Additive manufacturing path planning algorithm based on dual laser heads

Through the global segmentation and parallel line scanning method combined with improved genetic algorithm, the multi-travel merchant problem solution is solved, and the printhead emptying and lifting problems in complex model path planning in additive manufacturing is achieved, achieving uniformity of dual laser head path planning Distribution and efficiency improvement.

CN113487101BActive Publication Date: 2024-12-06DALIAN UNIV OF TECH
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202110838963.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-07-23
Publication Date
2024-12-06
Estimated Expiration
2041-07-23

AI Technical Summary

Technical Problem

The existing additive manufacturing path planning algorithms are prone to problems such as large emptying distances, many take-offs and landings, and sharp corner printing accumulation when processing complex models. Especially when using printers with multiple laser heads, printing speed and efficiency are difficult. improve.

Method used

The polygon convex decomposition algorithm based on global segmentation is used to partition the two-dimensional contour after the three-dimensional model slices, and select parallel line scanning method inside each subpartition to fill it, which is converted into a multi-travel provider problem to connect the subpartition paths, and use the improved Genetic algorithm RLGA solves the problem of multi-travel merchants and realizes path planning with dual laser heads.

Benefits of technology

It effectively reduces the number of lifting and landing times of the print head and the empty travel, avoids sharp corner accumulation problems, improves printing efficiency and quality, and shortens printing time.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113487101B_ABST
    Figure CN113487101B_ABST
Patent Text Reader

Abstract

The present invention belongs to the field of artificial intelligence and 3D printing, and relates to an additive manufacturing path planning algorithm based on dual laser heads. First, the two-dimensional contour obtained after slicing the three-dimensional model is partitioned using a polygon convex decomposition algorithm based on global subdivision, and then a parallel line scanning method is selected inside each sub-partition for filling, and finally the original traveling salesman problem is regarded as a multi-traveling salesman problem to connect the sub-partition paths to implement the partition connection algorithm based on dual laser heads, and the simulation result of the path planning is obtained, and finally the printing process is completed. This method can effectively distribute the printing path to the two print heads more evenly, and the partition algorithm based on polygon convex decomposition can effectively reduce the number of print head rises and falls and the empty travel of the print head, which can effectively reduce the printing time and achieve the purpose of improving printing efficiency and printing quality.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of artificial intelligence and 3D printing, and designs an additive manufacturing path planning algorithm based on dual laser heads. Background Art

[0002] The main process of additive manufacturing is to slice the part model according to a certain thickness, and then import the path planning of each layer of the slice into the printer for printing, and finally form it into the required part. One of the most important links in the printing process is the planning and design of the printing path. The quality of the printing path determines the quality of the printing and also affects the printing speed and the amount of printing consumables. Common path planning methods now include parallel scanning method, contour offset method, spiral scanning method, etc. For the path planning algorithm for additive manufacturing of large metal parts, the parallel scanning method is commonly used. This method is relatively mature, simple and easy to use, has a fast printing speed, can save printing time, and is conducive to the next step of path connection.

[0003] For more complex model contours, using traditional path filling algorithms will result in problems such as large print head empty distance, frequent print head rise and fall times, and accumulation of sharp corner prints. To solve these problems, the partitioning method can be used to first decompose the complex contour graphics into multiple simple graphics, and then fill the path of each small partition to form the final path planning.

[0004] In recent years, the widespread application of additive manufacturing technology has made the manufacturing of printers diversified, and many different types of printers have emerged for different industries and research and manufacturing. In order to improve printing efficiency and printing quality, multi-laser head printers and corresponding technologies have emerged. Since there are still not many studies in this area, the path planning technology for multi-laser head additive manufacturing is not mature. In the paper "Design of a Strong Mixed Composite Material Twin Screw Double Nozzle 3D Printer", a printer for strongly mixed composite materials was designed by the Manufacturing Technology Laboratory of Qingdao University of Science and Technology. This printer will fix the two nozzles and print by moving the printing platform along the three-dimensional coordinate axis. The printer head of the main printer uses a double screw mechanism for feeding to improve the uniformity of the material. The function of the auxiliary machine's print head is to print support materials during the printing of complex structures. This printer does not work in a simultaneous manner with two print nozzles. To a certain extent, it can improve the stability of the whole machine and improve the printing accuracy, but it cannot effectively improve the printing speed. Summary of the invention

[0005] In order to solve the above problems, the present invention proposes an additive manufacturing path planning algorithm based on dual laser heads.

[0006] The technical solution of the present invention:

[0007] An additive manufacturing path planning algorithm based on dual laser heads is proposed. First, the two-dimensional contour obtained after slicing the three-dimensional model is partitioned using a polygon convex decomposition algorithm based on global subdivision. Then, the parallel line scanning method is selected to fill in each sub-partition. Finally, the original traveling salesman problem is regarded as a multi-traveling salesman problem to connect the sub-partition paths to implement the partition connection algorithm based on dual laser heads, and the simulation results of path planning are obtained, and finally the printing process is completed.

[0008] The specific steps are as follows:

[0009] The first step is preprocessing and model contour partitioning

[0010] The model to be printed is processed by 3D software to form an STL file recognized by the printer for slicing. The sliced ​​2D contour sequence point set is stored as an Excel file. The vertices of the polygon contour are arranged in a counterclockwise direction. The complex concave polygon is divided into simple convex polygons based on the global partitioning idea in the contour point set. Concave points should be eliminated by adding partitioning lines.

[0011] The second step is path planning calculation based on dual laser heads

[0012] (1) Sub-partition path planning algorithm

[0013] By comparing different path filling algorithms, we choose to use the parallel line scanning path planning algorithm to fill the path of each small partition. Among them, for the path filling of each small partition, we adaptively choose to perform parallel scanning filling along a certain initial direction. Compared with parallel scanning along a certain edge of the polygon, it can avoid problems such as material accumulation and incomplete filling at the sharp corners.

[0014] (2) Subpartition Path Connection Algorithm

[0015] After partitioning, the entire printing path planning problem can be transformed into a traversal problem between sub-partitions, and the traversal problem between sub-partitions can be transformed into a salesman problem. For the printing path partition connection of dual laser heads, the original traveling salesman problem can be regarded as a multi-traveling salesman problem.

[0016] The original traveling salesman problem is that a traveling salesman traverses all cities to find the shortest path, while the multi-traveling salesman problem is that multiple traveling salesmen traverse all current cities at the same time to obtain the shortest traversal path, which is more complicated than the traveling salesman problem. The present invention uses an optimized genetic algorithm RLGA to solve the multi-traveling salesman problem. The other parts of the path planning are still calculated according to the partitioning algorithm and sub-partition filling algorithm in the path planning algorithm based on a single laser head in the previous chapter.

[0017] Beneficial effects of the present invention:

[0018] The method of the present invention can solve the path planning of additive manufacturing based on dual laser heads, and can effectively distribute the printing path to the two print heads relatively evenly. At the same time, the partitioning algorithm based on polygon convex decomposition can effectively reduce the number of print head rises and falls and the empty travel of the print head. In the process of filling the sub-partitions, the parallel scanning method is selected to fill, which can effectively reduce the problem of accumulation of printing sharp corners. The path planning of dual laser head printing through the above three steps can effectively reduce the printing time and achieve the purpose of improving printing efficiency and printing quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is the overall flow chart of the present invention;

[0020] Figure 2 is the concave point p i Schematic diagram of the dissection;

[0021] Figure 3 Plan the graph for the sub-partition parallel line scanning method;

[0022] Figure 4 Flowchart for solving the multi-business travel problem using the improved genetic algorithm;

[0023] Figure 5 The double laser head prints the planning result map for the square plane contour;

[0024] Figure 6(a) and Figure 6(b) are the results of the five-pointed star plane contour planning, where Figure 6(a) is the planning result of the dual laser heads, and Figure 6(b) is the planning result of the single laser head;

[0025] Figure 7 The result of dual laser head path planning for a circular plane contour with a hole;

[0026] Figure 8 This is the result of dual laser head path planning for a square plane contour with multiple holes;

[0027] Figures 9(a) and 9(b) are the path planning results of complex contours, where Figure 9(a) is the planning result of dual laser heads and Figure 9(b) is the planning result of a single laser head. DETAILED DESCRIPTION

[0028] The specific implementation of the present invention is further described below in conjunction with the accompanying drawings and technical solutions.

[0029] The flow chart of the path planning of the present invention is as follows Figure 1 As shown, the specific steps are as follows:

[0030] Step 1: Preprocessing and calculation of model partitions

[0031] According to the model that needs to be printed by additive manufacturing, a 3D simulation model is established and sliced. For each two-dimensional slice surface, its contour is output as a sequential point set and stored in an Excel file. Taking one layer as an example, the model partitioning process is as follows:

[0032] The first step is to traverse the polygon vertices of the two-dimensional slice surface contour and store them in the point set to establish a concave point string. Take the vertices from the polygon point set one by one, determine whether they are concave points, and if they are concave points, store them in the concave point string, and record the predecessor and successor of the point and mark them as unvisited. If there are no concave points after the traversal, the polygon is a convex polygon, and the path planning process of Step 2 is directly carried out; otherwise, continue to partition the polygon.

[0033] In the second step, if the lines between two vertices in a concave polygon can all fall inside the polygon, then the two vertices are called visible to each other. Traverse the unvisited concave points in the concave point string one by one, use the weighted method to guide the polygon partitioning process, and establish a weight function w for each set of visible point pairs. i As shown in formula (1), each time we connect the two visible point pairs with the largest weight, so that the number of convex polygons is minimized. Figure 2 As shown, p i is a concave point of the polygon, p i-1 、p i+1 are the neighboring points of this point. Assume that p j For p i At the other end of the starting splitting line, α and β are vectors p i p j With vector p i p i-1 , p i p i+1 Angle.

[0034] The third step is to traverse the concave point string and use the weight function of formula (1) to find the optimal segmentation point, weight and segmentation type from each concave point and save them. When the segmentation point is a concave point, the segmentation type value is 1, when it is a convex point, the segmentation type value is 2, and when it is an auxiliary point, the segmentation type value is 3.

[0035] w i =f(α,β)=|cosα-cosβ| (1)

[0036] The fourth step is to traverse the concave point string and find the concave point with the smallest weight. The search is based on the following rules: the segmentation type value is 1, which is the best segmentation; if there is no segmentation type value, find a convex point with a segmentation type value of 2; finally, find auxiliary points for segmentation.

[0037] The fifth step is to decompose the original polygon into two new polygons according to the concave points and the subdivision points, and add them to the polygon storage list. At the same time, delete the original polygon from the polygon list, and recursively perform concave polygon convex decomposition on the newly generated polygons in turn until there are no concave points.

[0038] Step 2: Path planning based on dual laser heads

[0039] (1) Sub-partition path planning

[0040] By comparing different additive manufacturing path planning algorithms, we choose to use the parallel line scanning path planning method to plan the path for each sub-partition. For each sub-partition, we adaptively select parallel scanning along a certain initial direction. Compared with parallel scanning along a certain edge of the polygon, it can avoid problems such as material accumulation and incomplete filling at the sharp corners. The filling results of the sub-partition path planning are shown in the figure below. Figure 3 shown.

[0041] (2) Connection of sub-partition paths

[0042] After the slice surface is segmented into regions through Step 1, the path planning for the sub-partitions is to print each closed sub-region one by one, and the problem of solving the connection order of the sub-partitions is regarded as the classic TSP multi-traveling salesman problem, that is, the laser head is regarded as a traveling salesman in the multi-traveling salesman problem, and each sub-partition is regarded as a city node in the multi-traveling salesman problem for solution. At the same time, the tasks should be evenly distributed to each traveling salesman (laser head). The multi-traveling salesman problem with M paths for N points can be divided into the following two cases: one, evenly distribute the number of city nodes visited by m traveling salesmen; two, evenly distribute the length of the paths visited by m traveling salesmen. Both cases have their own solutions and objective functions. For the present invention based on the use of the multi-traveling salesman problem in path planning, the second method of evenly distributing the length of the access path should be selected to solve the problem of printing path planning, so that the two printing laser heads can be fully utilized to print at the same time, thereby effectively shortening the printing time and improving the printing efficiency.

[0043] The specific process is as follows:

[0044] Point a 0 represents the starting point of the laser head, A={a 0 ,a 1 ,…,a n-1 ,a n} represents m traveling salesmen b 1 ,b 2 ,…,b m The sub-partition to be accessed. Define two variables as shown in the following formulas (2) and (3):

[0045]

[0046]

[0047] c ij Represents subpartition a i and subpartition a j The objective function of the mathematical model of the multi-traveling salesman problem for path planning with evenly distributed workload is shown in formula (4):

[0048] Z = min(max(z 1 ,z 2 ,…,z m )) (4)

[0049] in:

[0050]

[0051] The constraints are as follows:

[0052]

[0053]

[0054]

[0055]

[0056] Where S is the branch elimination constraint, that is, to eliminate some solutions that constitute incomplete routes. Formula (4) represents the minimization of the longest path among m laser heads; Formula (5) represents the path length of each laser head; Formula (6) represents the 0 , all sub-partitions have been visited by a laser head and only visited once; Formula (7) means that the end sub-partition of any path has one and only one starting city connected to it; Formula (8) means that the starting sub-partition of any path has one and only one ending city connected to it; Formula (9) means eliminating the solution set that constitutes an incomplete route.

[0057] For the multi-traveling salesman problem of evenly sharing the workload, the present invention adopts an improved genetic algorithm (RLGA) to solve the problem, that is, to find the solution of formula (4) and formula (5), so as to obtain a set of path sequences that traverse all sub-partitions, and connect the sub-partitions in Step 1 in sequence to obtain the final path planning. The flowchart of the process is shown in Figure 4As shown. The improved genetic algorithm is proposed based on the defects of the genetic algorithm and the advantages of the invasive weed optimization algorithm. The breeding mechanism based on fitness in the invasive weed optimization algorithm is added to the genetic algorithm to produce offspring and perform genetic operations, which improves the adaptability of the population and thus improves the efficiency of the algorithm. At the same time, a mixed local optimization strategy is proposed as a mutation operator to exchange all traveling salesmen between groups and adopt a two-element optimization (2-opt) exchange strategy for a single traveling salesman, so as to improve the local search ability of the algorithm and improve the solution accuracy of the algorithm.

[0058] Step 3: Output planning G-code

[0059] Output the final path planning data as G-code. G-code can be recognized by 3D printers and is generally stored as a G-code file. The G-code file contains relevant information such as printing speed, idle stroke speed, printing size, and coordinates of each printed point. Outputting G-code makes it easier to view the results of the planned path and use the planning results.

[0060] In order to verify the feasibility of the dual laser head path planning algorithm proposed in this invention, several groups of part slice 2D contour graphics were selected for experiments. First, a simplest square plane contour was selected for path planning. The results are shown in the figure below. Figure 5 As shown. The entire path planning is divided into two parts. The areas marked ① and ② represent the areas printed by laser heads 1 and 2 respectively. For simple graphics that do not require partitioning, choose to use the method of equalizing the printing path to divide the entire printing path length equally. S represents the printing start point and E represents the printing end point. As can be seen from the figure, the two print heads have different printing start positions, and their respective printing paths will not cross and collide, and the entire path will be printed in the end.

[0061] A five-pointed star cross-section figure contour is selected for path planning, as shown in Figure 6(a). For the five-pointed star figure, there is still a situation where the print head travels empty during the intermediate printing process, where the dotted line e is the empty travel of print head 1, and the dotted line f is the empty travel of print head 2. Compared with the printing path planning of a single laser head, the distance of the empty travel (as shown in the dotted path in the path of Figure 6(b)) shows that the empty travel is shorter in the path planning based on the dual laser head. The two use the same partitioning algorithm, so the partition scale is the same. The partition connection planning is regarded as a multi-traveling salesman problem with equal path division, and the path length of the path connection distance (empty travel) is relatively evenly distributed.

[0062] In addition to the two relatively regular graphics above, dual laser head path planning can also be performed on complex graphics with holes. Select a copper coin-shaped part cross-section, with a small square in the middle of the circle. The path planning result is as follows: Figure 7 A square with three holes was selected for path planning. The result is as follows Figure 8 By comparison Figure 7 and Figure 8 It can be seen that the two graphics based on dual laser head path planning have no empty travel distance (empty stroke) of the print head, and no print head lifting times, which can reduce the drawing of printing materials and the operation time caused by the lifting and lowering of the print head. It can make full use of the two print heads to improve printing efficiency and print quality.

[0063] In order to further verify the feasibility of the algorithm, a more complex fan-shaped sawtooth part was selected for dual laser head path planning. The experimental results are shown in Figure 9(a). According to the rule of path equalization, different sub-partitions are assigned to two print heads for printing. Print head 1 is assigned four partitions and print head 2 is assigned six partitions. The printing path lengths of the two laser heads are almost equal, which can effectively reduce the printing time. Compared with the path planning result of a single laser head (as shown in Figure 9(b)), the number of print head rises and falls is reduced. In the dual laser head path planning, the two print heads need to be lifted a total of seven times, print head 1 is lifted three times, print head 2 is lifted four times, and the two laser heads print at the same time, which is equivalent to consuming four print head lift times. In the single laser head printing path planning, the print head needs to be lifted 8 times, and a total of 8 print head rises and falls are consumed. It can be seen that the use of dual laser head path planning can effectively reduce printing time and improve printing efficiency. The experimental data related to Figure 9 are shown in Table 1.

[0064] Table 1 Results of fan-shaped serrated parts

[0065]

[0066] Through the above experiments, it can be seen that the present invention can effectively solve the path planning problem based on the dual laser head 3D printer, distribute the printing path more evenly to the two print heads, and the partitioning algorithm based on polygon convex decomposition can effectively reduce the number of print head rises and falls and the print head empty strokes. In the sub-partition filling process, the parallel scanning method is selected to fill, which can effectively reduce the problem of accumulation of printing corners. The above three steps are used to carry out the path planning of dual laser head additive manufacturing, which can effectively reduce the printing time and achieve the purpose of improving printing efficiency and printing quality.

Claims

1. The additive manufacturing path planning algorithm based on dual laser heads is characterized by: The specific steps are as follows: Step 1: Preprocessing and calculation of model partitions According to the model to be printed by additive manufacturing, a 3D simulation model is established and sliced; for each two-dimensional slice surface, its contour is output as a sequential point set and stored in an Excel file. Taking one layer as an example, the model partitioning process is as follows: The first step is to traverse the polygon vertices of the two-dimensional slice surface contour and store them in the point set to establish a concave point string; take out the vertices from the polygon point set in turn, determine whether they are concave points, and if they are concave points, store them in the concave point string, and record the predecessor and successor of the point and mark them as unvisited; if there is no concave point after the traversal, the polygon is a convex polygon, and the path planning process of Step 2 is directly carried out; Otherwise continue partitioning the polygon; Step 2: If the lines between two vertices in a concave polygon can all fall inside the polygon, then the two vertices are called visible to each other; Traverse the unvisited concave points in the concave point string one by one, use the weighted method to guide the polygon partitioning process, and establish the weight function w for each set of visible point pairs i As shown in formula (1), each time the two visible point pairs with the largest weight are connected to minimize the number of convex polygons obtained; Among them, p i is a concave point of the polygon, p i-1 、p i+1 is the neighboring point of the point; assuming p j For p i At the other end of the starting splitting line, α and β are vectors p i p j With vector p i p i-1 , p i p i+1 The angle of The third step is to traverse the concave point string and use the weight function of formula (1) to find the optimal segmentation point, weight and segmentation type from each concave point and save them. When the segmentation point is a concave point, the segmentation type value is 1, when it is a convex point, the segmentation type value is 2, and when it is an auxiliary point, the segmentation type value is 3; w i =f(α,β)=|cosα-cosβ| (1) The fourth step is to traverse the concave point string and find the concave point with the smallest weight; search according to the following rules: the segmentation type value is 1, which is the best segmentation; if there is no segmentation type value, find a convex point with a segmentation type value of 2; finally, find auxiliary points for segmentation; Step 5: Decompose the original polygon into two new polygons according to the concave points and the splitting points, and add them to the polygon storage list. At the same time, delete the original polygon from the polygon list, and recursively perform concave polygon convex decomposition on the newly generated polygons in turn until there are no concave points. Step 2: Path planning based on dual laser heads (1) Sub-partition path planning Use parallel line scanning path planning method to plan the path for each sub-partition, and adaptively select parallel scanning along a certain initial direction for each sub-partition; (2) Connection of sub-partition paths After the slice surface is segmented by Step 1, the path planning for the sub-partitions is to print each closed sub-region one by one. The problem of solving the connection order of the sub-partitions is regarded as the classic TSP multi-traveling salesman problem, that is, the laser head is regarded as the traveling salesman of the multi-traveling salesman problem, and each sub-partition is regarded as a city node in the multi-traveling salesman problem for solution; at the same time, the task should be evenly distributed to each traveling salesman, i.e., the laser head. For the multi-traveling salesman problem with M paths for N points, the length of the m traveling salesman access paths needs to be evenly distributed to solve the problem of printing path planning; The specific process is as follows: Point a0 represents the starting point of the laser head, A={a0,a1,…,a n-1 ,a n } represents m traveling salesmen b1, b2, ..., b m The sub-partition to be accessed; define two variables as shown in the following formulas (2) and (3): c ij Represents subpartition a i and subpartition a j The distance between them; the objective function of the mathematical model of the path planning multi-traveling salesman problem with evenly distributed workload is shown in formula (4): Z=min(max(z1,z2,…,z m )) (4) Among them: The constraints are as follows: Among them, S is the branch elimination constraint, that is, eliminating some solution sets that constitute incomplete routes; formula (4) represents the minimization of the longest path among m laser heads; formula (5) represents the length of the path traversed by each laser head; formula (6) indicates that from the starting point a0, all sub-partitions have been visited by a certain laser head and only visited once; formula (7) indicates that the end sub-partition of any path has one and only one starting city connected to it; formula (8) indicates that the starting sub-partition of any path has one and only one ending city connected to it; formula (9) represents the elimination of solution sets that constitute incomplete routes; For the multi-traveling salesman problem of evenly sharing the workload, an improved genetic algorithm RLGA is used to solve it, that is, to find the solutions of formula (4) and formula (5), so as to obtain a set of path sequences that traverse all sub-partitions. The sub-partitions in Step 1 are connected in sequence to obtain the final path planning. Step 3: Output planning G-code The final path planning data is output as G-code code, which facilitates the viewing and use of the planning results.

Citation Information

Patent Citations

  • Partition-based 3D printing filling path generation method

    CN104325638A

  • Intelligent 3D printing path planning method based on genetic algorithm

    CN110209363A