Towards faster convergence of evolutionary multi-criterion optimization algorithms using Karush Kuhn Tucker optimality based local search

Evolutionary multi-criterion optimization (EMO) algorithms emphasize non-dominated and less crowded solutions in a population iteratively until the population converges close to the Pareto optimal set. During the search process, non-dominated solutions are differentiated only by their local crowding or contribution to hypervolume or using a similar other metric. Thus, during evolution and even at the final iteration, the true convergence behavior of each non-dominated solutions from the Pareto optimal set is unknown. Recent studies have used Karush Kuhn Tucker (KKT) optimality conditions to develop a KKT Proximity Measure (KKTPM) for estimating proximity of a solution from Pareto optimal set for a multi-objective optimization problem. In this paper, we integrate KKTPM with a recently proposed EMO algorithm to enhance its convergence properties towards the true Pareto optimal front. Specifically, we use KKTPM to identify poorly converged non-dominated solutions in every generation and apply an achievement scalarizing function based local search procedure to improve their convergence. Assisted by the KKTPM, the modified algorithm is designed in a way that maintains the total number of function evaluations as low as possible while making use of local search where it is most needed. Simulations on both constrained and unconstrained multi- and many objectives optimization problems demonstrate that the hybrid algorithm significantly improves the overall convergence properties. This study brings evolutionary optimization closer to mainstream optimization field and should motivate researchers to utilize KKTPM measure further within EMO and other numerical optimization algorithms.     

A soft-computing Pareto-based meta-heuristic algorithm for a multi-objective multi-server facility location problem

In this paper, a novel multi-objective location model within multi-server queuing framework is proposed, in which facilities behave as M/M/m queues. In the developed model of the problem, the constraints of selecting the nearest-facility along with the service level restriction are considered to bring the model closer to reality. Three objective functions are also considered including minimizing (I) sum of the aggregate travel and waiting times, (II) maximum idle time of all facilities, and (III) the budget required to cover the costs of establishing the selected facilities plus server staffing costs. Since the developed model of the problem is of an NP-hard type and inexact solutions are more probable to be obtained, soft computing techniques, specifically evolutionary computations, are generally used to cope with the lack of precision. From different terms of evolutionary computations, this paper proposes a Pareto-based meta-heuristic algorithm called multi-objective harmony search (MOHS) to solve the problem. To validate the results obtained, two popular algorithms including non-dominated sorting genetic algorithm (NSGA-II) and non-dominated ranking genetic algorithm (NRGA) are utilized as well. In order to demonstrate the proposed methodology and to compare the performances in terms of Pareto-based solution measures, the Taguchi approach is first utilized to tune the parameters of the proposed algorithms, where a new response metric named multi-objective coefficient of variation (MOCV) is introduced. Then, the results of implementing the algorithms on some test problems show that the proposed MOHS outperforms the other two algorithms in terms of computational time.     

Two parameter tuned multi-objective evolutionary algorithms for a bi-objective vendor managed inventory model with trapezoidal fuzzy demand

This paper presents a bi-objective vendor managed inventory (BOVMI) model for a supply chain problem with a single vendor and multiple retailers, in which the demand is fuzzy and the vendor manages the retailers’ inventory in a central warehouse. The vendor confronts two constraints: number of orders and available budget. In this model, the fuzzy demand is formulated using trapezoidal fuzzy number (TrFN) where the centroid defuzzification method is employed to defuzzify fuzzy output functions. Minimizing both the total inventory cost and the warehouse space are the two objectives of the model. Since the proposed model is formulated into a bi-objective integer nonlinear programming (INLP) problem, the multi-objective evolutionary algorithm (MOEA) of non-dominated sorting genetic algorithm-II (NSGA-II) is developed to find Pareto front solutions. Besides, since there is no benchmark available in the literature to validate the solutions obtained, another MOEA, namely the non-dominated ranking genetic algorithms (NRGA), is developed to solve the problem as well. To improve the performances of both algorithms, their parameters are calibrated using the Taguchi method. Finally, conclusions are made and future research works are recommended.     

Multi-objective optimization algorithms applied to the class integration and test order problem

In the context of object-oriented software, a common problem is the determination of test orders for the integration test of classes, known as the class integration and test order (CITO) problem. The existing approaches, based on graphs, usually generate solutions that are sub-optimal, and do not consider the different factors and measures that can affect the construction of stubs. To overcome this limitation, solutions based on genetic algorithms (GA) have presented promising results. However, the determination of a cost function, which is able to generate the best solution, is not always a trivial task, mainly for complex systems. Therefore, to better represent the CITO problem, we introduce, in this paper, a multi-objective optimization approach, to generate a set of good solutions that achieve a balanced compromise between the different measures (objectives). Three different multi-objective optimization algorithms (MOA) were implemented: Pareto ant colony, multi-objective Tabu search and non-dominated sorting GA. The approach is applied to real programs and the obtained results allow comparison with the simple GA approach and evaluation of the different MOA.     

改进的约束优化多目标遗传算法及工程应用

利用多目标法处理约束条件,提出一种改进的基于多目标优化的遗传算法用于求解约束优化问题。该算法将约束优化问题转化为两个目标的多目标优化问题; 利用庄家法构造非劣个体,将种群分为支配子种群和非支配子种群,以一定概率分别从支配子种群和非支配子种群中选择个体进行算术交叉操作,引导个体逐步向极值点靠近,增强算法的局部搜索能力,对非支配子种群进行多样性变异操作。8个标准测试函数和3个工程应用的仿真实验结果表明了该算法的有效性。     

A self-organized speciation based multi-objective particle swarm optimizer for multimodal multi-objective problems

This paper proposes a self-organized speciation based multi-objective particle swarm optimizer (SS-MOPSO) to locate multiple Pareto optimal solutions for solving multimodal multi-objective problems. In the proposed method, the speciation strategy is used to form stable niches and these niches/subpopulations are optimized to search and maintain Pareto-optimal solutions in parallel. Moreover, a self-organized mechanism is proposed to improve the efficiency of the species formulation as well as the performance of the algorithm. To maintain the diversity of the solutions in both the decision and objective spaces, SS-MOPSO is incorporated with the non-dominated sorting scheme and special crowding distance techniques. The performance of SS-MOPSO is compared with a number of the state-of-the-art multi-objective optimization algorithms on fourteen test problems. Moreover, the proposed SS-MOSPO is also employed to solve a real-life problem. The experimental results suggest that the proposed algorithm is able to solve the multimodal multi-objective problems effectively and shows superior performance by finding more and better distributed Pareto solutions.     

A bi-objective continuous review inventory control model: Pareto-based meta-heuristic algorithms

In this paper, a bi-objective multi-product (r,Q) inventory model in which the inventory level is reviewed continuously is proposed. The aim of this work is to find the optimal value for both order quantity and reorder point through minimizing the total cost and maximizing the service level of the proposed model simultaneously. It is assumed that shortage could occur and unsatisfied demand could be backordered, too. There is a budget limitation and storage space constraint in the model. With regard to complexity of the proposed model, several Pareto-based meta-heuristic approaches such as multi-objective vibration damping optimization (MOVDO), multi-objective imperialist competitive algorithm (MOICA), multi-objective particle swarm optimization (MOPSO), non-dominated ranked genetic algorithm (NRGA), and non-dominated sorting genetic algorithm (NSGA-II) are applied to solve the model. In order to compare the results, several numerical examples are generated and then the algorithms were analyzed statistically and graphically.     

多目标资源优化分配问题的Memetic 算法

针对传统算法求解多目标资源优化分配问题收敛慢、Pareto解不能有效分布在Pareto 前沿面的问题, 提出一种新的Memetic 算法. 在遗传算法的交叉算子中引入模拟退火算法, 加强了遗传算法的局部搜索能力, 加快了收敛速度. 为了使Pareto 最优解均匀分布在Pareto 前沿面, 在染色体编码中引入禁忌表, 增加了种群的多样性, 避免了传统遗传算法后期Pareto 解集过于集中的缺点. 通过与已有的遗传算法、蚁群算法、粒子群算法进行比较, 仿真实验表明了所提出算法的有效性, 并分析了禁忌表长度和模拟退火参数对算法收敛性的影响.

     

Comparisons of some improving strategies on MOPSO for multi-objective (r, Q) inventory system

This paper presents comparisons of some recent improving strategies on multi-objective particle swarm optimization (MOPSO) algorithm which is based on Pareto dominance for handling multiple objective in continuous review stochastic inventory control system. The complexity of considering conflict objectives such as cost minimization and service level maximization in the real-world inventory control problem needs to employ more exact optimizers generating more diverse and better non-dominated solutions of a reorder point and order size system. At first, we apply the original MOPSO employed for the multi-objective inventory control problem. Then we incorporate the mutation operator to maintain diversity in the swarm and explore all the search space into the MOPSO. Next we change the leader selection strategy used that called geographically-based system (Grids) and instead of that, crowding distance factor is also applied to select the global optimal particle as a leader. Also we use ε-dominance concept to bound archive size and maintain more diversity and convergence in the MOPSO for optimizing the inventory control problem. Finally, the MOPSO algorithms created using these strategies are evaluated and compared with each other in terms of some performance metrics taken from the literature. The results indicate that these strategies have significant influences on computational time, convergence, and diversity of generated Pareto optimal solutions.     

An iterated local search for the multi-objective permutation flowshop scheduling problem with sequence-dependent setup times

Due to its simplicity yet powerful search ability, iterated local search (ILS) has been widely used to tackle a variety of single-objective combinatorial optimization problems. However, applying ILS to solve multi-objective combinatorial optimization problems is scanty. In this paper we design a multi-objective ILS (MOILS) to solve the multi-objective permutation flowshop scheduling problem with sequence-dependent setup times to minimize the makespan and total weighted tardiness of all jobs. In the MOILS, we design a Pareto-based variable depth search in the multi-objective local search phase. The search depth is dynamically adjusted during the search process of the MOILS to strike a balance between exploration and exploitation. We incorporate an external archive into the MOILS to store the non-dominated solutions and provide initial search points for the MOILS to escape from local optima traps. We compare the MOILS with several multi-objective evolutionary algorithms (MOEAs) shown to be effective for treating the multi-objective permutation flowshop scheduling problem in the literature. The computational results show that the proposed MOILS outperforms the MOEAs.     

Hybridizing a multi-objective simulated annealing algorithm with a multi-objective evolutionary algorithm to solve a multi-objective project scheduling problem

In this paper, a multi-objective project scheduling problem is addressed. This problem considers two conflicting, priority optimization objectives for project managers. One of these objectives is to minimize the project makespan. The other objective is to assign the most effective set of human resources to each project activity. To solve the problem, a multi-objective hybrid search and optimization algorithm is proposed. This algorithm is composed by a multi-objective simulated annealing algorithm and a multi-objective evolutionary algorithm. The multi-objective simulated annealing algorithm is integrated into the multi-objective evolutionary algorithm to improve the performance of the evolutionary-based search. To achieve this, the behavior of the multi-objective simulated annealing algorithm is self-adaptive to either an exploitation process or an exploration process depending on the state of the evolutionary-based search. The multi-objective hybrid algorithm generates a number of near non-dominated solutions so as to provide solutions with different trade-offs between the optimization objectives to project managers. The performance of the multi-objective hybrid algorithm is evaluated on nine different instance sets, and is compared with that of the only multi-objective algorithm previously proposed in the literature for solving the addressed problem. The performance comparison shows that the multi-objective hybrid algorithm significantly outperforms the previous multi-objective algorithm.     

零等待约束下多产品间歇过程的多目标调度

针对零等待约束下多产品间歇过程的总流程时间和完工时间最小化问题,提出一种多目标离散组搜索算法求解.在采用启发式规则产生初始解的基础上,通过发现者、追随者和巡逻者的操作设计,算法不断更新Pareto前沿,同时,混合了基于插入邻域的多目标局部搜索方法.大量计算实验表明,所提出的算法获得的非支配解集在IGD和Set Coverage指标上优于非支配排序遗传算法和模拟退火算法,可为多目标决策者提供更好的决策依据,利于间歇生产过程的优化运行.     

A multi-phase covering Pareto-optimal front method to multi-objective scheduling in a realistic hybrid flowshop using a hybrid metaheuristic

This paper considers the problem of sequence-dependent setup time hybrid flowshop scheduling with the objectives of minimizing the makespan and sum of the earliness and tardiness of jobs, and present a multi-phase method. In initial phase, the population will be decomposed into several subpopulations. In this phase we develop a random key genetic algorithm and the goal is to obtain a good approximation of the Pareto-front. In the second phase, for improvement the Pareto-front, non-dominant solutions will be unified as one big population. In this phase, based on the local search in Pareto space concept, we propose multi-objective hybrid metaheuristic. Finally in phase 3, we propose a novel method using e-constraint covering hybrid metaheuristic to cover the gaps between the non-dominated solutions and improve Pareto-front. Generally in three phases, we consider appropriate combinations of multi-objective methods to improve the total performance. The hybrid algorithm used in phases 2 and 3 combines elements from both simulated annealing and a variable neighborhood search. The aim of using a hybrid metaheuristic is to raise the level of generality so as to be able to apply the same solution method to several problems. Furthermore, in this study to evaluate non-dominated solution sets, we suggest several new approaches. The non-dominated sets obtained from each phase and global archive sub-population genetic algorithm presented previously in the literature are compared. The results obtained from the computational study have shown that the multi-phase algorithm is a viable and effective approach.     

An adaptive multi-population genetic algorithm to solve the multi-objective group scheduling problem in hybrid flexible flowshop with sequence-dependent setup times

In this paper we consider a multi-objective group scheduling problem in hybrid flexible flowshop with sequence-dependent setup times by minimizing total weighted tardiness and maximum completion time simultaneously. Whereas these kinds of problems are NP-hard, thus we proposed a multi-population genetic algorithm (MPGA) to search Pareto optimal solution for it. This algorithm comprises two stages. First stage applies combined objective of mentioned objectives and second stage uses previous stage’s results as an initial solution. In the second stage sub-population will be generated by re-arrangement of solutions of first stage. To evaluate performance of the proposed MPGA, it is compared with two distinguished benchmarks, multi-objective genetic algorithm (MOGA) and non-dominated sorting genetic algorithm II (NSGA-II), in three sizes of test problems: small, medium and large. The computational results show that this algorithm performs better than them.     

Multi-objective scheduling and a resource allocation problem in hospitals

This study addresses the issue of scheduling medical treatments for resident patients in a hospital. Schedules are made daily according to the restrictions on medical equipment and physicians who are being assigned at the same time. The problem is formulated as a multi-objective binary integer programming (BIP) model. Three types of metaheuristics are proposed and implemented to deal with the discrete search space, numerous variables, constraints and multiple objectives: a variable neighborhood search (VNS)-based method, scatter search (SS)-based methods and a non-dominated sorting genetic algorithm (NSGA-II). This paper also provides the results of computational experiments and compares their ability to find efficient solutions to the multi-objective scheduling problem.     

NSGA-Ⅱ算法的改进策略研究

带精英策略的非支配排序遗传算法(NSGA-Ⅱ)在多目标优化领域具有广泛的应用,但该算法种群收敛分布不均匀,全局搜索能力较弱,算法运行速度较慢。针对这些局限性提出了改进的排序适应度策略、算术交叉算子策略、按需分层策略和设定阈值选择策略。在典型的测试函数集上的数值实验结果表明,根据这些策略改进的算法得到的非劣解集具有较好的分布性,同时收敛速度更快。     

A comparative study of local search within a surrogate-assisted multi-objective memetic algorithm framework for expensive problems

A comparative study of the impacts of various local search methodologies for the surrogate-assisted multi-objective memetic algorithm (MOMA) is presented in this paper. The base algorithm for the comparative study is the single surrogate-assisted MOMA (SS-MOMA) with the main aim being to solve expensive problems with a limited computational budget. In addition to the standard weighted sum (WS) method used in the original SS-MOMA, we studied the capabilities of other local search methods based on the achievement scalarizing function (ASF), Chebyshev function, and random mutation hill climber (RMHC) in various test problems. Several practical aspects, such as normalization and constraint handling, were also studied and implemented to deal with real-world problems. Results from the test problems showed that, in general, the SS-MOMA with ASF and Chebyshev functions was able to find higher-quality solutions that were more robust than those found with WS or RMHC; although on problems with more complicated Pareto sets SS-MOMA-WS appeared as the best. SS-MOMA-ASF in conjunction with the Chebyshev function was then tested on an airfoil-optimization problem and compared with SS-MOMA-WS and the non-dominated sorting based genetic algorithm-II (NSGA-II). The results from the airfoil problem clearly showed that SS-MOMA with an achievement-type function could find more diverse solutions than SS-MOMA-WS and NSGA-II. This suggested that for real-world applications, higher-quality solutions are more likely to be found when the surrogate-based memetic optimizer is equipped with ASF or a Chebyshev function than with other local search methods.     

A machine-learning based memetic algorithm for the multi-objective permutation flowshop scheduling problem

In recent years, the historical data during the search process of evolutionary algorithms has received increasing attention from many researchers, and some hybrid evolutionary algorithms with machine-learning have been proposed. However, the majority of the literature is centered on continuous problems with a single optimization objective. There are still a lot of problems to be handled for multi-objective combinatorial optimization problems. Therefore, this paper proposes a machine-learning based multi-objective memetic algorithm (ML-MOMA) for the discrete permutation flowshop scheduling problem. There are two main features in the proposed ML-MOMA. First, each solution is assigned with an individual archive to store the non-dominated solutions found by it and based on these individual archives a new population update method is presented. Second, an adaptive multi-objective local search is developed, in which the analysis of historical data accumulated during the search process is used to adaptively determine which non-dominated solutions should be selected for local search and how the local search should be applied. Computational results based on benchmark problems show that the cooperation of the above two features can help to achieve a balance between evolutionary global search and local search. In addition, many of the best known Pareto fronts for these benchmark problems in the literature can be improved by the proposed ML-MOMA.     

基于DNA遗传算法的CR多载波参数优化

提出了一种基于DNA计算的非支配排序多目标遗传算法(DNA-GA)来对CR多载波传输参数进行优化。该算法通过非支配排序计算个体适应度,结合克隆操作使算法收敛于全局最优,并引入DNA基因级操作,以提高算法的搜索性能,保持种群的多样性。通过在不同服务需求情况下得到的仿真参数结果,证明了DNA-GA可以有效地优化CR传输参数。     

Simple heuristics-based selection of guides for multi-objective PSO with an application to electrical distribution system planning

In multi-objective particle swarm optimization (MOPSO), a proper selection of local guides significantly influences detection of non-dominated solutions in the objective/solution space and, hence, the convergence characteristics towards the Pareto-optimal set. This paper presents an algorithm based on simple heuristics for selection of local guides in MOPSO, named as HSG-MOPSO (Heuristics-based-Selection-of-Guides in MOPSO). In the HSG-MOPSO, the set of potential guides (in a PSO iteration) consists of the non-dominated solutions (which are normally stored in an elite archive) and some specifically chosen dominated solutions. Thus, there are two types of local guides in the HSG-MOPSO, i.e., non-dominated and dominated guides; they are named so as to signify whether the chosen guide is a non-dominated or a dominated solution. In any iteration, a guide, from the set of available guides, is suitably selected for each population member. Some specified proportion of the current population members follow their respective nearest non-dominated guides and the rest follow their respective nearest dominated guides. The proposed HSG-MOPSO is firstly evaluated on a number of multi-objective benchmark problems along with investigations on the controlling parameters of the guide selection algorithm. The performance of the proposed method is compared with those of two well-known guide selection methods for evolutionary multi-objective optimization, namely the Sigma method and the Strength Pareto Evolutionary Algorithm-2 (SPEA2) implemented in PSO framework. Finally, the HSG-MOPSO is evaluated on a more involved real world problem, i.e., multi-objective planning of electrical distribution system. Simulation results are reported and analyzed to illustrate the viability of the proposed guide selection method for MOPSO.