2. 1.Content
• Definition
• Basic Algorithm and formulation of DEA
• Implementation in MATLAB
• Introduction to Simplex Algorithm
3. 2.Defination
DEA is easy and population-based algorithm.
It is a direct search algorithm used to optimize
functions globally, including defined and real-
valued design parameters, especially in fully
regulated space.
4. The basic steps of the development-based
algorithm are as follows:
1. Establishment of the initial population
2. Evaluation
REPEAT
3. Mutation
4. Recombination
5. Selection
UNTIL (until the status criterion is available)
6. Differential evolution algorithm is an
evolution algorithm developed especially for
numerical optimization. This algorithm carries
out a new, but not only simple, but also very
effective, a mutation process. In contrast to the
development-based algorithms, such as the
genetic algorithm that runs on the basis of the
previously defined, probability distribution
function, DEA uses a mutation process based
on the difference of randomly selected vector
pairs
7. The simple mutation process used in the DEA improves
the performance of the algorithm and makes it more
robust. Besides this feature;
The advantages of DE are
• its simple structure,
• ease of use,
• speed and robustness,
• DE is one of the best genetic type algorithms for solving
problems with the real valued variables.
• Differential Evolution is a design tool of great utility
that is immediately accessible for practical
applications.
• DE has been used in several science and engineering
applications to discover effective solutions to nearly
intractable problems without appealing to expert
knowledge or complex design algorithms.…..
And so on…
8. 1.Propulation Structure and Parameter Limits
DGA uses a fixed size population. The population is defined in the NPXD floating-point array form. Two sequences of
goal vectors are stored in memory. The first set is used for the present parent population and the other is used to
create the next generation containing new vectors. Initial objective vectors are generated by selecting random
elements from a search space with well defined limitations. In practice, initial parameter limits are usually determined
by physical conditions. If the physical conditions are not determinant then the initial lower and upper parameter limits
should be chosen large enough to cover the regions containing the global optimization.
After using the lower and upper parameter limits for the initial parameters, it can be omitted to allow the algorithm to
conduct research beyond the initial limits. However, limitations are necessary to prevent interminableness, especially
when working with even or close to even surfaces. Initially, if a good solution could be identified or already present,
the algorithm could begin to search for a population of purpose vectors scattered around this nominal solution.
9. 2.Mutation
In terms of parameter optimization, the mutation process corresponds to the addition of a randomly generated
increment on one or more parameters of an existing purpose vector. The goal of a good mutation process is to
produce the increments, ie the steps to move the present vectors in the right direction with the right amount at the
right time. Achieving this goal depends primarily on the characteristics of the distribution that will produce the steps.
The DGA is performed by adding the weighted difference of the pair of purpose vectors to a purpose vector of the
randomly selected goal vectors for a purpose vector.
Briefly, DGA is an approach based on normal development using two simple operations, such as mutation and
recombination, which represent the linear combinations of random purpose vectors for actual parameter
optimization.
10. 3.Selection
The selection process is the criteria that define the conditions under which the newly produced vectors can enter the
population. The development strategy alternatively uses a deterministic selection process in which the best
performing members of the combined parent-child population are allowed to develop. The disadvantage is the cost of
aligning the population according to the quality value at the end of each generation.
11. 4.Recombination (Cross)
The mutation process is mainly responsible for making a population robust and investigating new research regions.
The collation process is a complementary process and its aim is to help the research to be successful by creating a new
vector by using the vector parameters. It increases the cost of the research but gives the research considerable speed.