Genetic Algorithm is a random search method for the evolution of evolution of the evolution of the biological industry (the survival of the survival, the genus genetic mechanism). It is first proposed by Professor of J. Hollland in the United States, first proposed in 1975. Its main feature is to direct structural objects, there is no definition of guidance and function continuity; there is an inherent hidden behavior and better global optimization capability. Adopt the probability optimization method, automatically acquire and guide the optimized search space, adaptively adjust the search direction, and do not need to be determined. These properties of the genetic algorithm have been widely used in combination optimization, machine learning, signal processing, adaptive control, and artificial life. It is one of the key technologies in modern intelligence calculations.
Genetic algorithm and natural selection
Darwin's natural choice is a broad accepted chemistry that is widely accepted. This kind of learning believes that organisms must be survived. Survival struggles include three aspects of struggle between struggles, struggles, and bio-with inorganic environments. In the survival struggle, individuals with favorable variations are easy to survive, and more opportunities will be favored to future generations; individuals with unfavorable variations are easily eliminated, and there are fewer opportunities for future generations. Therefore, all individuals who win in the survival struggle are stronger than environmental adaptability. Darwin's process of survived this in survival struggle and immigrants was called nature. It shows that genetic and variations are the intrinsic factors that determine biological evolution. The variations of various organisms in the natural world can be adapted to the environment, and it is not open to the genetic and variant life. It is this genetic characteristics of the organism to keep the species of the biological industry to maintain relative stability; the variation characteristics of the biological body produce new traits, so that the new species is formed, which promotes the evolution and development of the organism.
The genetic algorithm is the calculation model of simulating the genetic choice of Darwin and the biological evolution process of nature. Its ideology is derived from the natural laws of biological genetic and adaptive survival, and is a search algorithm for the iterative process with "survival detection". The genetic algorithm is subject to all individuals in a group, and uses a randomized technology to guide a high-efficiency search for a coded parameter space. Among them, the selection, crossing and variation constitute the genetic operation of the genetic algorithm; the parameter coding, the initial group setting, the design of the adaptation function, the genetic operation design, the control parameter setting five elements constitute the core content of the genetic algorithm. As a new global optimization search algorithm, the genetic algorithm is widely used in various fields in various fields, and has achieved good results in various fields with its simple and universal, robust, strong, practical, and other significant features. One of the important intelligent algorithms.
2. Basic steps for genetic algorithms
We are used to calling Holland1975 GAs called traditional GA. Its main steps are as follows:
Encoding: GA Prior to the search, the decomposition data of the solution is used to generate genotype string structural data in genetic space, which constitutes a different point of different combinations.
Generation of initial groups: randomly generate N initial string structure data, each string structure data is called an individual, and n individual constitutes a group. GA starts it iteration as the initial point as the initial point as the N-string structure data.
Adaptivity value evaluation detection: The adaptive function indicates the advantages and degradation of individuals or solutions. Different problems, the definition of adaptive functions is also different.
Select: The purpose of choice is to select an excellent individual from the current group, so that they have the opportunity to be a parent as the next generation of reproductive descendants. The genetic algorithm reflects this idea by selecting the process, and the principle of choice is the probability of adaptive individuals to contribute one or more progeny. Choosing the principle of the survival of the Advantages of Darwin.
Exchange: The exchange operation is the most important genetic operation in the genetic algorithm. The new generation of individuals can be obtained by switching operations, and new individuals combine their parent individual characteristics. Exchange reflects the idea of information exchange.
Variation: Variation First, randomly select an individual in the group, and randomly change the value of a string in the string structure data at a certain probability of selecting the individual. Like the same biological community, the probability of variation in GA is very low, usually between 0.001 to 0.01. Variation provides opportunities for the production of new individuals. The calculation process of GA is:
Select encoding method
Generate initial groups
Calculate the adaptability of the initial group
If the condition {selects the exchange variation to calculate the adaptive value of the new generation group}
3. Features of genetic algorithm
The genetic algorithm is used as a fast, simple, fault-tolerant algorithm, shown in the optimization of various types of structural objects.
Obvious advantage. The genetic algorithm has the following characteristics compared to the traditional search method:
The search process does not act directly on the variable, but the individual encoded individuals in the parameter set. This encoding operation allows the genetic algorithm to operate directly to the structural object (set, sequence, matrix, tree, diagram, chain, and table).
The search process is from a set of unparalleled to another, using a method of processing multiple individuals in the group, lowers the possibility of falling into local optimal solutions and is easy to simultaneously.
Use the probability of change rules to guide the search direction without using certainty search rules. There is no special requirements (such as connectivity, convexity, etc.) for search space, only use other auxiliary information such as derivatives, and has a wider adaptation range.
4. Research history and status quo of genetic algorithm
The rise of genetic algorithm is in the late 1980s and early 1990s, but its historical origin can be traced back to the 60s.
Early. Early studies are mostly based on computer simulations of the natural system. As Fraser simulation studies, he proposed a concept and idea that is very similar to the current genetic algorithm. The creative research results of Holland and Dejong have changed the lack of non-target and theoretical guidance of early genetic algorithm research. Among them, Holland published in 1975 << Natural System and Artificial System Adaptation >> Systematically expounds the basic theories and methods of genetic algorithm, and proposes the theoretical research and development of the theoretical research and development of genetic algorithms. . This theory first confirmed the importance of structural recombinant genetic operations to obtain hidden compliance.
In the same year, DEJON's important papers << Genetic adaptive system to behavior analysis >> Combining Holland's model theory combines with his calculation experiment, and proposes new genetic operation technologies such as generation ditch. It can be considered that the research work made by DEJONG is a milestone in the development of the genetic algorithm.
In the 1980s, the genetic algorithm ushered in the development of the prosperity, whether it is theoretical research or application research, has become a hot topic. In particular, the application field of genetic algorithms is also expanding. At present, the main fields involved in the genetic algorithm have automatic control, planning, design, combined optimization, image processing, signal processing, artificial life. It can be seen that the application research of genetic algorithms has expanded many updates from the initial combination optimization. More engineering applications.
