis model for the joint during welding was established. This model uses the temperature field as a thermal load and solves the three-dimensional stress-strain distribution of the joint through finite element analysis. This provides key thermal stress change data for subsequent process parameter optimization.

[0150] Next, the game theory method is used to perform multi-objective optimization of the welding process parameters. Specifically, the uniformity of temperature distribution and the minimization of thermal stress changes are taken as two contradictory optimization goals, and a game model is constructed. In this model, the modulator pursues uniform temperature distribution, while the modulated party pursues minimum thermal stress. By determining appropriate constraints, such as temperature changes not exceeding the melting point of the material, thermal stress changes not exceeding the yield strength of the material, etc., a multi-objective optimization algorithm is used to solve the game model and obtain the optimal process parameter solution set that meets the requirements. Compared with the existing single-objective optimization, this multi-objective optimization method can better balance the two goals of temperature distribution and thermal stress and improve welding quality.

[0151] Finally, based on the determined optimal process parameters, a prediction model of process parameter-welding joint energy input was established. The model was trained using a machine learning algorithm, and process parameters such as welding current, speed, angle, etc. were used as input features to predict the corresponding welding joint energy input. In this way, the optimized welding joint energy input results can be obtained quickly and accurately, providing reliable decision support for production practice.

[0152] In summary, the comprehensive method proposed in this invention makes full use of advanced technologies such as finite element analysis, game theory optimization and machine learning to solve the key problems of determining the line energy of contact fluidized bed welding joints in the following aspects:

[0153] 1. A more accurate and comprehensive temperature distribution model inside the fluidized bed was established to provide reliable basic data for subsequent welding process optimization.

[0154] 2. Through the coupled analysis of temperature field and stress field, the thermal stress variation law of the welding joint was deeply explored, laying a theoretical foundation for multi-objective optimization.

[0155] 3. By using the game theory optimization method, a better balance was achieved between the two conflicting goals of temperature distribution uniformity and thermal stress minimization, and the optimization results are more in line with actual process requirements.

[0156] 4. A fast and high-precision process parameter-welding joint line energy prediction model was established using machine learning technology, providing strong decision-making support for production practice.

[0157] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

[0158] In order to further illustrate the technical solution of the present invention, the following specific embodiments are provided.

[0159] 1. Implementation Environment

[0160] This embodiment studies a contact fluidized bed device for coal gasification. The geometric dimensions of the fluidized bed are: length L = 2m, width W = 1m, height H = 1.5m. The fluidized bed material is stainless steel, and its thermal physical parameters are: thermal conductivity k = 16.3W / (m·K), heat capacity c p =490J / (kg·K), density ρ=7900kg / m 3 , thermal expansion coefficient α = 1.8 × 10 -5 / K, elastic modulus E = 200 GPa, Poisson's ratio v = 0.3.

[0161] In this fluidized bed device, it is necessary to manufacture a welded joint. The welding process parameters are set as follows: welding current I = 220A, welding speed v = 12cm / min, welding gun angle θ = 30 ° The material is Q235 carbon steel, and its yield strength σ yield =345MPa.

[0162] (II) Modeling of temperature distribution inside the fluidized bed

[0163] First, the finite element analysis method was used to establish a three-dimensional transient temperature distribution model inside the fluidized bed. The fluidized bed geometry model was discretized into 25,600 octahedral unit grids, and the temperature degree of freedom was set at each grid node. The size and material parameters of the above fluidized bed were input into the model, as well as the temperature boundary conditions T0 = 800 °C at the top and bottom of the fluidized bed, and the internal heat source distribution q (x, y, z, t) = 5 × 10 5 W / m 3 .

[0164] Solving the three-dimensional transient heat conduction equation by finite element analysis The temperature distribution T(x, y, z, t) of each point inside the fluidized bed that changes with time is obtained.

[0165] (III) Modeling of thermal stress changes in welded joints

[0166] Based on the temperature distribution model established in step (ii), the three-dimensional stress-strain distribution of the welded joint is solved by using the thermal-stress coupled finite element analysis method.

[0167] Specifically, the temperature field T(x, y, z, t) calculated in step (ii) is used as the thermal load and input into the stress-strain analysis model. and stress-strain relationship in The three-dimensional thermal stress distribution σ of the joint is obtained by combining the material constitutive relationship and geometric boundary conditions. ij (x, y, z, t).

[0168] Further analysis of the thermal stress change process at the joint shows that the thermal stress increases rapidly in a short period of time after the start of welding, reaches a peak value in about 20 seconds, and then slowly decreases and stabilizes. This transient thermal stress change characteristic will inevitably have an important impact on the quality and performance of the welded joint, and corresponding optimization measures need to be taken.

[0169] (IV) Optimization of welding process parameters based on game theory

[0170] According to the temperature distribution model and thermal stress change model established in steps (ii) and (iii), the game theory method is used to perform multi-objective optimization of welding process parameters. Specifically, the uniformity of temperature distribution and minimization of thermal stress change are taken as two contradictory optimization goals, and the following game model is constructed:

[0171] Modulation objective function (temperature distribution uniformity):

[0172]

[0173] in, is the average temperature.

[0174] Modulated objective function (minimizing thermal stress change):

[0175]

[0176] Among them, σ j is the jth stress component.

[0177] At the same time, the following constraints are determined:

[0178] 1. Temperature change constraint: T max -Tmin ≤300℃

[0179] 2. Thermal stress change constraint: max{σ1,σ2,σ3}≤300MPa

[0180] 3. Process parameter constraints: 180A≤I≤260A, 10cm / min≤v≤18cm / min, 20 ° ≤θ≤40 °

[0181] The NSGA-II multi-objective optimization algorithm is used to solve the game model and obtain the optimal process parameter solution set that meets the constraints. Among them, one set of optimal solutions is: welding current I = 230A, welding speed v = 14cm / min, welding gun angle θ = 25 ° .

[0182] (V) Establishment of process parameter-welding joint energy prediction model

[0183] Based on the optimal process parameters determined in step (iv), a prediction model of process parameters-welding joint line energy was established. The model adopts an artificial neural network structure, with the input being process parameters x = [I, v, θ] and the output being the welding joint line energy E line .

[0184] The specific structure of the neural network is as follows:

[0185] Input layer: 3 neurons, corresponding to 3 process parameters

[0186] Hidden layer: 3 fully connected hidden layers, with 50, 30, and 20 nodes respectively

[0187] Output layer: 1 neuron, corresponding to the welding joint line energy

[0188] The network training uses the back propagation algorithm, and the loss function is the mean square error. The training data sources include:

[0189] 1. Step (iv) optimizes 20 sets of process parameters and corresponding line energy values;

[0190] 2. 50 sets of process parameter-line energy correspondence data calculated by finite element analysis and empirical formula.

[0191] After 5-fold cross validation, the average prediction accuracy of the prediction model reached more than 95%, meeting the needs of practical applications.

[0192] (VI) Actual welding test and model verification

[0193] The optimal process parameters and prediction model determined in steps (iv) and (v) were tested and verified in the actual fluidized bed welding process.

[0194] First, temperature sensors and strain gauges are arranged at the joint of the fluidized bed to record the temperature distribution T in real time during the welding process. 实测 and thermal stress σ 实测 data.

[0195] Compare and analyze the measured results with the predicted results of the temperature distribution model and thermal stress change model established in steps (ii) and (iii), and calculate the prediction accuracy index:

[0196] Temperature distribution prediction accuracy:

[0197]

[0198] Thermal stress prediction accuracy:

[0199]

[0200] The above prediction accuracy indicators all meet the threshold requirements set in step (VII) (η T ≥90%,η σ ≥85%).

[0201] Finally, the optimal process parameters determined in step (iv) are input into the prediction model established in step (v), and the optimized welding joint energy input is calculated as The actual measurement results are The relative error between the two is less than 3%, which meets the requirements of production practice.

[0202] (VII) Model parameter optimization and final results

[0203] Through the above actual test verification, it was found that the prediction accuracy of the temperature distribution model and thermal stress change model established in steps (ii) and (iii) was slightly lower than the preset threshold. Therefore, these two models were further optimized:

[0204] 1. Temperature distribution model optimization: Based on finite element analysis, the influence of turbulent convection heat transfer inside the fluidized bed was introduced, the boundary conditions and heat source distribution items were modified, and the physical description ability of the model was improved. After optimization, the temperature distribution prediction accuracy η T Increased to 94.2%.

[0205] 2. Optimization of thermal stress change model: In the thermal-stress coupling analysis, the influence of plastic deformation generated during welding on the stress field is considered, the stress-strain relationship equation is modified, and the accuracy of thermal stress prediction is improved. After optimization, the thermal stress prediction accuracy η σ Increased to 91.6%.

[0206] Based on the optimized model, the game theory optimization of step (iv) was performed again, and the new optimal process parameter combination was obtained as follows: welding current I = 235A, welding speed v = 13cm / min, welding gun angle θ = 27 ° .

[0207] This set of optimal parameters is input into the prediction model trained in step (V), and the optimized weld joint energy is calculated as: Compared with the actual measurement results The deviation is only 1.5%, which meets the needs of production practice.

[0208] In summary, the comprehensive method based on finite element analysis, game theory optimization and machine learning proposed in this invention can accurately describe the temperature distribution inside the fluidized bed and the thermal stress changes at the joint, and determine the optimal welding process parameters that meet the requirements through multi-objective optimization, thereby obtaining the optimized welding joint line energy results. This method has shown obvious technical advantages in the following aspects:

[0209] 1. The three-dimensional transient temperature distribution model established by finite element analysis can more comprehensively reflect the complex temperature field change law inside the fluidized bed and provide reliable basic data for subsequent welding process optimization.

[0210] 2. Through the thermal-stress coupling analysis, the thermal stress variation characteristics of the welded joint were deeply explored, laying a theoretical foundation for multi-objective game theory optimization. The optimization results not only meet the requirements of temperature distribution uniformity, but also minimize the thermal stress variation, which can effectively improve the welding quality.

[0211] 3. The process parameter-joint line energy prediction model established using machine learning technology can quickly and accurately provide decision support for production practice, greatly improving production efficiency.

[0212] 4. Through continuous optimization and iteration of model parameters, the accuracy of temperature distribution and thermal stress prediction has reached more than 90%, ensuring the reliability of the optimization results.

Claims

1. A method for determining the linear energy of welding joints of upper and lower sections of a contact fluidized bed, characterized in that: The following steps are involved: S10. Using finite element simulation or experimental measurement, establish a temperature distribution model inside the fluidized bed to describe the temperature variation in space and time; S20, combining material performance parameters and welding process parameters, establishing a thermal stress change model of the joint during welding; S30, taking temperature distribution and thermal stress as two objective functions, using game theory to find the optimal combination of process parameters and establish a game model so that the temperature distribution is uniform and the thermal stress change is minimal; S40, determining the constraint conditions of the game model, using the vector consisting of the temperature change and thermal stress change caused by the welding joint line energy as the mediation factor, and solving the game model using a multi-objective optimization algorithm to obtain an optimal solution set; S50, calculating the process parameters corresponding to each solution in the optimal solution set, wherein the process parameters at least include welding current, welding speed, and welding gun angle; S60, according to the process parameters corresponding to each solution, obtaining the actual temperature distribution and thermal stress change of the welding joint part of the process parameters during the actual welding test, and comparing the actual test results with the calculation results of the game model to evaluate the prediction accuracy of the game model; S70, if the prediction accuracy of the game model is less than a preset threshold, the thermal stress change model and the parameters of the game model are adjusted and optimized, and steps S10 to S70 are iteratively performed to guide the prediction accuracy of the game model to be no less than the preset threshold, and the process parameters corresponding to the optimal solution set at this time are the pending process parameters; S80, inputting the undetermined process parameters into a pre-trained process parameter-welding joint line energy model to obtain an optimized welding joint line energy result and output it to an operator; The constraints of the game model include: temperature change constraints, thermal stress change constraints and process parameter constraints; Among them, the temperature change constraint: T max -T min ≤ΔT max ; Among them, T max and T min are the highest and lowest temperatures of the joint, ΔT max is the maximum allowable temperature change; Thermal stress change constraint: max{σ1,σ2,σ3}≤σ yield ; Among them, σ1, σ2, σ3 are the three principal stress components, σ yield is the yield strength of the material; Process parameter constraints include: I min ≤I≤I max in min ≤v≤v max i min ≤θ≤θ max They correspond to the value ranges of welding current, speed and angle respectively; The above constraints are combined into a constraint vector g, which is input into the game model as a mediation factor. Then, a multi-objective optimization algorithm is used to solve the optimal solution set of the game model. Each solution Corresponding to a set of optimal welding process parameters.

2. The method for determining the linear energy of welding joints of upper and lower sections of a contact fluidized bed according to claim 1 is characterized in that: The specific steps of S10 are as follows: Step 1: Establish a three-dimensional transient heat conduction finite element model inside the fluidized bed using the finite element analysis method, discretize the fluidized bed geometry model into grid units, and set the temperature degree of freedom at each grid node; input the geometry size, material properties, temperature boundary conditions at the top and bottom of the fluidized bed, and internal heat source distribution into the model; obtain the temperature distribution of each point inside the fluidized bed that changes with time through finite element analysis; Step 2: Arrange temperature sensors at different locations inside the fluidized bed to measure and record the temperature changes over time; analyze and fit the experimental measurement data to establish a mathematical model that describes the temperature distribution inside the fluidized bed; Step 3: Comprehensively analyze the temperature distribution data obtained from finite element analysis and experimental measurement to establish a temperature distribution model that describes the spatial and temporal variation of the temperature inside the fluidized bed.

3. The method for determining the linear energy of welding joints of upper and lower sections of a contact fluidized bed according to claim 1 is characterized in that: The specific steps of S20 are as follows: Step 1: Combine welding process parameters, including welding current, welding speed, welding gun angle, and material performance parameters, including thermal conductivity, heat capacity, and thermal expansion coefficient, to establish a finite element model of thermal stress changes at the joint during welding; Step 2: Input the temperature field calculated by the temperature distribution model established in step S10 as a thermal load into the thermal-stress coupling analysis model, and solve the stress-strain field distribution of the joint part that changes with time by combining the material constitutive relationship and the geometric boundary conditions to obtain a thermal stress change model of the welding joint part that changes with time.

4. The method for determining the linear energy of welding joints of upper and lower sections of a contact fluidized bed according to claim 1 is characterized in that: The algorithm used to solve the game model is a multi-objective genetic algorithm.

5. The method for determining the line energy of welding joints of upper and lower sections of a contact fluidized bed according to claim 1 is characterized in that: The objective function of the temperature distribution of the game model is: in, is the average temperature, n is the number of discrete points in the temperature distribution, T i is the temperature of the ith discrete point in the temperature distribution.

6. The method for determining the linear energy of welding joints of upper and lower sections of a contact fluidized bed according to claim 1 is characterized in that: The objective function of the thermal stress in the game model is: Among them, σ j is the jth stress component, and m is the number of stress components.

7. The method for determining the linear energy of welding joints of upper and lower sections of a contact fluidized bed according to claim 1, characterized in that: The preset threshold corresponding to the prediction accuracy of the game model is 0.

9.

8. The method for determining the line energy of welding joints of upper and lower sections of a contact fluidized bed according to claim 1, characterized in that: The process parameter-welding joint energy prediction model is a neural network model, and its training data comes from: The process parameters and joint line energy data recorded during the actual welding test and the corresponding relationship data between the process parameters and joint line energy calculated by finite element simulation or empirical formula.

9. The method for determining the linear energy of welding joints of upper and lower sections of a contact fluidized bed according to claim 1, characterized in that: The game model is a multi-objective game model, and the objective function of the game model also includes: Weld seam quality objective function: Among them, h k is the weld height at the kth measuring point, h0 is the ideal weld height, and p is the number of weld measuring points; Heat affected zone size objective function: Among them, d l is the size of the heat affected zone at the lth measuring point, d0 is the ideal heat affected zone size, and q is the internal heat source distribution.

Images