LAUSR

This paper presents an Enhanced PSO (EPSO) method to address the weaknesses associated with the traditional PSO such as high sensitivity to the initial conditions, fast convergence, decline of solutions’ variety, and trapping in the local optimum. For this, several strategies are suggested including segmentation of search space, modification of solution’s updating rule, accepting poor solutions using an intelligent probabilistic function, searching in the regions with poor solutions, gradually removing the regions with poor solutions, and focusing the solutions on the local search after removing all regions with poor solutions. The performance of the EPSO was investigated using the 30 CEC 2014 test functions, 30 CEC 2017 test functions, 10 standard optimization algorithms, and a challenging optimization problem called Probabilistic Load Flow (PLF) in a distribution network. The results of the Wilcoxon signed-rank test on the 2014 and 2017 test functions revealed the superiority of the EPSO over nearly 80% of cases compared to the other algorithms. The obtained results of solving 10 benchmark functions by the proposed EPSO and the other six improved PSO algorithms indicated advantages of the proposed EPSO compared to the other algorithms in finding the optimal value. Meanwhile, the proposed EPSO took the average time of 0.66 s to solve the 10 test functions; it was the shortest time compared to other improved PSO algorithms. According to the results of implementing the EPSO algorithm and other algorithms with random agents for 61 times, in more than 65% of the test functions, the proposed EPSO could find the global optimal solution in a shorter time than the other algorithms. The results of solving the probabilistic load flow problem indicated 89% similarity of the results of the proposed EPSO to those of the most accurate method i.e. Monte Carlo Simulation (MCS). Comparison of the obtained results with the other algorithms as well as outcomes of several improved versions of the PSO indicated the competitive proficiency of the proposed EPSO in various optimization circumstances.