Optimal control optimization problems are relative complex and widespread in aerospace field. The evolutionary algorithms are useful methods in solving those problems and Multi Agent Collaborative Search Algorithm (MACS) is a effective method for them. It's a mix of evolution approach and gradient-based method. The evolution part includes individualistic and social heuristics with different characteristics. The latest version of MACS is Multi-Agent Collaborative Search optimal control algorithm (MACSoc) which has good performance on multi-objective optimal control problems. However, this algorithm has two important limitations: The two kinds of heuristics are performed evenly on each individuals which may weaken the algorithm efficiency; The Pareto Front in objective space may be uneven but the initial weight vectors of MACSoc is even. This may lead to diversity losing. We proposed a new algorithm MACSoc with adaptive heuristics and weight vectors (MACSoc-AH) which contains the following improvements: The two kinds of heuristics are executed on individuals based on the their characteristics; A weight vector adjusting process with a original trigger is added to improve diversity. The new algorithm is compared with MACSoc and shows competitive results.

Abstract This paper presents a new version of Multi Agent Collaborative Search (MACS) with Adaptive Weights (named MACS‐AW). MACS is a multi‐agent memetic scheme for multi‐objective optimization originally developed to mix local and population‐based search. MACS was proven to perform well on a number of test cases but had three limitations: (i) the amount of computational resources allocated to each agent was not proportional to the difficulty of the sub‐problem the agent had to solve; (ii) the population‐based search (called social actions in the following) was using only one differential evolution (DE) operator with fixed parameters; (iii) the descent directions were not adapted during convergence, leading to a loss of diversity. In this paper, we propose an improved version of MACS, that implements: (i) a new utility function to better manage computational resources; (ii) new social actions with multiple adaptive DE operators; (iii) an automatic adaptation of the descent directions with an innovative trigger to initiate adaptation. First, MACS‐AW is compared against some state‐of‐art algorithms and its predecessor MACS2.1 on some standard benchmarks. Then, MACS‐AW is applied to the solution of two real‐life optimization problems and compared against MACS2.1. It will be shown that MACS‐AW produces competitive results on most test cases analysed in this paper. On the standard benchmark test set, MACS‐AW outperforms all other algorithms in 11 out of 30 cases and comes second in other 8 cases. On the two real engineering test set, MACS‐AW and its predecessor obtain same results.

Large scale optimization is a very challenging task in optimization area. The variable interaction in non-separable problems is a primary source of performance loss, especially for large scale problems. Cooperative Coevolution framework is a popular approach to deal with large scale optimization. It is based on a divide-and-conquer manner. This paper proposes a novel algorithm called DISCC to tackle large-scale optimization problems. It adopts a function-based grouping decomposition strategy called Dependency Identification Grouping to distinguish the interactive variables in the decision space. The grouping strategy aims to find the most suitable arrangement for the variables in order to minimize the limitation that occurs when they are grouped into different groups. The experimental results show this new algorithm is more effective than the existing Cooperative-Coevolution-based algorithms.

In this paper, we propose a new algorithm, named HACC-D, for large scale optimization problems. The motivation is to improve the optimization method for the subcomponents in the cooperative coevolution framework. In the new HACC-D algorithm, an algorithm selection method named hybrid adaptive optimization strategy is used. It is aimed to hybridize the superiority of two very efficient differential evolution algorithms, JADE and SaNSDE, as the subcomponent optimization algorithm of the cooperative coevolution. In the beginning stage, the novel strategy evolves the initial population with JADE and SaNSDE as the subcomponent optimization algorithm for a certain number of iterations separately. Then the one obtained better fitness value will be chosen to be the subcomponent optimization algorithm for the following evolution process. In the later stage of evolution, the selected algorithm may be trapped in a local optimum or lose its ability to make further progress. So it exchanges the subcomponent optimization algorithm with the other one when there is no improvement in the fitness every certain number of iterations. The proposed HACC-D algorithm is evaluated on CEC'2010 benchmark functions for large scale global optimization.

Based on the regularity that: the Pareto set of a continuous m-objectives problem is a piecewise continuous (m-1)-dimensional manifold, a novel hybrid multi-objective optimization algorithm is proposed in this paper. In the early evolutionary stage, traditional crossover and mutation operations are used to produce offspring, in addition, the locally linear embedding (LLE) with small neighbor parameter approach is introduced to learn the local geometry of the manifold. When certain regularity in population's distribution is detected, new offspring are sampled from the probability models created by the statistical distribution information. An entropy-based criterion is imported to determine the switching time of the two different phases of evolutionary search. The proposed hybrid multi-objective estimation of distribution algorithm combining locally linear embedding (HMOEDA_LLE) adopts several widely used test problems to conduct the comparison experiments with two state-of-the-art multi-objective evolutionary algorithms NSGA-II and RM-MEDA. The simulated results show the effectiveness of the entropy-based criterion and the proposed algorithm has better optimization performance.