Searching With Direction Awareness: Multi-Objective Genetic Algorithm Based on Angle Quantization and Crowding Distance MOGA-AQCD

Abstract

Multi-objective optimization (MOO) is widely used for solving various engineering real-life problems. Meta-heuristic optimization has been regarded as an effective solution for such problems because it enables the successful examination of a broad range of candidate solutions and the selection of optimal ones. However, there is a high probability of the algorithms becoming ensnared in local minima due to the complex optimization surface and the unlimited number of viable solutions. Therefore, to provide the decision maker with the optimal non-dominated set of solutions, significant improvements must be made to the search process, where the efficient exploration of the population has a vital role in maintaining a good non-dominated solution in evolutionary algorithms. NSGA-II is regarded as the state of the art for the meta-heuristic MOO. NSGA-II and its variants have adopted the concept of crowding distance as a measure that can leverage the characteristics of the distribution of solutions in the search space and provide a highlevel of exploration. However, this method is not sufficient to effectively explore the search space because it ignores the direction of the exploration. In this paper, the angle quantization of solutions is combined with the crowding distance to create the MOGA-AQCD algorithm, which preserves equal exploration in all directions and aims at finding equal distribution of solutions within the search space. This approach is applied to a set of standard benchmark MOO functions. The results show that MOGA-AQCD is superior to NSGA-II and NSGA-III on the most evaluation measures for MOO.

Hyper-Angle Exploitative Searching for Enabling Multi-Objective Optimization of Fog Computing

“…This section presents a hyper angle exploitive searching HAES algorithm. Firstly, we present its working principle and the difference between HAES and MOGA-AQCD [ 30 ] in Section 3.2.1 . Secondly, we present the objective partitioning in Section 3.2.2 .…”

“…Then, from previous approaches, the concept of crowd distance, when combined with angle searching, achieves the extensive scope of the search. Specifically, authors in [ 30 ] have used range angle as a criterion to balance the search, then using it in finding criterion solutions as the goal of the study.…”

A Multi-Objective Optimization Model Using Improved NSGA-II for Optimizing Metal Mines Production Process

“…The hypervolume reflects the approximation degree of the Pareto optimal solutions to the true Pareto optimal front. The hypervolume can evaluate the performance of convergence and diversity simultaneously [38], [39]. The greater the hypervolume is, the better the convergence and diversity performance will be.…”

Shared-Mode Resource Allocation for Cloud-Based Load Testing

“…The more optimized dominator plans are ranked at higher levels. Then, we adopt the crowding distance [33] (the crowding distance of a resource allocation plan is the sum of the differences between its two adjacent allocation plans on each objective function) to sort the resource allocation plans at the same level. The larger crowding distance, the better for ensuring the diversity of the population.…”