A Survey of Evolutionary Continuous Dynamic Optimization Over Two Decades—Part A

Abstract

Please refer to published version for the most recent bibliographic citation information. If a published version is known of, the repository item page linked to above, will contain details on accessing it.

A Survey of Evolutionary Continuous Dynamic Optimization Over Two Decades—Part B

“…(i) A generated landscape by (PB-5). The height (uniformly randomized in [30,70]), width (uniformly randomized in [1,12]), and location of all peak center positions (uniformly randomized with the same random seed numbers in [−50, 50] for (P B -2), (P B -3), (P B -4) and in [−5, 5] for (P B -1) and (P B -5)) are similar in all landscapes.…”

“…PMPB works based on a pendulum length parameter pl and a direction parameter dir. It first generates pl environments f (1) , f (2) , • • • , f (pl) . Then, it archives the parameters of these environments for future reappearances.…”

“…The organization of the rest of this paper is as follows. The commonly used benchmark problems in the DOP liter-ature are reviewed in Section ( §) P B -II 1 . The performance indicators and plots used for analyzing performance of DOAs are reviewed in § P B -III.…”

“…Generated landscapes with five peaks by the baseline functions (P B -1), (P B -2), (P B -3) with β = 0, (P B -4), and (P B -5). Each landscape's contour plot in the left column is shown in the same row of the right column.The height (uniformly randomized in[30,70]), width (uniformly randomized in[1,12]), and location of all peak center positions (uniformly randomized with the same random seed numbers in [−50, 50] for (P B -2), (P B -3), (P B -4) and in [−5, 5] for (P B -1) and (P B -5)) are similar in all landscapes.…”

A special point-based transfer component analysis for dynamic multi-objective optimization

“…In recent years, many studies have made great achievements by taking evolutionary algorithms and swarm intelligence algorithms as the common frameworks in DMOPs [1,2,17,18,21]. According to the response strategy after changing, the existing algorithms can be divided into four categories: diversity-based, memory-based, multi-populationbased, and prediction-based method [1].…”