Waam layer-in directional bead collaborative planning method and system based on thermal state perception

Abstract

The application discloses a WAAM layer in-situ directed bead collaborative planning method and system based on thermal state perception, and the method comprises the following steps: generating a plurality of to-be-deposited bead segments and extracting bead geometry and topological information; constructing a sequence-direction hybrid chromosome, determining actual arcing points and arc collecting points and effective cooling intervals, and establishing a thermal history inheritance model between the beads; generating an initial population; using a thermal-mechanical field proxy model to quickly predict different path schemes, and constructing a multi-objective fitness function for comprehensive evaluation; iterative optimization is performed, and a dynamic penalty function is introduced in the iteration process to constrain and determine the continuity of the WAAM welding gun movement, the depositability and the interference risk of adjacent beads; and the optimal bead deposition sequence and deposition direction are output. Through the establishment of the thermal history transmission relationship between the sequence and the direction, the collaborative optimization of the layer temperature field and the residual stress distribution is realized, so that the additive manufacturing forming quality and stability are improved.

Description

Technical Field

[0001] This invention belongs to the field of additive manufacturing and relates to path planning technology in line-arc additive manufacturing. Specifically, it relates to a method and system for directional weld bead collaborative planning within a WAAM layer based on thermal state perception. Background Technology

[0002] Wire arc additive manufacturing (WAAM) is a high-efficiency metal additive manufacturing technology that uses an electric arc as a heat source and metal wire as a filler material. Compared with traditional powder additive manufacturing technology, WAAM has advantages such as high deposition efficiency, high material utilization, low equipment cost, and suitability for manufacturing large and complex metal structural components. Therefore, it has received widespread attention and application in aerospace, energy equipment, marine engineering, and the manufacturing of large structural components.

[0003] In the WAAM (Wastewater Aluminum Metallization) manufacturing process, components are typically formed into three-dimensional structures by depositing weld beads layer by layer. The deposition path planning of weld beads within each layer has a significant impact on the heat input distribution during manufacturing. The weld bead deposition sequence and direction not only determine the trajectory of the welding heat source on the component surface but also affect heat transfer between adjacent weld beads, the cooling rate, and the spatial distribution of the temperature field within the layer. Therefore, rational path planning is crucial for controlling temperature field uniformity, reducing residual stress levels, and improving component forming accuracy.

[0004] When the weld path within a layer is poorly planned, excessive heat accumulation in localized areas can easily occur, leading to localized overheating of the component. Simultaneously, excessive temperature gradients can also cause significant residual stress and structural warping. Furthermore, an inappropriate deposition path can result in unstable weld geometry, degrading interlayer forming quality and further affecting the uniformity of the material's microstructure and mechanical properties. Therefore, optimizing the weld deposition path to control the thermal history distribution during manufacturing is one of the key issues in WAAM process research.

[0005] Existing WAAM path planning methods are mostly based on geometric infill strategies, such as parallel scan paths, reciprocating scan paths, or contour offset paths. These methods are mainly designed from the perspective of geometric coverage efficiency or processing efficiency, usually treating each weld bead as an independent processing unit and optimizing only the path length or printing time. However, these methods often ignore the differences in heat diffusion of the welding heat source under different deposition directions, as well as the heat history transfer relationship between adjacent weld beads.

[0006] In recent years, some studies have attempted to introduce intelligent optimization algorithms such as genetic algorithms to optimize the weld bead deposition sequence in order to reduce heat accumulation or minimize idle path travel. However, existing studies generally only optimize the weld bead deposition sequence, treating the weld bead deposition direction as a fixed parameter or independent variable, failing to establish a coupling relationship between the deposition sequence and the deposition direction. Furthermore, existing methods generally lack modeling of the thermal history inheritance effect between weld beads, making it difficult to effectively control the evolution of the temperature field within the layer. Summary of the Invention

[0007] Purpose of the invention: In order to overcome the shortcomings of the existing technology, a method and system for directional weld bead collaborative planning within a WAAM layer based on thermal state perception is provided. By establishing the thermal history transfer relationship between sequence and direction, the method achieves collaborative optimization of the temperature field and residual stress distribution within the layer, thereby improving the quality and stability of additive manufacturing.

[0008] Technical Solution: To achieve the above objectives, this invention provides a directed weld bead collaborative planning method within a WAAM layer based on thermal state awareness, comprising the following steps:

[0009] S1: Obtain the single-layer two-dimensional cross-sectional profile of the target component, discretize the cross-section according to the set weld width and overlap rate, generate several weld segments to be deposited, and extract the weld geometry and topology information.

[0010] S2: Based on the geometric and topological information of the weld bead, a hybrid chromosome structure containing sequential coding segments and directional coding segments is constructed. The actual arc start point and arc end point of each weld bead are determined according to the directional coding. The effective cooling interval is determined according to the endpoint transfer distance and transition time of adjacent deposited weld beads. A thermal history inheritance model between weld beads is established so that the deposition direction has a coupled influence on the initial thermal state of subsequent weld beads.

[0011] S3: Based on the hybrid chromosome structure, an initial population is generated using a random path and heuristic scanning strategy;

[0012] S4: For the initial population, the initial thermal state of each weld in the candidate path is determined according to the thermal history inheritance model. The initial thermal state, weld geometry, transition motion and process parameter characteristics are input into the thermal-mechanical field proxy model to predict the temperature field distribution, thermal gradient and residual stress index of different path schemes. After dimensionless processing, a multi-objective evaluation function is constructed for comprehensive evaluation.

[0013] S5: Under the genetic algorithm framework, iterative optimization is performed through selection, crossover and mutation operations, and the dynamic penalty coefficient is updated according to the current generation to constrain and determine the continuity of WAAM welding torch movement, deposition executability and the risk of interference between adjacent weld beads.

[0014] S6: Outputs the optimal weld bead deposition sequence and direction, and generates WAAM welding torch deposition path and process parameter instructions.

[0015] Further, step S1 includes:

[0016] A1: Read the CAD model file of the target component, which contains complete geometric information and 3D coordinate data of the component; the system slices the model at equal intervals along the Z-axis according to a preset layer thickness, generating a set of 2D cross-sectional contours for each slice. ,in , Total number of floors;

[0017] A2: For each two-dimensional cross-section, the system determines the weld width based on the set weld width. and overlap rate The cross-sectional region is discretized to generate a set of weld bead segments to be deposited:

[0018]

[0019] Where N is the total number of weld beads generated in this layer section; each weld bead segment includes: weld bead centerline coordinates weld length Adjacency topology set Record the spatial adjacency information between the current weld bead and adjacent weld beads;

[0020] A3: After the weld bead is discretized, the system further establishes the weld bead heat-affected adjacency matrix. ,in:

[0021]

[0022] in, As the topological adjacency indicator factor, if the weld bead With weld Spatial adjacency, then ,otherwise ; Indicates weld bead With weld Euclidean distance between centers; matrix It is used to quantify the heat-affected zone intensity between weld passes and to provide a data basis for subsequent thermal history inheritance calculations; The characteristic length of the heat effect is used to characterize the rate at which the heat effect decays as the weld bead spacing increases.

[0023] Further, step S2 includes:

[0024] B1: Constructing a hybrid chromosome;

[0025] In the genetic algorithm, a hybrid chromosome of length 2N is generated for each weld bead layer:

[0026]

[0027] Among them: sequential coding segment Used to indicate the weld bead deposition access order, each Corresponding to the weld bead number of the k-th deposition execution; direction coding segment Used to indicate the direction of weld bead deposition;

[0028] B2: Sequence and direction binding mapping;

[0029] In actual implementation, the k-th weld bead deposition was numbered as follows: Its corresponding direction is determined by the direction coding segment:

[0030]

[0031] Actual arc starting point of weld With the end of the arc The calculation formula is:

[0032]

[0033] in, These are the geometric start and end coordinates of the weld bead, respectively.

[0034] B3: Establish a heat history inheritance model between weld passes;

[0035] Thermal history inheritance of weld bead k Defined as:

[0036]

[0037] Directional coupling factor:

[0038]

[0039] Initial temperature of weld bead k:

[0040]

[0041] in, For the first weld bead and the first The heat-affected zone coefficient between weld beads; For the first The second deposition ended until the [number]th [deposition / deposition]. The cumulative time interval between the start of each deposition includes all transfer, waiting, and intermediate weld bead deposition time between two depositions; The thermal decay coefficient; For the first Unit vector in the direction of the weld bead; This represents the actual arc termination point of the j-th deposition weld bead. This is the actual arc starting point of the k-th deposition weld bead; It is a very small positive number; This is the thermal inheritance gain coefficient; The ambient temperature; This is the directional coupling weighting coefficient, used to adjust the influence of weld bead deposition direction on the amount of thermal history inherited.

[0042] Further, step S3 includes:

[0043] C1: Random path generation;

[0044] Based on the constructed hybrid chromosome structure Several individuals are randomly generated as the initial population. The sequential coding segment P of each individual is randomly arranged to form the weld bead number, and the direction coding segment D is randomly assigned a value of 0 or 1. The generated individuals ensure that each weld bead number is unique, and the direction coding corresponds to the sequential coding.

[0045] C2: Heuristic path generation;

[0046] Reference paths are generated using heuristic rules, including: reciprocating scanning path: scanning sequentially along the cross-sectional contour to generate weld bead sequence in turn; thermal equilibrium priority path: calculating the heat accumulation value based on the thermal influence matrix A and prioritizing the weld bead sequence with lower heat accumulation; each individual generated by the heuristic path contains a sequence code P and a direction code D;

[0047] C3: Initial population composition;

[0048] The initial population is composed of individuals using random paths and individuals using heuristic paths, with a total population size of M.

[0049] .

[0050] Furthermore, the prediction method for the thermo-mechanical field surrogate model in step S4 includes:

[0051] D1: Input vector construction;

[0052] For each chromosome individual Construct the input vector for the thermal-mechanical field prediction model:

[0053]

[0054] Where L is the weld length vector; A is the heat-affected adjacency matrix; This represents the initial thermal state vector for each weld bead. The characteristic vector of inter-bead transition motion and process parameters includes at least one or more of the following: endpoint transfer distance, effective cooling interval, deposition rate, arc power, or equivalent heat input;

[0055] D2: Model prediction;

[0056] By calling machine learning or physically constrained neural network models to make predictions, thermal-dynamic field indices can be obtained.

[0057]

[0058] in, This is an indicator of temperature uniformity. The maximum thermal gradient index; This is a residual stress index;

[0059] D3: Initial thermal state update;

[0060] Based on thermal history inheritance Calculate the initial temperature of the k-th weld bead:

[0061] .

[0062] Furthermore, in step D2, the physical constraint neural network model satisfies the heat conduction residual constraint:

[0063] in, To predict the temperature field for the model, As an equivalent heat source input, , , These are the material density, specific heat capacity, and thermal conductivity, respectively. It is a gradient operator used to characterize the rate of change of the temperature field in the spatial direction.

[0064] Furthermore, the multi-objective fitness function in step S4 is expressed as follows:

[0065]

[0066] in, , , The weighting coefficients for temperature uniformity, maximum thermal gradient, and residual stress are respectively, satisfying:

[0067]

[0068] This is a function for evaluating temperature uniformity. This is the thermal gradient evaluation function. This is the residual stress evaluation function. All the above evaluation indicators have been dimensionless; a smaller F-value indicates a better overall evaluation result for the candidate path.

[0069] Furthermore, the constraint determination process in step S5 includes:

[0070] E1: WAAM welding torch movement continuity and deposition feasibility determination;

[0071] Determine the target pose of the welding torch corresponding to the k-th weld bead. Is it reachable?

[0072]

[0073] in, Let K be the target pose of the welding torch corresponding to the k-th weld pass. For the attitude variables of the WAAM actuator, For WAAM actuator motion mapping, For the device to execute motion domain;

[0074] E2: Welding torch collision detection;

[0075] Calculate the minimum safe distance:

[0076]

[0077] in, Let be the geometric envelope set of the welding torch and end effector during the k-th deposition. It is a collection of obstacles consisting of the workpiece, the fixture, and the deposited area;

[0078] like If so, the path scheme is determined to have a collision risk; It is the safe distance threshold;

[0079] E3: Dynamic penalty value calculation;

[0080] For the i-th candidate individual, define the comprehensive constraint violation quantity. :

[0081]

[0082] in, These are the non-negative constraint weight coefficients. This is a non-negative process constraint violation quantity; the reachability, collision distance, and process constraint terms in this formula are all calculated for the i-th candidate individual;

[0083] Based on the comprehensive constraint violation amount With the dynamic penalty coefficient of the g-th generation Calculate the dynamic penalty value for this individual. :

[0084]

[0085]

[0086] in, Let g be the maximum number of generations, and g be the current number of generations. ,and ; i is the candidate individual index, distinguished from the evolutionary generation g.

[0087] Furthermore, the execution of the selection, crossover, and mutation operations in step S5 includes:

[0088] Select operation:

[0089] For each individual in the current population Calculate its comprehensive objective function value:

[0090]

[0091] in Let be the multi-objective evaluation function value for the i-th candidate individual. This represents the dynamic penalty value for the individual in the g-th generation;

[0092] The selection operation is performed according to the comprehensive objective function value from smallest to largest;

[0093] Cross operation:

[0094] The interleaving operation is performed on the sequential coded segment P and the directional coded segment D respectively:

[0095] Sequential encoding segment partial matching crossover: randomly select two parent individuals , Randomly determine the intersection interval [u,v]; retain the sequential code of parent generation 1 in interval [u,v], and fill the other positions sequentially according to the remaining weld order of parent generation 2; generate the child generation sequential code. ;

[0096] Single-point intersection of directional coding segments: for parent directional coding segments Randomly select a single intersection position q; offspring direction encoding :

[0097]

[0098] After the crossover operation is completed, a child set is generated. ;

[0099] Mutation operation:

[0100] Mutation operations are performed separately for sequential encoding and directional encoding:

[0101] Sequential code swap mutation: Randomly select two positions r and s in the sequential code segment and swap the weld bead numbers.

[0102]

[0103] Directional coding bit flip mutation: Perform bit flipping on the i-th bit of the directional coding segment:

[0104]

[0105] Based on the heat-sensitive region of the weld, the probability of orientation variation is weighted:

[0106]

[0107] in, The absolute value of the change in heat history inheritance of the weld bead after fixing the weld bead sequence and other directional positions and only flipping the direction of the i-th weld bead; Based on the probability of mutation, ; The thermal sensitivity enhancement coefficient, To avoid the denominator being zero, the same normalization benchmark is used for all directions, and the resulting probability of variation does not exceed 1; the basic probability of variation is taken when all changes are zero.

[0108] This invention also provides a directed weld bead collaborative planning system within a WAAM layer based on thermal state awareness, comprising:

[0109] The layer section discretization modeling module obtains the single-layer two-dimensional section profile of the target component, discretizes the section according to the set weld width and overlap rate, generates several weld segments to be deposited, and extracts the geometric and topological information of the weld.

[0110] The chromosome construction module constructs a hybrid chromosome structure containing sequential coding segments and directional coding segments. Based on the directional coding, it determines the actual arc start point and arc end point of each weld bead, and determines the effective cooling interval based on the endpoint transfer distance and transition time of adjacent deposited weld beads. It establishes a thermal history inheritance model between weld beads, so that the deposition direction has a coupled influence on the initial thermal state of subsequent weld beads.

[0111] The initial population generation module uses a random path and heuristic scanning strategy to generate the initial population;

[0112] The path scheme evaluation module determines the initial thermal state of each weld bead based on the thermal history inheritance model, and inputs the initial thermal state, weld bead geometry, transition motion and process parameter characteristics into the thermo-mechanical field proxy model to predict the temperature field distribution, thermal gradient and residual stress index of different path schemes. After dimensionless processing, a multi-objective evaluation function is constructed for comprehensive evaluation.

[0113] The constraint determination module introduces a dynamic penalty function to determine the constraints on the continuity of WAAM welding torch movement, deposition feasibility, and interference risk between adjacent weld beads.

[0114] The genetic evolution search module introduces a dynamic penalty term during the genetic evolution process and performs iterative optimization through selection, crossover, and mutation operations within the genetic algorithm framework.

[0115] The optimal path output module outputs the optimal weld bead deposition sequence and direction, and generates WAAM welding torch deposition path and process parameter instructions.

[0116] Beneficial effects: Compared with the prior art, the present invention has the following advantages:

[0117] 1. Establishing a Sequence and Direction Coupled Optimization Mechanism: This invention, for the first time, jointly encodes and couples the weld bead deposition sequence and direction using a genetic algorithm. Through the binding mapping between direction encoding and weld bead number, and the calculation of thermal history inheritance, the direct influence of the deposition direction of the preceding weld bead on the initial temperature and heat accumulation of subsequent weld beads is realized. This mechanism ensures that path optimization no longer only considers geometric constraints, but also the thermal state transfer relationship between weld beads, thereby more realistically simulating the heat conduction and accumulation effects in the WAAM process, making the optimized path more consistent with the actual welding physics process.

[0118] 2. Path planning is upgraded from geometric optimization to thermal history control: Unlike existing methods that only perform geometric optimization on the weld bead sequence, this invention extends the path planning problem to a thermal history optimization problem by constructing a rapid thermo-mechanical field prediction model and thermal history inheritance. During the planning process, not only the weld bead deposition sequence is optimized, but also the deposition direction of each weld bead is optimized to control the evolution of local temperature distribution, thermal gradient, and residual stress. This technological advancement enables path planning to simultaneously consider geometric coverage, temperature uniformity, and component mechanical properties, providing a guarantee for high-precision additive manufacturing.

[0119] 3. Reduce the risk of temperature gradient and residual stress within the layer: By coupling and optimizing the weld sequence and direction, this invention can adjust the heat input sequence and endpoint connection relationship between welds, reduce the continuous accumulation of local heat and uneven cooling, thereby helping to reduce the local temperature gradient and suppress residual stress concentration, and improve the stability of the thermal state and forming process within the layer.

[0120] 4. Improved Forming Stability and Dimensional Accuracy of WAAM Components: This invention effectively suppresses warping, deformation, and dimensional deviations in components during manufacturing by optimizing the weld deposition path and reducing heat accumulation and temperature gradient fluctuations. The coupled optimization of deposition sequence and orientation ensures controllable heat input for each weld layer, and a dynamic penalty mechanism guarantees path feasibility under robot kinematic constraints. This method significantly improves the manufacturing stability and final forming accuracy of WAAM components while ensuring optimized thermo-mechanical performance.

[0121] 5. Strong engineering robustness and wide applicability: The path optimization method of this invention has good engineering scalability and adaptability, and parameters can be adjusted for components of different sizes and shapes, including weld width, overlap ratio, heat-affected zone coefficient, and robot numerical control constraints. Simultaneously, the adaptive dynamic penalty function ensures that the generated weld path is fully feasible under robot kinematic reachability and collision constraints, making this invention applicable to various industrial WAAM systems, including multi-joint robotic arms and additive manufacturing of large structural components.

[0122] 6. Improved Process Efficiency and Controllability: While ensuring thermal equilibrium and minimizing residual stress, this invention efficiently generates paths through a genetic algorithm, combined with a heuristic initial population and a weighted strategy for heat-sensitive regions, enabling the optimization process to quickly converge to a high-quality solution. The optimized weld bead sequence and direction combination not only improves the controllability of thermo-mechanical synergistic control but also effectively shortens path planning time, providing a fast and reliable path generation solution for actual production. Attached Figure Description

[0123] Figure 1 This is a flowchart illustrating the overall process of the method of the present invention.

[0124] Figure 2 This is a schematic diagram of a mixed chromosome structure;

[0125] Figure 3 This is a schematic diagram illustrating the relationship between thermal history inheritance and temperature field coupling during the deposition process of adjacent weld beads.

[0126] Figure 4 This is a flowchart of the iterative process for thermal state-aware collaborative planning based on multi-objective fitness function evaluation.

[0127] Figure 5 The images show a comparison of the macroscopic morphology of arc additive manufacturing (WAAM) samples prepared using the method of the present invention and samples prepared using the method of the present invention. In each image, (a) is a sample prepared using the method of the present invention and (b) is a sample prepared using the method of the present invention. The upper half of each image shows the top morphology and the lower half shows the side morphology. Detailed Implementation

[0128] The present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. After reading this invention, any modifications of the invention in various equivalent forms by those skilled in the art will fall within the scope defined by the appended claims.

[0129] Example 1:

[0130] This embodiment provides a directed weld bead collaborative planning method within a WAAM layer based on thermal state perception. Thermal state perception refers to estimating, updating, and evaluating the initial thermal state of the subsequent weld bead, as well as the temperature field, thermal gradient, and residual stress risk within the layer, based on weld bead geometry and topology information, actual arc initiation and termination points, effective cooling intervals between weld beads, and thermal history inheritance. The thermal state estimation calculated by the model is the basic method, and it can be corrected by combining temperature detection data or forming process state data.

[0131] like Figure 1 As shown, the directed weld bead collaborative planning method within the WAAM layer includes the following steps:

[0132] S1: Obtain the single-layer two-dimensional cross-sectional profile of the target component, discretize the cross-section according to the set weld width and overlap rate, generate several weld segments to be deposited, and extract the weld geometry and topology information.

[0133] Step S1 includes:

[0134] A1: Read the CAD model file of the target component, which contains complete geometric information and 3D coordinate data of the component; the system slices the model at equal intervals along the Z-axis according to a preset layer thickness, generating a set of 2D cross-sectional contours for each slice. ,in , Total number of floors;

[0135] Each two-dimensional cross-sectional profile consists of a series of continuous closed curves, representing the boundary shape of the current layer. The system records the coordinates of each node of the curve for subsequent weld planning. During the slicing process, the system controls the cross-sectional point density according to the preset layer thickness and resolution to ensure that the weld can accurately cover the cross-sectional area and meet the geometric constraints during the subsequent discretization process.

[0136] A2: For each two-dimensional cross-section, the system determines the weld width based on the set weld width. and overlap rate The cross-sectional region is discretized to generate a set of weld bead segments to be deposited:

[0137]

[0138] Where N is the total number of weld beads generated in this layer section; each weld bead segment includes: weld bead centerline coordinates weld length Adjacency topology set Record the spatial adjacency information between the current weld bead and adjacent weld beads;

[0139] The specific discretization process is as follows: Equally spaced scan lines are generated along the cross-sectional contour, with a scan line spacing of... Ensure the overlap area meets process requirements; generate the start and end coordinates of the weld bead segment between the intersection of the scan line and the contour; calculate the coordinates of the weld bead centerline. and weld length The adjacency relationship of weld beads is determined based on spatial distance and stored as a set. Repeat the above process for the entire cross-section until the entire weld bead layer is generated. This step ensures complete weld bead coverage and reasonable spacing, providing basic geometric data for subsequent optimization of weld bead sequence and direction.

[0140] A3: After the weld bead is discretized, the system further establishes the weld bead heat-affected adjacency matrix. ,in:

[0141]

[0142] in, As the topological adjacency indicator factor, if the weld bead With weld Spatial adjacency, then ,otherwise ; Indicates weld bead With weld Euclidean distance between centers; matrix It is used to quantify the heat-affected zone intensity between weld passes and to provide a data basis for subsequent thermal history inheritance calculations; The characteristic length of the heat effect is used to characterize the rate at which the heat effect decays as the weld bead spacing increases.

[0143] During discretization, not only were the centerline coordinates, length parameters, and adjacency topology of each weld bead segment extracted, but a thermal impact adjacency matrix was also established to quantify the potential thermal interference between adjacent weld beads. In this way, the deposition strategy for each weld bead considers not only spatial constraints but also the transmission and accumulation of thermal states, providing fundamental data for sequential and directional coupled optimization. This ensures that the deposition decisions for each weld bead in subsequent optimization processes can balance geometric constraints and thermal field control.

[0144] S2: Based on the geometric and topological information of the weld bead, a hybrid chromosome structure containing sequential coding segments and directional coding segments is constructed. The actual arc start point and arc end point of each weld bead are determined according to the directional coding. The effective cooling interval is determined according to the endpoint transfer distance and transition time of adjacent deposited weld beads. A thermal history inheritance model between weld beads is established so that the deposition direction has a coupled influence on the initial thermal state of subsequent weld beads.

[0145] Step S2 includes:

[0146] B1: Constructing a hybrid chromosome;

[0147] In the genetic algorithm, a hybrid chromosome of length 2N is generated for each weld bead layer:

[0148]

[0149] Among them: sequential coding segment Used to indicate the weld bead deposition access order, each Corresponding to the weld bead number of the k-th deposition execution; direction coding segment Used to represent the weld bead deposition direction; the chromosome structure ensures that the weld bead deposition order and direction can be adjusted simultaneously during genetic algorithm operations (crossover, mutation), achieving thermal history coupling optimization;

[0150] The hybrid chromosome structure constructed in this embodiment is as follows: Figure 2 As shown. Figure 2 As shown, the hybrid chromosome consists of sequential coding segments and directional coding segments: the sequential coding segments indicate the deposition access order of each weld bead, and the directional coding segments indicate the arc start and arc end directions of each weld bead; the red deposition bars in the figure represent the weld bead deposition segments selected by the current genetic individual, and the arrows indicate the execution direction of the corresponding weld bead.

[0151] B2: Sequence and direction binding mapping;

[0152] In actual implementation, the k-th weld bead deposition was numbered as follows: Its corresponding direction is determined by the direction coding segment:

[0153] Actual arc starting point of weld With the end of the arc The calculation formula is:

[0154]

[0155] in, These are the geometric start and end coordinates of the weld bead, respectively; this mapping ensures that the deposition direction of each weld bead is bound to its number and provides input for thermal history inheritance calculations;

[0156] Directional encoding is not only used to represent the geometric deposition direction of the weld bead, but also directly determines the arc start and arc end positions of the weld bead. It also affects the cooling behavior and thermal history inheritance process between weld beads through the spatial transfer distance between the arc end point and the arc start point of the next weld bead, thus making directional encoding a thermal state coupling optimization variable, rather than a simple geometric execution variable.

[0157] B3: Establish a heat history inheritance model between weld passes;

[0158] Thermal history inheritance of weld bead k Defined as:

[0159]

[0160] Directional coupling factor:

[0161]

[0162] Initial temperature of weld bead k:

[0163]

[0164] in: For the first weld bead and the first The heat-affected zone coefficient between weld beads; For the first The second deposition ended until the [number]th [deposition / deposition]. The cumulative time interval between the start of each deposition includes all transfer, waiting, and intermediate weld bead deposition time between two depositions; The thermal decay coefficient; For the first Unit vector in the direction of the weld bead; This represents the actual arc termination point of the j-th deposition weld bead. This is the actual arc starting point of the k-th deposition weld bead; It is a very small positive number; This is the thermal inheritance gain coefficient; The ambient temperature; This is the directional coupling weighting coefficient, used to adjust the influence of weld bead deposition direction on the amount of thermal history inherited.

[0165] In this embodiment, the actual deposition order is used as the index. The calculation formula is:

[0166]

[0167]

[0168] in, ; The distance between the arc-ending point of the r-th deposition and the arc-starting point of the (r+1)-th deposition is denoted as . Transitional movement speed; This transfer does not include the waiting time for movement and weld bead deposition, as well as the start and end time of the arc. Let be the deposition duration of the r-th weld bead. All time terms are non-negative, and the summation range is 0 when empty. Two consecutive depositions satisfy . The thermal history contribution of earlier deposited weld beads is attenuated based on the cumulative time mentioned above.

[0169] B4: Heat-sensitive weld treatment;

[0170] After chromosome construction is completed, the population generation strategy is adjusted based on the weld heat sensitivity, including: increasing the probability of random mutation in directional encoding for heat-sensitive welds; and applying heuristic initial path configuration to sequential encoding segments with significant heat impact. This step ensures that the initial population covers key heat-sensitive regions, providing a feasible solution basis for subsequent genetic algorithm iterations.

[0171] Directional encoding not only marks the geometric start and end positions of the welding arc, but also serves as a boundary condition input for the initial temperature of the next weld pass, forming a mechanism for the inheritance and coupling of thermal states between weld passes. This mechanism can upgrade traditional path planning from simple geometric optimization to thermal history optimization, enabling the genetic algorithm to simultaneously control temperature uniformity and residual stress distribution when searching for the optimal deposition path, thereby significantly improving the forming quality and manufacturing stability of WAAM components.

[0172] like Figure 3 As shown, the arc-ending position of the previous weld bead, the arc-starting position of the next weld bead, and the spatial adjacency relationship between weld beads jointly determine the amount of thermal history inheritance. Therefore, changes in the deposition direction will alter the cooling time and initial thermal state of adjacent weld beads, thereby affecting the subsequent temperature field and residual stress distribution.

[0173] S3: Based on the hybrid chromosome structure, an initial population is generated using a random path and heuristic scanning strategy;

[0174] Step S3 includes:

[0175] C1: Random path generation;

[0176] Based on the constructed hybrid chromosome structure Several individuals are randomly generated as the initial population. The sequential coding segment P of each individual is randomly arranged to form the weld bead number, and the direction coding segment D is randomly assigned a value of 0 or 1. The generated individuals ensure that each weld bead number is unique, and the direction coding corresponds to the sequential coding.

[0177] C2: Heuristic path generation;

[0178] Reference paths are generated using heuristic rules, including: reciprocating scanning path: scanning sequentially along the cross-sectional contour to generate weld bead sequence in turn; thermal equilibrium priority path: calculating the heat accumulation value based on the heat influence matrix A and prioritizing the weld bead sequence with lower heat accumulation; each individual generated by the heuristic path contains a sequence code P and a direction code D to ensure chromosome legality;

[0179] C3: Initial population composition;

[0180] The initial population is composed of individuals using random paths and individuals using heuristic paths, with a total population size of M.

[0181]

[0182] Ensure that the population covers the global search space and includes feasible reference solutions.

[0183] Random generation ensures global search space coverage, while heuristic path rules generate initial individuals based on reciprocal scanning or a thermal equilibrium priority strategy, guaranteeing the existence of feasible reference solutions in the population. Each individual in the initial population contains a joint configuration of sequential and directional encodings, enabling subsequent genetic operations to perform thermal-mechanical field optimization while ensuring path feasibility.

[0184] S4: For the initial population, the initial thermal state of each weld in the candidate path is determined according to the thermal history inheritance model. The initial thermal state, weld geometry, transition motion and process parameter characteristics are input into the thermal-mechanical field proxy model to predict the temperature field distribution, thermal gradient and residual stress index of different path schemes. After dimensionless processing, a multi-objective evaluation function is constructed for comprehensive evaluation.

[0185] In this embodiment, the rapid prediction model for thermal-mechanical fields is pre-trained through the following steps:

[0186] 1) Construct a parameterized weld bead sequence dataset;

[0187] Each sample in the dataset contains at least the weld sequence code P, the orientation code D, the weld geometry feature L, the thermal influence adjacency matrix A, and the initial thermal state vector generated based on the thermal history inheritance model. ;

[0188] To enable the model to learn the thermal boundary changes caused by orientation reversal, different orientation combinations of samples are generated for the same weld bead set when constructing the dataset, so that the model can explicitly learn the mapping relationship of "orientation-final arc position-thermal inheritance-stress evolution".

[0189] 2) Obtain temperature and stress field labels for each sequence based on finite element analysis, including temperature variance labels. Maximum thermal gradient label and residual stress label ;

[0190] 3) Train the neural network surrogate model until the prediction error is less than the threshold.

[0191] In this embodiment, training samples can be obtained through one or more of the following methods: finite element simulation, infrared thermometry, thermocouple thermometry, or molten pool image inversion. Each sample corresponds to a candidate deposition path. Input features include at least sequential encoding, directional encoding, weld length, weld endpoint coordinates, weld center distance, thermal influence adjacency matrix, endpoint transfer distance, effective cooling interval, welding current, arc voltage, wire feed speed, deposition rate, material thermal properties, and the initial temperature state obtained from the thermal history inheritance model. Output labels include at least the intralayer temperature variance and the maximum thermal gradient, and, if necessary, equivalent residual stress or residual stress risk indicators.

[0192] During model training, samples can be divided into training and validation sets, and the prediction errors of temperature variance, maximum thermal gradient, and residual stress risk can be used as stopping or filtering criteria. Error thresholds can be determined based on the allowable temperature fluctuation range of the WAAM process, heat source calibration error, historical test repetition error, or simulation mesh convergence error. Through these limitations, the thermal state prediction model in this invention has clearly defined input, output, and validation methods, rather than abstractly calling machine learning models.

[0193] Prediction methods for thermo-mechanical field surrogate models include:

[0194] D1: Input vector construction;

[0195] For each chromosome individual Construct the input vector for the thermal-mechanical field prediction model:

[0196]

[0197] Where L is the weld length vector; A is the heat-affected adjacency matrix; This represents the initial thermal state vector for each weld bead. The characteristic vector of inter-bead transition motion and process parameters includes at least one or more of the following: endpoint transfer distance, effective cooling interval, deposition rate, arc power, or equivalent heat input;

[0198] D2: Model prediction;

[0199] By calling a physical constraint neural network model to make predictions, the thermo-mechanical field indices are obtained:

[0200]

[0201] in, This is an indicator of temperature uniformity. The maximum thermal gradient index; This is a residual stress index;

[0202] Temperature uniformity index is used to characterize the degree of dispersion of temperature distribution after deposition of each weld bead within a single layer, and is defined as follows:

[0203]

[0204] Where M is the total number of sampling points, Let m be the temperature at the m-th sampling point. The average temperature across all sampling points;

[0205] The maximum thermal gradient index is defined as:

[0206]

[0207] in, This is the sampling area for the temperature field of the current layer. The temperature gradient at the m-th sampling point;

[0208] The residual stress index is defined as:

[0209]

[0210] or

[0211]

[0212] in, Let m be the residual stress at the m-th sampling point. The equivalent residual stress at the m-th sampling point;

[0213] The physical constraint neural network model satisfies the heat conduction residual constraint:

[0214] in, To predict the temperature field for the model, As an equivalent heat source input, , , These are the material density, specific heat capacity, and thermal conductivity, respectively. It is a gradient operator used to characterize the rate of change of the temperature field in the spatial direction.

[0215] When the model output includes residual stress indices, the residual stress prediction branch uses thermo-mechanical coupled finite element results or measured residual stress data as supervision labels; the thermal conduction residual is used to constrain the temperature field prediction branch. Alternatively, residual stress risk indices can be calculated based on the predicted temperature field and material thermal strain parameters to maintain the correspondence between model inputs, prediction objects, and evaluation indices.

[0216] D3: Initial thermal state update;

[0217] Based on thermal history inheritance Calculate the initial temperature of the k-th weld bead:

[0218]

[0219] The multi-objective fitness function is expressed as follows:

[0220]

[0221] in, , , The weighting coefficients for temperature uniformity, maximum thermal gradient, and residual stress are respectively, satisfying:

[0222]

[0223] This is a function for evaluating temperature uniformity. This is the thermal gradient evaluation function. This is the residual stress evaluation function. All the above evaluation indicators have been dimensionless; a smaller F-value indicates a better overall evaluation result for the candidate path.

[0224] The fitness function integrates temperature uniformity, maximum thermal gradient and residual stress indices by weighting, and the weights can be adjusted according to specific process requirements to achieve a balance between thermal control and component forming quality.

[0225] S5: As Figure 4 As shown, iterative optimization is performed through selection, crossover and mutation operations within the genetic algorithm framework, and the dynamic penalty coefficient is updated according to the current generation to constrain and determine the continuity of WAAM welding torch movement, deposition executability and interference risk between adjacent weld beads.

[0226] The constraint determination process includes:

[0227] E1: WAAM welding torch movement continuity and deposition feasibility assessment

[0228] Determine the target pose of the welding torch corresponding to the k-th weld bead. Is it reachable?

[0229]

[0230] in, Let K be the target pose of the welding torch corresponding to the k-th weld pass. For the attitude variables of the WAAM actuator, For WAAM actuator motion mapping, For the device to execute motion domain;

[0231] E2: Welding torch collision detection

[0232] Calculate the minimum safe distance:

[0233]

[0234] in, Let be the geometric envelope set of the welding torch and end effector during the k-th deposition. It is a collection of obstacles consisting of the workpiece, the fixture, and the deposited area;

[0235] like If so, the path scheme is determined to have a collision risk; It is the safe distance threshold;

[0236] E3: Dynamic Penalty Value Calculation

[0237] For the i-th candidate individual, define the comprehensive constraint violation quantity. :

[0238]

[0239] in, These are the non-negative constraint weight coefficients. This is a non-negative process constraint violation quantity; the reachability, collision distance, and process constraint terms in this formula are all calculated for the i-th candidate individual;

[0240] Based on the comprehensive constraint violation amount With the dynamic penalty coefficient of the g-th generation Calculate the dynamic penalty value for this individual. :

[0241]

[0242]

[0243] in, Let g be the maximum number of generations, and g be the current number of generations. ,and ; i is the candidate individual index, which is distinguished from the evolutionary generation g. Slight violations are allowed in the early stage, and the penalty is strengthened in the later stage.

[0244] During the genetic iteration process, each individual undergoes checks on the continuity of the WAAM welding torch movement, the feasibility of deposition, and the risk of interference between adjacent weld passes. Individuals that violate the constraints are subject to dynamic penalties or are eliminated. The dynamic penalty function gradually increases with the number of iterations. In the early stages, the algorithm is allowed to explore paths with slight collision risks but excellent thermal performance, while in the later stages, infeasible solutions are forcibly eliminated, thereby achieving a balance between thermal optimization and engineering feasibility.

[0245] The inputs to the executability constraints include the target pose of the welding torch, the geometric envelope of the welding torch and end effector, the set of obstacles in the workpiece and fixture, the area of ​​deposited weld beads, and the motion domain of the equipment. For each candidate path, the system generates a welding torch pose sequence based on the arc initiation point, arc termination point, and transition trajectory, and determines whether the pose sequence is within the reachable domain of the equipment. Simultaneously, it calculates the minimum distance between the welding torch envelope and the obstacle set. This process ensures that the optimization results can be translated into actual WAAM control commands, rather than remaining merely as geometric paths or mathematical ranking results.

[0246] The execution of selection, crossover, and mutation operations includes:

[0247] Select operation:

[0248] For each individual in the current population Calculate its comprehensive objective function value:

[0249]

[0250] in Let be the multi-objective evaluation function value for the i-th candidate individual. This represents the dynamic penalty value for the individual in the g-th generation;

[0251] The selection operation is performed according to the comprehensive objective function value from smallest to largest;

[0252] Cross operation:

[0253] The interleaving operation is performed on the sequential coded segment P and the directional coded segment D respectively:

[0254] Sequential encoding segment partial matching crossover: randomly select two parent individuals , Randomly determine the intersection interval [u,v]; retain the sequential code of parent generation 1 in interval [u,v], and fill the other positions sequentially according to the remaining weld order of parent generation 2; generate the sequential code of the offspring generation. ;

[0255] Single-point intersection of directional coding segments: for parent directional coding segments Randomly select a single intersection position q; offspring direction encoding :

[0256]

[0257] After the crossover operation is completed, a child set is generated. Ensure that the sequence code is valid and that the direction code corresponds to the weld number;

[0258] Mutation operation:

[0259] Mutation operations are performed separately for sequential encoding and directional encoding:

[0260] Sequential code swap mutation: Randomly select two positions r and s in the sequential code segment and swap the weld bead numbers.

[0261]

[0262] Directional coding bit flip mutation: Perform bit flipping on the i-th bit of the directional coding segment:

[0263]

[0264] Based on the heat-sensitive region of the weld, the probability of orientation variation is weighted:

[0265]

[0266] in, The absolute value of the change in heat history inheritance of the weld bead after fixing the weld bead sequence and other directional positions and only flipping the direction of the i-th weld bead; Based on the probability of mutation, ; The thermal sensitivity enhancement coefficient, To avoid the denominator being zero, the same normalization benchmark is used for all directions, and the resulting probability of variation does not exceed 1; the basic probability of variation is taken when all changes are zero.

[0267] During evolution, the elite individual retention strategy ensures the inheritance of high-fitness paths. Simultaneously, crossover operations recombine deposition order and direction information among different superior individuals, while mutation operations break local optima traps and enhance the population's search capabilities. Since direction encoding participates in the weld bead thermal history inheritance process, mutations in direction positions not only change the start and end positions of a single weld bead's deposition but also alter the thermal state transfer relationship between it and its adjacent weld beads. This allows the genetic algorithm to search the coupled solution space that existing path planning methods that only optimize the order cannot cover during evolution. Through continuous iterative updates of the genetic algorithm, the system continuously optimizes the weld bead deposition order and direction combination, achieving a comprehensive optimization goal of thermal equilibrium, thermal gradient suppression, and residual stress minimization. Through iterative updates, the system continuously optimizes the weld bead deposition order and direction combination, achieving thermal equilibrium and residual stress minimization.

[0268] S6: Output the optimal weld bead deposition sequence and direction, and generate WAAM welding gun deposition path and process parameter instructions;

[0269] Step S6 includes:

[0270] F1: Termination condition determination;

[0271] Determine if the genetic iteration meets the termination condition: reaching the maximum number of generations. When the population fitness converges and the change in optimal fitness is below a threshold for several consecutive generations, the genetic iteration stops if the condition is met.

[0272] F2: Optimal individual selection;

[0273] Select the individual with the smallest overall objective function value from the final population. As the optimal weld bead deposition sequence and direction, among which For optimal weld sequence and Encode the optimal direction;

[0274] F3: WAAM deposition control command generation;

[0275] In this embodiment, the WAAM deposition control commands may include the welding torch movement trajectory points, welding torch attitude angles, welding current, welding voltage, wire feed speed, deposition speed, arc initiation commands, arc termination commands, and idle movement commands between adjacent weld passes. For different control platforms, these commands can be converted into trajectory files recognizable by G-code, robot control programs, or controllers. This step ensures that the optimization results of this invention are implemented at the WAAM device execution level, avoiding algorithmic results that can only be evaluated offline but cannot control the deposition process.

[0276] Generate WAAM device execution instructions based on the optimal individual:

[0277] 1) Generate the arc starting point for each weld bead. With the end of the arc ;

[0278] 2) Generate a sequence of weld transition trajectory points;

[0279] 3) Generate a sequence for controlling the welding torch posture and deposition rate;

[0280] 4) Outputs a complete motion control instruction file, which can directly drive the WAAM equipment to perform weld bead deposition within the WAAM layer.

[0281] When the termination condition is met, the system outputs the optimal weld bead deposition sequence and corresponding start and end directions, and generates WAAM equipment deposition motion control commands for direct WAAM printing. Through sequence and direction coupling optimization and thermal history inheritance modeling, the method of this invention can effectively reduce intralayer temperature gradient and residual stress, suppress local heat accumulation, improve component forming accuracy and manufacturing stability, and ensure that the generation path is continuous and feasible during the WAAM deposition process, thus possessing high engineering applicability and scalability.

[0282] like Figure 5 As shown, this embodiment compares the macroscopic morphology of WAAM samples prepared using the method of the present invention with those prepared using the method of the present invention. Figure 5The top weld bead of the sample shown in (a) exhibits uneven width and irregular edge contour along its length. The side deposit layer shows obvious undulations and local protrusions, indicating poor consistency in interlayer morphology.

[0283] Compared to Figure 5 (a) Figure 5 The top weld width of the sample shown in (b) is more uniform, the edge contour is smoother, and the regularity of the surface deposition texture is better; the local protrusions and undulations of the deposition layer on the side of the sample are reduced, and the consistency of the interlayer morphology is improved.

[0284] The above comparison results show that, under the experimental conditions of this embodiment, the directional weld bead collaborative planning method based on thermal state perception within the WAAM layer proposed in this invention helps to improve the uniformity of weld bead formation and the consistency of multilayer deposition morphology, improve the macroscopic forming quality of the sample, and provide experimental evidence for the effectiveness of the method of this invention.

[0285] It should be noted that the path optimization ideas for thermal uniformity, thermal gradient, or residual stress are not the exclusive content of this invention. Existing technologies can adjust the heat distribution by scanning the path and evaluate the thermal state of candidate paths through path-level thermal history models. The difference between this invention and such solutions is that it does not only use temperature uniformity or residual stress as fitness indicators, but further transforms the changes in the arc initiation and arc termination caused by the weld bead orientation into the endpoint transfer distance, effective cooling interval, and thermal history inheritance between adjacent weld beads, so that the orientation variable is included in the calculation chain of thermal state update.

[0286] This invention addresses the problem of how to enable genetic algorithms to identify the differences in thermal state caused by different deposition directions within the same weld bead sequence. Only when the orientation position can change the initial temperature state of the subsequent weld bead, and this initial temperature state continues to participate in path-level thermal evaluation, does the sequence-orientation coupled coding cease to be a simple binary extension and become a substantial optimization variable affecting the thermal process within the WAAM layer.

[0287] It is important to emphasize that the innovation of this invention does not lie in general equipment motion control, but rather in the thermal-path coupling planning problem of weld beads within a Wire-Arc Additive Manufacturing (WAAM) layer. During WAAM deposition, the arc heat source moves along the weld bead, and the arc termination position, cooling time, and heat diffusion range of the previous weld bead directly affect the initial temperature state of the subsequent weld bead. Therefore, even with the same set of weld beads and the same deposition sequence, changes in the arc initiation and termination points of each weld bead will significantly alter the heat input distribution, temperature gradient, and residual stress within the layer.

[0288] Compared to existing methods that only optimize scanning sequence or idle travel distance, the distinctive technical concept of this invention lies in unifying weld bead deposition sequence, weld bead deposition direction, inter-weld thermal influence adjacency relationship, cooling time, and thermo-mechanical field evaluation indicators into a single genetic optimization framework. This framework transforms intra-layer path planning from simply finding the shortest geometric path into a coupled optimization problem of "sequence-direction-thermal history-temperature field-residual stress," directly addressing the optimization of WAAM forming quality.

[0289] From a mathematical perspective, the intra-layer path optimization problem of this invention can be summarized as follows:

[0290]

[0291] in, This indicates the weld bead number of the k-th deposition. This indicates the deposition direction of the i-th weld bead. This expression illustrates that the present invention optimizes not a single sequence variable, but a combined path that simultaneously determines "which weld bead to deposit first" and "from which end to begin deposition".

[0292] Based on the above, the innovative points of the method of this invention can be summarized as follows:

[0293] Core Innovation Point 1: Constructing a Sequence-Direction Coupled Hybrid Chromosome Encoding System. Existing path planning within the WAAM layer often employs serpentine scanning, reciprocating scanning, or partitioned scanning, or abstracts the weld bead centerline as a node to obtain a shorter idle path. These methods share the common premise that a weld bead only needs to be arranged in a certain deposition order; the starting and ending points of the arc are usually considered execution-level details. This premise applies to general geometric path sorting but cannot fully reflect the thermal process of deposition within the WAAM layer. For WAAM, the geometric length of the same weld bead starting from the left end is the same as starting from the right end, but the location of the heat concentration at the end of the arc, the starting distance of the next weld bead, the idle cooling time, and the thermal superposition relationship between adjacent weld beads all change. This invention does not simply apply genetic algorithms to weld bead sorting, but defines each weld bead as a directed deposition unit with directional attributes, so that the genetic individual simultaneously contains decision information at both the weld bead access order and weld bead deposition direction levels.

[0294] Compared to chromosomes containing only sequential arrangement, the hybrid chromosome of this invention advances the directional selection, originally hidden in post-processing or artificial process rules, to the optimization variable layer. Thus, the object evaluated by the genetic algorithm is no longer a string of abstract numbers, but a complete deposition scheme that can be directly expanded into an arc initiation point, arc termination point, idle connection segment, and in-layer heat input path. The creativity brought by this processing lies not in increasing the number of variables, but in changing the physical meaning of the path optimization problem: traditional schemes search for "which weld bead goes first," while this invention searches for "which weld bead enters the next weld bead at what time, from which end, and with what endpoint relationship." Only under such encoding can subsequent thermal history inheritance, cooling intervals, and residual stress risks have a calculable, comparable, and evolvable basis.

[0295] From a creative proof perspective, simply adding directional bits after arranging chromosomes is easily misunderstood as an extension of ordinary path coding. This invention emphasizes not "adding another variable," but rather that this variable has a clear physical transmission path: the directional bit first determines the actual arc initiation and termination points, then changes the endpoint transfer distance between two adjacent depositions, further altering the idle cooling time and thermal history inheritance, ultimately affecting temperature variance, maximum thermal gradient, and residual stress risk assessment. This chain transforms the directional bit from an execution command parameter into a thermal state evaluation parameter.

[0296] Specifically, mixed chromosomes can be represented as:

[0297]

[0298] in For sequential coded segments, This is the direction encoding segment. The direction variable further determines the actual arc start and end points of the i-th weld bead, and their relationship can be expressed as:

[0299]

[0300] While the geometry of a weld bead remains unchanged, the direction of heat source movement, the location of heat concentration at the end of the arc, and the connection relationship of the next idle path will change with the orientation. Existing methods that only optimize the sequence can at most distinguish whether weld bead i is before or after weld bead j, but cannot distinguish whether weld bead i ends at an end close to or far from weld bead j. Therefore, even if two candidate paths have the exact same weld bead sequence, their actual thermal cycles may differ depending on the combination of directions. This invention utilizes this point to transform the "endpoint thermal effect" that is easily ignored by path planning in WAAM into a discrete decision variable that can be handled by a genetic algorithm.

[0301] The second core innovation is the establishment of a direction-sensitive weld thermal history inheritance model. Existing solutions, when considering thermal effects, typically employ fixed cooling times, fixed adjacency weights, or perform temperature field simulation verification after path generation. These approaches have two shortcomings: first, thermal evaluation lags behind path search, making it impossible to continuously refine candidate paths during the genetic iteration process; second, they approximate the thermal effect between adjacent welds as a direction-independent constant, failing to reflect the actual distance difference between the end of the arc and the beginning of the next arc. This invention uses spatial adjacency, weld center distance, directional endpoints, and effective cooling intervals together in the thermal history inheritance calculation, enabling different directional combinations within the same weld sequence to yield different thermal state evaluations.

[0302] The weld beads are renumbered according to the actual deposition order of the candidate paths. The thermal influence adjacency strength of weld bead i on weld bead j can be expressed as:

[0303]

[0304] Where χij indicates whether two weld beads are topologically adjacent, Dij represents the distance between the centerlines of the weld beads, and l0 is the length of the heat-affected zone feature. This formula uses the same notation as the aforementioned heat-affected zone adjacency matrix to describe the attenuation of heat-affected zone as the weld bead distance increases.

[0305] For two consecutive weld beads deposited in the candidate path ( Its effective cooling interval can be expressed as:

[0306]

[0307] in, This excludes waiting time for idle travel and start / end delay time. This is the actual arc termination point of the previous weld pass. This is the actual arc starting point for the next weld pass. This represents the idle travel speed. Direction position. Changing the endpoint transfer distance alters the effective cooling interval. For two weld beads with discontinuous deposition sequences, the summation of the thermal history... The cumulative time described in Example B3 includes the deposition time of the intermediate weld bead and all transfer and waiting time, and is not replaced by the direct movement time of the two endpoints.

[0308] The aforementioned effective cooling interval is not a simple time constant, but is determined by the idle distance between the arc termination point of the previous weld bead and the arc initiation point of the next weld bead. In other words, this invention incorporates "natural cooling caused by idle movement" into the evaluation of inter-weld heat inheritance. If the arc termination point of the previous weld bead is closer to the end of the next weld bead, although the idle distance may be shorter, the local area of ​​the next weld bead is still in a high-temperature state at the start, increasing the risk of heat superposition. If the arc termination point is far from the next weld bead after the direction is reversed, the idle distance time increases, local cooling is more sufficient, and the temperature gradient and heat accumulation may decrease. Existing solutions based primarily on the shortest path usually favor the former outcome, while this invention can quantitatively balance idle distance length and thermal state.

[0309] When the order of weld bead access is fixed, traditional sequence optimization models typically provide the same evaluation of candidate paths. However, in this invention, different directional combinations lead to changes in the arc termination point of the preceding weld bead and the arc initiation point of the following weld bead, resulting in different effective cooling intervals. Thus, even if two individuals have the same sequence encoding segment, as long as the direction encoding segments are different, the system will obtain different heat inheritance amounts and different initial temperature states. This difference is the key difference between this invention and methods that only optimize the weld bead sequence.

[0310] The amount of heat history inherited and the initial thermal state of the subsequent weld pass can be expressed as follows:

[0311]

[0312]

[0313] Where β is the cooling attenuation coefficient, γ is the directional coupling weight, Φij is the directional coupling factor, α is the thermal inheritance gain coefficient, and Tenv is the ambient temperature. The above symbols are consistent with the thermal history inheritance model in the claims. Using this model, the initial thermal state of the next weld pass can be updated based on the endpoint relationships of the candidate paths and the effective cooling interval, and path-level thermal indices with lower computational cost can participate in the genetic iterative evaluation.

[0314] This model also avoids the high computational cost of directly embedding complete thermo-mechanical finite element calculations into each generation of genetic search. For each candidate individual in the genetic algorithm, the system first uses endpoint mapping and effective cooling intervals to update the path-level thermal state, and then inputs the updated starting temperature state into subsequent thermal state prediction or fitness evaluation stages. Thus, the thermal history is not passively verified after the path is determined, but continuously influences the selection, crossover, and mutation results of individuals during the candidate path competition process.

[0315] The third core innovation is the construction of a multi-objective fitness function for WAAM forming quality. Existing path optimization methods, which only consider idle travel distance, scan continuity, or weld bead sorting length, tend to result in geometrically short paths with severe localized heat accumulation. For WAAM in-layer deposition, path quality is not equivalent to the length of travel distance; it is more important to consider whether the temperature field within the layer is uniform, whether excessive temperature gradients form between adjacent weld beads, whether local thermal cycling leads to residual stress concentration, and whether the weld bead overlap area remains stable. This invention incorporates temperature field uniformity, maximum temperature gradient, residual stress risk, idle travel distance, and process continuity into the fitness function simultaneously, enabling the genetic algorithm to search with thermo-mechanical forming quality as the objective.

[0316] The fitness function can be expressed as:

[0317]

[0318] Where f1(T) represents the normalized in-layer temperature field discrete index, f2(G) represents the normalized maximum temperature gradient index, and f3(σ) represents the normalized residual stress or residual stress risk index; welding torch movement continuity, collision risk, and process constraints are incorporated into the comprehensive objective function through the aforementioned dynamic penalty values.

[0319] The residual stress risk index can be approximately characterized by the thermal strain constraint relationship, for example:

[0320]

[0321] Where E is the elastic modulus, α is the coefficient of thermal expansion, ν is Poisson's ratio, and ΔT is the characteristic temperature difference or thermal cycling amplitude within the layer. This expression does not simplify the complex thermo-elastic-plastic process into a single constant, but rather is used to quickly identify candidate paths with large temperature differences, steep gradients, and high risk of residual stress concentration in genetic iteration.

[0322] This fitness function embodies the fundamental difference between this invention and general path planning algorithms. General path planning prioritizes motion efficiency and tends to reduce non-depositional movement; this invention allows for slight increases in the idle distance of certain candidate paths, as long as this increase can be traded for lower heat accumulation, gentler temperature gradients, or less residual stress risk. In other words, the optimization objective shifts from "minimizing motion cost" to "making the in-layer thermal state more favorable to forming quality while satisfying deposition coverage and process continuity." This objective shift directly correlates the algorithm output with the deformation, cracking risk, and dimensional accuracy of WAAM components, rather than remaining at the geometric level of route shortening.

[0323] Therefore, the multi-objective fitness function in this invention is not a simple sum of several common process indices. Its evaluation object originates from the aforementioned direction-sensitive thermal history inheritance model: the initial temperature change caused by the directional endpoint relationship will continue to propagate to the intra-layer temperature dispersion, local thermal gradient peak value, and residual stress risk index. This causal relationship prevents the algorithm from pursuing only short idle paths while ignoring local thermal accumulation caused by continuous deposition.

[0324] The fourth core innovation is the design of a heat-sensitive direction mutation mechanism. Conventional genetic algorithms use random swapping, insertion, or reversal for mutation, primarily to maintain population diversity. However, this random perturbation doesn't identify which direction has a greater impact on the thermal state. This invention adjusts the mutation probability of the direction position based on the change in the inherited thermal history before and after the direction reversal, causing the algorithm to prioritize exploring weld bead direction combinations that may significantly alter the heat distribution. This mechanism transforms mutation from a simple random search into a targeted exploration guided by differences in thermal state.

[0325] The function of this mechanism should not be understood as random parameter tuning in a typical genetic algorithm. For a given orientation, the system can compare the amount of thermal history inherited before and after the orientation is flipped, the initial temperature state of the next weld pass, or the changes in the local thermal gradient. If the change is significant, it indicates that the orientation is more sensitive to the thermal state within the layer, and the genetic search should grant it a higher exploration probability. If the change is small, the basic mutation probability can be maintained. In this way, mutation resources are concentrated in thermally sensitive areas, rather than indiscriminately perturbing all orientations.

[0326]

[0327] in, The absolute value of the change in heat history inheritance of the weld bead after fixing the weld bead sequence and other directional positions and only flipping the direction of the i-th weld bead; Based on the probability of mutation, ; The thermal sensitivity enhancement coefficient, To avoid zero denominators, the same normalization benchmark is used for all orientation positions, resulting in a mutation probability not exceeding 1; the base mutation probability is taken when all changes are zero. Through the above mechanism, this invention forms a closed loop of "hybrid chromosome encoding - thermal history inheritance calculation - thermal quality evaluation - thermally sensitive genetic evolution". Its innovation lies not in simply using a genetic algorithm, but in transforming the physical and thermal process of weld bead deposition within the WAAM layer into mathematical variables that can be identified and iteratively optimized by the genetic algorithm: orientation positions change endpoint relationships, endpoint relationships change effective cooling intervals, effective cooling intervals change the amount of heat inheritance, and the amount of heat inheritance further affects the temperature field, residual stress risk, and fitness value. The genetic algorithm thus obtains a search basis for WAAM process quality.

[0328] The innovative effect is further manifested in the clear functional progression among the aforementioned technical features, rather than simply listing several known methods side by side. Hybrid chromosomes first address the problem of "how to incorporate weld direction into heritable expression"; direction endpoint mapping further addresses the problem of "how direction variables correspond to actual arc initiation and termination positions"; the thermal history inheritance model addresses the problem of "how direction changes affect the initial thermal state of the next weld"; the multi-objective fitness function addresses the problem of "how to unify temperature field, residual stress risk, and process continuity as the basis for genetic selection"; and the heat-sensitive direction mutation mechanism addresses the problem of "how to make genetic search prioritize directions more sensitive to heat distribution." Each step is interdependent, with each step providing the computational foundation for the next, and the next step guiding the search results of the previous step.

[0329] Based on existing technologies, even if those skilled in the art consider using genetic algorithms to optimize paths within the WAAM layer, they typically follow the modeling methods of the Traveling Salesman Problem or the shortest path, treating weld beads as undirected nodes or line segments. This approach makes it difficult to naturally derive the directional endpoint thermal inheritance model of this invention, because in undirected node models, the differences in cooling time between the arc-ending endpoint, the arc-starting endpoint, and after direction reversal are not clearly defined variables. On the other hand, simply using thermal simulation to evaluate the path cannot directly yield the solution of this invention, because thermal simulation is merely an evaluation tool and does not necessarily provide a direction-sensitive optimization structure that is heritable, crossover-mutable, and rapidly evaluated in iterations. This invention organizes path planning, endpoint direction, thermal history inheritance, and a genetic search mechanism into a unified model, forming a technical route different from existing geometric path optimization and post-processing thermal verification.

[0330] Furthermore, even though existing technologies have disclosed thermal uniformity path planning, path-level thermal history evaluation, or WAAM state feedback control, the complete solution of this invention cannot be directly obtained. This is because the aforementioned technologies typically handle path generation, thermal evaluation, or equipment feedback separately, while this invention uses weld bead direction endpoint mapping as a prerequisite variable for thermal state updates, integrating it throughout cooling interval calculation, thermal inheritance calculation, fitness evaluation, and direction variation selection. This interconnected relationship creates a continuous computational dependency between the various technical features, rather than simply listing known technologies side-by-side.

[0331]

[0332] Therefore, this invention differs substantially from traditional serpentine scanning, reciprocating scanning, and genetic path optimization methods that only target path length in at least four aspects. First, the optimization variable is expanded from a single weld bead sequence to a directional deposition scheme determined by both sequence and direction, capable of expressing the thermal differences caused by different start and end points of the same weld bead. Second, thermal evaluation is transformed from static adjacency or post-simulation verification to direction-sensitive thermal history inheritance calculation embedded in genetic iteration. Third, the fitness objective is expanded from space length or geometric continuity to a comprehensive evaluation of temperature uniformity, temperature gradient, residual stress risk, space distance, and process constraints. Fourth, the mutation mechanism is transformed from indiscriminate random perturbation to directional exploration guided by changes in thermal inheritance. Traditional methods mainly answer the question "In which order should the weld beads be deposited to achieve the shortest possible length?" This invention further answers the question "From which end should the weld bead begin deposition to achieve a more uniform thermal state, a lower temperature gradient, and a lower residual stress risk?"

[0333] This sequence-direction-thermal history coupled optimization method can regulate the thermal history distribution by reorganizing the in-layer weld bead execution sequence and deposition direction without changing the basic process parameters of WAAM deposition equipment, welding current, voltage, wire feed speed, etc. This method can reduce heat accumulation caused by continuous heating in local areas, mitigate excessive temperature gradients between adjacent weld beads, reduce the risk of residual stress concentration caused by uneven thermal strain, and improve weld bead overlap stability, in-layer forming uniformity, and component dimensional accuracy. The effectiveness of this technique stems from the synergistic effect of the encoding method, thermal model, fitness function, and genetic operators, rather than a conventional replacement or simple application of known genetic algorithms. Therefore, it more fully demonstrates the creative contribution of this invention to in-layer path optimization in WAAM.

[0334] In terms of equivalent implementation methods, the heat-affected zone adjacency coefficient between weld beads is not limited to a single distance attenuation form, but can also be determined based on the weld bead topological adjacency relationship, weld bead center distance, heat source influence radius, local heat dissipation conditions, or material thermal diffusivity. The effective cooling interval is not limited to the ratio of the idle distance to the welding torch transition speed, but can also further consider the arc ignition waiting time, arc termination dwell time, local preheating state within the layer, or shielding gas cooling conditions.

[0335] The path-level thermal state prediction model is not limited to physically constrained neural networks; it can also employ regression models, response surface models, simplified finite element models, lookup table models, or semi-empirical models modified from experimental data. As long as the model can receive the initial thermal state obtained from the directional endpoints and thermal history inheritance, and output indices such as intralayer temperature dispersion, maximum thermal gradient, or residual stress risk, the technical objective of this invention can be achieved.

[0336] The orientation variation probability is not limited to a specific function form. It can be set as a function that monotonically increases with the change in thermal history inheritance, the change in the initial temperature of the next weld pass, the change in the local thermal gradient, or the change in the residual stress risk.

[0337] The above equivalent embodiments all retain the core chain of the present invention, namely, the directional endpoint mapping affects the effective cooling interval, the effective cooling interval affects the thermal history inheritance amount, and the thermal history inheritance amount further affects the path-level thermal state evaluation and the genetic search direction.

[0338] Example 2:

[0339] This embodiment provides a directed weld bead collaborative planning system within a WAAM layer based on thermal state awareness, including:

[0340] The layer section discretization modeling module obtains the single-layer two-dimensional section profile of the target component, discretizes the section according to the set weld width and overlap rate, generates several weld segments to be deposited, and extracts the geometric and topological information of the weld.

[0341] The chromosome construction module constructs a hybrid chromosome structure containing sequential coding segments and directional coding segments. Based on the directional coding, it determines the actual arc start point and arc end point of each weld bead, and determines the effective cooling interval based on the endpoint transfer distance and transition time of adjacent deposited weld beads. It establishes a thermal history inheritance model between weld beads, so that the deposition direction has a coupled influence on the initial thermal state of subsequent weld beads.

[0342] The initial population generation module uses a random path and heuristic scanning strategy to generate the initial population;

[0343] The path scheme evaluation module determines the initial thermal state of each weld bead based on the thermal history inheritance model, and inputs the initial thermal state, weld bead geometry, transition motion and process parameter characteristics into the thermo-mechanical field proxy model to predict the temperature field distribution, thermal gradient and residual stress index of different path schemes. After dimensionless processing, a multi-objective evaluation function is constructed for comprehensive evaluation.

[0344] The constraint determination module introduces a dynamic penalty function to determine the constraints on the continuity of WAAM welding torch movement, deposition feasibility, and interference risk between adjacent weld beads.

[0345] The genetic evolution search module introduces a dynamic penalty term during the genetic evolution process and performs iterative optimization through selection, crossover, and mutation operations within the genetic algorithm framework.

[0346] The optimal path output module outputs the optimal weld bead deposition sequence and direction, and generates WAAM welding torch deposition path and process parameter instructions.

Claims

1. A directed weld bead collaborative planning method within a WAAM layer based on thermal state awareness, characterized in that, Includes the following steps: S1: Obtain the single-layer two-dimensional cross-sectional profile of the target component, discretize the cross-section according to the set weld width and overlap rate, generate several weld segments to be deposited, and extract the weld geometry and topology information. S2: Based on the geometric and topological information of the weld bead, a hybrid chromosome structure containing sequential coding segments and directional coding segments is constructed. The actual arc start point and arc end point of each weld bead are determined according to the directional coding. The effective cooling interval is determined according to the endpoint transfer distance and transition time of adjacent deposited weld beads. A thermal history inheritance model between weld beads is established so that the deposition direction has a coupled influence on the initial thermal state of subsequent weld beads. S3: Based on the hybrid chromosome structure, an initial population is generated using a random path and heuristic scanning strategy; S4: For the initial population, the initial thermal state of each weld in the candidate path is determined according to the thermal history inheritance model. The initial thermal state, weld geometry, transition motion and process parameter characteristics are input into the thermo-mechanical field proxy model to predict the temperature field distribution, thermal gradient and residual stress index of different path schemes. After dimensionless processing, a multi-objective evaluation function is constructed for comprehensive evaluation. S5: Under the genetic algorithm framework, iterative optimization is performed through selection, crossover and mutation operations, and the dynamic penalty coefficient is updated according to the current generation to constrain and determine the continuity of WAAM welding torch movement, deposition executability and the risk of interference between adjacent weld beads. S6: Outputs the optimal weld bead deposition sequence and direction, and generates WAAM welding torch deposition path and process parameter instructions.

2. The directed weld bead collaborative planning method within the WAAM layer based on thermal state awareness as described in claim 1, characterized in that, Step S1 includes: A1: Read the CAD model file of the target component, which contains complete geometric information and 3D coordinate data of the component; the system slices the model at equal intervals along the Z-axis according to a preset layer thickness, generating a set of 2D cross-sectional contours for each slice. ,in , This represents the total number of floors. A2: For each two-dimensional cross-section, the system determines the weld width based on the set weld width. and overlap rate The cross-sectional region is discretized to generate a set of weld bead segments to be deposited: ； Where N is the total number of weld beads generated in this layer section; each weld bead segment includes: weld bead centerline coordinates weld length Adjacency topology set Record the spatial adjacency information between the current weld bead and adjacent weld beads; A3: After the weld bead is discretized, the system further establishes the weld bead heat-affected adjacency matrix. ,in: ； in, As the topological adjacency indicator factor, if the weld bead With weld Spatial adjacency, then ,otherwise ; Indicates weld bead With weld Euclidean distance between centers; matrix It is used to quantify the heat-affected zone intensity between weld passes and to provide a data basis for subsequent thermal history inheritance calculations; The characteristic length of the heat effect is used to characterize the rate at which the heat effect decays as the weld bead spacing increases.

3. The directed weld bead collaborative planning method within the WAAM layer based on thermal state awareness according to claim 2, characterized in that, Step S2 includes: B1: Constructing a hybrid chromosome; In the genetic algorithm, a hybrid chromosome of length 2N is generated for each weld bead layer: ； Among them: sequential coding segment Used to indicate the weld bead deposition access order, each Corresponding to the weld bead number of the k-th deposition execution; direction coding segment Used to indicate the direction of weld bead deposition; B2: Sequence and direction binding mapping; In actual implementation, the k-th weld bead deposition was numbered as follows: Its corresponding direction is determined by the direction coding segment: ； Actual arc starting point of weld With the end of the arc The calculation formula is: ； in, These are the geometric start and end coordinates of the weld bead, respectively. B3: Establish a heat history inheritance model between weld passes; Thermal history inheritance of weld bead k Defined as: ； Directional coupling factor: ； Initial temperature of weld bead k: ； in, For the first weld bead and the first The heat-affected zone coefficient between weld beads; For the first The second deposition ended until the [number]th [deposition]. The cumulative time interval between the start of each deposition includes all transfer, waiting, and intermediate weld bead deposition time between two depositions; The thermal decay coefficient; For the first Unit vector in the direction of the weld bead; This represents the actual arc termination point of the j-th deposition weld bead. This represents the actual arc initiation point of the k-th deposition weld bead; It is a very small positive number; This is the thermal inheritance gain coefficient; Ambient temperature; This is the directional coupling weighting coefficient, used to adjust the influence of weld bead deposition direction on the amount of thermal history inherited.

4. The directed weld bead collaborative planning method within the WAAM layer based on thermal state awareness according to claim 3, characterized in that, Step S3 includes: C1: Random path generation; Based on the constructed hybrid chromosome structure Several individuals are randomly generated as the initial population. The sequential coding segment P of each individual is randomly arranged to form the weld bead number, and the direction coding segment D is randomly assigned a value of 0 or 1. The generated individuals ensure that each weld bead number is unique, and the direction coding corresponds to the sequential coding. C2: Heuristic path generation; Reference paths are generated using heuristic rules, including: reciprocating scanning path: scanning sequentially along the cross-sectional contour to generate weld bead sequence in turn; thermal equilibrium priority path: calculating the heat accumulation value based on the thermal influence matrix A and prioritizing the weld bead sequence with lower heat accumulation; each individual generated by the heuristic path contains a sequence code P and a direction code D; C3: Initial population composition; The initial population is composed of individuals using random paths and individuals using heuristic paths, with a total population size of M. 。 5. The directed weld bead collaborative planning method within the WAAM layer based on thermal state awareness according to claim 4, characterized in that, The prediction method for the thermo-mechanical field surrogate model in step S4 includes: D1: Input vector construction; For each chromosome individual Construct the input vector for the thermal-mechanical field prediction model: ； Where L is the weld length vector; A is the heat-affected adjacency matrix; This represents the initial thermal state vector for each weld bead. The characteristic vector of inter-bead transition motion and process parameters includes at least one or more of the following: endpoint transfer distance, effective cooling interval, deposition rate, arc power, or equivalent heat input; D2: Model prediction; By calling machine learning or physically constrained neural network models to make predictions, thermal-dynamic field indices can be obtained. ； in, This is an indicator of temperature uniformity. The maximum thermal gradient index; This refers to the residual stress index; D3: Initial thermal state update; Based on thermal history inheritance Calculate the initial temperature of the k-th weld bead: 。 6. The directed weld bead collaborative planning method within the WAAM layer based on thermal state awareness according to claim 5, characterized in that, In step D2, the physical constraint neural network model satisfies the heat conduction residual constraint: ; in, To predict the temperature field for the model, As an equivalent heat source input, , , These are the material density, specific heat capacity, and thermal conductivity, respectively. It is a gradient operator used to characterize the rate of change of the temperature field in the spatial direction.

7. The directed weld bead collaborative planning method within the WAAM layer based on thermal state awareness according to claim 6, characterized in that, The multi-objective fitness function in step S4 is expressed as follows: ； in, , , The weighting coefficients for temperature uniformity, maximum thermal gradient, and residual stress are respectively, satisfying: ； This is a function for evaluating temperature uniformity. This is the thermal gradient evaluation function. This is the residual stress evaluation function.

8. The directed weld bead collaborative planning method within the WAAM layer based on thermal state awareness according to claim 7, characterized in that, The constraint determination process in step S5 includes: E1: WAAM welding torch movement continuity and deposition feasibility determination; Determine the target pose of the welding torch corresponding to the k-th weld bead. Is it reachable? ； in, Let the target pose of the welding torch be the position of the welding torch corresponding to the k-th weld pass. For the attitude variables of the WAAM actuator, For WAAM actuator motion mapping, For the device to execute motion domain; E2: Welding torch collision detection; Calculate the minimum safe distance: ； in, Let be the geometric envelope set of the welding torch and end effector during the k-th deposition. It is a collection of obstacles consisting of the workpiece, the fixture, and the deposited area; like If so, the path scheme is determined to have a collision risk; It is the safe distance threshold; E3: Dynamic penalty value calculation; For the i-th candidate individual, define the comprehensive constraint violation quantity. : ； in, These are the non-negative constraint weight coefficients. This is a non-negative process constraint violation quantity; Based on the comprehensive constraint violation amount With the dynamic penalty coefficient of the g-th generation Calculate the dynamic penalty value for each individual. : ； ； in, Let g be the maximum number of generations, and g be the current number of generations. ,and ; i is the candidate individual index, distinguished from the evolutionary generation g.

9. The directed weld bead collaborative planning method within the WAAM layer based on thermal state awareness according to claim 8, characterized in that, The execution of the selection, crossover, and mutation operations in step S5 includes: Select operation: For each individual in the current population Calculate its comprehensive objective function value: ； in Let be the multi-objective evaluation function value for the i-th candidate individual. is the dynamic penalty value for an individual in the g-th generation; The selection operation is performed according to the comprehensive objective function value from smallest to largest; Cross operation: The interleaving operation is performed on the sequential coded segment P and the directional coded segment D respectively: Sequential encoding segment partial matching crossover: randomly select two parent individuals , Randomly determine the intersection interval [u,v]; retain the sequential code of parent generation 1 in interval [u,v], and fill the other positions sequentially according to the remaining weld order of parent generation 2; generate the child generation sequential code. ; Single-point intersection of directional coding segments: for parent directional coding segments Randomly select a single intersection position q; offspring direction encoding : ； After the crossover operation is completed, a child set is generated. ; Mutation operation: Mutation operations are performed separately for sequential encoding and directional encoding: Sequential code swap mutation: Randomly select two positions r and s in the sequential code segment and swap the weld bead numbers. ； Directional coding bit flip mutation: Perform bit flipping on the i-th bit of the directional coding segment: ； Based on the heat-sensitive region of the weld, the probability of orientation variation is weighted: ； in, The absolute value of the change in the heat history inheritance of the weld bead after fixing the weld bead sequence and other directional positions and only flipping the direction of the i-th weld bead; Based on the probability of mutation, ; The thermal sensitivity enhancement coefficient, Used to avoid a denominator of zero.

10. A directed weld bead collaborative planning system within a WAAM layer based on thermal state perception, characterized in that, The system for implementing the method according to any one of claims 1 to 9 comprises: The layer section discretization modeling module obtains the single-layer two-dimensional section profile of the target component, discretizes the section according to the set weld width and overlap rate, generates several weld segments to be deposited, and extracts the geometric and topological information of the weld. The chromosome construction module constructs a hybrid chromosome structure containing sequential coding segments and directional coding segments. Based on the directional coding, it determines the actual arc start point and arc end point of each weld bead, and determines the effective cooling interval based on the endpoint transfer distance and transition time of adjacent deposited weld beads. It establishes a thermal history inheritance model between weld beads, so that the deposition direction has a coupled influence on the initial thermal state of subsequent weld beads. The initial population generation module uses a random path and heuristic scanning strategy to generate the initial population; The path scheme evaluation module determines the initial thermal state of each weld bead based on the thermal history inheritance model, and inputs the initial thermal state, weld bead geometry, transition motion and process parameter characteristics into the thermo-mechanical field proxy model to predict the temperature field distribution, thermal gradient and residual stress index of different path schemes. After dimensionless processing, a multi-objective evaluation function is constructed for comprehensive evaluation. The constraint determination module introduces a dynamic penalty function to determine the constraints on the continuity of WAAM welding torch movement, deposition feasibility, and interference risk between adjacent weld beads. The genetic evolution search module introduces a dynamic penalty term during the genetic evolution process and performs iterative optimization through selection, crossover, and mutation operations within the genetic algorithm framework. The optimal path output module outputs the optimal weld bead deposition sequence and direction, and generates WAAM welding torch deposition path and process parameter instructions.

Images