A state transition-based process planning method for complex sheet metal parts
By using state transfer matrix calculation and reverse planning, the problems of few feasible solutions and low efficiency in the process planning of complex sheet metal parts are solved, and the optimal process planning is obtained efficiently.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NANJING UNIV OF AERONAUTICS & ASTRONAUTICS
- Filing Date
- 2022-11-04
- Publication Date
- 2026-05-19
AI Technical Summary
Existing technologies suffer from a lack of feasible solutions, difficulty in planning, and low efficiency in the process planning of complex sheet metal parts. In particular, for highly complex sheet metal parts, existing methods are prone to getting trapped in local optima.
An efficient process planning method based on state transfer is adopted to reverse plan from final forming to blank thin plate. Feasible processes are calculated by using the state transfer matrix, and the optimal process is evaluated by using genetic algorithms.
It effectively obtains all feasible solutions for complex sheet metal parts, avoiding redundant calculations and local optima, thus improving planning efficiency and accuracy.
Smart Images
Figure CN115809519B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to sheet metal technology, particularly sheet metal bending process planning technology, for quickly obtaining the optimal process of sheet metal parts in sheet metal bending processing to achieve high efficiency. Specifically, it is a process planning method for complex sheet metal parts based on state transfer. Background Technology
[0002] Sheet metal bending is a typical cold working process for sheet metal, widely used in home appliances, automobiles, shipbuilding, aerospace, and other fields. Sheet metal bending is typically performed by CNC bending machines. Before processing, the operator needs to determine the bending sequence to avoid problems such as collisions and interference between the workpiece and the die. This manual planning method places high demands on the operator's process experience, especially when dealing with complex sheet metal parts, often requiring repeated planning, trials, and multiple test bends to obtain a relatively ideal processing sequence. With the widespread application of modern CNC technology and CAD / CAM technology, the CNCization of sheet metal bending machines and the rapid popularization of CAD for sheet metal parts have led to the emergence of sheet metal bending programming and simulation (CAM) software. Internationally, there are companies such as Radan from Plaint (UK), MBend from Metalix (Israel), and high-end bending machine control systems from Delem (Netherlands). Although there has been some research in China, there is still no mature commercial sheet metal bending programming software.
[0003] Bending process planning is the core function of a sheet metal bending programming system. Its main goal is to obtain the optimal or near-optimal bending sequence for a given part. The possible process space for a part with N bending operations is... This may involve multiple feasible steps. How to... In this approach, infeasible solutions with collision interference are eliminated, and the optimal solution among the feasible solutions is searched. Researchers have tried numerous methods over the years. Branch and bound search algorithms, genetic algorithms, and backtracking algorithms have all been applied to some extent in the planning of sheet metal bending processes. These methods perform well for sheet metal parts with low to medium complexity. However, highly complex sheet metal parts often have multiple bends, resulting in a very large process solution space. In addition, the complexity of the shape leads to a very low proportion of feasible solutions, making existing methods difficult to search and prone to getting trapped in local optima.
[0004] Therefore, an efficient process planning algorithm is needed to solve the problems of few feasible solutions, difficult planning, and low efficiency in the process planning of complex sheet metal parts. Summary of the Invention
[0005] The purpose of this invention is to address the problems of limited feasible solutions, difficult planning, and low efficiency in the existing complex sheet metal part process planning. It proposes an efficient process planning method based on state transfer, which adopts a reverse planning approach from final forming to blank sheet metal, and improves planning efficiency by leveraging the shape complexity of the sheet metal part itself.
[0006] The technical solution of this invention is:
[0007] A method for planning the operation of complex sheet metal parts based on state transfer, characterized by the following steps:
[0008] Step 1: Based on the number of bends in the sheet metal part create The state transfer matrix;
[0009] Step 2: Calculate and fill the state transfer matrix based on the state transfer mechanism;
[0010] Step 3: Obtain all feasible processes for the sheet metal part from the row represented by the termination state in the matrix;
[0011] Step 4: Evaluate the processing efficiency of all feasible processes and select the optimal process.
[0012] The step one mentioned above is based on the number of bends of the sheet metal parts. create The state transfer matrix is a matrix that... The bend locations are numbered, and the planning process is then divided into states, specifically the major state represented by the number of bends. The small state represented by the number of bends it is processed in. The specific state formed by combination . Each element in the state transfer matrix represents a specific state. , This state reflects the feasibility of bending sheet metal parts during processing, and is reflected in the experience stored within it regarding this state, i.e., at the bending position. Conduct the first Feasible partial process for bending.
[0013] In step two, the state transfer matrix is calculated and filled based on the state transfer mechanism, where the state transfer mechanism allows for unidirectional transfer of experience between large states, while small states isolate each other from each other.
[0014] There are two schemes for unidirectional transfer of experience between large states:
[0015] Option 1: Forward propagation, indicating the first... Second bend The direction can be represented as the first Second bend Passing on experience, that is Each The calculation is from The experience gained is based on the preceding bend, which is the bend position that has already been processed, and the bend position. For the first The bending position of the second bending process is used to generate test units for testing. Bending collision situation during secondary bending process;
[0016] Option 2: Reverse propagation, indicating the first... Second bend The direction can be represented as the first Second bend Passing on experience, that is Each The calculation is from The experience gained is for subsequent bends, i.e., bends that have not yet been processed, based on the bend position. For the first The bending position of the second bending process is used to generate test units for testing. Bending collision situation during secondary bending process.
[0017] The experience of isolation between smaller states refers to Each The calculation from the previous The experience gained should not include bend locations. .
[0018] Step two, which involves calculating the filled state transfer matrix based on the state transfer mechanism, is a calculation process using the reverse state transfer mechanism and includes the following steps:
[0019] Step 1: Initial state, for the current Each bend position The meaning Based on its bending position As a test unit, the test checks for bending collisions during the final bending process. If the test result is no collision, then the test unit is filled with [missing information]. experience;
[0020] Step 2: Non-initial state, for the current The position of each bend in the middle represents , from the previous one Get excluding the current bend position Experience, combined with the current bending position Combined into test units, and after deduplication, the test is performed. If the test result for the bending collision during the second bending process is no collision, then the test unit will be filled with... Experience.
[0021] In the calculation process, step 2, deduplication of combinations, involves classifying test units by combination. If the bending collision detection result of a certain test unit is no collision, then that test unit is filled with... Based on experience, and by marking the combination to which the test unit belongs as a collision-free combination, subsequent test units belonging to this collision-free combination can be directly filled without bending collision detection. Based on experience; if the bending collision detection result of a test unit is a collision, then the test unit is discarded and not filled. Based on experience, and by marking the combination to which the test unit belongs as a collision-prone combination, subsequent test units belonging to that collision-prone combination can be discarded directly without bending collision detection.
[0022] In step three, obtaining all feasible processes for the sheet metal part from the row represented by the termination state in the matrix is from the termination state. Each Gaining experience from the bend position The first feasible partial process, that is, the process feasible solution of the sheet metal part, then all the experience gained is all the feasible processes of the sheet metal part.
[0023] In step four, the processing efficiency of all feasible processes is evaluated, and the optimal process is selected. This involves evaluating factors such as the number of times the process is flipped and the number of molds used, and using genetic algorithms, priority search algorithms, etc., to select the optimal process.
[0024] During the reasoning process from the initial state to the final state, experience is gained through state propagation to create a minimal test set to reduce invalid computations. Combined representations of this experience are used to avoid redundant calculations. Finally, all feasible bending solutions for the sheet metal part can be obtained at the final state. Further, based on processing efficiency, accuracy requirements, etc., the optimal solution for the sheet metal bending process can be obtained.
[0025] The beneficial effects of this invention are:
[0026] (1) This invention introduces the concept of state and the theory of state transfer, which avoids the problem of losing existing experience in other process planning algorithms, where the process is treated as a single planning individual.
[0027] (2) Compared with existing process planning algorithms, the present invention adopts a reverse planning method from final forming to blank sheet. It takes advantage of the fact that the sheet metal parts in the later bending process are more complex than those in the earlier process, and the probability of collision is higher. Early testing and early elimination can be achieved, and experience can be inherited to effectively avoid invalid calculations.
[0028] (3) In the state transfer process, the present invention uses an experience-based combination representation, which avoids the problem of repeated testing and calculation of experience for different permutations of the same combination in the state transfer process;
[0029] (4) The present invention can obtain all feasible solutions for complex sheet metal parts, so that subsequent evaluations such as processing efficiency can be performed to obtain the optimal solution, thus avoiding the local optimal solution problem of existing algorithms. Attached Figure Description
[0030] Figure 1 This is an example of a bent piece with four bends.
[0031] Figure 2 yes Figure 1 The state transfer matrix for the sheet metal part process planning is shown.
[0032] Figure 3 yes Figure 2 The diagram shows some bending collision detection during the state transfer matrix calculation process. Detailed Implementation
[0033] The present invention will be further described below with reference to the accompanying drawings and embodiments.
[0034] like Figure 1-3 As shown.
[0035] A method for planning complex sheet metal parts based on state transfer, using Figure 1 Taking a bent component with four bends as an example, the specific steps include:
[0036] Step 1: Based on the number of bends in the sheet metal part create The state transfer matrix;
[0037] Step 2: Calculate and fill the state transfer matrix based on the state transfer mechanism;
[0038] Step 3: Obtain all feasible processes for the sheet metal part from the row represented by the termination state in the matrix;
[0039] Step 4: Evaluate the processing efficiency of all feasible processes and select the optimal process.
[0040] The step one mentioned above is based on the number of bends of the sheet metal parts. create The state transfer matrix is a matrix that... The bend locations are numbered, and the planning process is then divided into states, specifically the major state represented by the number of bends. The small state represented by the number of bends it is processed in. The specific state formed by combination . Each element in the state transfer matrix represents a specific state. , This state reflects the feasibility of bending sheet metal parts during processing, and is reflected in the experience stored within it regarding this state, i.e., at the bending position. Conduct the first Feasible partial process for bending.
[0041] In step two, the state transfer matrix is calculated and filled based on the state transfer mechanism, where the state transfer mechanism allows for unidirectional transfer of experience between large states, while small states isolate each other from each other.
[0042] There are two schemes for unidirectional transfer of experience between large states:
[0043] Option 1: Forward propagation, indicating the first... Second bend , This can represent the first Second bend Passing on experience, that is Each The calculation is from The experience gained is based on the preceding bend, which is the bend position that has already been processed, and the bend position. For the first The bending position of the second bending process is used to generate test units for testing. Bending collision situation during secondary bending process;
[0044] Option 2: Reverse propagation, indicating the first... Second bend , This can represent the first Second bend Passing on experience, that is Each The calculation is from The experience gained is for subsequent bends, i.e., bends that have not yet been processed, based on the bend position. For the first The bending position of the second bending process is used to generate test units for testing. Bending collision situation during secondary bending process.
[0045] The experience of isolation between smaller states refers to Each The calculation from the previous The experience gained should not include bend locations. .
[0046] Step two, which involves calculating the filled state transfer matrix based on the state transfer mechanism, is a calculation process using the reverse state transfer mechanism and includes the following steps:
[0047] Step 1: Initial state, for the current Each bend position The meaning Based on its bending position As a test unit, the test checks for bending collisions during the final bending process. If the test result is no collision, then the test unit is filled with [missing information]. experience;
[0048] Step 2: Non-initial state, for the current , The position represented by each bend in the text , from the previous one Get excluding the current bend position Experience, combined with the current bending position Combined into test units, and after deduplication, the test is performed. If the test result for the bending collision during the second bending process is no collision, then the test unit will be filled with... experience;
[0049] In the calculation process, step 2, deduplication of combinations, involves classifying test units by combination. If the bending collision detection result of a certain test unit is no collision, then that test unit is filled with... Based on experience, and by marking the combination to which the test unit belongs as a collision-free combination, subsequent test units belonging to this collision-free combination can be directly filled without bending collision detection. Based on experience; if the bending collision detection result of a test unit is a collision, then the test unit is discarded and not filled. Based on experience, and by marking the combination to which the test unit belongs as a collision-prone combination, subsequent test units belonging to that collision-prone combination can be discarded directly without bending collision detection.
[0050] In step three, obtaining all feasible processes for the sheet metal part from the row represented by the termination state in the matrix is from the termination state. Each Gaining experience from the bend position The first feasible partial process, that is, the process feasible solution of the sheet metal part, then all the experience gained is all the feasible processes of the sheet metal part.
[0051] In step four, the processing efficiency of all feasible processes is evaluated, and the optimal process is selected. This involves evaluating factors such as the number of times the process is flipped and the number of molds used, and using genetic algorithms, priority search algorithms, etc., to select the optimal process.
[0052] Taking a sheet metal part with four bends as an example, the specific processing procedure involves marking the four bend locations with serial numbers, such as... Figure 1 As shown, the process planning for this sheet metal part is as follows.
[0053] First, create State transfer matrix, such as Figure 2 As shown, each element in the matrix ,in Each state represents a specific state. After planning is completed, each state will contain corresponding feasible partial processes.
[0054] When using backward reasoning programming It is the initial state, and cannot be changed from the previous one. Inheriting experience can only rely on each Data is used for calculation. For It is the bending position As the processing location for the final bend, its processing feasibility is tested. For example... Figure 3 As shown in (a)(d), in middle and Collisions will occur in all cases; such as Figure 3 As shown in (b)(c). and It can be bent normally; therefore, and No feasible local process, and The feasible partial processes are as follows: and ,like Figure 2 middle As shown. In This indicates that the fourth bend at bend position 2 is safe to process. This bend feasibility data can be used as experience to guide subsequent processes.
[0055] For non-initial states, it is necessary to start from the previous one. China is different from the present of Get the value excluding the current bend position. Experience, combined with the current bending position The experience gained from combining and deduplicating test units, and then passing the test, is transformed into the current state.
[0056] for ,from The test set obtained from it is ,and After combination, they become two test units. .in This indicates that the third bend should be made at bend position 1, and the fourth bend should be made at bend position 2. like Figure 3 (e) shows a collision-free machining process. Then as Figure 3 The machining shown in (f) involves collisions. Therefore, The partially feasible process is ultimately This indicates that the third bend is processed at bend position 1 and the subsequent processes are as follows. It can be bent without collision. (And) The calculation process is similar, and the following can be obtained: The partially feasible process is as follows , The partially feasible process is as follows , The partially feasible process is as follows ,like Figure 2 In As shown.
[0057] for , from The test set obtained from it is ,and After combination, it becomes three test units. .for and The test involved a second bend performed at bend position 1, and the subsequent procedures were as follows: and The bending and collision situation, due to subsequent processes and Both indicate that at the time of testing, bend positions 2 and 3 had not yet been bent, and belong to the same combination. Therefore, these two test units only require one bending collision detection, such as Figure 3 (g) shows no collisions. The discrepancy between the number of test units and the number of bending collision detections is due to empirical combination deduplication. like Figure 3 (h) shows a collision. Therefore, The partially feasible process is ultimately This indicates that the second bend is processed at bend position 1 and the subsequent processes are as follows. or It can be bent without collision. (And) The calculation process is similar, and the following can be obtained: The partially feasible process is as follows , The partially feasible process is as follows , The partially feasible process is as follows ,like Figure 2 In As shown.
[0058] The calculation method is the same as and Just like, in the end Figure 2 In As shown. Among them. All locally feasible processes are the feasible solutions for all bending processes of the sheet metal part, for example. One of the feasible partial processes in This indicates that the first bend is processed at bend position 1 and subsequent processes are as follows. It can be bent without collision, which also indicates a feasible solution for a process of the sheet metal part.
[0059] from Figure 2 middle It can be known Figure 1 The sheet metal part shown has 6 feasible solutions for each process. The optimal process can be selected based on information such as the number of flips and the number of molds used, combined with common search algorithms such as genetic algorithms and priority search algorithms. The relevant content is not within the scope of this patent.
Claims
1. A method for planning the process of complex sheet metal parts based on state transfer, characterized in that: It includes the following steps: Step 1: Based on the number of bends in the sheet metal part create The state transfer matrix; Step 2: Calculate and fill the state transfer matrix based on the state transfer mechanism; Step 3: Obtain all feasible processes for the sheet metal part from the row represented by the termination state in the state transition matrix; Step 4: Evaluate the processing efficiency of all feasible processes and select the optimal process; The step one mentioned above is based on the number of bends in the sheet metal part. create The state transfer matrix is a matrix that... The bend locations are numbered, and the planning process is then divided into states, specifically the major state represented by the number of bends. The small state represented by the number of bends it is processed in. The specific state formed by combination ; Each element in the state transfer matrix represents a specific state. , This state reflects the feasibility of bending sheet metal parts during processing, and is reflected in the experience stored within it regarding this state, that is, at the bending position. Conduct the first Feasible partial processes for secondary bending; The state transfer mechanism used in step two to calculate and fill the state transfer matrix is a one-way transfer of experience between large states and an isolation of experience between small states. There are two schemes for unidirectional transfer of experience between large states: Option 1: Forward propagation, indicating the first... Second bend The direction can be represented as the first Second bend Passing on experience, that is Each The calculation is from The experience gained is based on the preceding bend, which is the bend position that has already been processed, and the bend position. For the first The bending position of the second bending process is used to generate test units for testing. Bending collision situation during secondary bending process; Option 2: Reverse propagation, indicating the first... Second bend To indicate the first Second bend Passing on experience, that is Each The calculation is from The experience gained is for subsequent bends, i.e., bends that have not yet been processed, based on the bend position. For the first The bending position of the second bending process is used to generate test units for testing. Bending collision situation during secondary bending process; The experience of mutual isolation between small states refers to Each The calculation from the previous The experience gained should not include bend locations. ; Step two, which involves calculating the filling state transfer matrix based on the state transfer mechanism, is a calculation process using the reverse state transfer mechanism and includes the following steps: Step 1: Initial state, for the current Each bend position The meaning Based on its bending position As a test unit, the test checks for bending collisions during the final bending process. If the test result is no collision, then the test unit is filled with [missing information]. experience; Step 2: Non-initial state, for the current The position of each bend in the middle represents , from the previous one Get excluding the current bend position Experience, combined with the current bending position Combined into test units, and after deduplication, the test is performed. If the test result for the bending collision during the second bending process is no collision, then the test unit will be filled with... experience; In step three, obtaining all feasible processes for the sheet metal part from the row represented by the termination state in the state transition matrix starts from the termination state. Each Gaining experience from the bend position The first feasible partial process, that is, the process feasible solution of the sheet metal part, then all the experience gained is all the feasible processes of the sheet metal part.
2. The process planning method according to claim 1, characterized in that: The deduplication process in step 2 involves classifying test units by combination. If the bending collision detection result of a certain test unit is no collision, then that test unit is filled with... Based on experience, and by marking the combination to which the test unit belongs as a collision-free combination, subsequent test units belonging to this collision-free combination can be directly filled without bending collision detection. Based on experience; if the bending collision detection result of a test unit is a collision, then the test unit is discarded and not filled. Based on experience, and by marking the combination to which the test unit belongs as a collision-prone combination, subsequent test units belonging to that collision-prone combination can be discarded directly without bending collision detection.