A Systematic Review on Harmony Search Algorithm: Theory, Literature, and Applications

Table 7

Improvement in HSA with hybrid algorithms.

Ref.	Hybrid algorithm	Description	Application	Performance
[51]	SA	PAR is modified by using the cooling concept of SA	Travelling salesman problem	Improve reliability
[52]	PSO	BW is replaced with the concept of global-best particle	Optimization problems	Improve performance
[53]	DPSO	PAR is dynamically updated by using the concept of DPSO	PID controller	Improve performance
[54]	CSA	CSA is employed to improve the harmony vectors in HSA	Fuzzy classification system	Better classification accuracy
[55]	GA, SA, AIS	GA, SA, and AIS are used to improve the solutions stored in HM	Task scheduling	Better convergence, avoid the local optima
[56]	GA, PSO	Genetic mutation and position updating mechanism are used to improve the performance	Reliability problems	Higher exploration capability
[A44]	SQP	SQP is employed to improve the local search	Optimization problems	Improve the effectiveness and robustness of HSA
[57]	FCM	FCM is integrated into HSA for improving the convergence	Clustering	Superior performance
[58]	DE	DE is employed to fine tune the harmony vectors	Optimization problems	Improvement in performance
[59]	Taguchi	Taguchi is used to initialize the harmony memory	Shape optimization	Avoid premature convergence
[60]	NM-SA	NM-SA is employed to enhance the search capability of HSA	Optimization problems	Better exploration capability
[71]	GDL-FLANN	Search capability of GDL is used to optimize the weight of FLANN	Classification	Enhance the classification accuracy
[61]	Fuzzy logic	Fuzzy rules are used to select the best rules in fuzzy-based system	Medical diagnosis	Improve the efficiency
[62]	CS	Mutation operator is employed to improve the harmony in HSA	Optimization problems	Improve the convergence
[63]	ACO	The concepts of ACO are incorporated into HSA	Optimal location of structural dampers	Improve the convergence rate
[65]	FA	HSA is used to mutate the fireflies for escaping the solutions, which are being trapped into local optima	Optimization problems	Less computational cost