Large module main bearing node collaborative welding process parameter optimization method

Abstract

The invention discloses a large-scale module main bearing node collaborative welding process parameter optimization method, which comprises the steps of collecting historical welding data of a large-scale module main bearing node, establishing a historical welding data set of the large-scale module main bearing node, inputting the historical welding data set into a BP neural network model to obtain an optimized BP neural network model with better training output welding process parameters, and a back propagation algorithm is adopted to repeatedly adjust and optimize welding process parameters, a multi-robot collaborative welding analysis model of the main bearing node of the large module is established, a multi-robot collaborative welding path is optimized, and finally the minimum deformation value delta Smin in the deformation values of the main bearing node of the large module is determined. According to the method, the welding path during multi-robot collaborative welding of the main bearing node of the large module is optimized, the welding quality and welding efficiency of the main bearing node of the large module are improved, and meanwhile thermal deformation in the welding process is relieved.

Description

Technical Field

[0001] The invention relates to a method for welding large-scale structures in marine engineering, and in particular to a method for optimizing collaborative welding process parameters of main bearing nodes of large-scale modules. Background Art

[0002] At present, the large main bearing node of the marine structure is a very critical link in welding the deck of the large structure. The structure of the large main bearing node of the marine structure is a left-right front-back symmetrical structure, including an upper cover plate 1 and a lower cover plate 2 arranged at intervals up and down, and a support column 4 is connected to the middle position between the upper cover plate 1 and the lower cover plate, and the rib plate 3 connects the upper cover plate 1, the lower cover plate 2 and the support column. The weld is at the connection between the upper cover plate, the lower cover plate, the rib plate and the support column. The welding quality of the large main bearing node is directly related to the bearing weight of the deck of the large structure. Since the large main bearing node is welded by steel plates, the number of welds is large and the length is large when welding the steel plates, the welding strength is high, and the manufacturing process of the large module is complicated and the technical difficulty is high. Especially in the welding of the main bearing node, the requirements for welding process and parameters are particularly strict. At this stage, the welding task of manually welding the welds is still adopted. Whether it is traditional manual welding or semi-automatic welding, it faces the problems of high labor intensity and low welding efficiency, and welding deformation is easy to occur during the welding process, which affects the welding quality. Summary of the invention

[0003] The purpose of the present invention is to overcome the shortcomings of the existing technology and provide a method for optimizing the collaborative welding process parameters of the main load-bearing nodes of large modules. The method optimizes the welding process parameters of the main load-bearing nodes of large modules, improves the welding efficiency and welding quality of the main load-bearing nodes of large modules, and reduces welding deformation.

[0004] To achieve the above object, the technical solution adopted by the present invention is:

[0005] A method for optimizing the collaborative welding process parameters of a large module main bearing node of the present invention comprises the following steps:

[0006] Step 1: Collect historical welding data of the main bearing node of the large module, and establish a historical welding data set of the main bearing node of the large module, wherein the historical welding data set includes welding quality corresponding to welding current A, welding wire diameter ψ, and welding wire dry extension length s used for different welding materials and welding thicknesses, and the structure of the main bearing node of the large module is a left-right symmetrical structure and a front-back symmetrical structure, and the weld is set at the intersection of the upper cover plate, the lower cover plate, the rib plate, and the support column;

[0007] Step 2: Establish a BP neural network model, input the historical welding data set of the main bearing node of the large module into the BP neural network model, train the BP neural network model, use welding quality as an evaluation index, and obtain an optimized BP neural network model with better training output welding process parameters, wherein the welding process parameters include welding current A, welding wire diameter ψ, and welding wire dry extension length s;

[0008] Step 3: Set the initial input sample welding current A = 1A, welding wire diameter ψ = 1mm, welding wire extension length s = 1mm and maximum welding current A of the main load-bearing node of the large module to be welded. max ;

[0009] Step 4: Input the initial input sample welding current A, welding wire diameter ψ and welding wire extension length s of the main load-bearing node of the large module to be welded into the optimized BP neural network analysis model, and use the back propagation algorithm to repeatedly adjust the welding process parameter adjustment weight ω and deviation Δ of the optimized BP neural network model to obtain the welding process parameters with the shortest welding time T under the same weld length S1, and output the optimized welding process parameters

[0010] Step 5: Establish a multi-robot collaborative welding analysis model of the main bearing node of the large module in the 3D software, establish the welding coordinate system O-XYZ according to the right-hand rule, set the central axis of the support column of the main bearing node of the large module as the Z axis, the horizontal right as the X axis, the vertical plane inward as the Y axis, and the origin O of the coordinate system is located at the intersection of the bottom surface of the main bearing node of the large module and the central axis of symmetry;

[0011] Step 6: According to the multi-robot collaborative welding analysis model of the main load-bearing node of the large module, the main load-bearing node of the large module is divided into four identical parts with the X-axis and the Y-axis as the dividing line, marked as the first quadrant A, the second quadrant B, the third quadrant C and the fourth quadrant D, respectively. A robot is arranged at each quadrant position, and the starting point P of the welding route when each quadrant robot performs welding of different welds in the quadrant is set. i1 and the end point P i2 , starting from the starting point of the welding route, welding all the welds in the quadrant and stopping the welding work at the end point of the welding route;

[0012] Step 7: Set the starting point P of each quadrant according to the multi-robot collaborative welding analysis model i1 and the end point P i2 The welding paths of the robots in each quadrant are simulated and analyzed together, and the welding paths of multiple robots are optimized to obtain four robot welding optimization paths respectively;

[0013] Step 8. Establish a temperature field analysis model for multi-robot collaborative welding of the main bearing nodes of large modules, input the optimized welding process parameters obtained in step 4 and the optimized multi-robot collaborative welding path obtained in step 7, analyze the temperature field of the main bearing nodes of large modules during the multi-robot welding process, obtain the deformation value of the main bearing nodes of large modules during welding through temperature field analysis, and record the deformation value ΔS;

[0014] Step 9. Return to step 3 to modify the values ​​of the initial input sample welding current A, welding wire diameter ψ, and welding wire extension length s, and execute steps 4 to 9 until the initial input sample welding current A is equal to the maximum welding current A. max End, proceed to the next step;

[0015] Step 10: Determine the minimum deformation value ΔS among the deformation values ​​of the main bearing nodes of the large module min , the minimum deformation value ΔS min The corresponding welding parameters and welding paths are output as a collaborative welding plan.

[0016] Compared with the prior art, the present invention has the following beneficial effects:

[0017] The present invention establishes a multi-robot collaborative welding model for the main load-bearing nodes of large modules, uses a neural network algorithm to optimize the welding process parameters of the main load-bearing nodes of large modules, and optimizes the welding path during multi-robot collaborative welding of the main load-bearing nodes of large modules, thereby improving the welding quality and welding efficiency of the main load-bearing nodes of large modules and reducing thermal deformation during the welding process. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a flow chart of a method for optimizing the collaborative welding process parameters of a main load-bearing node of a large module according to the present invention;

[0019] Figure 2 It is a schematic diagram of the partition welding of the main bearing node of a large module of the present invention;

[0020] Figure 3 A schematic diagram of the welding path of the main bearing node of a large module of the present invention;

[0021] Figure 4 It is a schematic diagram of the three-dimensional structure of the main bearing node of the large module of the present invention. DETAILED DESCRIPTION

[0022] The present invention is described in detail below in conjunction with the accompanying drawings:

[0023] As shown in the figure, a method for optimizing the collaborative welding process parameters of a large module main bearing node of the present invention comprises the following steps:

[0024] Step 1: Collect historical welding data of the main load-bearing node of the large module, and establish a historical welding data set of the main load-bearing node of the large module. The historical welding data set includes welding quality corresponding to welding current A, welding wire diameter ψ, and welding wire dry extension length s used for different welding materials and welding thicknesses. The structure of the main load-bearing node of the large module is as shown in the attached figure. Figure 2 and 4 In the left-right and front-back symmetrical structure shown, the weld is set at the intersection of the upper cover plate 1, the lower cover plate 2, the rib plate 3 and the support column 4;

[0025] Step 2: Establish a BP neural network model, input the historical welding data set of the main bearing node of the large module into the BP neural network model, train the BP neural network model, use welding quality as an evaluation index, and obtain an optimized BP neural network model with better training output welding process parameters, wherein the welding process parameters include welding current A, welding wire diameter ψ, and welding wire dry extension length s;

[0026] Step 3: Set the initial input sample welding current A = 1A, welding wire diameter ψ = 1mm, welding wire extension length s = 1mm and maximum welding current A of the main load-bearing node of the large module to be welded. max ;

[0027] Step 4: Input the initial input sample welding current A, welding wire diameter ψ and welding wire dry extension length s of the main load-bearing node of the large module to be welded into the optimized BP neural network analysis model, and use the back propagation algorithm (see Cao Jinbi's "Analysis of Wind Power Prediction Model Based on PSO-BP Neural Network", Modern Industrial Economy and Informatization. 2024, 14 (08)) to repeatedly adjust the welding process parameter adjustment weight ω and deviation Δ of the optimized BP neural network model to obtain the welding process parameters with the shortest welding time T under the same weld length S1, and output the optimized welding process parameters. The specific optimization steps are as follows:

[0028] The first step is to set the expected welding time T0 of the optimized BP neural network and the threshold ε for stopping optimization;

[0029] Step 2: Set the initial input sample of the optimized BP neural network, welding current A=1A, welding wire diameter ψ=1mm, welding wire extension length s=1mm, and welding process parameter adjustment weight ω=1, deviation Δ=1;

[0030] Step 3: Input the initial input sample welding current A=1A, welding wire diameter ψ=1mm, welding wire extension length s=1mm, weight ω=1, and deviation Δ=1 set in step 2 into the optimized BP neural network to obtain different welding times T under the same weld length S1;

[0031] Step 4: Compare the output welding time T with the expected welding time T0 to obtain the error value Δε;

[0032] Step 5: Use the back propagation algorithm to adjust and optimize the weight ω and deviation Δ of the BP neural network model, return to step 2, reset the initial input sample welding current A, wire diameter ψ, and wire extension length s, and repeat steps 2 to 5 until the error value Δε is less than the set threshold ε;

[0033] Step 6. When the output error value is less than the set threshold, the corresponding welding current A, welding wire diameter ψ, and welding wire extension length s are used as the optimized welding process parameters.

[0034] Step 5: Establish a multi-robot collaborative welding analysis model of the main bearing node of the large module in the three-dimensional software, establish the welding coordinate system O-XYZ according to the right-hand rule, set the central axis of the support column 4 of the main bearing node of the large module as the Z axis, the horizontal right as the X axis, the vertical plane inward as the Y axis, and the origin O of the coordinate system is located at the intersection of the bottom surface of the main bearing node of the large module and the symmetric central axis;

[0035] Step 6: According to the multi-robot collaborative welding analysis model of the main load-bearing node of the large module, the main load-bearing node of the large module is divided into four identical parts with the X-axis and the Y-axis as the dividing line, marked as the first quadrant A, the second quadrant B, the third quadrant C and the fourth quadrant D, respectively. A robot is arranged at each quadrant position, and the starting point P of the welding route when each quadrant robot performs welding of different welds in the quadrant is set. i1 and the end point P i2 , starting from the starting point of the welding route, welding all the welds in the quadrant is carried out and the welding work is stopped at the end point of the welding route. The layout of all large main bearing nodes and robots is shown in the figure;

[0036] Step 7: Set the starting point P of each quadrant according to the multi-robot collaborative welding analysis model i1 and the end point P i2 The welding paths of the robots in each quadrant are simulated and analyzed together, and the multi-robot welding paths are optimized to obtain four robot welding optimization paths. The specific path optimization steps are as follows:

[0037] The first step is to extract the location of the welds between the main load-bearing node plates of the large modules in each quadrant and the length information of each weld, and plan all the welds in the same quadrant into a weld trajectory through the robot welding information simulation software, and obtain the weld path that the robot needs to weld in each quadrant;

[0038] Step 2: Set the welding priority of the welds. The welds in the same quadrant are ranked based on the starting point P. i1Plan the starting point for welding, give priority to welding short welds, then weld long welds. When multiple welds have the same welding priority, take the origin O of the coordinate system O-XYZ as the node center, give priority to welding the welds close to the node center, then weld the welds far from the node center, until the end point P i2 ;

[0039] The third step is to establish a multi-objective optimization function F(x) of the welding trajectory with welding path and welding time as optimization targets:

[0040] F(x)=Min(f1(x),f2(x))

[0041] Where f1(x) is the welding path optimization function, and f2(x) is the welding time optimization function.

[0042] The specific optimization function f1(x) of the welding path is expressed as:

[0043]

[0044] In the formula, (x i ,y i ,z i ) is the coordinate value of a point on the welding path, (x i+1 ,y i+1 ,z i+1 ) is the coordinate value of the adjacent points on the welding path, n is the number of points on the welding path, and i is the current sequence value of the welding point.

[0045] The specific optimization function f2(x) of welding time is expressed as:

[0046]

[0047] Where, L j is the length of a weld on the main bearing node of the large module, v j It is the welding speed set when the robot welds a certain weld, m is the total number of welds, and j is the current weld rank value.

[0048] Step 4: According to the starting point P of each quadrant i1 and the end point P i2 , taking the welding priority of the weld as the constraint condition, the multi-objective particle swarm algorithm is used to iterate the multi-objective optimization function F(x) of the welding trajectory, and the multi-robot collaborative welding optimization path of the main bearing node of the large module is obtained;

[0049] Step 5: Save and output the optimized path of multi-robot collaborative welding of the main bearing nodes of large modules.

[0050] Step 8. Establish a temperature field analysis model for multi-robot collaborative welding of the main bearing nodes of large modules, input the optimized welding process parameters obtained in step 4 and the optimized multi-robot collaborative welding path obtained in step 7, analyze the temperature field of the main bearing nodes of large modules during the multi-robot welding process, obtain the deformation value of the main bearing nodes of large modules during welding through temperature field analysis, and record the deformation value ΔS;

[0051] Step 9. Return to step 3 to modify the values ​​of the initial input sample welding current A, welding wire diameter ψ, and welding wire extension length s, and execute steps 4 to 9 until the initial input sample welding current A is equal to the maximum welding current A. max End, proceed to the next step;

[0052] Step 10: Determine the minimum deformation value ΔS among the deformation values ​​of the main bearing nodes of the large module min , the minimum deformation value ΔS min The corresponding welding parameters and welding paths are output as a collaborative welding plan.

[0053] The above-described embodiments are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A method for optimizing the collaborative welding process parameters of a large module main bearing node, characterized in that The following steps are involved: Step 1: Collect historical welding data of the main bearing node of the large module, and establish a historical welding data set of the main bearing node of the large module, wherein the historical welding data set includes welding quality corresponding to welding current A, welding wire diameter ψ, and welding wire dry extension length s used for different welding materials and welding thicknesses, and the structure of the main bearing node of the large module is a left-right symmetrical structure and a front-back symmetrical structure, and the weld is set at the intersection of the upper cover plate, the lower cover plate, the rib plate, and the support column; Step 2: Establish a BP neural network model, input the historical welding data set of the main bearing node of the large module into the BP neural network model, train the BP neural network model, use welding quality as an evaluation index, and obtain an optimized BP neural network model with better training output welding process parameters, wherein the welding process parameters include welding current A, welding wire diameter ψ, and welding wire dry extension length s; Step 3: Set the initial input sample welding current A = 1A, welding wire diameter ψ = 1mm, welding wire extension length s = 1mm and maximum welding current A of the main load-bearing node of the large module to be welded. max ; Step 4: Input the initial input sample welding current A, welding wire diameter ψ and welding wire extension length s of the main load-bearing node of the large module to be welded into the optimized BP neural network analysis model, and use the back propagation algorithm to repeatedly adjust the welding process parameter adjustment weight ω and deviation Δ of the optimized BP neural network model to obtain the welding process parameters with the shortest welding time T under the same weld length S1, and output the optimized welding process parameters Step 5: Establish a multi-robot collaborative welding analysis model of the main bearing node of the large module in the 3D software, establish the welding coordinate system O-XYZ according to the right-hand rule, set the central axis of the support column of the main bearing node of the large module as the Z axis, the horizontal right as the X axis, the vertical plane inward as the Y axis, and the origin O of the coordinate system is located at the intersection of the bottom surface of the main bearing node of the large module and the central axis of symmetry; Step 6: According to the multi-robot collaborative welding analysis model of the main load-bearing node of the large module, the main load-bearing node of the large module is divided into four identical parts with the X-axis and the Y-axis as the dividing line, marked as the first quadrant A, the second quadrant B, the third quadrant C and the fourth quadrant D, respectively. A robot is arranged at each quadrant position, and the starting point P of the welding route when each quadrant robot performs welding of different welds in the quadrant is set. i1 and the end point P i2 , starting from the starting point of the welding route, welding all the welds in the quadrant and stopping the welding work at the end point of the welding route; Step 7: Set the starting point P of each quadrant according to the multi-robot collaborative welding analysis model i1 and the end point P i2 The welding paths of the robots in each quadrant are simulated and analyzed together, and the welding paths of multiple robots are optimized to obtain four robot welding optimization paths respectively; Step 8. Establish a temperature field analysis model for multi-robot collaborative welding of the main bearing nodes of large modules, input the optimized welding process parameters obtained in step 4 and the optimized multi-robot collaborative welding path obtained in step 7, analyze the temperature field of the main bearing nodes of large modules during the multi-robot welding process, obtain the deformation value of the main bearing nodes of large modules during welding through temperature field analysis, and record the deformation value ΔS; Step 9. Return to step 3 to modify the values ​​of the initial input sample welding current A, welding wire diameter ψ, and welding wire extension length s, and execute steps 4 to 9 until the initial input sample welding current A is equal to the maximum welding current A. max End, proceed to the next step; Step 10: Determine the minimum deformation value ΔS among the deformation values ​​of the main bearing nodes of the large module min , the minimum deformation value ΔS min The corresponding welding parameters and welding paths are output as a collaborative welding plan.

2. The method for optimizing the collaborative welding process parameters of the main bearing node of the molded module according to claim 1 is characterized in that: The specific optimization steps of step 4 are as follows: The first step is to set the expected welding time T0 of the optimized BP neural network and the threshold ε for stopping optimization; Step 2: Set the initial input sample of the optimized BP neural network, welding current A=1A, welding wire diameter ψ=1mm, welding wire extension length s=1mm, and welding process parameter adjustment weight ω=1, deviation Δ=1; Step 3: Input the initial input sample welding current A=1A, welding wire diameter ψ=1mm, welding wire extension length s=1mm, weight ω=1, and deviation Δ=1 set in step 2 into the optimized BP neural network to obtain different welding times T under the same weld length S1; Step 4: Compare the output welding time T with the expected welding time T0 to obtain the error value Δε; Step 5: Use the back propagation algorithm to adjust and optimize the weight ω and deviation Δ of the BP neural network model, return to step 2, reset the initial input sample welding current A, wire diameter ψ, and wire extension length s, and repeat steps 2 to 5 until the error value Δε is less than the set threshold ε; Step 6. When the output error value is less than the set threshold, the corresponding welding current A, welding wire diameter ψ, and welding wire extension length s are used as the optimized welding process parameters.

3. The method for optimizing the collaborative welding process parameters of the main bearing node of the molded module according to claim 1 is characterized in that: The specific path optimization steps of step seven are as follows: The first step is to extract the location of the welds between the main load-bearing node plates of the large modules in each quadrant and the length information of each weld, and plan all the welds in the same quadrant into a weld trajectory through the robot welding information simulation software, and obtain the weld path that the robot needs to weld in each quadrant; Step 2: Set the welding priority of the welds. The welds in the same quadrant are ranked based on the starting point P. i1 Plan the starting point for welding, give priority to welding short welds, then weld long welds. When multiple welds have the same welding priority, take the origin O of the coordinate system O-XYZ as the node center, give priority to welding the welds close to the node center, then weld the welds far from the node center, until the end point P i2 ; The third step is to establish a multi-objective optimization function F(x) of the welding trajectory with welding path and welding time as optimization targets: F(x)=Min(f1(x),f2(x)) Where, f1(x) is the welding path optimization function, and f2(x) is the welding time optimization function; The specific optimization function f1(x) of the welding path is expressed as: In the formula, (x i ,y i ,z i ) is the coordinate value of a point on the welding path, (x i+1 ,y i+1 ,z i+1 ) is the coordinate value of the adjacent points on the welding path, n is the number of points on the welding path, and i is the current sequence value of the welding point; The specific optimization function f2(x) of welding time is expressed as: Where, L j is the length of a weld on the main bearing node of the large module, v j is the welding speed set when the robot welds a certain weld, m is the total number of welds, and j is the current weld rank value; Step 4: According to the starting point P of each quadrant i1 and the end point P i2 , taking the welding priority of the weld as the constraint condition, the multi-objective particle swarm algorithm is used to iterate the multi-objective optimization function F(x) of the welding trajectory, and the multi-robot collaborative welding optimization path of the main bearing node of the large module is obtained; Step 5: Save and output the optimized path of multi-robot collaborative welding of the main bearing nodes of large modules.

Images