KR20100031218A - System and method for manufacturing mask using genetic algorithm and dna computing - Google Patents

Abstract

PURPOSE: A mask generation system and a method using GA and DNA computing are provided so that in the DNA(Deoxyribonucleic Acid) computing, mask is created based on the GA(Genetic Algorithm). A mask with high conformity is created. CONSTITUTION: A computing module(100) comprises a random number generator, a selecting unit, and a hybridization part and mutation part. A memory module(300) stores the address of population, and the fitness value of each mask and the mask which becomes with the random number. A playing module(400) receives a mask having the most big fit value from a memory module. Mask is regenerated. A GA controller(200) transfers the first and the second parent data to the hybridization part. The first and the second child data received from the hybridization part is transferred to the memory module. The GA controller obtains the fit value of mask. The mask which becomes as described above and fit value are transferred to the memory module.

Description

Mask generation system and method using genetic algorithm and DNA computing {SYSTEM AND METHOD FOR MANUFACTURING MASK USING GENETIC ALGORITHM AND DNA COMPUTING}

The present invention relates to a mask generation system and method using genetic algorithm and DNA computing.

In modern engineering, large-scale optimization problems with complex constraints among various applications are very difficult to find an optimal solution in a short computation time using a combination of general mathematical programs or optimization algorithms. Evolutionary Algorithm is used as the main mechanism to solve these problems in a reasonable computation time.

Generating a mask using Genetic Alogorithm in DNA computing has the disadvantage that it takes a long time to make a random and suitable mask, and also to increase the suitability. By applying the genetic algorithm, which is an evolutionary algorithm, to DNA computing (DNAC), it is possible to generate a highly suitable mask through evolution. In addition, it was possible to increase the speed of mask generation. Genetic algorithms have a problem in that many generations need to be calculated repeatedly, and since the mask generation must be performed in real time in parallel, there is a need to implement the hardware.

On the other hand, DNAC implemented in software generally has two problems. The first is that it is slow. This is caused by simply comparing all the values for comparison, which can be implemented in hardware and processed in parallel to speed up. However, a simple count does not take much time, but when using a large DNAC as a GA there is a disadvantage that parallel processing is not possible. Secondly, you can use masks that are bad when using random. This can be improved by using a genetic algorithm of survival of the fittest.

Genetic algorithms, on the other hand, are an optimization algorithm based on Darwin's law of natural selection, which suggests that individuals with higher fitness are more likely to survive the next generation. It is widely used in areas such as pattern recognition, neural network learning, speech recognition, and robot control because of its excellent global search and convergence speed through parallelism.

The genetic algorithm, first introduced by John Holland, is a stochastic algorithm that encodes several individuals into a gene-like structure and then evolves in parallel using a genetic operator to achieve optimal conditions. It's a search technique.

1 is a diagram illustrating a flowchart of a genetic algorithm according to the prior art. Referring to the operation principle of the algorithm, the candidate solution to the problem is expressed in the form of chromosomes, and the population is initialized with randomly generated chromosomes (S10). Individuals initialized to random values are assessed for their suitability according to their relative problem-solving ability (S20), and by varying the degree of parental gene replication to the next generation according to their suitability, individuals with dominant traits have recessive traits. Induced to create more children than objects. This selection randomly selects an object to evolve to the next generation among the provided elements (S30), and these selected cloned individuals combine to create a new factor by combining two selected factors (S40) and mutation (S50). Goodness-of-fit is evaluated by various genetic operators such as (S60), and when the termination condition is satisfied (S70), it is recombined to generate a population of the next generation. This generational change is repeated until the desired level of solution exists in the population or other termination conditions are met.

This algorithm, however, has the disadvantage of being time consuming because a large number of iterative operations must be performed to evolve a large number of objects over generations. In order to solve this problem, parallel processing is necessary to perform evolutionary operations using several different groups.

Accordingly, in order to solve the above problems, the present invention provides a mask generation system using a genetic algorithm and DNA computing capable of generating a mask having high suitability by generating a mask using a genetic algorithm in DNA computing, and It is an object to provide a production method.

In addition, by implementing a process of generating a mask using a genetic algorithm in DNA computing in hardware and parallel processing, provides a genetic algorithm and a mask generation system and a method for generating a mask using DNA computing to increase the speed of mask generation. It aims to do it.

It is also an object of the present invention to provide a mask generation system and a method using a genetic algorithm and DNA computing that can improve the efficiency of generating a mask by using a random, but using a genetic algorithm of the survival of the fittest. .

The mask generating system using the genetic algorithm and DNA computing which concerns on Claim 1 produces a random value, selects the predetermined address through the random number generator which generate the random number address, and the random number generator, The first child receives the first data and the second data corresponding to the address, selects the first parent data and the second parent data, and exchanges the genetic characteristics of the first parent data and the second parent data, respectively. Population unit for generating data and second eye data, a computation module including a mutation unit for receiving a random number generated from a random number generator, and performing a calculation to modify the genotype by comparing the random number random mutation probability (population) Memory module for storing each address of), mask generated by random number and fitness value of each mask, memory The module receives a mask having the largest goodness-of-fit value among the masks, and transfers the reproduction module, the first parent data, and the second parent data, which reproduce the mask, to the mating unit, and the first eye data and the second eye data from the mating unit. Receives and passes to the memory module, generates a mask with random numbers at each address of the population, obtains the mask's goodness-of-fit value, and transfers the created mask and its goodness-of-fit value to the memory module. It includes a GA controller (Genetic Algorithm Controller) for controlling to deliver a mask having a to the playback module.

The mask generation system using the genetic algorithm and DNA computing which concerns on Claim 2, The mask generation system using the genetic algorithm and DNA computing which concerns on Claim 1 WHEREIN: The arithmetic module consists of the 1st parent data and the 1st parent data transmitted from a GA controller. 2, further comprising: a parent buffer for storing the parent data for a predetermined time, and transmitting the parent buffer to the mating unit, and storing the first eye data and the second eye data generated from the mating unit for a predetermined time, and then transferring the parent buffer to the GA controller. The GA controller obtains a goodness of fit value of the mask, passes the generated mask and the goodness of fit value to the memory module, and initializes the parent buffer.

The mask generation system using the genetic algorithm and DNA computing of the invention of Claim 3 WHEREIN: In the mask generation system using the genetic algorithm and DNA computing of Claim 1, a crossing part produces | generates eye data using a simple crossover method.

The mask generation system using the genetic algorithm and DNA computing which is invention of Claim 4, The mask generation system using the genetic algorithm and DNA computing which is invention of Claim 1 WHEREIN: Mutation part uses the simple mutation which only one mutation of each individual generate | occur | produces To modify the genetic trait.

In the method of generating a mask using an inventive genetic algorithm and DNA computing according to claim 5, a random value is generated, an address for a random number is generated, and first data and second data corresponding to the address are respectively generated from the first parent data. And a first step of selecting as the second parent data, a second step of generating a mask with random numbers at each address of the population, obtaining a fitness value of the mask, and storing the generated mask and the fitness value in a memory module. A third step of modifying the genotype by exchanging the genotypes of the parent data and the second parent data to generate the first eye data and the second eye data, and comparing the random number with a predetermined mutation probability; In the fourth step of storing the eye data and the modified genotype in the memory module, the mask having the largest goodness-of-fit value among the masks is transferred to the reproduction module. A fifth step of reproducing the disk.

In the method of generating a mask using the genetic algorithm and DNA computing according to claim 6, in the method of generating a mask using the genetic algorithm and DNA computing according to the fifth aspect, the second step includes storing the generated mask and its fitness value as a memory. After storing in the module, the method further includes initializing the parent buffer.

As described above, according to the mask generating system and method using the genetic algorithm and the DNA computing according to the present invention, a mask having high suitability can be generated by generating a mask using the genetic algorithm in the DNA computing. .

In addition, according to the present invention, the process of generating a mask using a genetic algorithm in DNA computing by implementing a parallel processing in hardware, it is possible to speed up the mask generation speed.

In addition, according to the present invention, by using a random, but by using a genetic algorithm of the survival of the fittest, it is possible to improve the efficiency of generating the mask.

Specific matters other than the problem to be solved, the problem solving means, and the effects of the present invention as described above are included in the following embodiments and the drawings. Advantages and features of the present invention and methods for achieving them will be apparent with reference to the embodiments described below in detail with the accompanying drawings. Like reference numerals refer to like elements throughout.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. However, the accompanying drawings are only described in order to more easily disclose the contents of the present invention, but the scope of the present invention is not limited to the scope of the accompanying drawings that will be readily available to those of ordinary skill in the art. You will know.

On the other hand, in the embodiment of the present invention described below, in order to see the performance improvement of the genetic algorithm, the simulation using a Visual studio (Visual studio) program, the gene programmed in the C language to compare the sensory speed of software and hardware Implemented in algorithms and Verilog hardware.

FIG. 2 is a block diagram illustrating a mask generation system using genetic algorithm and DNA computing according to the present invention, and FIG. 3 is a block diagram illustrating a calculation module of a mask generation system using genetic algorithm and DNA computing according to the present invention. 4 is a view for explaining a comparison method of the fitness function used in the mask generation system using the genetic algorithm and DNA computing according to the present invention, Figures 5a to 5e is a genetic algorithm and DNA computing according to the present invention It is a figure for demonstrating the experiment result of the mask generation system and generation method which were used.

2 is a block diagram illustrating a mask generation system using a genetic algorithm and DNA computing according to the present invention.

As shown in FIG. 2, the mask generation system using the genetic algorithm and the DNA computing according to the present invention includes an operation including a random number generator 110 and a selector 120, a crossing unit 130, and a mutation unit 140. Module 100, memory module 300, playback module 400, and GA algorithm (Genetic Algorithm Controller) 200.

The calculation module 100 selects a predetermined address through the random number generator 110, receives the first data and the second data corresponding to the address from the random number generator 110, and respectively, the first parent data and the second data. From the selection unit 120 to select the parent data, the cross-linking unit 130 for generating the first and second child data by exchanging the genetic characteristics of the first and second parent data, the random number generator 110 Receiving the generated random number, and compares the random number with a predetermined mutation probability comprises a mutation unit 140 for performing an operation for modifying the genotype. In other words, the operation module 100 is a part for selecting a parent entity and also arbitrarily selecting an object to be evolved to the next generation. In this case, the selection is to determine the survival distribution of the individual who crosses the population according to the distribution of the degree of adaptation in the next step, which is based on the distribution of the degree of adaptability. Individuals tend to leave more offspring. Data transmission between the memory module 300 and the random number generator 110 is performed by a handshaking protocol. In addition, the selector 120 performs a function of selecting a parent object for crossover and mutation, which is controlled by the GA controller 200 as a whole.

The memory module 300 stores each address of a population, a mask generated by a random number, and a fitness value of each mask.

The regeneration module 400 receives a mask having the largest goodness-of-fit value among the masks from the memory module and reproduces the mask.

The GA algorithm 200 receives the first data and the second data corresponding to the address from the random number generator 110, and receives the first data and the second data from the first parent data and the second parent, respectively. The data is selected and controlled to be transferred to the mating unit 120. Also, the first eye data and the second eye data are received from the crossing unit 120, and the first eye data and the second eye data are transferred to the memory module 300. Then, a mask is generated with random numbers at each address of the object group, the mask has a goodness-of-fit value, the generated mask and the goodness-of-fit value are transmitted to the memory module 300, and a mask having the largest goodness-of-mask value among the masks is reproduced. Control to pass to 400.

3 is a block diagram illustrating an operation module of a mask generation system using genetic algorithm and DNA computing according to the present invention.

As shown in FIG. 3, the operation module 100 of the mask generation system using the genetic algorithm and the DNA computing according to the present invention includes a selection unit 120, a mating unit 130, and a mutation unit 140. do.

In addition, the operation module 100 may store the first parent data and the second parent data transmitted from the GA controller (not shown) for a predetermined time, and then transfer the parent buffer 125 and the mating unit to the mating unit 130. The apparatus further includes an eye buffer 135 that stores the first eye data and the second eye data generated from the 130 for a predetermined time, and then transfers the first eye data and the second eye data to the GA controller. In addition, the GA controller obtains a goodness of fit value of the mask, transfers the generated mask and the goodness of fit value to the memory module, and initializes the parent buffer 125.

In brief, the selector 110 selects two objects from the memory module 300 to generate two parent data, and the hybridization unit 120 and the mutation unit 140 have two parents for fast operation. Data is used to create two eye data on one clock and pass it to the next module. Crossover, on the other hand, refers to altering a gene between two chromosomes to generate a new individual. Mutation refers to forcibly changing the value of a portion of a gene.

The operation module 100 is a module designed to perform two genetic operators step by step. First, crosses were used by using a one-point crossover method, which is basically used, and a single-point mutation method in which a mutation occurs in only one of each individual.

3 illustrates crossover and mutation in hardware, and receives two parent objects selected from the selection unit 120 to perform the first crossover. If not, a crossover is made by creating a new object, a child object using the parent object as it is. In addition, in the next mutation method, the genetic operation was designed to be performed in a combination of crosses and mutations. That is, after the selection is completed in the selection unit 120, the random bit is compared with a random probability received from the random number generator 110 and a mutation rate. If the random probability is large, the bit is modified.

4 is a view for explaining a method of comparing the fitness function used in the mask generation system using genetic algorithm and DNA computing according to the present invention.

Fitness function, on the other hand, is a preparatory work required to find a solution to a problem using a genetic algorithm, and determines an evaluation function to measure how good each chromosome is in solving a problem. The selection method used in the embodiment of the present invention is as follows.

This, as the fitness proportionate selection (proportionate selection) method, the fit of each object _{s j f (s j) (} > 0), j = 1, ... , Sum N, and determine the probability of selection for each object s _i .

This is how you decide.

In other words, the goodness-of-fit function value in the present invention refers to a value expressing how close to the optimal solution, and the same method used in the numerical recognition of DNAC was used. Using this method to find the fitness value of the mask for zero, increase the value if it is equal to 0 using 3760 initial trains and 376 corresponding to zero. The value is stored in memory with the mask created.

5A to 5E are diagrams for explaining experimental results of a mask generation system and a generation method using genetic algorithm and DNA computing according to the present invention.

Meanwhile, the mask generation method using genetic algorithm and DNA computing will be briefly described. First, a mask is generated using random numbers generated from a random number generator at each address of a population. Then, the suitability value of each mask is obtained and stored together with the mask in a memory. Then, the one with the largest goodness of fit among the masks is selected as the best mask. Then initialize the parent to '0'. Then, the selection unit executes the selection of the random method under the control of the GA controller. Each address is generated from the random number generator, and the first data (data A) and the second data (data B) corresponding to the address are obtained, and then the first parent data (parent A) and the second parent data (parent B) are obtained. Select. Then, a crossover operation is performed to exchange genotypes. Crossing uses a simple crossover method, which selects a point (ie, a given address) through a random number generator, and then exchanges the genotypes of the first and second parent data. Then, the mutagenesis is performed to generate mutations in the genetic trait. Mutations modify genetic traits by comparing random numbers generated using random number generators with mutation probabilities. At this time, the mutation type (mutation type) uses a simple mutation method in which only one of each individual mutation occurs. Then, the first eye data (child progeny A) and the second eye data (child progeny B) are saved and left in the next generation. If you do this repeatedly, 100 objects converge to one, and one mask is created by selecting a mask having a large number of them.

The waveform of FIG. 5A shows a waveform from a random number generator that produces a 64-bit mask, and the waveform of FIG. 5B obtains a goodness-of-fit value of a mask from the random number generator to create the first 100 masks, and stores it in a memory. This is a waveform that contains a series of steps.

In the case of the waveform of FIG. 5C, the waveform counts the number of 1s in the mask. In the case of the waveform of FIG. 5D, the waveform is a crossover result waveform. After the result is directly entered into the mutant part, as in the waveform of FIG. 5E, the random number generator Using the mutation point (mutation point) generated in the 1-> 0, 0-> 1 will be transformed.

Mask generation system and method using a genetic algorithm and DNA computing according to the present invention to perform the above operation, in the simulation using visual studio (visual studio) to generate a mask using GA, corresponding to 0 It took more than 21 hours to make a mask with the Dual core 2 duo 6600, but it took about 200s to implement it in hardware.

Accordingly, the mask generation system and generation method using the genetic algorithm and DNA computing according to the present invention as described above, to select a mask having a high randomness and excellent DNA using the GA, parallel GA and hardware implementation The mask generation time used can be reduced. And since this reduction in computation time and the number of iterations of genetic algorithms are small, the system can be applied in real time. In addition, the use of GA is not performed only once, but when a new pattern comes in and you want to add it to a train, it is added to generate a mask using GA, thereby improving the mask generation efficiency, By constructing a system with a firmware program of the fitness function module that occupies, and making only the GA operation modules that are repeatedly calculated in hardware, a large area of the hardware can be reduced, thereby forming an efficient system.

As such, the technical configuration of the present invention described above can be understood by those skilled in the art that the present invention can be implemented in other specific forms without changing the technical spirit or essential features of the present invention.

Therefore, the exemplary embodiments described above are to be understood as illustrative and not restrictive in all respects, and the scope of the present invention is indicated by the following claims rather than the detailed description, and the meaning and scope of the claims and their All changes or modifications derived from equivalent concepts should be construed as being included in the scope of the present invention.

1 is a view for explaining a flow chart of a genetic algorithm according to the prior art.

2 is a block diagram illustrating a mask generation system using a genetic algorithm and DNA computing according to the present invention.

Figure 3 is a block diagram for explaining the operation module of the mask generation system using a genetic algorithm and DNA computing in accordance with the present invention.

4 is a view for explaining a method of comparing the fitness function used in the mask generation system using genetic algorithm and DNA computing according to the present invention.

5A to 5E are diagrams for explaining experimental results of a mask generation system and generation method using genetic algorithm and DNA computing according to the present invention.

Claims (6)

A random number generator for generating a random value and generating an address for the random number; A selection unit for selecting a predetermined address through the random number generator, receiving first data and second data corresponding to the address from the random number generator, and selecting the first parent data and the second parent data, respectively; A mating unit for generating first eye data and second eye data by exchanging the genotypes of the first parent data and the second parent data; And Receiving a random number generated from the random number generator, a mutation unit for performing the operation of modifying the genotype by comparing the random number with a predetermined probability of mutation A calculation module comprising a; A memory module for storing each address of a population, a mask generated by the random number, and a fitness value of each mask; A reproduction module which receives a mask having the largest goodness of fit value among the masks from the memory module and reproduces the mask; And Delivering the first and second parent data to the mating unit; Receiving first eye data and second eye data from the mating unit, and transferring the first eye data and the second eye data to the memory module; In addition, a mask is generated with the random number at each address of the population, the fitness value of the mask is obtained, the generated mask and the fitness value are transferred to a memory module, and a mask having the largest fitness value among the masks is reproduced. GA controller (Genetic Algorithm Controller) to control the transfer to; Mask generation system using a genetic algorithm and DNA computing comprising a. The method of claim 1, The calculation module, A parent buffer which stores the first parent data and the second parent data transmitted from the GA controller for a predetermined time and then transfers them to the mating unit; And And an eye buffer configured to store the first eye data and the second eye data generated from the mating unit for a predetermined time, and then deliver the first eye data and the second eye data to the GA controller. The GA controller obtains a goodness-of-fit value of the mask, transfers the generated mask and the goodness-of-fit value to a memory module, and initializes the parent buffer. The mask generation system using genetic algorithm and DNA computing. The method according to claim 1 or 2, The mating unit generates mask data using a genetic algorithm and DNA computing, characterized in that for generating the eye data using a simple cross method. The method of claim 1, The mutation unit mask generation system using a genetic algorithm and DNA computing, characterized in that for performing the operation of modifying the genotype using a simple mutation method in which only one of each individual mutation occurs. Generating a random value, generating an address for the random number, and selecting first data and second data corresponding to the address as first parent data and second parent data, respectively; Generating a mask with the random number at each address of a population, obtaining a fitness value of the mask, and storing the generated mask and the fitness value in a memory module; A third step of exchanging the genotypes of the first parent data and the second parent data to generate first eye data and second eye data, and modifying the genotype by comparing the random number with a predetermined mutation probability; A fourth step of storing the first eye data and the second eye data and the modified genotype in the memory module; And Transmitting a mask having the largest goodness-of-fit value among the masks to a reproducing module to regenerate the mask; Gene generation method comprising a mask and a mask using DNA computing. The method of claim 5, The second step may further include initializing the parent buffer after storing the generated mask and its goodness-of-fit value in the memory module.

Images