摘要：

蜜獾算法(HBA)是一种新型智能优化算法，通过模拟蜜獾觅食行为进行寻优，具有结构简单且收敛速度快等特点。针对HBA在解决高维复杂问题时收敛精度低、收敛速度慢以及全局寻优能力不足等问题，提出一种多策略改进的蜜獾算法(MSHBA)。设计一种限制反向学习机制，随着算法迭代生成限制反向解更新种群，提高种群质量，加快算法收敛速度，引入自适应权重因子，随着迭代次数的变化调节不同寻优路径上的寻优步长，协调算法不同探索阶段，提升算法稳定性，加快收敛速度，构建一种新的饥饿搜索策略，根据种群能量以及全局最差位置改变寻优路径上的寻优步长，避免算法陷入早熟。基于9个标准测试函数对MSHBA、HBA、鲸鱼优化、哈里斯鹰、单一策略等算法在不同维度上进行仿真实验，结果表明，MSHBA具有更优的稳定性和收敛精度，将算法应用于机械设计优化问题并进行结果比较，MSHBA对比原HBA性能优化了88%，适用于求解高维复杂问题。

关键词: 蜜獾算法, 限制反向学习机制, 自适应权重因子, 饥饿搜索策略, 机械设计

Abstract:

The Honey Badger Algorithm(HBA) is a new type of intelligent optimization algorithm that simulates the foraging behavior of honey badgers. It has the characteristics of a simple structure and fast convergence speed. A multi-strategy improved Honey Badger algorithm(MSHBA) is proposed to address the issues of low convergence accuracy, slow convergence speed, and insufficient global optimization ability of the HBA to solve high-dimensional complex problems. It designs a restricted reverse learning mechanism that updates the population with restricted reverse solutions generated through algorithm iteration, improved population quality, and accelerated algorithm convergence speed. MSHBA introduces adaptive weight factors to adjust the optimization step size for different optimization paths as the number of iterations changes, thus coordinating different exploration stages of the algorithm, improving stability, and accelerating convergence speed; and construct a new hungry search strategy that changes the optimization step size for the optimization path based on population energy and global worst-case position to prevent premature convergence. Based on nine standard test functions, simulation experiments are conducted on the MSHBA, HBA, Whale Optimization, Harris Hawks, and single strategy in different dimensions. The results show that the MSHBA has better stability and convergence accuracy. The algorithm is applied to mechanical design optimization problems and the results are compared. Compared with the original HBA, the MSHBA achieved 88% performance optimization, confirming its suitability for solving high-dimensional complex problems.

Key words: Honey Badger Algorithm(HBA), restricted reverse learning mechanism, adaptive weight factor, hungry search strategy, mechanical design