Advances in Computer Science Research, volume 78

3rd International Workshop on Materials Engineering and Computer Sciences (IWMECS 2018)

The Application Research of Genetic Algorithm

Jumei Zhang
Weinan Normal University, Weinan, Shaanxi, 714000

Keywords: Genetic Algorithm, Application Research, Geophysical Prospecting

Abstract: Genetic algorithm is now an optimization method widely used in geophysical

prospecting. This paper summarizes the basic principle and method of genetic algorithm and its
developing direction. It analyzes the implement process of the method, and introduces the genetic
process, the current situation and the application prospect of the genetic algorithm.

1. Introduction
Genetic Algorithm (GA) was proposed by American scholar Holland in 1975. It is a kind of
optimization and efficient search algorithm that simulates the theory of natural selection and
genetics based on the survival of the fittest. Because of its strong ability to solve problems and its
wide adaptability, it has penetrated into various fields of research and engineering in recent years
and achieved good results. Genetic algorithm is a kind of random global search algorithm, which
searches the target space randomly. It treats possible solutions in the problem domain as an
individual or chromosome of a population and encodes each individual in binary notation or
floating-point notation to achieve parameterization of the model, One by one mapped to the
chromosome space chromosomes, repeated genetic groups based on the operation, according to the
predetermined objective function of each individual to be evaluated through the basic genetic
operation process, and repeated iterative optimization of breeding to produce new Generation,
continue to get a better group, while the global parallel search to search for the best individuals in
the optimization group, to obtain the optimal solution to meet the requirements.

2. Genetic algorithm concept and calculation method

Chromosome: When using GA, the problem needs to be solved into a fixed symbol string, each
of which represents a gene. A chromosome represents a solution to a problem. Each chromosome is
called an individual. Population (Population): the total number of chromosomes generated per
generation. A group contains a collection of some solutions to the problem in this generation. ③
Fitness: Each individual corresponds to a solution to a specific problem. The function value
corresponding to each solution is the fitness function. It is an indicator to measure the fitness of the
chromosome to the environment and an objective function that reflects the actual problem.
The process of solving with GA is based on the parameter set of the problem to be solved, a
population is generated randomly, the fitness function and the selection rate are calculated, and
genetic operations such as selection, crossover and mutation are performed. If the iterative
convergence condition is satisfied, the population is the best Individuals, otherwise, the generation
of a new generation of groups to re-genetic operation, reciprocating cycle until the conditions are
met.
The basic genetic operations are: Select, according to a certain probability from the previous
generation to select M individuals as parents, directly copied to the next generation, the
chromosome does not change. One of the most common and simplest choice probabilities is: Ps (xi)
= f (xi) / Eif (xi) where f (xi) is the fitness of the model xi. Crossover (Crossover) is the process of
randomly selecting two individuals from the old population, exchanging genetic information and
generating offspring. Mutation (Mutation) That is, the process of generating new genes, select the
group of individuals (chromosomes), random selection of a bit to be inverted operation. This can
prevent the genetic algorithm converges to the local optimal solution, help to expand the scope of

Copyright © 2018, the Authors. Published by Atlantis Press.

This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/). 138
 Advances in Computer Science Research, volume 78

optimization, and enhance search capabilities. Especially in the latter part of the genetic algorithm,
when the individuals and fitness values in the population are similar, the further evolution of the
population depends on the mutation operation. In summary, the process of finding the optimal
solution using genetic algorithm can be divided into the following steps: 1) coding the problem to
be solved; 2) randomly giving a set of initial solutions X (0) = (X1, X2, ..., Xn); (3) evaluating the
performance of the current group, and calculating fitness f (xi) for each individual x in the current
group x (t); selecting a certain number of solutions from the current solution as the objects of the
genetic operation according to the evaluation result of; genetic operation of the selected solution to
obtain a new set of solutions; return to evaluate the new solution; if the current solution to meet the
requirements or evolutionary process reaches a certain value, the calculation is completed, or
continue. Genetic algorithm in the first few iterations, the emergence of individuals are good and
bad coexist, the fitness is not high, with the increase of the number of iterations, individuals with
high fitness were followed by genetic. Genetic algorithms have many advantages for solving
optimization problems, such as selection of conditions. Especially, GA has high search order and
search has exploratory and self-evolving capabilities.

3. GA application status and prospects

Genetic algorithm is the result of multidisciplinary integration and infiltration, and has developed
into a self-organizing and adaptive integrated technology. As an effective global search method, it
has been widely used in many fields including engineering design, manufacturing, artificial
intelligence, computer science, bioengineering, petroleum exploration, automatic control, social
science, commerce and finance. Oil exploration is mainly used in many fields such as predicting
oilfield production, optimizing oilfield development, optimizing logging interpretation, and
inferring the distribution of formation permeability. Genetic algorithms make up for the lack of
traditional optimization techniques, and show its characteristics and attractiveness in the solution
and application of many problems in the field of oil and gas exploration and development. However,
at present, there are still some shortcomings and deficiencies in both theory and application of
genetic algorithms. In practice, genetic algorithms are often prone to premature convergence and
poor convergence, and genetic algorithms are not universal, In terms of problem solving, they can
not completely replace the existing optimization techniques in a particular field and they all have
their own scope of application. For a particular area, genetic algorithms often can not compare with
algorithms that deal with problems in this area.
The reactive power optimization compensation problem in power system is a multivariable and
multi-constrained mixed nonlinear programming problem. The manipulated variables include
continuous variables (such as node voltage and generator reactive power), and discrete variables
Load tap changer tap gear, compensation capacitor switching group), making the optimization
process is very complicated. However, the traditional linear programming method, non-linear
programming method and mixed integer programming method, which are applied to reactive power
optimization compensation, have the problems of approximate processing of discrete variables and
do not meet the actual requirements of reactive power planning. Therefore, However, there are still
some common limitations that are easy to fall into the local extreme area and can not achieve the
accurate processing of discrete variables. Because of its own characteristics, genetic algorithm is
especially suitable for solving the problem of multi-objective hybrid optimization and has stable
convergence characteristics. It is an advanced global optimization method. Reactive power
optimization is a set of initial solutions under the condition of power system. The genetic algorithm
is limited by various constraints, and its advantages and disadvantages are evaluated by the fitness
function. The low fitness value is abandoned. Only the high fitness value Have the opportunity to
iterate their characteristics to the next round of solutions, through the selection, crossover, mutation
and other genetic operations, the population gradually tends to be the best, and finally get the
optimal solution. The disadvantage of simple genetic algorithm is that it takes a long time to
compute and can easily converge to the local extreme if the population and the genetic algebra are
not large enough. For this reason, an improved genetic algorithm, or a combination of genetic

139
 Advances in Computer Science Research, volume 78

algorithm and other algorithms, is proposed to improve computational speed and computational
accuracy. In this paper, annealing genetic algorithm is used in combination with simulated
annealing selection algorithm, that is to use annealing selection as an individual replacement
strategy to avoid falling into the local optimal solution. The first part, reactive optimization
algorithm has been given earlier. The other part, that is, the goal of real-time control system of
reactive power and voltage, is to use EMS application software to remotely control the transformer
taps and capacitor banks in the substation in the regional dispatch center so as to realize the safe and
economic operation of the regional power grid. As a real-time control system, not only to ensure
good enough optimal solution, but also to ensure optimal solution corresponding to the control of
the program's implementation. Therefore, in addition to solving the problems of theory and
algorithm, a large number of technical problems in project implementation should also be solved,
including handling of constraints, starting methods, dead zone handling, safety considerations and
program implementation.

4. Evolutionary algorithm in genetic algorithm to solve the path optimization problem

specific methods
Identify a specific set of locations to visit and call a certain number of vehicles from the
warehouse to access these locations. So how to choose is the optimal path of the vehicle so that the
vehicles can access each position in a certain order, not only satisfying specific constraints, but also
minimizing the time and the optimal path of the vehicle. I propose an improved genetic algorithm,
the genetic algorithm coding mechanism to natural number coding mechanism, the advantages of
this mechanism is very intuitive. Re-use three replica championship approach to evolutionary
solution. Experiments show that the improved algorithm can better solve the optimal path
The new coding scheme adopted makes coding very straightforward and responds to problems of
all sizes. And to set up a special position, as a genetic position. The order of visiting a specific
location is the required solution (including infeasible solutions). Each locus has its own constraints
that need to be met. Such as location, time of arrival, etc. For example, if the solution of the
problem is: 1: 0 → 2 → 6 → 5 → 0 Route 2: 0 → 8 → 3 → 7 → 1 → 0 When looking for the
optimal solution, the head of the position to be passed by the vehicle And the tail connected together,
so that the chromosome can be encoded as 256173849. The disadvantage of doing so is the lack of a
route demarcation point, but later it is more convenient to cross the operation. As long as the
optimal solution in decoding, in accordance with the order to insert the node to the route, you can
solve it. If you can not meet some time or can not satisfy other constraints, you can insert a new
path from the insertion point. According to the evolutionary thinking, when the route is optimal, the
total cost and time of the visit will be reduced. Therefore, as far as possible, the location will be
inserted into the original route during the evolutionary process.
Through the improved genetic algorithm, we can solve the problem of finding the optimal path
in the process of enterprise, postal delivery and freeway picking. Other industries on the optimal
path of the problem can be solved. Studying the genetic algorithm can help to improve the profits of
the related enterprises, and the enterprises can also improve the better service for the customers, and
the demands of the people can also be met to the greatest extent. Therefore, the promotion and
application of the algorithm is very meaningful.

5. Conclusion
The basic theory of the genetic algorithm is weak, there are many problems need to explore
improvements in practice, its continuous development in recent years and its successful application
in many areas has shown its unique advantages, with the genetic algorithm to further improve And
its development and application in the field of oil and gas exploration and development, its
extensive potential for application will be further tapped and its scope of application will continue
to expand. This will provide a powerful tool for solving some oil exploration and development
problems.

140
 Advances in Computer Science Research, volume 78

Acknowledgements
Shaanxi Provincial Department of Education 2017 Special Scientific Research Program
(17JK0259)
Weinan Teachers College International Cooperation Project on Education Science (17GJHZ17)
Weinan Teachers College 2016 Humanities and Social Science Project (16SYB15)

References
[1] Jin Cong.Heuristic Genetic Algorithm and Its Application [J]. Journal of Numerical Computing
and Computer Applications. 2003 (01)
[2] Yang Xiaohua, Lu Guihua, Yang Zhifeng, Li Jianqiang. Gray code accelerated genetic algorithm
and its theoretical research [J]. Systems Engineering Theory and Practice. 2003 (03)
[3] Zhuang Jian, Wang Sunan.Study on self-adjusting genetic algorithm [J]. Journal of System
Simulation. 2003 (02)
[4] Chen Xiaoping, Yu Shenglin. Improvement of the Cross Strategy of Real Genetic Algorithms
[J] .Journal of Electronics & Information Technology. 2003 (01)
[5] Wen Xinchao, Zhou Linxia, Niu Kai.A genetic algorithm based on hybrid coding [J] .Electronic
Technology, 2003 (01)
[6] Chen Chao, Chen Jialian, Yu Feng. Using bit manipulation of a genetic algorithm program
[J] .Application Research of Computers. 2002 (11)

141