We study the multi-objective route planning problem of an unmanned air vehicle (UAV) moving in a continuous terrain. In this problem, the UAV starts from a base, visits all targets and returns to the base in a continu...
详细信息
We study the multi-objective route planning problem of an unmanned air vehicle (UAV) moving in a continuous terrain. In this problem, the UAV starts from a base, visits all targets and returns to the base in a continuous terrain that is monitored by radars. We consider two objectives: minimizing total distance and minimizing radar detection threat. This problem has infinitely many Pareto-optimal points and generating all those points is not possible. We develop a general preference-based multi-objectiveevolutionary algorithm to converge to preferred solutions. Preferences of a decision maker (DM) are elicited through reference point(s) and the algorithm converges to regions of the Pareto-optimal frontier close to the reference points. The algorithm allows the DM to change his/her reference point(s) whenever he/she so wishes. We devise mechanisms to prevent the algorithm from producing dominated points at the final population. We also develop mechanisms specific to the UAV route planning problem and test the algorithm on several UAV routing problems as well as other well-known problem instances. We demonstrate that our algorithm converges to preferred regions on the Pareto-optimal frontier and adapts to changes in the reference points quickly. (C) 2019 Elsevier Ltd. All rights reserved.
In software industry, a common problem that the companies face is to decide what requirements should be implemented in the next release of the software. This paper aims to address the multi-objective next release prob...
详细信息
In software industry, a common problem that the companies face is to decide what requirements should be implemented in the next release of the software. This paper aims to address the multi-objective next release problem using search based methods such as multi-objectiveevolutionary algorithms for empirical studies. In order to achieve the above goal, a requirement-dependency-based multi-objective next release model (MONRP/RD) is formulated firstly. The two objectives we are interested in are customers' satisfaction and requirement cost. A popular multi-objectiveevolutionary approach (MOEA). NSGA-II, is applied to provide the feasible solutions that balance between the two objectives aimed. The scalability of the formulated MONRP/RD and the influence of the requirement dependencies are investigated through simulations as well. This paper proposes an improved version of the multi-objective invasive weed optimization and compares it with various state-of-the-art multi-objective approaches on both synthetic and real-world data sets to find the most suitable algorithm for the problem.
Coastal aquifer management (CAM) considering conjunctive optimization of pumping and injection system for seawater intrusion (SI) mitigation poses significant decision-making challenges. CAM needs to pose multiple obj...
详细信息
Coastal aquifer management (CAM) considering conjunctive optimization of pumping and injection system for seawater intrusion (SI) mitigation poses significant decision-making challenges. CAM needs to pose multiple objectives and massive decision variables to explore tradeoff strategies between the conflicting resources, economic, and environmental requirements. Here, we investigate a joint artificial injection scheme for ameliorating SI by establishing an evolutionarymulti-objective decision-making framework that combines simulation-optimization (S-O) modelling with a cost-benefit analysis, and demonstrate the framework on a large-scale CAM case in Baldwin County, Alabama. First, a SI numerical model, using SEAWAT, was configured to predict the vulnerable region as an SI encroachment area with the scenarios of minimum and maximum pumping capacity. As a result, a smaller number of candidate sites were selected in the SI encroachment area for implementing groundwater injection to avoid the computationally infeasible SI optimization with an inordinate number of injection related decision variables. Second, the effective S-O methodology of niched Pareto tabu search combined with a genetic algorithm (NPTSGA), which considers the moving-well option, was applied to discover optimal pumping/injection (P/I) strategies (including P/I rates and injection well locations) between three conflicting management objectives under complicated SI constraints. Third, for practical operation of the P/I schemes, a cost-benefit analysis provides judgment criteria to allow decision-makers to implement more sustainable P/I strategies to capture the different realistic preferences. The implementation of three extreme optimization solutions for the case study indicates that, compared to the initial unoptimized scheme, a maximum increase of a factor of 3 in groundwater extraction rates, a maximum reduction of 17% in extent of SI, and a maximum 82.3 million US dollars in comprehensive benefits
Inspired by the mechanism of generation and restriction among five elements in Chinese traditional culture, we present a novel multi-objective Five-Elements Cycle optimization algorithm (MOFECO). During the optimizati...
详细信息
Inspired by the mechanism of generation and restriction among five elements in Chinese traditional culture, we present a novel multi-objective Five-Elements Cycle optimization algorithm (MOFECO). During the optimization process of MOFECO, we use individuals to represent the elements. At each iteration, we first divide the population into several cycles, each of which contains several individuals. Secondly, for every individual in each cycle, we judge whether to update it according to the force exerted on it by other individuals in the cycle. In the case of an update, a local or global update is selected by a dynamically adjustable probability Ps;otherwise, the individual is retained. Next, we perform combined mutation operations on the updated individuals, so that a new population contains both the reserved and updated individuals for the selection operation. Finally, the fast non-dominated sorting method is adopted on the current population to obtain an optimal Pareto solution set. The parameters' comparison of MOFECO is given by an experiment and also the performance of MOFECO is compared with three classic evolutionary algorithms Non-dominated Sorting Genetic Algorithm II (NSGA-II), multi-objective Particle Swarm optimization algorithm (MOPSO), Pareto Envelope-based Selection Algorithm II (PESA-II) and two latest algorithms Knee point-driven evolutionary Algorithm (KnEA) and Non-dominated Sorting and Local Search (NSLS) on solving test function sets Zitzler et al's Test suite (ZDT), Deb et al's Test suite (DTLZ), Walking Fish Group (WFG) and Many objective Function (MaF). The experimental results indicate that the proposed MOFECO can approach the true Pareto-optimal front with both better diversity and convergence compared to the five other algorithms.
Estimation of Distribution Algorithm (EDA) is a kind of new evolutionary algorithm which updates and samples from probabilistic model in evolutionary computation. Recently it is used to solve multi-objective problems....
详细信息
ISBN:
(纸本)9781424467129
Estimation of Distribution Algorithm (EDA) is a kind of new evolutionary algorithm which updates and samples from probabilistic model in evolutionary computation. Recently it is used to solve multi-objective problems. The key is how to construct probability model suitable for real distribution and how to keep diversity of solutions. In this paper a new multi-objectiveevolutionary of distribution algorithm using kernel density estimation model is presented. It used kernel density estimation method to obtain probability density of samples and generate new population with stochastic universal sampling method. In order to get pareto front of multi-objective problems, fitness sharing method is used. 5 bi-objective test problems are selected to test the performance of the new algorithm. The results show that multi-objectiveevolutionary of distribution algorithm using kernel density estimation model has better suitable performance for test problems comparing with non-dominated sorting genetic algorithm II, multi-objective particle swarm optimization and multi-objective estimation of distribution algorithm.
This paper studies the possibility to use efficient multi-modal optimizers for multi-objectiveoptimization. In this paper, the application area considered for such new approach is the optimal dispatch of energy sourc...
详细信息
ISBN:
(纸本)9781627483889
This paper studies the possibility to use efficient multi-modal optimizers for multi-objectiveoptimization. In this paper, the application area considered for such new approach is the optimal dispatch of energy sources in smart microgrids. The problem indeed shows a non uniform Pareto front and requires efficient optimal search methods. The idea is to exploit the potential of agents in population-based heuristics to improve diversity in the Pareto front, where solutions show the same rank and are thus equally weighted. Since Pareto dominance is at the basis of the theory of multi-objectiveoptimization, most algorithms show the non dominance ranking as quality indicator, with some problem in finding sufficiently diverse solutions. Other algorithms, such as the Indicator Based evolutionary Algorithm, use most commonly the Hypervolume indicator which also intrinsically shows diversity preserving problems. In this paper, the Glow-worm swarm optimizer is used as multimodal optimization method over a set of solutions ordered based on non dominance. After the introduction of this algorithm, its multiobjective implementation is briefly outlined. Then some tests are carried out on test functions taken from the literature giving quite encouraging results. Finally, the problem of optimal energy dispatch in smart microgrids is described and different applications are shown comparing the results with those obtained emplying the Non Dominated Sorting Genetic Algorithm II.
暂无评论