A Method Of Complementary Bi-system Co-evolutionary Design And Its Application

The idea of divide-and-conquer is an underlying philosophy frequently employed to solve complex engineering system (such as satellite design). Existing research primarily focus on two prespected area, Multidisciplinary Design Optimization(MDO) and Co-evolutionary Algorithm(CEA). Inspired by the co-evolutionary phenomena occurring in ecosytem, this paper develops a co-evolutionary design methodology called complementary bi-system cooperative optimization algorithm (CBCEA), against the background of the layout design of satellite module. The chief aim of proposed algorithm is to solve the large-scale weak-coupled optimization problem in the complex engineering system, characterized by nonlinearity and high-dimension, more efficiently and effectively.Firstly, the description of the complex engineering system is presented, and then, this paper brieves the state of the art of MDO and CEA. Based on the overview above, complementary bi-system cooperative optimization algorithm (CBCEA) is given in detail. The innovation and work in this paper are as follows:(1) Present the model, main idea and implimentation of CBCEA. The main points of CBCEA are as follows: employ two populations, system A and system B, to co-evolve the complex system parallelly. For the system A, fitness function of the original problem and the grain variation strategy of the design variables are used. For the system B, which is decomposed into several sub-systems further, simplified fitness function is applied and the sub-systems also evolve concurrently. System A and system B collaborate with each other through inter-system migration of the individuals. The complementary framework is constructed by the two systems with complementary strategies, which are different in the number of design variables and the complexity of fitness function.(2) Build the prototype of proposed method, which is used to solve a simplified layout design problem of satellite module. The simulation proofs that compared to tradictional GA and cooperative coevolutionary algorithm, CBCEA can solve the layout design problem more efficiently and effectively. It should be noted that, because the simulation is limited, further verification of the generality of proposed method is waiting to be made.The drawbacks of CBCEA and the direction of future work are pointed out finally.It's hoped that this work can be beneficial to the theoretical discussion of co-evolutionary algorithm and its application on weak-coupled complex engineering system.