Depends upon a number of factors including:

1. Computational complexity

2. Number of iterations

3. Mutation factors/speed of particles/ initial weights.

Personally, I have found that Differential Evolution tends to give a more accurate solution to problems if computational complexity is not an issue. Keep in mind that all these solutions are for offline tuning, as online tuning may not be practical solution as far as implementation is concerned.

Agree with Syed Ahmed Raza

There is not a single algorithm which is best for all optimization problems. You should use some popular meta-heuristic optimization algorithms to solve your optimization problem and choose the one which gives the overall minimum (maximum) fitness value.

There are various optimization control techniques for power system oscillation damping such as Genetic Algorithm, Tabu- Search, Simulated Annealing, Differential Evaluation, Partical Swarm Optimization etc. But they have their own advantages and limitations with respect to speed of convergence, dimensionality, complexity, precision. This can overcome by use of hybrid techniques.

Dear Nahid Haque

According to my experience and the results that I obtained, I see that the particle swarm algorithm is the best, and the best among them is hybrid algorithms that use several algorithms to obtain the optimal solution

Best regards

There is a theory called "no free lunch theorem".

It says that there is no 1 algorithm that works best for all purposes. no 1 size fits all.

Most commonly used ones are PSO, GA and ant-colony.

Furthermore, the novelty & contribution of your paper should not be based upon using a new algorithm. I mean you can not claim that "this problem was solved previously with PSO, now we'll solve it with GA". that's not a contribution

Even though the number of optimization techniques are available, hybrid technique is better

Many algorithms exist for this purpose, but it depends on application and system characteristics such as the number of parameters, type of system (linear or non-linear) and etc.

Therefore, it is not possible to comment on a specific algorithm in general.

Generally, each optimization algorithms has its own uniqueness, strengths and weaknesses, no single algorithm is "ALL DO IT" algorithm. It is advisable you employ different algorithms and use your performance metrics to decide which one best suited your set standard(s).

It's hard to tell which algorithm is the best beforehand. I suggest that you initially simplify the problem you intend to solve and then use a number of optimization algorithms in solving it. Thereafter, the results you'll obtain may then help you in determining the most suitable algorithm.

You may refer to my paper

https://www.researchgate.net/publication/339540623_Performance_Analysis_of_Six_Search_Algorithms_in_Minimizing_the_Cost_of_Real_Power_Losses_and_Shunt_Capacitors_Purchase_through_Optimal_Shunt_Capacitors_Sizing

(on optimal shunt capacitors sizing) for guidance. I compared six metaheuristic search algorithms and then selected the best amongst them. Currently, I am using that best algorithm in my thesis.

The most appropriate algorithm to solve an optimization problem is mostly dependent on the problem type. A linear programming problem can be solved efficiently using linear solver such as the Simplex method. A nonlinear convex programming problem can be solved using a convex programming algorithm such as the interior-point method. Similarly, a nonconvex nonlinear programming problem can be solved efficiently using a suitable nonconvex algorithm. However, when a problem is very complex and difficult to identify the nature of the problem then such a problem can be solved using metaheuristic algorithms such as genetic algorithm, particle swarm optimization, differential evolution, etc. In brief, the best optimization algorithm is problem specific.

I think the meta-heuristics algorithm such as GA, PSO, BA, ABC and ACO are appropriate. But, I suggest you to use two algorithm for more assurance.

Depends upon a number of factors including:

1. Computational complexity

2. Number of iterations

3. Mutation factors.

For large scale non linear optimization problems I suggest using interior point method or one of the available solvers that uses this resolution method like MINOS.

State of art algorithmes use interior-point initialized PSO.