Beam type chip mounter surface mounting process optimization method based on simulated annealing

Abstract

The invention discloses a beam type chip mounter surface mounting process optimization method based on simulated annealing, solves the problem of non-ideal mounting optimization effect of the existing chip mounter mounting optimization method, and belongs to the field of path planning optimization in a printed circuit board (PCB) assembly process. According to the method, firstly, the mounting path optimization problem of the beam-type chip mounter is decomposed into two sub-problems of element distribution and feeder distribution, then a mathematical model is established for the two sub-problems by establishing constraints, a traditional simulated annealing method is improved, a dynamic temperature adjustment mechanism is introduced, the solving capability of an algorithm is improved, and the optimization of the mounting path of the beam-type chip mounter is realized. And then a convex hull greedy algorithm is combined to serve as an evaluation function to solve the two sub-problems, and the path is optimized. The method can be suitable for various complicated circuit board mounting tasks, and meets the real mounting operation requirements under variable industrial scenes.

Description

Technical Field

[0001] The present invention relates to an optimization method for the surface mounting process of a beam type pick-and-place machine based on simulated annealing, and belongs to the field of path planning optimization in the printed circuit board (PCB) assembly process. Background Art

[0002] As a crucial link in the production process of the entire electronic device, the mounting rate of the pick-and-place machine will directly affect the production efficiency of the entire electronic device industry chain. Among them, the process of the pick-and-place machine picking up components and placing them at the designated positions on the PCB printed circuit board occupies most of the entire production process. Therefore, path optimization of the pick-and-place machine mounting process is the main means to improve the mounting rate of the pick-and-place machine.

[0003] The beam type pick-and-place machine is one of the most widely used pick-and-place machine types in the current market. Multiple mounting heads of the beam type pick-and-place machine can move along the crossbeam together. This design allows multiple mounting heads to perform picking operations simultaneously, thus greatly improving the production efficiency. During the production process, the beam type pick-and-place machine first supplies various electronic components to the machine for picking through a feeder. The feeder is usually installed on both sides of the machine and can accommodate various types and sizes of components to meet different production requirements. In the work process, the mounting head picks up components from the feeder through a vacuum suction nozzle. To ensure the accurate picking of components, the device is also equipped with a high-precision vision system to monitor the position and angle of the components in real time, ensuring that there will be no deviation during the subsequent mounting process after the mounting head picks up the components. Then the mounting head moves the picked-up components to the designated positions on the PCB and accurately places them on the pads.

[0004] The main problems existing in the existing beam type pick-and-place machine in the PCB mounting process optimization technology are as follows:

[0005] (1) The mounting optimization effect is not ideal: For traditional pick-and-place machine mounting optimization algorithms, the optimized performance is limited and the convergence is poor. Therefore, a longer solution time is required, resulting in an unsatisfactory solution optimization effect.

[0006] (2) Lack of consideration for actual production constraints: In the process of constructing models and solving problems, traditional optimization algorithms do not fully consider various constraint conditions in actual production, such as the mechanical structure limitations of the pick-and-place machine and the physical characteristics of components. As a result, the optimization scheme may not be implementable in actual applications.

[0007] (3) Insufficient consideration for problem coupling: The pick-and-place machine mounting optimization problem can usually be decomposed into a component allocation problem and a feeder allocation problem. However, traditional optimization algorithms solve these two sub-problems separately, ignoring their coupling. As a result, the optimization results are not comprehensive and accurate enough, and the synergistic effect of each factor cannot be fully exerted, thus affecting the overall optimization effect.

[0008] Based on the above analysis, there are many deficiencies in the traditional placement optimization algorithm of pick-and-place machines, which to a certain extent limit the performance of pick-and-place machines. Summary of the Invention

[0009] Aiming at the problem of unsatisfactory placement optimization effect in the existing placement optimization methods of pick-and-place machines, the present invention provides an optimization method for the surface mounting process of a gantry pick-and-place machine based on dynamic temperature regulation simulated annealing.

[0010] The optimization method for the surface mounting process of a gantry pick-and-place machine based on simulated annealing of the present invention includes:

[0011] Step 1: Obtain the pick-and-place machine equipment information and PCB data;

[0012] Step 2: Calculate the total number of minimum placement cycles according to the pick-and-place machine equipment information and PCB data;

[0013] Step 3: Initialize the parameters of the simulated annealing algorithm: k, d, S, and randomly generate an initial component allocation result according to the total number of minimum placement cycles as the current solution; where k is the coefficient of the Sigmoid function, d is the translation distance of the Sigmoid function, and S is the monitoring variable;

[0014] Step 4: If the iteration stop condition is reached, go to Step 10, otherwise go to Step 5;

[0015] Step 5: Update based on the current solution to obtain an updated component allocation result as the new solution;

[0016] Step 6: Solve the placement location path and the total path length according to the updated component allocation result;

[0017] Step 7: Compare the total path lengths corresponding to the new solution and the current solution. If the total path length corresponding to the new solution is less than the total path length corresponding to the current solution, update the current solution to the new solution, and S = 0, then go to Step 9, otherwise go to Step 8;

[0018] Step 8: Judge whether Exp(-Δd / T) is greater than rand. If so, update the current solution to the new solution, and S = 0, then go to Step 9, otherwise, S = S + 1, then go to Step 9;

[0019] Δd represents the difference between the total path lengths corresponding to the new solution and the current solution, and the value of rand is randomly generated, ranging from 0 to 1;

[0020] Step 9: Update the simulated annealing temperature T using the Sigmoid function, then go to Step 4;

[0021] The updated simulated annealing temperature T is:

[0022]

[0023] Step 10. Output the optimized result of the optimal placement location path.

[0024] Preferably, in Step 5, update is performed on the basis of the current solution by using component constraints, feeder constraints, and nozzle constraints;

[0025] Among them, the component constraints and nozzle constraints are used to reasonably allocate the placement sequence of components on the premise of ensuring a relatively small placement cycle, so as to minimize the total path length within the entire cycle;

[0026] The feeder constraint is used to ensure the component picking and placement sequence.

[0027] Preferably, the component constraints include:

[0028] The sum of the number of components placed in each cycle must be equal to the total number of all components to be placed;

[0029] Each component can only be placed once during the entire placement process;

[0030] The number of components that can be placed in each cycle cannot exceed the number of placement heads;

[0031] Each component has a clear placement position and a corresponding placement head;

[0032] Each placement head can only process one component in each cycle.

[0033] Preferably, the nozzle constraints include:

[0034] In any one cycle, each placement head can correspond to at most one nozzle;

[0035] In any one cycle, the empty space of each nozzle library can correspond to at most one nozzle;

[0036] In any one cycle, each nozzle must have an empty space in a nozzle library corresponding to it;

[0037] Each component corresponds to at least one type of nozzle.

[0038] Preferably, the feeder constraints are:

[0039] Each feeder can correspond to a specific slot;

[0040] Each slot can correspond to at most one feeder;

[0041] One placement head can correspond to at most one feeder in one placement cycle.

[0042] Preferably, Step 6 includes:

[0043] Solve for the distance of the component positions in each cycle:

[0044]

[0045] where (X 1 , Y 1 ), (X 2 , Y 2 ) are the coordinates of components p 1 and p 2 respectively;

[0046] First, use the divide-and-conquer method to find the convex hull for the component sequence in each cycle to obtain the basic sequence seq. Then, insert the points closest to the convex hull into the basic sequence seq in order, and update the component mounting order in the component allocation result for each cycle to the order in step eq. Then, calculate μ m , ν m , ξ m in sequence; μ m represents the mounting distance in the m-th cycle, v m represents the picking distance in the m-th cycle, and ξ m represents the nozzle change distance in the m-th cycle;

[0047]

[0048] where, δ m represents the number of components mounted in the m-th cycle, δ m (i) represents the i-th component mounted in the m-th cycle represents the distance between the i-th component and the (i + 1)-th component in the m-th cycle, ζ m represents the total number of feeders in the m-th cycle, represents the distance between the i-th feeder and the (i + 1)-th feeder in the m-th cycle, represents the distance between the last feeder and the first component in the m-th cycle, β m represents whether a nozzle needs to be changed in the m-th cycle. If a change is needed, the value is 1; if not, it is 0, represents the distance between the nozzle magazine and the first feeder in the m-th cycle, represents the distance between the last component in the (m - 1)-th cycle and the nozzle magazine;

[0049] Calculate the total path length D:

[0050] D = μ m + μ m + ξ m .

[0051] Preferably, the method of the present application further includes:

[0052] According to the obtained component allocation result, execute according to the steps from step 3 to step 10 to obtain the feeder allocation result.

[0053] The beneficial effects of the present invention are as follows. Aiming at the problem of unsatisfactory mounting optimization effect existing in the existing mounting optimization method for pick-and-place machines, the present invention uses the simulated annealing algorithm to update the allocation result and optimize the path. The present invention introduces the concept of dynamic temperature adjustment, aiming to enhance the exploration ability of the algorithm during the solution process while ensuring the global convergence of the algorithm. Specifically, the improved algorithm adds a monitoring variable S on the basis of the traditional simulated annealing algorithm. When the solution generated in the current iteration is worse than the previous solution and is not accepted through the acceptance probability, this variable S is incremented. Each time an iteration is performed, the variable S is substituted into the Sigmoid function that has been translated to the right. The reason for choosing the Sigmoid function is that the Sigmoid function has the characteristics of slow growth in the early stage, rapid growth in the later stage and being bounded, which is consistent with the temperature adjustment characteristics required in the simulated annealing process. Aiming at the problems that the traditional mounting optimization method for pick-and-place machines lacks consideration of actual production constraints and does not fully consider problem coupling, etc., the present invention comprehensively optimizes the two mounting sub-problems of component allocation and feeder allocation, fully considers various constraint conditions in actual production, improves the solution speed of the optimization path, gives full play to the synergistic effect of various factors, thereby effectively improving the assembly efficiency of the gantry pick-and-place machine, making it more suitable for the actual application background, and having important theoretical value and application value. The present invention also uses a greedy path solving algorithm based on convex hull to solve the mounting path length, which can better reflect the effect of path optimization and accelerate the convergence speed of the algorithm. Brief Description of the Drawings

[0054] Figure 1 It is a schematic diagram of the surface mounting process flow;

[0055] Figure 2 It is a comparison diagram of the surface mounting component allocation optimization results of the gantry pick-and-place machine described in the present invention;

[0056] Figure 3 It is a comparison diagram of the surface mounting feeder allocation optimization results of the gantry pick-and-place machine described in the present invention. Detailed Embodiments

[0057] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0058] It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0059] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, but it is not limited to the present invention.

[0060] The method for optimizing the surface mounting process of the beam type pick-and-place machine in this embodiment includes:

[0061] Step 1: Obtain pick-and-place machine equipment information and PCB data:

[0062] Step 1-1: Import pick-and-place machine equipment data, including: the number of placement heads H, the index number of each head is h ∈ [1,..., H], the total number of feeder slots L, and the feeder slot number l ∈ [1,..., L].

[0063] Step 1-2: Import PCB data, including: the total number of components C, the index number of each type of component is c ∈ [1,..., C], the total number of feeders is F, the index number of each type of feeder is f ∈ [1,..., F], the total number of nozzles N, and the index number of each type of nozzle is n ∈ [1,..., N]. The number of nozzle library holes S, and the nozzle hole position number is s ∈ [1,..., S];

[0064] Step 2: Calculate the total number of minimum placement cycles based on the pick-and-place machine equipment information and PCB data:

[0065] Step 2-1: If the number and type of nozzles are sufficient, specifically, there is a situation where all heads are utilized within each cycle, then calculate the total number of cycles M according to the ratio of the number of components C to be placed and the number of placement heads H. The index number of the cycle is M ∈ [1,..., M].

[0066] Step 2-2: If the quantity and type do not meet the above requirements, add the number of cycles to the ratio of the number of components C to the number of placement heads H until all components are placed.

[0067] Step 3: Initialize the parameters of the dynamic temperature regulation simulated annealing algorithm:

[0068] Step 3-1: Initialize the parameters of the simulated annealing algorithm: k, d, S, E, where k is the coefficient of the Sigmoid function, d is the translation distance of the Sigmoid function, S is the monitoring variable, and E is the number of iterations;

[0069] Step 3-2: Randomly generate an initial component allocation result δ according to the total number of minimum placement cycles M.

[0070] Step 4: If the iteration stop condition is reached, go to Step 10; otherwise, go to Step 5. The iteration stop condition in this embodiment is the number of iterations E;

[0071] Step 5: Update based on the current solution to obtain the updated component allocation result as the new solution:

[0072] In Step 5 of this embodiment, the component constraint, feeder constraint, and nozzle constraint are used to update based on the current solution. The component constraint and nozzle constraint are used to reasonably allocate the component placement sequence on the premise of ensuring a small placement cycle, so as to minimize the total path length within the entire cycle. The feeder constraint is used to ensure the component picking and placement sequence.

[0073] Step 5-1: The component constraints are as follows:

[0074] where δ m represents the number of components placed in the m-th cycle. This constraint stipulates that the sum of the number of components placed in each cycle must be equal to the total number of all components to be placed. That is to say, all components must be completely placed within one or more cycles to ensure that no component is missed.

[0075] where σ m (p) is 1 indicating that component p is placed in cycle m, otherwise it is 0. This constraint ensures that each component can only be placed once during the entire placement process. This constraint prevents repeated placement and ensures the uniqueness and correctness of each component on the PCB.

[0076] limits the number of components that can be placed in each cycle and cannot exceed the number of placement heads. This constraint condition reflects the capacity limitation of the mounter in actual operation, that is, within each cycle, the number of mounter heads determines the number of components that can be processed simultaneously.

[0077] where γ h (p) is 1 indicating that each component is ensured to be assigned to a specific placement head, otherwise it is 0. This means that each component has a clear placement position and the corresponding operating head, ensuring the sequentiality and orderliness of the placement process.

[0078] This constraint ensures that each placement head can only process one component in each cycle. This ensures that the placement head will not be assigned multiple tasks in the same cycle, thus avoiding operation conflicts.

[0079] Component constraints fully consider multiple key factors to ensure the accuracy and effectiveness of the model. First, it is considered that the number of placement heads of the mounter is limited, which means that during the actual placement process, the allocation of components must be carried out according to cycles. Since the placement head can only process a certain number of components in each cycle, how to reasonably arrange these cycles to minimize the time of the entire placement process has become one of the core issues in model optimization. Second, during the actual placement process, the distance to pick up components is often much greater than the placement distance. This is because the placement head needs to move from the feeder position to the corresponding component position on the PCB and then return to the feeder position to pick up the next component. Therefore, when allocating components, the number of pickups should be minimized as much as possible to shorten the placement path within each cycle, thereby reducing the overall moving distance. Based on the above two points, the solution objective of component allocation can be summarized as: on the premise of ensuring a small placement cycle, reasonably allocate the placement order of components to minimize the total moving distance within the entire cycle. This problem can be analogized to the Traveling Salesman Problem (TSP), that is, to find a path between multiple given component positions so that the placement head can complete the placement of all components with the shortest total moving distance.

[0080] Step 5-2. The feeder constraints are as follows:

[0081] where θ l (f) being 1 means that feeder f is assembled at the l position of the feeder slot, otherwise it is 0. This constraint limits that each feeder can correspond to a specific slot position. That is, during the feeder allocation process, a suitable slot position must be found for each feeder for installation, so as to ensure that components can be picked up and placed smoothly;

[0082] limits that each slot can correspond to at most one feeder. This means that a slot cannot be shared by multiple feeders, ensuring that the feeder installed on each slot is unique, avoiding resource conflicts and installation problems;

[0083] where ω h (f) being 1 means that in cycle m, head h picks up materials from feeder f, otherwise it is 0. This constraint stipulates that a placement head can correspond to at most one feeder in one placement cycle. This limitation reflects that in actual operation, a placement head can only process a specific feeder in each cycle, ensuring the sequentiality and efficiency of the operation and avoiding the conflict situation where multiple feeders are allocated to the same placement head in the same cycle;

[0084] The feeder constraints of this embodiment comprehensively consider the relationship between component picking and placement order. Generally speaking, within a placement cycle, the component picking operation is performed first, and then the component placement operation is carried out. Therefore, the actual picking path starts from the last component of the previous cycle, passes through each feeder, and finally reaches the position of the first component of this cycle. During the picking process, according to the mechanical structure of the placement head, simultaneous picking is ensured as much as possible to further reduce the picking path.

[0085] Step 5-3. The nozzle constraints are as follows:

[0086] The following restrictions are all within any one cycle. Among them is 1, indicating that nozzle n is installed on head h, otherwise it is 0. This constraint limits that within any one cycle, each placement head can correspond to at most one nozzle. This means that within a placement cycle, a placement head can only use one specific nozzle type, ensuring the specificity and sequentiality of the operation.

[0087] where η n (s) is 1, indicating that nozzle n corresponds to nozzle hole position s, otherwise it is 0. This constraint limits that within any one cycle, each empty position in the nozzle library can correspond to at most one nozzle. This ensures that the nozzle stored in each empty position in the nozzle library is unique, avoiding resource conflicts and ensuring that each nozzle has a clear storage position during the replacement process.

[0088] It is stipulated that within any one cycle, each nozzle must correspond to an empty position in the nozzle library. This means that when each nozzle is not in use, it has a predetermined position stored in the nozzle library, ensuring that the required nozzle can be accurately found during the replacement operation.

[0089] where α s (p) is 1, indicating that component p is placed by nozzle s, otherwise it is 0. This constraint stipulates that each component corresponds to at least one nozzle type. That is to say, each component in the placement process must be able to find a suitable nozzle for operation, ensuring that the component can be correctly picked up and installed, and avoiding placement failures caused by nozzle mismatch.

[0090] Step 5-4. Randomly update based on the original allocation result according to the above constraints to obtain the updated component allocation result as the new solution.

[0091] Step 6 decomposes the placement path optimization problem of the beam-type placement machine into two sub-problems: component allocation and feeder allocation, and then establishes mathematical models for these two sub-problems respectively. Subsequently, these two problems

[0092] Step 6. Solve the placement location path and the total path length according to the updated component allocation result:

[0093] Step 6-1. Solve the distance between points. In this embodiment, the Chebyshev distance is adopted, and its calculation formula is as follows

[0094]

[0095] where (X 1 , Y 1 ), (X 2 , Y 2 ) are the coordinates of component p 1 and p 2 respectively; first, use the divide-and-conquer method to find the convex hull for the component sequence of each cycle to obtain the basic sequence seq, then insert the points closest to the convex hull into the basic sequence seq in order, and update the component placement order in the component allocation result of each cycle to the order in seq, and then calculate μ m , v m , ξ m in turn;

[0096] Step 6-2. Calculate the distance for the obtained allocation result:

[0097] D = μ m + μ m + ξ m

[0098]

[0099] where μ m represents the placement distance of the m-th cycle, v m represents the pickup distance of the m-th cycle, ξ m represents the nozzle change distance of the m-th cycle, D is the total distance of the mounter placement; δ m represents the number of components placed in the m-th cycle, δ m (i) represents the i-th component placed in the m-th cycle represents the distance between the i-th component and the (i + 1)-th component in the m-th cycle, ζ m represents the total number of feeders in the m-th cycle, represents the distance between the i-th feeder and the (i + 1)-th feeder in the m-th cycle, represents the distance between the last feeder and the first component in the m-th cycle, β m represents whether a nozzle change is required in the m-th cycle. If a nozzle change is required, the value is 1; if not, the value is 0, represents the distance between the nozzle magazine and the first feeder in the m-th cycle, denotes the distance from the last component in the (m - 1)-th cycle to the nozzle magazine;

[0100] In step 6 of this embodiment, a greedy path solving algorithm based on convex hull is used to solve the placement path length. This is because when solving the distances for a given component sequence, directly using the default order for calculation often cannot directly reflect the allocation effect of the sequence, and may even miss a better solution, causing the solution direction to develop in the wrong direction. For each placement cycle, it can be regarded as the distance from the feeder to the PCB component position plus the distance to the nozzle magazine for nozzle replacement according to the situation in the middle, which can be approximated as a TSP problem. Using the greedy path solving algorithm based on convex hull can better reflect the effect of path optimization and accelerate the convergence speed of the algorithm.

[0101] Step 7: Compare the total path lengths corresponding to the new solution and the current solution. If the total path length corresponding to the new solution is less than the total path length corresponding to the current solution, update the current solution to the new solution, set S = 0, and transfer to step 9; otherwise, transfer to step 8.

[0102] Step 8: Determine whether Exp(-Δd / T) is greater than rand. If so, update the current solution to the new solution, set S = 0, and transfer to step 9; otherwise, set S = S + 1 and transfer to step 9.

[0103] Δd represents the difference between the total path lengths corresponding to the new solution and the current solution, and the value of rand is randomly generated, ranging from 0 to 1.

[0104] Step 9: Update the simulated annealing temperature T using the Sigmoid function, and transfer to step 4.

[0105] The updated simulated annealing temperature T is:

[0106]

[0107] This formula ensures that the temperature adjustment is relatively slow in the early stage of the algorithm, and only when no better solution is found after multiple iterations will the temperature adjustment play a significant role, thus increasing the exploratory nature of the algorithm.

[0108] Step 10: Output the optimized result of the placement location path.

[0109] Step 11: According to the output component allocation result, execute according to the steps from step 3 to step 10 to obtain the feeder allocation result.

[0110] In this embodiment, the problem of optimizing the placement path of the beam-type mounter is first decomposed into two sub-problems: component allocation and feeder allocation. Then, constraints are established to build a mathematical model for these two sub-problems. This embodiment also improves the traditional simulated annealing method by introducing a dynamic temperature adjustment mechanism to enhance the algorithm's solving ability. Then, combined with the convex hull greedy algorithm as the evaluation function, the above two sub-problems are solved to optimize the path. This embodiment can be applied to various complex circuit board mounting tasks and meet the real mounting operation requirements in changing industrial scenarios.

[0111] Verification test:

[0112] The beneficial effects of the present invention are verified by the following embodiments, specifically:

[0113] The algorithm proposed in this embodiment is implemented on a desktop computer with Python 3.11 and an Intel Core i5 3.5-GHz CPU, and the simulation data is obtained from a self-developed simulation platform.

[0114] First, the experimental data is introduced. In this embodiment, 6 PCB data with different numbers of components are selected for mounting simulation experiments. The experimental data are all from actual mounting data, and the specific information is shown in Table 1 below.

[0115] Table 1 PCB information selected for the experiment

[0116]

[0117] The data used tries to ensure the distinguishability in the number of components, the number of component types, and the number of placement heads as much as possible to verify the reliability of the algorithm as much as possible. In addition, the unimproved simulated annealing algorithm, genetic algorithm, and particle swarm algorithm are respectively selected for comparative experiments. The parameters of each group of experiments are kept as consistent as possible to ensure the credibility of the experimental data. The experimental comparison is divided into two levels: placement optimization and pick-up optimization. At the same time, the placement distance and pick-up distance obtained by the algorithm, as well as the time taken for the algorithm to solve, are also compared.

[0118] From Figure 2It can be seen that during the optimization process of the placement path, all algorithms show a rapid convergence trend in the initial iteration. In particular, the genetic algorithm and the ant colony algorithm quickly approach the optimal solution in the first few hundred iterations. However, as the iteration progresses, the optimization effects of the genetic algorithm and the ant colony algorithm gradually become stable, showing obvious lack of exploration ability and being prone to falling into local optimal solutions. Although the traditional simulated annealing algorithm maintains a certain degree of convergence throughout the process, its exploration ability weakens in the middle and late stages, resulting in a significant decrease in the improvement speed of the solution. In contrast, the method of the present invention maintains good convergence throughout the iterative process and continues to steadily improve the solution in the middle and late iterations, demonstrating strong global search ability and continuous exploration ability for better solutions.

[0119] Furthermore, from Figure 3 It can be seen that during the optimization process of feeder allocation, similar to the placement path optimization, the genetic algorithm and the simulated annealing algorithm quickly find the optimal solution in the initial iteration, showing good convergence speed. However, the genetic algorithm reaches a bottleneck at an earlier stage and then shows almost no further improvement, indicating that it has strong local search ability but insufficient global exploration ability for complex problems. The performance of the ant colony algorithm is even less satisfactory, with a slow convergence speed in the early stage and failing to find a better solution throughout the process. In contrast, the method of the present invention significantly improves its global search ability by introducing a dynamic temperature adjustment mechanism into the algorithm. During the middle and late iteration process, the method of the present invention can effectively avoid falling into local optima and continuously explore better solutions, and finally also achieves the best results in feeder allocation optimization.

[0120] Compared with other algorithms, the method of the present invention demonstrates strong global search ability and excellent convergence in both the placement path optimization and feeder allocation optimization tasks. By introducing a dynamic temperature adjustment mechanism, the method of the present invention effectively solves the problem of insufficient exploration ability of the traditional simulated annealing algorithm in the late iteration, enabling it to continuously find better solutions when facing a complex solution space. This improvement makes the method of the present invention significantly superior to the genetic algorithm, the simulated annealing algorithm, and the ant colony algorithm in terms of both optimization effect and solution speed in the experiment, verifying its superiority and reliability in the placement algorithm. Next, the performance data of different algorithms in the PCB placement optimization and pick-up optimization processes will be analyzed, including placement distance, placement calculation time, pick-up distance, and pick-up calculation time.

[0121] Table 1 Placement Distance

[0122]

[0123]

[0124] As can be seen from Table 1, the method of the present invention shows the optimal placement distance on all PCB data. Especially on PCB6 with a large amount of data, its placement distance is 5814.91, which is significantly lower than other algorithms. This indicates that the improved simulated annealing algorithm can effectively reduce the placement path length, thereby improving the placement efficiency.

[0125] Table 2 Placement Calculation Time

[0126]

[0127] Table 2 shows the calculation times of each algorithm on different PCB data. While maintaining the optimal distance, the method of the present invention also has a relatively short calculation time. And the difference in calculation efficiency increases with the increase in the amount of data. For example, on PCB1, the calculation time of the method of the present invention is 11.1 seconds, the calculation time of the simulated annealing algorithm is 11.56 seconds, and the calculation times of the genetic algorithm and the ant colony algorithm are 143.27 seconds and 15.45 seconds respectively. On PCB6, the calculation time of the method of the present invention is 64.69 seconds, the calculation time of the simulated annealing algorithm is 69.44 seconds, and the calculation times of the genetic algorithm and the ant colony algorithm are 1893.58 seconds and 407.42 seconds respectively. This shows that the exploration efficiency of the method of the present invention is stronger than other algorithms.

[0128] Table 3 Pick-up Distance

[0129]

[0130]

[0131] Table 3 shows the distance results of different algorithms in pick-up optimization. Similar to the placement distance, the improved simulated annealing algorithm shows the best pick-up distance on all PCB data, with a maximum improvement of 16.8%.

[0132] Table 4 Pick-up Calculation Time

[0133]

[0134] In terms of the pick-up optimization calculation time (Table 4), the method of the present invention also performs outstandingly. Its calculation time on PCB6 is 1.65 seconds, which is significantly lower than that of the genetic algorithm (51.17 seconds) and the ant colony algorithm (92.30 seconds), and there is also an obvious improvement compared with the simulated annealing algorithm (1.9 seconds).

[0135] Generally speaking, the dynamic temperature regulation simulated annealing algorithm proposed by the present invention shows significant advantages in the PCB placement and pick-up optimization problems. Whether from the quality of the optimization results (distance) or the efficiency of the solution (time), the method of the present invention is superior to the traditional simulated annealing algorithm, genetic algorithm, and ant colony algorithm. Especially in the processing of large-scale PCB data, the method of the present invention demonstrates its better convergence and exploration capabilities, and is suitable for efficiently solving complex placement and pick-up optimization problems. Although the present invention has been described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the present invention. Therefore, it should be understood that many modifications can be made to the exemplary embodiments, and other arrangements can be designed, as long as they do not depart from the spirit and scope of the present invention as defined by the appended claims. It should be understood that the different dependent claims and the features described herein can be combined in a manner different from that described in the original claims. It should also be understood that the features described in connection with a single embodiment can be used in other described embodiments.

Claims

1. A beam mounter surface mounting process optimization method based on simulated annealing, characterized in that: The method comprises: Step 1: Get the SMT equipment information and PCB data; Step 2: Calculate the minimum total number of placement cycles based on the placement machine equipment information and PCB data; Step 3, initialize the parameters of the simulated annealing algorithm: k, d, S, and randomly generate the initial component allocation result as the current solution according to the minimum total number of placement cycles; where k is the coefficient of the Sigmoid function, d is the translation distance of the Sigmoid function, and S is the monitoring variable; Step 4: If the iteration stop condition is reached, go to step 10, otherwise go to step 5; Step 5: Update the current solution to obtain an updated component allocation result as a new solution; Step 6: Solve the placement location path and total path length according to the updated component allocation result; Step 7: Compare the total path lengths corresponding to the new solution and the current solution. If the total path length corresponding to the new solution is smaller than the total path length corresponding to the current solution, update the current solution to the new solution, and S=0, and proceed to step 9; otherwise, proceed to step 8. Step 8: Determine whether Exp(-Δd / T) is greater than rand. If so, update the current solution to the new solution, and S=0, and go to step 9. Otherwise, S=S+1, and go to step 9. Δd is the difference between the total path lengths of the new solution and the current solution. The value of rand is randomly generated and ranges from 0 to 1. Step 9: Use the Sigmoid function to update the simulated annealing temperature T and go to step 4; The updated simulated annealing temperature T is: Step 10: Output the optimal placement location path optimization result.

2. The surface mounting process optimization method of beam mounter based on simulated annealing according to claim 1 is characterized in that: In step 5, the component constraints, feeder constraints, and nozzle constraints are used to update the current solution; Among them, component constraints and nozzle constraints are used to reasonably allocate the placement order of components while ensuring a small placement cycle, so as to minimize the total path length in the entire cycle; Feeder constraints are used to ensure component picking and placement order.

3. The method for optimizing the surface mounting process of a beam mounter based on simulated annealing according to claim 2, characterized in that: Component constraints include: The sum of the number of components mounted in each cycle must be equal to the total number of all components that need to be mounted; Each component can only be mounted once during the entire mounting process; The number of components that can be placed in each cycle cannot exceed the number of placement heads; Each component has a clear placement position and corresponding placement head; Each placement head can only process one component per cycle.

4. The method for optimizing the surface mounting process of a beam mounter based on simulated annealing according to claim 2, characterized in that: Nozzle constraints include: In any cycle, each placement head can only correspond to one nozzle at most; In any cycle, each empty space in the nozzle library can only correspond to one nozzle at most; In any cycle, each nozzle must have an empty position in the nozzle library corresponding to it; Each component corresponds to at least one nozzle type.

5. The method for optimizing the surface mounting process of a beam mounter based on simulated annealing according to claim 2, characterized in that: The feeder constraints are: Each feeder can correspond to a specific slot; Each slot can only correspond to one feeder at most; A placement head can only correspond to one feeder in a placement cycle.

6. The method for optimizing the surface mounting process of a beam mounter based on simulated annealing according to claim 1, characterized in that: The step 6 comprises: Solve for the distance for the element position in each period: Where (X1, Y1) and (X2, Y2) are the coordinates of components p1 and p2 respectively; First, the convex hull of the component sequence of each cycle is calculated by the divide-and-conquer method to obtain the basic sequence seq, and then the points closest to the convex hull are inserted into the basic sequence seq in order, and the component placement order in the component allocation result of each cycle is updated to the order in step eq, and then μ is calculated in sequence. m 、v m , m ;μ m represents the mounting distance of the mth cycle, v m represents the pickup distance of the mth cycle, ξ m Indicates the nozzle replacement distance in the mth cycle; Among them, δ m represents the number of components mounted in the mth cycle, δ m (i) represents the i-th component mounted in the m-th cycle represents the distance from the ith element to the i+1th element in the mth period, ζ m represents the total number of feeders in the mth cycle, represents the distance from the ith feeder to the i+1th feeder in the mth cycle, represents the distance from the last feeder to the first element in the mth cycle, β m Indicates whether the nozzle needs to be replaced in the mth cycle. If it needs to be replaced, the value is 1, and if it does not need to be replaced, the value is 0. Indicates the distance from the nozzle library to the first feeder in the mth cycle, Indicates the distance from the last component to the nozzle library in the m-1th cycle; Calculate the total path length D: D=μ m +m m +ξ m 。 7. The method for optimizing the surface mounting process of a beam mounter based on simulated annealing according to claim 1, characterized in that: The method further comprises: According to the output component allocation result, follow the steps from step 3 to step 10 to obtain the allocation result of the feeder.

8. A computer-readable storage device storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for optimizing the surface mounting process of a beam mounter based on simulated annealing as claimed in any one of claims 1 to 7 are implemented.

9. A device for optimizing the surface mounting process of a beam mounter based on simulated annealing, comprising a storage device, a processor, and a computer program stored in the storage device and executable on the processor, characterized in that: The processor executes the computer program to implement the steps of the beam mounter surface mounting process optimization method based on simulated annealing as claimed in any one of claims 1 to 7.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method for optimizing the surface mounting process of a beam mounter based on simulated annealing as claimed in any one of claims 1 to 7 are implemented.

Images