Cluster intelligent optimization algorithm
Overview
Algorithm essence
The optimization algorithm is a traversal in a given direction
The advantage of intelligent theoretical algorithm is that it does not require prior knowledge, does not need to analyze the internal laws and internal relations of the data, only needs to learn the data itself, self-organize and self-adaptively complete the optimization problem solving, the disadvantage is the convergence of this model The speed is often extremely slow, and it will occupy a lot of computer running memory when processing a large number of sample data, which has high requirements on the running speed of the computer.
Definition The
swarm intelligence optimization algorithm mainly simulates the group behavior of insects, animals, birds and fish. These groups find food in a cooperative way. Each member of the group learns its own experience and the experience of other members. To constantly change the direction of the search. Any algorithm or distributed problem-solving strategy inspired by insect groups or other animal social behavior mechanisms belongs to swarm intelligence (Swarm Intelligence)
Vernacular explanation
imitating the foraging or other behaviors of insects or some animals. These animal groups search for food in a cooperative way, constantly exchange food information, and can quickly find more food. An algorithm abstracted from the study of their behavior is the swarm intelligence optimization algorithm. (For example, if an individual finds food, it will notify other individuals to come to this place with food. This is a behavior)
Algorithm principle
- Proximity principle: the group can perform simple space and time calculations;
- Quality principle: the group can respond to quality factors in the environment;
- Diversity response principle: the scope of group action should not be too narrow;
- The principle of stability: the group should not change its behavior every time the environment changes;
- The principle of adaptability: The group can change its behavior at the appropriate time when the cost is not too high.
Genetic Algorithm (GA)
The global search ability is strong, the local search ability is weak, and often only sub-optimal solutions can be obtained instead of optimal solutions.
Studies have found that genetic algorithms can reach more than 90% of the optimal solution at a very fast speed, but it takes a long time to reach the true optimal solution, that is, the local search ability is insufficient.
Particle Swarm Optimization (PSO)
Produce premature convergence and prove that the algorithm is not globally convergent.
Unweighted-fast convergence speed but easy to fall into local optimal solution
Ant Colony Algorithm
The parameter setting is complicated. If the parameter setting is improper, it is easy to deviate from the high-quality solution.
Simulated annealing algorithm
Global optimization, suitable for collocation with particle swarms, whale optimization algorithms, etc. that are easy to fall into local optimal solutions
Fish school algorithm
The parameter setting is complicated. If the parameter setting is improper, it is easy to deviate from the high-quality solution. (Same as ant colony)
Whale optimization algorithm
The algorithm is trapped in local extremum and convergence speed. The algorithm is trapped
in a local optimal solution.
Combined with the simulated annealing algorithm, the global optimization capability is improved by accepting the inferior points. The
convergence speed is slow.
Adaptive weights are introduced to improve the local optimization capability of the algorithm.
(The following is not summarized)
Leapfrog Algorithm
The idea of the leapfrog algorithm is: a group of frogs live in a wetland. Many rocks are scattered in the wetland, and frogs jump to find places with more food by searching for different rocks. Each frog individual realizes the exchange of information through cultural exchanges. Each frog has its own culture. The culture of each frog is defined as a solution to the problem.
The entire frog population in the wetland is divided into different sub-groups, each sub-group has its own culture and executes a local search strategy. Each individual in the subgroup has its own culture, and influences other individuals, is also influenced by other individuals, and evolves with the evolution of the subgroup. When the sub-groups evolve to a certain stage, each sub-group exchanges ideas (global information exchange) to realize the mixed operation between the sub-groups, until the set conditions are met.
Artificial bee colony algorithm
Artificial bee colony algorithm is an optimization method proposed by imitating the behavior of bees. It is a specific application of the idea of cluster intelligence. Its main feature is that it does not need to know the special information of the problem. It only needs to compare the pros and cons of the problem. The local optimization behavior of individual worker bees finally makes the global optimal value stand out in the group, which has a faster convergence speed.