Particle swarm optimization (PSO) is a population-based evolutionary technique. Advancements in the PSO development over the last decade have made it one of the most promising optimization algorithms for a wide range of complex engineering optimization problems which traditional derivative-based optimization techniques cannot handle. The most attractive features of PSO are its algorithmic simplicity and fast convergence. However, PSO tends to suffer from premature convergence when applied to strongly multi-modal optimization problems. This paper proposes a method of incorporating a real-valued mutation (RVM) operator into the PSO algorithms, aimed at enhancing global search capability. Three variants of PSO algorithms are considered. The resultant hybrid PSO-RVM algorithms are experimentally investigated along with the PSO variants and an existing PSO with Gaussian mutation using six typical benchmark functions. It is interesting to see that the effectiveness of RVM varies for different PSO variants as well as different kinds of functions. It has been found that one of the hybrid algorithms, CBPSO-RVM, which is an integration of the PSO with the constriction factor and inertia weight (CBPSO) and the RVM operator, exhibits significantly better performance in most of the test cases compared to the other algorithms under consideration. Furthermore, this algorithm is superior to most of the existing algorithms used in this study when applied to two practical ED problems with non-smooth cost function considering the multiple fuel type and/or valve-point loading effects.