Sheet metal bending automatic process planning based on improved genetic algorithm


Because sheet metal bending parts have the advantages of high strength, light weight, low cost, simple processing and high production efficiency, they have been more and more widely used in the fields of machinery, communication, electronics and electrical appliances. During the processing of sheet metal parts, the order of bending seriously affects the processing efficiency and part accuracy. The traditional bending process planning requires manual planning by designers, and requires a long time for testing and adjustment. It is difficult to judge the feasibility of the process plan, so the efficiency and reliability are low. Fast, efficient and reliable automatic process planning is an urgent problem to be solved in sheet metal bending.

At home and abroad, extensive and in-depth research has been carried out on the methods and theories of bending process optimization. M. I nui et al. used the topology of the part as a constraint to filter out inappropriate bending processes and improve the planning efficiency ¨J. Duf lou J proposed a solution to the traveling salesman problem based on the priority constraint method and the branch-and-bound method. J. c. RI CO, J. M. GONZALEZ et al. proposed a method to solve the bending process planning, which divides the bending parts into some basic shape units, and then performs partial process planning for them respectively, and finally combines the sequence of these subunits into a complete bending T sequence. J. Thanapandi C. M et al. proposed a genetic algorithm for process planning preprocessing to reduce search time. Duf lou J et al. made a detailed review of bend sorting. Kannan TR et al. proposed a genetic algorithm to obtain bending close to the optimal solution ;

Based on the above problems, this paper uses an improved genetic algorithm to plan the sheet metal bending process, so as to ensure the rationality and efficiency of the bending process planning.

1 Improvement of Bending Process Planning Genetic Algorithm Bending process planning is a process evaluation process composed of multiple work steps. When using genetic algorithm to plan a bending process, it is necessary to select an appropriate adaptation function for process evaluation, and then select a set of The initial population evolves better individuals through replication, crossover and mutation operations. The improvement of genetic algorithm in this paper is mainly improved from three aspects: adaptation function, initial population and evolution process.

1.1 Selection of adaptation function

The selection of the fitness function should be considered to improve the processing efficiency of the workpiece as much as possible without interference. The main factors are the number of molds, the number of mold removals, the number of turns and turns of the workpiece, and the number of operation balances. Take the fitness function: in the formula: 渆 represents the fitness value; N! Represents the total number of molds NI, the number of mold disassembly (mold replacement and change of direction)/V2, the number of sheet turnover, the number of sheet turns N4, and the number of unbalanced operations N5; it represents the weight of the corresponding factors. The weights are respectively taken as: 100, island: 50, ground 3: Qiao “4: 10 “5: 5, where the operation imbalance means that when the worker grabs the workpiece, the distance from the bending line to the back gauge is greater than that of the worker. the distance. The smaller the appropriate value, the better the bending process scheme. However, for the bending process with interference, there are also advantages and disadvantages. In addition, the bending process with interference is an infeasible solution, and the appropriate value must be larger than that without interference. In order to quickly evolve the non-interference bending process in the genetic algorithm, this paper defines the adaptive function of the interference as: F=Ew,N+(N-p)max(w (2) In the formula: N represents the total workpiece The number of bends Yin < N) indicates the location where the bend occurs. 〗N > 】some N! , so “0 can ensure that the appropriate value of the process in which the interference occurs is greater than the appropriate value of the non-interference, 0 is a dead body)

[email protected]) is the best way to define the interference process according to the position where the interference occurs.inferior.

1.2 Initial population optimization

For sheet metal parts with a large number of bends and a complex structure, the randomness in the selection of the initial population is too strong, and the appropriate value is generally too large, which makes the selection of the initial population too large.

Convergence speedslower.for thisAfter the initial population is generated, the optimal individuals in the population are subjected to corresponding optimization operations to obtain individuals with smaller fitness values.

picture1The bending process shown (3,4,6,9,8,2,1,5,7) as an example, the interference detection is performed for each step in turn, and for the step where interference occurs,

it willin turn with the following foldBend the horns and recheck for interference.If there is interference in a step no matter how it is exchanged, the detection process is stopped at this time, although no

Does not interfere with the individual, but at least one relatively superiorthe individual.

1.3 Evolutionary process improvement

The evolutionary process requires replication, crossover and mutation operations to continuously generate new populations.The copy operation in this paper adopts the runner method, and takes the reciprocal of the appropriate value to calculate.

Crossover operation with partial matching(PMX), set theand the probability is0.9the mutation adopts the swap operation, and the mutation probability is set to 0.1.

Since in the evolution process, the next generation solution after the crossover operation is likely not to have a better solution, resulting in the loss of the current optimal solution and slowing down the convergence speed, so the best solution in each generation can be used.

The optimal solution is saved directly to the next generation.At the same time, in order to avoid the local optimal solution, the optimal solution is forced to undergo mutation operation, that is, the mutation probability is1if the variation is changed to

Excellent is directly saved to the next generation, otherwise the original current optimal solution is still preserved. This process allows each generation of the population to evolve in a better direction.

2Bending process automatic planning process

The bending process planning needs to select a suitable mold for each process step under the premise of comprehensively considering factors such as interference collision, bending efficiency and operational balance, so that the appropriate value is as small as possible.

2.1Automatic mold selection

The mold includes an upper mold and a lower mold. The selection of the mold is determined by the characteristic parameters of the workpiece. When selecting the mold, it is necessary to ensure that there is no interference between the mold and the sheet metal.

At the same time, as much as possible to ensure the discountThe bent workpiece meets the actual requirements.

First, according to the bending angle, bending radius and sheet thickness specified by the bending characteristics of the workpiece, all molds that meet the conditions are initially screened for each process step.

The molds are then screened again by priority during the process evaluation process.There are two main aspects of priority: first, try to avoid the use of special molds;

First select the standard mold; secondThe angle of the die in sheet metal bending is generally selected to be slightly smaller than the bending angle. The closer the angle is to the bending angle, the higher the priority of the die. For each bend

It is necessary to ensure that the angle of the upper die and the lower die are the same.

2.2 Automatic planning of bending process

The specific process of planning the bending process using the above improved genetic algorithm is shown in the figure2shown.

(1) Pre-processing of process planning Select a set of molds for each bending from the mold library, and define the fitness function and initial population.

(2) The process evaluation is mainly to calculate the appropriate value.The fitness value is an important basis for the evolution of the genetic algorithm population, which determines whether an individual is is eliminated or copied, and is also a review

CertainlyThe only indicator of the quality of the process.

In the process of calculating the appropriate value, it should be ensured that the bending process of each step does not interfere as much as possible. Interference needs to detect the interference between the workpiece and the machine tool, mold, and back gauge respectively.As shown3As shown, the upper die is taken as an example, and the upper die is rotated counterclockwise and clockwise to contact the workpiece by the reversal method, respectively.PRotate half of the bending angle to enclose the shaded area shown in the figure. If the workpiece intersects with the shaded area, it means that there is interference.

In addition, when the workpiece has dimensional accuracy requirements, the selection of the positioning point of the back gauge must consider the accuracy requirements of the workpiece, so that the dimensional deviation accumulates on the bending section with low accuracy requirements.

<3) population evolution directly performs forced mutation operation on the optimal individual in each generation, compares it with the pre-mutation adaptive value, and saves the better individual into the next generation population. For non-current optimal individuals, copy, crossover and mutation operations are performed directly to generate the next generation population. After several generations, a relatively better bending process can be evolved.

3Example verification and application

picture4The shown sheet metal part is taken as an example to verify the effectiveness of the genetic algorithm, where the wall thickness of the workpiece is2灬]the bending radius is2 nun,width is400mm

According to the feature parameters of the workpiece, the qualified molds are automatically selected from the library, and then the priority order is set according to their angle size. Fig.5Shown are all the dies that can be selected for this workpiece, and each dies are numbered separately.in4 Number punch and3No. The die angle is greater than90. , only for sheet metal parts 120.of4,5number bend.

Define the initial population size as20the evolutionary algebra is100set the default positive direction of the workpiece and the mold, use the optimized genetic algorithm above to plan the process, arrange the final results according to the appropriate value and take5 programs, as shown in the table1shown.

For the non-interfering process scheme, the negative sign in the bending process in the tableThe direction of the displayed workpiece is opposite to the default direction, and the workpiece direction is not defined during interference.last in the table

PieceThe process plan is in the7a workerWhen the step interferes, the stationCould not find a suitable mold.

The change of the appropriate value in the whole genetic algorithm process is shown in the figure6shown in the first generationThe fitness of the optimal solution is895no interference occurred, indicating that the initial populationoptimization process

non-interfering individuals,Make the entire genetic algorithm convergefaster.This bending part is in thetwenty oneThe approximate optimal solution can be reached inshared use2,7stime.

There may also be no feasible tradeoffs during the initial population optimization process.Bend solution, as shown in the figure7As shown, the fitness of the optimal solution in the first generation is 1 800,Although interference occurs,

But its convergence speed is the samevery fast, in theThe approximate optimal solution is reached in 29 generations, and it takes time3,8sabout.

If the first two improved conditions of the genetic algorithm in this paper are not adopted, that is,Without considering the fitness function during interference and the optimization of the initial population, onlyImprove the evolution process to save the current optimal solution, the result is shown in the figure8PlaceShow.to the70After the generation, the non-interfering bending process appeared, reachingApproximate time to optimal solution is approaching16smost of its time is wasted inIn the process of evolving a feasible bending solution, it is necessary to set a largerEvolutionary generation or population size.

In addition, if the3An improvement condition saves the current optimal solution,The fitness curve fluctuates greatly.100Generation is likely not toA feasible bending operation appears.

Develop a set of process planning and simulation software, with the first bendingprocess plan (1)3,8,6,1,9,7,2,10,one4,one5) as an example, use this software to show its bending process, the results are as followspicture9shown.

4 Epilogue

Using the improved genetic algorithm to carry out process planning for sheet metal workpieces,Introduce the interference position to the adaptive function when the interference occurs to ensure fastRapidly evolve feasible non-interfering processes; optimize initial population to obtain phaseFor better individuals, the evolutionary starting point of the genetic algorithm is better;Save the current optimal solution to the next generation to ensure that the current optimal solution is not lost,Make the population evolve in a more and more optimal direction.Results vs. PosteriorThe effectiveness of the algorithm is proved, and the approximate optimal solution can be obtained quickly, andAvoid local optima.

Sheet fabrication services for mild steel, high strength low alloy (HSLA) steel, cold/hot rolled steel, galvanized steel, stainless steel, aluminum, copper and brass. Capable of fabricating parts up to 12 ft. length and +/-0.001 in. tolerance. Various capabilities include contract manufacturing,custom stamping,edge rolling, forming,top laser cutting, roll bending and welding. Finishing and secondary services such as hardware installation, tapping, deburring, cleaning, heat treating, plating, anodizing and painting available. Sheet Metal Prototype and low to high volume production runs offered. Suitable for commercial/residential architectural, aluminum brake shape parts, wall panel systems, brackets, general flashings, rails, call button plates and ship building component parts.

Related Posts