Success-history based parameter adaptation is very useful.

if an algorithm proceeds by making a sequence of decisions, and if an unsatisfactory solution is obtained, starts again with a new sequence of decisions then it is likely that the algorithm can be improved by adding a method for improving the sequence of decisions as described in System and method for effecting optimization of a sequential arrangement of …US US8417652B2

Don't worry it is admissible to use a patented method for research.

and also described in Genetic operators for sequencing problems

BR Fox, MB McMahon

Foundations of genetic algorithms 1, 284-300

I use a a carefully crafted variation of this method (in all of my work) that is substantially better than one and two point, order and location preserving genetic algorithms.

The attached pdf file illustrates the comparative value of this method.

Let me know if you would like to discuss.

you can see the modification in following papers:

A Modified Grey Wolf Optimizer by Individual Best Memory and Penalty Factor for Sonar and Radar Dataset Classification

Improved whale trainer for sonar datasets classification using neural network

Design and implementation of a neighborhood search biogeography-based optimization trainer for classifying sonar dataset using multi-layer perceptron neural network

Chaotic fractal walk trainer for sonar data set classification using multi-layer perceptron neural network and its hardware implementation

Improved migration models of biogeography-based optimization for sonar dataset classification by using neural network

https://www.researchgate.net/publication/353040925_Balancer_Genetic_Algorithm-A_Novel_Task_Scheduling_Optimization_Approach_in_Cloud_Computing

By using the dynamic FDB selection method, you can transform meta-heuristic search algorithms into a more powerful and efficient method. For this, you should analyze the guide selection process in the meta-heuristic search algorithm and integrate the dynamic FDB selection method into this process. In the uploaded source codes, we applied the dFDB method in the guide selection process of the MRFO algorithm and achieved a unique improvement in the performance of the MRFO. If you read the article that introduces the source codes and the dFDB-MRFO algorithm, you can redesign different algorithms with the dFDB selection method and improve their performance.

Manta Ray Foraging Optimizer has been redesigned using the dFDB method, and thus the dFDB-MRFO algorithm has been developed with improved search performance. dFDB-MRFO is an up-to-date and powerful meta-heuristic search algorithm that can be used to solve single-objective optimization problems.

FDB Selection Method: Fitness Distance Balance was first introduced in the following link:

https://se.mathworks.com/matlabcentral/fileexchange/72311-fdb-sos

Kahraman, H. T., Bakir, H., Duman, S., Kati, M., Aras, S., Guvenc, U. Dynamic FDB selection method and its application: modeling and optimizing of directional overcurrent relays coordination. Appl Intell (2021). https://doi.org/10.1007/s10489-021-02629-3

https://www.mathworks.com/matlabcentral/fileexchange/96113-dfdb-mrfo-a-powerful-meta-heuristic-optimization-algorithm?s_tid=srchtitle

Sometimes modifying how the algorithm handles the population can considerably improve the performance of an algorithm. I have tried this in my paper attached. We used the population handling mechanism of SCE-UA algorithm with the GWO (with some tweaks here and there). However, the result was a more complex algorithm and required more attention to its tuning.

In addition to the suggested answers, you can try the concept of depth search proposed in https://link.springer.com/article/10.1007/s00500-021-06062-z

This solution is a good way to balance between exploration and exploitation.

Radhwan ALI ABDULGHANI Saleh

Thanks and can you share this code

Use hybrid Meta-heuristic by adapting the parameter settings of the metaheuristic method by another one, but there are numerous papers did this

Premature convergence is a challenge for several population-based or weak optimization algorithms,

some general strategies to reduce probability of its occurrence may be listed as:

- applying larger population size

- limiting the use of "Greedy Selection" operator in the algorithm main loop.

- frequent re-initiation of the population (in partial or full), note that the best-so-far individual be saved beside the main population in order not to loose it.

such an strategy can be easily applied in the framework given in the following link as it starts from a given population

https://www.mathworks.com/matlabcentral/fileexchange/105315-escaping-bird-search-for-constrained-optimization

- employing special operators that search diverse parts of design spaces (such as OBL)

- ensemble or sequential application of various algorithms in a unified framework

- check diversity variation vs.iterations on test problems to compare algorithms regarding better diversity maintenance behavior, refer to: https://www.researchgate.net/project/Escaping-bird-search-EBS

Integrate in your metaheuristics the tangent flight operator

X=X+tan(rand*pi) % large flight

or

X=X+ sign(rand-0.5)*tan(rand*pi/2.5) % small flight

This operator has great capacity of exploitation

Layeb, Abdesslem. "Tangent search algorithm for solving optimization problems." Neural Computing and Applications (2022): 1-32.

https://www.mathworks.com/matlabcentral/fileexchange/89897-tangent-search-algorithm-for-solving-optimization-problem?s_tid=srchtitle

https://ieeexplore.ieee.org/document/8746627

Parallel mirror technique for opposition based strategy

It isn't necessarily an operator per se, but you can try using "reinforcement learning" to better balance algorithmic exploration and exploitation. This article demonstrated its successful use in challenging optimization problems: https://www.sciencedirect.com/science/article/abs/pii/S0950705121003075

I consider the question to be meaningless. According to the No Free Lucnh Theorem (Wolpert and Macready, 1997), there is no search algorithm that performs better than all other algorithms when applied to all possible optimization problems. This implies that, for certain problems, a search algorithm (with its respective operators) will perform well, but for other problems it will perform poorly. So there is no single answer to your question since it depends on what kind of problem you are trying to solve and the algorithm you use.

Wolpert, D. H. and Macready, W. G. (1997) ‘No free lunch theorems for optimization’, IEEE Transactions on Evolutionary Computation, 1(1), pp. 67–82. doi: 10.1109/4235.585893.

There may be, recently I've been working on a weighting method that does improve performance in some heuristically intelligent algorithms.

By using the dynamic FDB selection method, you can transform meta-heuristic search algorithms into a more powerful and efficient method. For this, you should analyze the guide selection process in the meta-heuristic search algorithm and integrate the dynamic FDB selection method into this process. In the uploaded source codes, we applied the dFDB method in the guide selection process of the MRFO algorithm and achieved a unique improvement in the performance of the MRFO. If you read the article that introduces the source codes and the dFDB-MRFO algorithm, you can redesign different algorithms with the dFDB selection method and improve their performance.

Manta Ray Foraging Optimizer has been redesigned using the dFDB method, and thus the dFDB-MRFO algorithm has been developed with improved search performance. dFDB-MRFO is an up-to-date and powerful meta-heuristic search algorithm that can be used to solve single-objective optimization problems.

paper link: https://link.springer.com/article/10.1007/s10489-021-02629-3

paper link: https://rdcu.be/cqFmD

https://www.mathworks.com/matlabcentral/fileexchange/96113-dfdb-mrfo-a-powerful-meta-heuristic-optimization-algorithm?s_tid=srchtitle

To address the problem that metaheuristic algorithms easily fall into local optimum, the article "Improved Harris hawks optimization for non-convex function optimization and design optimization problems" (DOI: 10.1016/ j.matcom.2022.09.010) proposed a novel mutation strategy based on Brownian motion, and the experimental results show that the strategy can effectively help the algorithm jump out of the local optimum