A Dynamic Hierarchical Improved Tyrannosaurus Optimization Algorithm with Hybrid Topology Structure

Shihong Zhang,Hu Shi,Baizhong Wang,Chunlu Ma,Qinghua Li

DOI: https://doi.org/10.3390/math12101459

IF: 2.4

2024-05-09

Mathematics

Abstract:Aiming at the problems of the Tyrannosaurus optimization algorithm, of poor search accuracy, insufficient global search capability, and ease of falling into local optimality, a dynamic hierarchical improved Tyrannosaurus optimization algorithm (DHTROA) with hybrid topology structure is proposed. Initially, a chaotic opposition-based learning approach is selected to start the population, ensuring a more uniform distribution of prey across the solution area and boosting population diversity; later, a dynamic hybrid bi-population strategy is introduced to divide the initial population into an 'advantaged group' and a 'disadvantaged group' to improve the efficiency of individual information exchange. Finally, the 'advantaged group' and 'disadvantaged group' are hunted synchronously; for the 'advantaged group', the position update is carried out using the cellular ring topology strategy, and for the 'disadvantaged group', the original algorithm is run in accordance with the main loop process. For the problem of the constant running rate of the Tyrannosaurus in the original algorithm, an adaptive running rate strategy is proposed, which enhances the ability of global optimization, and at the same time, the shortcomings of the original algorithm's 'failure' strategy are improved in order to enhance the original algorithm to jump out of extrema. DHTROA was tested for performance with nine optimization algorithms in different dimensions of the CEC2017 test function. The efficiency of these enhancements was confirmed through the Wilcoxon rank sum test and Friedman test, while DHTROA was utilized for six engineering optimization challenges of differing complexities. The experimental results show that DHTROA has improved greatly in convergence speed, optimality search accuracy, global search ability, and stability, and the excellent engineering optimization performance also proves the excellent robustness of DHTROA.

mathematics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the problems existing in the existing Tyrannosaurus optimization algorithm (TROA), such as poor search accuracy, insufficient global search ability, and being prone to fall into local optimal solutions. To overcome these limitations, the author proposes a dynamic hierarchical improved Tyrannosaurus optimization algorithm (DHTROA) with a hybrid topology structure. Specifically, the paper mainly focuses on the following aspects:

1. **Initialization phase**:
- Use the chaotic opposition - based learning method to initialize the population, so as to ensure that the prey is more evenly distributed in the solution space and enhance the population diversity.

2. **Dynamic mixed dual - population strategy**:
- Divide the initial population into "dominant group" and "inferior group" to improve the efficiency of individual information exchange. The dominant group uses the cellular ring topology strategy for position update, while the inferior group runs according to the main loop process of the original algorithm.

3. **Adaptive running speed**:
- Aiming at the problem of the constant running speed of Tyrannosaurus in the original algorithm, an adaptive running speed strategy is proposed to enhance the global optimization ability, and the "failure" strategy of the original algorithm is improved, so that it can better jump out of the extreme points.

4. **Hunting "failure" improvement**:
- Improve the position update strategy when hunting fails, so as to improve the ability of the algorithm to jump out of local optima.

5. **Cellular ring topology structure**:
- Introduce a cellular ring topology structure guided by the optimal value and far away from the worst solution, so as to reduce the possibility of premature convergence of the algorithm.

Through these improvements, DHTROA has been performance - tested on CEC2017 test functions of different dimensions and compared with nine other optimization algorithms. The experimental results show that DHTROA has been significantly improved in terms of convergence speed, optimization accuracy, global search ability and stability, and in particular, it has shown excellent robustness in engineering optimization problems.