8 Genetic Algorithm Variants
This chapter covers
- Introducing Gray-coded genetic algorithm
- Understanding real-valued GA and its genetic operators
- Understanding permutation-based GA and its genetic operators
- Understanding multi-objective optimization
- Adapting genetic algorithms to strike a balance between exploration and exploitation
- Solving continuous and discrete problems using GA
This chapter serves as an extension of the prior chapter, and we will introduce various forms of genetic algorithms (GAs) and delve deeper into their real-world applications. A number of case studies and exercises are included such as the traveling salesman problem, Proportional Integral Derivative (PID) controller design, political districting, cargo bike loading problem, manufacturing planning, facility allocation, and the opencast mining problem.
8.1 Gray-coded GA
In binary genetic algorithms, the crossover and mutation operations can significantly impact the solution, especially when the bits that are to be changed are among the most significant bits in the binary string leading to premature convergence due to the Hamming cliff effect. This effect refers to the fact that small changes in the chromosome can result in large changes in the fitness, which can lead to a sharp drop-off in the fitness landscape and cause the algorithm to converge prematurely. To mitigate the Hamming cliff effect, Gray-code GA uses a Gray-code encoding scheme for the chromosomes.