Adaptive Aquila Optimizer Combining Niche Thought with Dispersed Chaotic Swarm

Yue Zhang, Xiping Xu, Ning Zhang, Kailin Zhang, Weida Dong, Xiaoyan Li2023 · DOI

TL;DR

This paper introduces the Adaptive Aquila Optimizer Combining Niche Thought with Dispersed Chaotic Swarm (NCAAO), a novel meta-heuristic algorithm designed to overcome the limitations of the original Aquila Optimizer (AO). NCAAO enhances AO's global exploration and local exploitation capabilities by incorporating chaotic mapping for population initialization, an adaptive strategy for search preferences, and a niche-based communication mechanism for improved convergence and robustness.

What Problem Does This Paper Address?

The original Aquila Optimizer (AO) is a powerful meta-heuristic algorithm inspired by the hunting behavior of eagles. While it demonstrates strong global exploration capabilities, it suffers from insufficient local exploitation and a tendency to converge slowly or get stuck in local optima. This limits its effectiveness in solving complex optimization problems where fine-tuning solutions is crucial. The authors aim to address these shortcomings by developing a more robust and efficient optimization algorithm.

Methods

The NCAAO algorithm builds upon the AO by introducing three key enhancements:

  • Dispersed Chaotic Swarm Initialization: Instead of random initialization, NCAAO uses the DLCS chaotic mapping to generate the initial population. This method is chosen for its superior uniformity, stability, and ability to traverse the search space more effectively, leading to a better initial exploration state and avoiding premature convergence.
  • Adaptive Adjustment Strategy for De-Searching Preferences: This strategy balances global exploration and local exploitation. It employs an adaptive probability threshold that dynamically adjusts between 0 and 1 throughout the iterations, guiding the algorithm to explore broadly in the early stages and focus on local refinement later. Additionally, an adaptive position weight parameter is introduced to perturb individual positions, enhancing local exploitation.
  • Communication Strategy Based on Niche Thought: Inspired by biological niche theory, this module uses a sharing mechanism to compare individuals within a "habitat." It penalizes individuals with lower fitness, encouraging them to explore other regions, while rewarding high-fitness individuals. This promotes population diversity and accelerates convergence towards the global optimum by facilitating information exchange and selecting elite individuals.

Datasets

The paper evaluates the NCAAO algorithm using:

  • Fifteen standard benchmark functions: These include unimodal, high-dimensional multimodal, and fixed-dimensional multimodal functions, commonly used to test the performance of optimization algorithms.
  • Three engineering design problems:
    • Tension/Compression Spring Design Problem
    • Pressure Vessel Design Problem
    • Three-Bar Truss Design Problem

Key Insights and Ideas

  • Chaotic Initialization is Key: The use of DLCS chaotic mapping for initial population generation significantly improves the uniformity and distribution of individuals, providing a better starting point for the optimization process and aiding in avoiding local optima.
  • Adaptive Strategies are Crucial for Balance: The adaptive adjustment of search preferences and position weights is vital for effectively balancing exploration and exploitation. This dynamic approach allows the algorithm to adapt its strategy based on the current stage of the optimization process.
  • Niche Theory Enhances Convergence: The incorporation of a niche-based communication strategy effectively promotes information exchange among individuals, leading to faster convergence and improved accuracy by focusing on elite solutions.
  • Addressing "Zero Search Preference": The paper highlights the issue of "zero search preference" in AO, where the algorithm might overly favor certain solutions. NCAAO's combined strategies aim to mitigate this, leading to more robust performance.

Key Results

The NCAAO algorithm demonstrated superior performance compared to several well-known meta-heuristic algorithms (GWO, WOA, HHO, ALO, AO) across various benchmark functions and engineering problems.

  • Benchmark Functions: NCAAO achieved competitive or superior results on unimodal, multimodal, and fixed-dimensional multimodal functions, often converging to the global optimum with high accuracy. Statistical analysis using the Wilcoxon rank sum test confirmed the significant advantage of NCAAO in most cases.
  • Engineering Problems: NCAAO showed very competitive results in solving the tension/compression spring design, pressure vessel design, and three-bar truss design problems, often outperforming other algorithms and demonstrating its capability in handling constrained optimization tasks.
  • Convergence: Convergence curves indicated that NCAAO generally converges faster and to better solutions than the compared algorithms, especially on multimodal functions.

Limitations and Future Work

  • Limitations: The paper acknowledges that while NCAAO shows significant improvements, the convergence speed for some functions (like F8) might be slower compared to others. The authors also note that the effectiveness of chaotic mappings can be influenced by their homogeneity.
  • Future Work: The authors plan to investigate the application of NCAAO to problems such as image enhancement, feature extraction, and camera calibration. They also intend to further explore the impact of introducing chaotic mappings for initial population generation on the algorithm's convergence performance.

Summary updated

BibTeX
@misc{zhang2023adaptive,
  author = {Yue Zhang and Xiping Xu and Ning Zhang and Kailin Zhang and Weida Dong and Xiaoyan Li},
  title = {{Adaptive Aquila Optimizer Combining Niche Thought with Dispersed Chaotic Swarm}},
  year = {2023},
  doi = {10.3390/s23020755},
}
Figure