A LED die sorting path optimization method based on hybrid ant colony algorithm

Abstract

The present invention discloses a method for optimizing LED grain sorting paths based on a hybrid ant colony algorithm, which mainly includes the following steps: reading the map file of the current wafer disk to obtain the grade information and coordinate information of all grains; selecting all grains of one grade as a set of grains to be sorted, calculating the distance between each grain and storing it in a distance matrix; using the iteration of the hybrid ant colony algorithm to generate multiple feasible sorting path sequences; selecting the shortest sorting path as the optimized path. The present invention discloses a method for optimizing LED grain sorting paths based on a hybrid ant colony algorithm, which optimizes the grain sorting path sequence with the help of reasonable path planning, reduces the time consumption of sorting equal amounts of grains by minimizing the total moving distance of the wafer table, and improves the working efficiency of the sorting equipment as a whole.

Description

Technical Field

[0001] The present invention relates to an LED and Mini-LED crystal sorting technology, and in particular to an LED crystal sorting path optimization method based on a hybrid ant colony algorithm. Background Art

[0002] With the development of industrial technology, the application fields of LED and Mini-LED are constantly expanding. Related products have expanded from home monitors to commercial monitors, industrial control monitors and even military monitors, and have been integrated into all aspects of people's lives and social development. Due to production process reasons, the LED grains on a single Wafer disk will have inconsistent physical parameters such as luminous wavelength, voltage and power. When such grains are applied to the production of display devices, the products will have problems such as uneven brightness and color, which cannot meet people's demand for high-quality display. Therefore, before large-scale industrial production, sorting manufacturers need to use grain sorting equipment to transfer grains of the same grade / specification to the corresponding Bin disk according to the specified sorting path. With the advancement of semiconductor technology, the size of wafers is gradually transitioning from 4 inches, 6 inches, and 8 inches to 12 inches. At the same time, the size of grains is gradually developing towards miniaturization, from the original millimeter level to 50-200 microns, so the number of grains on a single wafer will increase by about 5-10 times. However, the sorting speed has not been greatly improved, resulting in an exponential increase in the cost of grain production, which has greatly hindered the development of the industry. Without increasing the speed of the sorting motor, the sorting efficiency can be improved and the production cost of grains can be reduced by optimizing the sorting path reasonably. Sorting path optimization has become a key technology to increase the sorting speed.

[0003] There are two main sorting path solutions:

[0004] (1) The Z-shaped sorting scheme plans a sorting path that is roughly "Z-shaped". The distance between chips is not considered during the operation, and the sorting rule is strictly followed row by row and sequence. As a result, there is a serious path redundancy problem, which increases the moving distance of the wafer table as a whole, greatly affecting the working efficiency of the equipment.

[0005] (2) A sorting scheme based on the ant colony algorithm. This scheme uses the iteration of the ant colony algorithm to generate a better sorting path. However, the optimization performance of the algorithm will decrease as the problem scale increases. Then the convergence speed of the solution will decrease and it will be easy to fall into the local optimum, thus obtaining a sorting path with low optimization quality.

[0006] In view of this, the present invention is proposed. Summary of the invention

[0007] The purpose of the present invention is to optimize and improve the existing technology and provide an LED die sorting path optimization method based on a hybrid ant colony algorithm with high solution quality and suitable for large-scale path optimization problems, so as to solve the above-mentioned technical problems existing in the prior art.

[0008] The objective of the present invention is achieved through the following technical solutions:

[0009] The LED die sorting path optimization method based on the hybrid ant colony algorithm of the present invention comprises the following steps:

[0010] Step 1: Read the map file of the current wafer disk to obtain the level information and coordinate information of all grains.

[0011] Step 2: Select all the grains of one grade as the grain set to be sorted, calculate the distance between each grain and store it in the distance matrix.

[0012] Step 3: Use the iteration of the hybrid ant colony algorithm to generate multiple feasible sorting path sequences.

[0013] Step 4: Select the shortest sorting path as the optimized path.

[0014] Compared with the prior art, the LED die sorting path optimization method based on the hybrid ant colony algorithm provided by the present invention has the advantages of high solution quality and good convergence reliability. It can find a high-quality path optimization solution within a reasonable time and is suitable for large-scale LED chip path planning problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A schematic flow chart of a method for optimizing LED die sorting paths based on a hybrid ant colony algorithm provided in an embodiment of the present invention;

[0016] Figure 2 Schematic diagram of the path for sorting using a zigzag path and an optimized path for five sorting examples of the present invention with the number of grains ranging from 80 to 300. DETAILED DESCRIPTION

[0017] The following is a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention; it is obvious that the described embodiments are only part of the embodiments of the present invention, not all of the embodiments, which does not constitute a limitation of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the protection scope of the present invention.

[0018] First, the terms that may be used in this article are explained as follows:

[0019] The term “and / or” means that either or both of them can be realized at the same time. For example, X and / or Y means both “X” or “Y” and “X and Y”.

[0020] The terms "include", "comprises", "contains", "has" or other descriptions with similar semantics should be interpreted as non-exclusive inclusion. For example, including certain technical feature elements (such as raw materials, components, ingredients, carriers, dosage forms, materials, dimensions, parts, components, mechanisms, devices, steps, procedures, methods, reaction conditions, processing conditions, parameters, algorithms, signals, data, products or products, etc.) should be interpreted as including not only certain technical feature elements explicitly listed, but also other technical feature elements known in the art that are not explicitly listed.

[0021] The term "consisting of..." means excluding any technical feature elements not explicitly listed. If this term is used in a claim, it will make the claim closed, so that it does not contain technical feature elements other than the technical feature elements explicitly listed, except for the conventional impurities related to them. If this term only appears in a clause of a claim, it only limits the elements explicitly listed in the clause, and the elements recorded in other clauses are not excluded from the overall claim.

[0022] Unless otherwise specified or limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in this article can be understood according to specific circumstances.

[0023] The orientation or position relationship indicated by terms such as "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", etc. are based on the orientation or position relationship shown in the drawings and are only for the convenience and simplification of description, and do not explicitly or implicitly indicate that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation of this document.

[0024] The contents not described in detail in the examples of the present invention belong to the prior art known to professionals in the field. If no specific conditions are specified in the examples of the present invention, the conditions are carried out according to the conventional conditions in the field or the conditions recommended by the manufacturer. If the manufacturer is not specified in the reagents or instruments used in the examples of the present invention, they are all conventional products that can be purchased commercially.

[0025] The LED die sorting path planning method based on the hybrid ant colony algorithm of the present invention comprises the following steps:

[0026] Step 1: Read the map file of the current wafer disk to obtain the grade information and coordinate information of all grains;

[0027] Step 2: Select all the grains of one grade as the grain set to be sorted, calculate the distance between each grain and store it in the distance matrix;

[0028] Step 3: Generate multiple feasible sorting path sequences by using the iteration of the hybrid ant colony algorithm;

[0029] Step 4: Select the shortest sorting path as the optimized path.

[0030] The distance matrix in step 2 is used to store the distance between each grain. If the number of grains to be sorted is n, the size of the matrix is ​​n*n.

[0031] Assume that the center coordinates of the grains i and j to be sorted are (x i ,y i )、(x j ,y j ), the calculation formula of the distance between grains is:

[0032]

[0033] Among them, Δx and Δy are the moving distances of the wafer stage in the horizontal and vertical directions respectively, Δx=|x i -x j |、Δy=|y i -y j |, to simplify the analysis, assuming Δx = Δy, the calculation formula for the grain spacing can be simplified to

[0034]

[0035] Further obtain the total distance d of the planned path of the grains to be sorted total

[0036]

[0037] The hybrid ant colony algorithm in step 3 is based on the ant colony algorithm and incorporates the ideas of the genetic algorithm and the 2-opt algorithm. It is essentially an improved ant colony algorithm and mainly includes the following steps:

[0038] Step 31: Generate γ initial paths using the greedy strategy, and select the shortest path as the suboptimal solution path;

[0039] Step 32: Initialize the number of ants m, initial pheromone τ0, number of iterations iter, α and β and other parameters, and set the selection ratio p s , crossover probability p c and mutation probability p m size;

[0040] Step 33: Differentiate the distribution of initial pheromones;

[0041] Step 3 and 4: Randomly generate the starting points of m ants and construct the solution space based on the state transition probability;

[0042] Step 3.5: After completing the path construction, calculate its fitness and select m*p according to the selection ratio. s Paths, on this basis, crossover and mutation operations are performed to obtain the population NewChrom;

[0043] Step 36: Insert (mm*ps) paths into NewChrom to form a new population Chrom;

[0044] Step 37: Update pheromones on all paths;

[0045] Step 38: Determine whether the maximum number of iterations iter is reached. If so, output the optimal solution. Otherwise, clear the path record table and return to step 34 for the next iteration.

[0046] The purpose of generating the suboptimal solution in step 31 is to complete the distribution of the differentiated initial pheromone in step 33 by means of the length of the suboptimal solution path. The suboptimal solution generation method is:

[0047] Step 3: randomly select a crystal grain as the starting point of the path and store the sequence number of the crystal grain in the path array;

[0048] Step 3-2: According to the distance matrix, select the grain with the shortest distance from the current starting point as the optimal point;

[0049] Step 313: Accumulate the shortest distance;

[0050] Step 3-4: Take the optimal point as the current starting point and store it in the path array;

[0051] Step 315: determine whether all the grains to be sorted have been traversed. If so, save the current path array and its length and start the next iteration. Otherwise, return to step 312.

[0052] Step 316: Determine whether the maximum number of iterations γ has been reached. If so, output the suboptimal solution, otherwise continue to the next iteration.

[0053] The parameters m, α, β and ρ in step 32 will greatly affect the optimization performance of the entire algorithm, so it is necessary to obtain the specific values ​​of the above parameters through multiple experiments.

[0054] The distribution of differentiated initial pheromones in step 33 is used to improve the optimization speed of the algorithm in the early stage. The formula for the initial distribution of pheromones is:

[0055]

[0056] Among them, i and j are the grain numbers on the suboptimal solution path, τ0 is the initial pheromone, τ add is the additional pheromone increment on the path, calculated according to the path length of the suboptimal solution;

[0057] By adding additional pheromones to the suboptimal solution path, the problem of ants blindly searching due to the same pheromone distribution in the early stage of the algorithm can be improved, thereby improving the algorithm's search speed for high-quality paths to a certain extent.

[0058] There are three ways of variation in step 35, specifically:

[0059] Method 1: Random exchange mutation: randomly generate two mutation points i and j in the range of 1 to n (n is the number of grains), exchange the grain numbers at the position and form a new path as the mutation path;

[0060] Method 2: The general reversal operation is: randomly generate two mutation points i and j in the range of 1 to n, and flip the path sequence between i and j to form a new path. To increase the effectiveness of mutation, the reversal operation is improved as follows: search for all feasible solutions of reversal and calculate their path lengths, and select the shortest path as the mutation path;

[0061] Method 3: Evolutionary reversal mutation based on 2-opt algorithm: Let the current optimal path be L best , perform the reversal operation to obtain a new mutation path, and compare L best and the length of the mutation path, and take the smaller solution path as the new L best , repeat the evolutionary reversal operation until no more than L can be found best Smaller value.

[0062] The pheromone update method in step 37 is specifically to add pheromones only to the optimal path in the current iteration. The pheromone update formula is:

[0063]

[0064] Among them, ρ is the pheromone volatility coefficient, t is the current iteration number, is the pheromone increment of ant k in the path sequence ij.

[0065] The shortest distance accumulation in step 313 is specifically the sum of the distances between each current starting point and the current optimal point;

[0066] If the number of grains is n, the number of the current starting point and the current optimal point is n-1, and n-1 shortest distance accumulations are required to obtain the total path length;

[0067] If the coordinates of the kth current starting point and the kth optimal point are (x k ,y k )、(X k ,Y k ), the sum of distances is L, then

[0068]

[0069] In summary, the LED die sorting path optimization method based on the hybrid ant colony algorithm in the embodiment of the present invention has the advantages of high solution quality and good convergence reliability. It can find a high-quality path optimization solution within a reasonable time and is suitable for large-scale LED chip path planning problems.

[0070] The advantages compared with the existing technology are:

[0071] The present invention integrates the search strategies of genetic algorithm and local search algorithm on the basis of ant colony algorithm to improve the algorithm's ability to explore high-quality sorting paths. The distributed parallel computing capability of ant colony algorithm is used to perform global search on solution paths, and the convergence speed of the algorithm in the early stage is improved by differentiating the initial distribution of pheromones. After all ants complete the path construction once, the selection strategy of genetic algorithm is used to improve the population quality. With the help of crossover and three mutation strategies based on different stages, the situation where the ant colony algorithm falls into the local optimum at different stages is improved, and the convergence of the algorithm is further accelerated. On this basis, the updating method of pheromones is adjusted, and the method of increasing pheromones only on the contemporary optimal path is adopted to improve the efficiency of the algorithm converging to the vicinity of the optimal solution.

[0072] The hybrid ant colony algorithm provided by the present invention is actually an improved ant colony algorithm. On the basis of the ant colony algorithm, the following changes are made: ① The initial pheromone is differentially distributed using a greedy strategy; ② After all ants complete the path construction, multiple paths are selected according to the selection ratio, and crossover and mutation operations are performed. ③ The global pheromone change mechanism of the basic ant colony algorithm is adjusted to only allow the current optimal path to add additional pheromones. The three improved strategies introduced greatly improve the convergence efficiency and search quality of the algorithm. After multiple iterations, the quality of the optimized path will be greatly improved. Compared with the above-mentioned Z-shaped sorting scheme and the sorting scheme based on the ant colony algorithm, the present invention has more excellent optimization performance, and the quality of the optimized path obtained is significantly higher. It greatly improves the path redundancy problem existing in the Z-shaped path, while overcoming the local optimal problem existing in the ant colony algorithm, and is more suitable for path optimization of large-scale grain sorting.

[0073] The present invention discloses a LED grain sorting path planning method based on a hybrid ant colony algorithm. By incorporating an improved strategy into the ant colony algorithm, the optimization performance of the algorithm is improved as a whole, thereby finding a better sorting path. When the response speed of the motor system is constant, using this path for grain sorting can greatly improve the grain sorting efficiency.

[0074] The experimental test results show that:

[0075] Compared with the Z-shaped sorting path, five sorting instances (with 80 to 300 grains and the same grain grade) completed the sorting using their respective optimized paths, and their path optimization rates could reach 72.1% to 82.6%, and the sorting efficiency could be increased by 10.49% to 28.56%.

[0076] In order to more clearly demonstrate the technical solution and technical effects provided by the present invention, the embodiments of the present invention are described in detail with specific embodiments below.

[0077] Examples, such as Figure 1 , Figure 2 As shown: Specific implementation method one:

[0079] The path planning flow chart of the LED die sorting path planning method based on the hybrid ant colony algorithm of the present invention is as follows: Figure 1 The figure shows a flow chart of a method of the present invention.

[0080] Specifically, a method for LED die sorting path planning based on a hybrid ant colony algorithm includes the following steps:

[0081] Step 1: Read the map file of the current wafer disk to obtain the grade information and coordinate information of all grains;

[0082] Step 2: Select all the grains of one grade as the grain set to be sorted, calculate the distance between each grain and store it in the distance matrix;

[0083] Step 3: Generate multiple feasible sorting path sequences by using the iteration of the hybrid ant colony algorithm;

[0084] Step 4: Select the shortest sorting path as the optimized path. Specific implementation method 2:

[0086] The present embodiment is different from the first embodiment in that:

[0087] In the step 2, all the grains of one level are selected as the grain set to be sorted, and the distance between each grain is calculated and stored in the distance matrix, specifically:

[0088] In step 1, the map file is obtained in the test phase, which includes the coordinate information, grade information and related physical parameters of each grain on the wafer. The general sorting order is: after completing the sorting of all grains of the current grade, manually or automatically select all grains of the next other grade and perform the next sorting.

[0089] Furthermore,

[0090] The distance matrix in step 2 is used to store the distances between each grain. If the number of grains to be sorted is n, the size of the matrix is ​​n*n. Suppose the center coordinates of the grains i and j to be sorted are (x i ,y i )、(x j ,y j ), the inter-grain distance formula can be simplified to:

[0091]

[0092] Where Δx and Δy are the moving distances of the wafer stage in the horizontal and vertical directions respectively, Δx=|x i -x j |, Δy=|y i -y j |.

[0093] To simplify the analysis, assuming that Δx = Δy, the calculation formula for the grain spacing can be expressed as

[0094]

[0095] Further obtain the total distance d of the planned path of the grain set to be sorted total

[0096]

[0097] The other steps and parameters are the same as those in the first embodiment. Specific implementation method three:

[0099] The present embodiment differs from any one of the first to second specific embodiments in that:

[0100] In step 3, the hybrid ant colony algorithm is iterated to generate multiple feasible sorting path sequences. Specifically:

[0101] The hybrid ant colony algorithm in step 3 is based on the ant colony algorithm and incorporates the ideas of the genetic algorithm and the 2-opt algorithm. It is essentially an improved ant colony algorithm and mainly includes the following steps:

[0102] Step 31: Generate γ initial paths using the greedy strategy, and select the shortest path as the suboptimal solution path;

[0103] Step 32: Initialize the number of ants m, initial pheromone τ0, number of iterations iter, α and β and other parameters, and set the selection ratio p s , crossover probability p c and mutation probability p m size;

[0104] Step 33: Differentiate the distribution of initial pheromones;

[0105] Step 3 and 4: Randomly generate the starting points of m ants and construct the solution space based on the state transition probability;

[0106] Step 3.5: After completing the path construction, calculate its fitness and select m*p according to the selection ratio. s Paths, on this basis, crossover and mutation operations are performed to obtain the population NewChrom;

[0107] Step 36: Insert (mm*p in NewChrom s ) paths form a new population Chrom;

[0108] Step 37: Update pheromones on all paths;

[0109] Step 38: Determine whether the maximum number of iterations iter is reached. If so, output the optimal solution. Otherwise, clear the path record table and return to step 34 for the next iteration.

[0110] The other steps and parameters are the same as those in Specific Embodiments 1 and 2. Specific implementation method four:

[0112] The present embodiment is different from the specific embodiments 1 to 3 in that: in step 31, a greedy strategy is used to generate γ initial paths, and the shortest path among them is selected as the suboptimal solution path.

[0113] The purpose of generating a suboptimal solution in step 31 is to complete the differentiated initial pheromone distribution in step 33 by means of the length of the suboptimal solution path. The specific method for generating a suboptimal solution is:

[0114] Step 3: randomly select a crystal grain as the starting point of the path and store the sequence number of the crystal grain in the path array;

[0115] Step 3-2: Obtain the grain with the shortest distance from the current starting point as the optimal point according to the distance matrix;

[0116] Step 313: Accumulate the shortest distance;

[0117] Step 3-4: Take the optimal point as the current starting point and store it in the route array;

[0118] Step 315: determine whether all the grains to be sorted have been traversed. If so, save the current path sequence and its length and start the next iteration. Otherwise, return to step 312.

[0119] Step 316: Determine whether the maximum number of iterations γ has been reached. If so, output the suboptimal solution, otherwise continue to the next iteration.

[0120] The shortest distance accumulation in step 313 specifically refers to: the sum of the distances between each current starting point and the optimal point. If the number of grains is n, the number of current starting points and current optimal points is n-1, and n-1 shortest distance accumulations are required to obtain the total path length. If the coordinates of the kth current starting point and the kth optimal point are (x k ,y k )、(X k ,Y k ), the total distance is L, and its calculation formula is:

[0121]

[0122] The other steps and parameters are the same as those in Specific Embodiments 1 to 3. Specific implementation method five:

[0124] The difference between this embodiment and the first embodiment of the embodiment 1 to 4 is that in step 32, the number of ants m, the initial pheromone τ0, the number of iterations iter, α and β and other parameters are initialized, and the selection ratio p is set. s , crossover probability p c and mutation probability p m The size is:

[0125] The hybrid ant colony algorithm provided by the present invention is actually an improved ant colony algorithm. The parameters such as m, α, β and ρ in step 32 will greatly affect the optimization ability of the entire algorithm. Therefore, it is necessary to obtain the specific values ​​of the above parameters with the help of multiple experiments to ensure that the hybrid ant colony algorithm has good optimization performance and obtains a relatively optimal optimization path.

[0126] The other steps and parameters are the same as those in Specific Embodiments 1 to 4.

[0127] Specific implementation method six: This implementation method is different from any one of specific implementation methods one to five in that: the differentiated initial pheromone distribution in step three three.

[0128] The differentiated initial pheromone distribution in step 33 is used to improve the optimization speed of the algorithm in the early stage. The formula for the initial pheromone distribution is:

[0129]

[0130] Where i and j are the grain numbers on the suboptimal solution path, τ0 is the initial pheromone, τ add is the additional pheromone increment on the path, which is obtained according to the path length of the suboptimal solution.

[0131] By adding additional pheromones to the suboptimal solution path, the problem of ants blindly searching due to the same pheromone distribution in the early stage of the algorithm can be improved, thereby improving the algorithm's search speed for high-quality paths to a certain extent.

[0132] The other steps and parameters are the same as those in Specific Embodiments 1 to 5. Specific implementation method seven:

[0134] The present embodiment is different from any one of the specific embodiments 1 to 6 in that: in the steps 3 and 4, starting points of m ants are randomly generated and a solution space is constructed according to the state transition probability.

[0135] The path construction in steps 3 and 4 is achieved with the help of state transition probability. The formula for state transition probability is:

[0136]

[0137] The pheromones and heuristic information on the path determine the probability of the ant transferring to the next path node. If the number of ants and the number of grains to be sorted are m and n respectively, after finishing steps 3 and 4, the size of the path record table storing the sorting path sequence is m*n. represents the transfer probability of the kth ant, τ ij (t) represents the pheromone left by the ant on the path ij in the tth iteration, η ij (t) is the heuristic function, which is the inverse of the distance, i.e. η ij(t) = 1 / d ij α and β represent the pheromone importance factor and the heuristic information importance factor respectively, and allowed k A collection of chips to be sorted.

[0138] The other steps and parameters are the same as those in Specific Embodiments 1 to 6. Specific implementation eight:

[0140] The difference between this embodiment and the first embodiment of the embodiment 1 to 7 is that after the path construction is completed in step 35, its fitness is calculated, and m*p is selected according to the selection ratio. s Paths are obtained by performing crossover and mutation operations on this basis to obtain the population NewChrom, specifically:

[0141] There are three ways of variation in step 35, specifically:

[0142] Method 1: Random exchange mutation: randomly generate two mutation points i and j in the range of 1 to n (n is the number of grains), exchange the grain numbers at the position and form a new path as the mutation path.

[0143] The first variation method can improve the situation where the ant colony algorithm falls into the local optimum in advance, so that the algorithm maintains a relatively good search state in the initial stage.

[0144] Method 2: The general reversal operation is: randomly generate two mutation points i and j in the range of 1 to n, and flip the path sequence between i and j to form a new path. Considering the scale of the grains to be sorted, the reversal operation is improved as follows: Let the current optimal path be X best , search for all feasible solutions for reversing the current path and calculate their path lengths, and select the shortest path as the mutation path.

[0145] The second mutation method can increase the effectiveness of the mutation. By comparing the path lengths of multiple mutations, a mutation path with higher quality can be found. While improving the algorithm from falling into a local stagnation state, it can also accelerate the convergence of the algorithm to a certain extent.

[0146] Method 3: Evolutionary reversal mutation based on 2-opt algorithm, specifically: let the current optimal path be L best , perform the reversal operation to obtain a new mutation path, and compare L best and the length of the mutation path, and take the smaller solution path as the new L best , repeat the evolutionary reversal operation until no more than L can be found best Smaller value.

[0147] The mutation mode 3 is essentially a local search. During the iteration process, if the path value keeps the same value for λ consecutive times, the mutation operation can prompt the algorithm to jump out of the local optimal state and continue to converge to the optimal path.

[0148] Selectively using the above three mutation methods in different stages of the algorithm can improve the problem of the algorithm falling into local extreme values ​​at different stages, maintain the algorithm's exploration trend for high-quality paths, and improve the overall optimization quality. The population NewChrom refers to the collection of sorting path sequences that are recovered after the mutation. NewChrom is a set of (m*p s )*n path record table.

[0149] The other steps and parameters are the same as those in Specific Embodiments 1 to 7. Specific implementation method nine:

[0151] The difference between this embodiment and the first embodiment of the embodiment 1 to 8 is that in step 36, (mm*ps) paths are inserted into NewChrom to form a new population Chrom, specifically:

[0152] The (mm*ps) paths inserted in step 36 are the first (mm*ps) paths with the highest fitness among the paths constructed by ants before performing the selection, crossover and mutation operations. The new population Chrom is a path record table of size m*n, which is used to store the m sorting path sequences after the current iteration is completed.

[0153] The other steps and parameters are the same as those in Specific Embodiments 1 to 8.

[0154] Specific implementation method ten: This implementation method is different from any one of specific implementation methods one to nine in that: in step thirty-seven, pheromones on all paths are updated.

[0155] The pheromone update method is specifically: only the pheromone addition operation is performed on the optimal path in the current iteration, and the pheromone update formula is:

[0156]

[0157] Among them, ρ is the pheromone volatility coefficient, t is the current iteration number, is the pheromone increment of ant k in the path sequence ij.

[0158] The other steps and parameters are the same as those in Specific Embodiments 1 to 9.

[0159] The following examples are used to verify the beneficial effects of the present invention:

[0160] The present invention uses a sorting machine of model ZKMY to perform a sorting experiment on a wafer disk with 12066 grains (including 39 grades of grains). Part of the map data of the disk is shown in Table 1.

[0161] Embodiment 1:

[0162] Table 1: Part of the map file data.

[0163] PosX PosY BIN TEST LOP1 VF1 VF2 VZ IR WLD WLP HW VF3 LOP2 WLD2 VFD 33 40 4 1 210.464 2.994 0 0 0 449.68 445.7 16.1 0 0 0 0 48 39 13 2 218.905 2.997 0 0 0 450.74 446.38 16.3 0 0 0 0 47 39 5 3 210.472 3.008 0 0 0 448.98 444.85 17.3 0 0 0 0 46 39 8 4 232.532 3.005 0 0 0 449.13 445.07 16.2 0 0 0 0 45 39 17 5 231.123 3.001 0 0 0 450.38 446.15 16.3 0 0 0 0 44 39 4 6 224.853 2.993 0 0 0 449.92 445.65 16.8 0 0 0 0 43 39 14 7 223.38 3.017 0 0 0 451.67 447.32 17.2 0 0 0 0 42 39 13 8 229.875 2.994 0 0 0 451.56 447.35 16.8 0 0 0 0 41 39 14 9 215.333 3.017 0 0 0 451.02 446.52 17 0 0 0 0 40 39 8 10 233.49 3.01 0 0 0 448.88 444.81 16 0 0 0 0

[0164] The first four columns of data in Table 2 are the horizontal coordinate, vertical coordinate, grade and test serial number of the grain, and the other parameters are the relevant physical parameters of the grain (including light emission wavelength, voltage and current, etc.).

[0165] From the complete map file, select some data and perform the die sorting operation according to the specific implementation method 1. The experimental results of five sorting examples with the number of chips ranging from 80 to 300, respectively, completing the sorting of equal amounts of die according to the Z-shaped sorting path and the optimized path are shown in Table 2.

[0166] Table 2: Test results of various scale sorting examples (grain grade is 13, number of grains: 80 to 300)

[0167] Grain size Zigzag path length Optimize the average path length Path optimization rate Z-shaped path sorting time / ms Optimized path sorting time / ms Improved sorting efficiency 80 709.36 198.22 72.1％ 15918 13205 17.04％ 110 883.76 243.27 72.5％ 20770.5 18346 11.67％ 170 1250.2 343.52 72.5％ 31688.5 28365 10.49％ 230 1952.9 507.13 74.0％ 43811.5 37940 13.40％ 300 5102.1 890.08 82.6％ 71258 50905.5 28.56％

[0168] According to Example 1, compared with the Z-shaped sorting path commonly used in existing equipment, the hybrid ant colony algorithm provided by the present invention can be used to perform sorting operations, and the total path length of sorting can be optimized by 72.1% to 82.6%. According to machine testing, the actual sorting time can be shortened by 2424.5 to 20352.5 ms when the same amount of grains are sorted using the optimized path, and the sorting efficiency can be improved by 17.04% to 28.56%.

[0169] The grain size tested on the machine is 17*34mil, the grain spacing in the x-direction and y-direction is 1.5mm and 0.75mm respectively, and the acceleration of the swing arm motor is 83333.33r / s 2 ; The acceleration of the Wafer motor and Bin motor is 100m / s 2 During the experiment, the ejector pins did not move upward to lift the grains, and the suction nozzles on both sides of the swing arm did not suck up the grains. Other operations were the same as the actual industrial sorting.

[0170] Different grain sizes will lead to differences in the time consumed by the algorithm to obtain the optimized path. In order to avoid the impact of path calculation time on sorting efficiency to a certain extent, the sorting operation can be carried out asynchronously in a combination of "online + offline" methods, using an independent computer to implement path planning, and store the optimized path in a map, which can be directly used in subsequent sorting links.

[0171] In actual industrial sorting, using an algorithm to obtain the sorting path of tens of thousands of grains at a time will result in an extremely unreasonable time loss. The wafer disk is divided into several sorting areas in a block manner, and the sorting of all grains on the disk is completed through multiple path planning and multiple sorting, which can reduce the impact of path calculation time on work efficiency. On this basis, the "online + offline" combination method is used to asynchronously sort grains, which can further improve the sorting efficiency.

[0172] The present invention provides a method for optimizing a grain sorting path, which uses an improved ant colony algorithm to obtain a high-quality optimized path solution, and minimizes the total movement distance of a wafer stage with the help of the path, thereby improving the overall sorting efficiency of the equipment.

[0173] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention, rather than to limit the scope of protection of the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the essence and scope of the technical solution of the present invention.

[0174] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by any technician familiar with the technical field within the technical scope disclosed in the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims. The information disclosed in the background technology section of this article is only intended to deepen the understanding of the overall background technology of the present invention, and should not be regarded as an admission or in any form that the information constitutes prior art known to those skilled in the art.

Claims

1. A LED die sorting path planning method based on a hybrid ant colony algorithm, characterized in that: Includes steps: Step 1: Read the map file of the current wafer disk to obtain the grade information and coordinate information of all grains; Step 2: Select all the grains of one grade as the grain set to be sorted, calculate the distance between each grain and store it in the distance matrix; Step 3: Generate multiple feasible sorting path sequences by using the iteration of the hybrid ant colony algorithm; Step 4: Select the shortest sorting path as the optimization path; The hybrid ant colony algorithm in step 3 is based on the ant colony algorithm and incorporates the ideas of the genetic algorithm and the 2-opt algorithm. It is essentially an improved ant colony algorithm and includes the following steps: Step 31: Generate γ initial paths using the greedy strategy, and select the shortest path as the suboptimal solution path; Step 32: Initialize the number of ants m, the initial pheromone τ0, the number of iterations iter, the α and β parameters, and set the selection ratio p s , crossover probability p c and mutation probability p m size; Step 33: Differentiate the distribution of initial pheromones; Step 3 and 4: Randomly generate the starting points of m ants and construct the solution space based on the state transition probability; Step 3.5: After completing the path construction, calculate its fitness and select m*p according to the selection ratio. s Paths, on this basis, crossover and mutation operations are performed to obtain the population NewChrom; Step 36: Insert (mm*ps) paths into NewChrom to form a new population Chrom; Step 37: Update pheromones on all paths; Step 38: Determine whether the maximum number of iterations iter is reached. If so, output the optimal solution. Otherwise, clear the path record table and return to step 34 for the next iteration.

2. The LED die sorting path planning method based on hybrid ant colony algorithm according to claim 1, characterized in that: The distance matrix in step 2 is used to store the distance between each grain. If the number of grains to be sorted is n, the size of the matrix is ​​n*n. Assume that the center coordinates of the grains i and j to be sorted are (x i ,y i )、(x j ,y j ), the calculation formula of the distance between grains is: Among them, Δx and Δy are the moving distances of the wafer stage in the horizontal and vertical directions respectively, Δx=|x i -x j |, Δy=|y i -y j |, to simplify the analysis, assuming Δx = Δy, the calculation formula for the grain spacing can be simplified to Further obtain the total distance d of the planned path of the grains to be sorted total 3. The LED die sorting path planning method based on hybrid ant colony algorithm according to claim 1, characterized in that: The purpose of generating the suboptimal solution in step 31 is to complete the distribution of the differentiated initial pheromone in step 33 by means of the length of the suboptimal solution path. The suboptimal solution generation method is: Step 3: randomly select a crystal grain as the starting point of the path and store the sequence number of the crystal grain in the path array; Step 3-2: According to the distance matrix, select the grain with the shortest distance from the current starting point as the optimal point; Step 313: Accumulate the shortest distance; Step 3-4: Take the optimal point as the current starting point and store it in the path array; Step 315: determine whether all the grains to be sorted have been traversed. If so, save the current path array and its length and start the next iteration. Otherwise, return to step 312. Step 316: Determine whether the maximum number of iterations γ has been reached. If so, output the suboptimal solution, otherwise continue to the next iteration.

4. The LED die sorting path planning method based on hybrid ant colony algorithm according to claim 1, characterized in that: The parameters m, α, β and ρ in step 32 will greatly affect the optimization performance of the entire algorithm, so it is necessary to obtain the specific values ​​of the above parameters through multiple experiments.

5. The LED die sorting path planning method based on hybrid ant colony algorithm according to claim 1, characterized in that: The distribution of differentiated initial pheromones in step 33 is used to improve the optimization speed of the algorithm in the early stage. The formula for the initial distribution of pheromones is: Among them, i and j are the grain numbers on the suboptimal solution path, τ0 is the initial pheromone, τ add is the additional pheromone increment on the path, calculated according to the path length of the suboptimal solution; By adding additional pheromones to the suboptimal solution path, the problem of ants blindly searching due to the same pheromone distribution in the early stage of the algorithm can be improved, thereby improving the algorithm's search speed for high-quality paths to a certain extent.

6. The LED die sorting path planning method based on hybrid ant colony algorithm according to claim 1, characterized in that: There are three ways of variation in step 35, specifically: Method 1: Random exchange mutation: randomly generate two mutation points i and j in the range of 1 to n, where n is the number of grains, exchange the grain numbers at the position and form a new path as the mutation path; Method 2: The reversal operation is as follows: randomly generate two mutation points i and j in the range of 1 to n, and flip the path sequence between i and j to form a new path; In order to increase the effectiveness of mutation, the reversal operation is improved as follows: all feasible solutions of reversal are searched and their path lengths are calculated, and the shortest path is selected as the mutation path; Method 3: Evolutionary reversal mutation based on 2-opt algorithm: Let the current optimal path be L best , perform the reversal operation to obtain a new mutation path, and compare L best and the length of the mutation path, and take the smaller solution path as the new L best , repeat the evolutionary reversal operation until no more than L can be found best Smaller value.

7. The LED die sorting path planning method based on hybrid ant colony algorithm according to claim 1, characterized in that: The pheromone update method in step 37 is specifically to add pheromones only to the optimal path in the current iteration. The pheromone update formula is: Among them, ρ is the pheromone volatility coefficient, t is the current iteration number, is the pheromone increment of ant k in the path sequence ij.

8. The LED die sorting path planning method based on hybrid ant colony algorithm according to claim 3, characterized in that: The shortest distance accumulation in step 313 is specifically the sum of the distances between each current starting point and the current optimal point; If the number of grains is n, the number of the current starting point and the current optimal point is n-1, and n-1 shortest distance accumulations are required to obtain the total path length; If the coordinates of the kth current starting point and the kth optimal point are (x k ,y k )、(X k ,Y k ), the sum of distances is L, then

Images