Multi-objective Cooperative Co-evolution Method For Equipment Layout Design Of Multi-cabin Satellite

Recently,China's State Council Information Office published the "2016 China Aerospace" white dissertation,proposed that China will accelerate the construction of aerospace power in the next five years,continue to enhance the basic capabilities of the aerospace industry,and strengthen the research of key and cutting-edge technologies.An important task in the field of manned spaceflight is to complete the development of each cabin in the space station and to carry out the assembly and operation of the space station in orbit.Each of these cabins can be seen as a sub-system of the space station,and form a complex system by interacting each other.The layout of such multiple compartments generally need to consider multiple design objectives,such as balance,inertia,stability,vibration,electromagnetic and temperature fields,etc.,and it needs to meet complex non-interference constraints.Therefore,this problem belongs to a class of multi-objective optimization problem on system-level.The multi-cabin spacecraft(includes satellites)equipment layout optimization problem(MLSELP for short)in this dissertation studies how to rationally place a set of given equipment,apparatus and device(components for short)in a specific 3D constrained space of spacecraft's each cabin.At the same time,the technical indicators(or performance)of layout design techniques satisfy the design allowable value or optimize to the greatest extent,and meet the constraints of non-interference or the properties in given space.Although MLSELP belongs to the multi-objective optimization problem,it is usually transferred into single objective optimization problem and solved by evolutionary algorithms or cooperative co-evolutionary algorithms.This method only finds the optimal solution under a given weight.The weight coefficient mainly depends on experience,and the Pareto optimal solution set is unable to obtain.Furthermore,MLSELP is in the field of multi-layer system multi-objective optimization problem in inconsistent space.With the increasing of the optimization objective and component size of satellite cabins(or bearing surface),single objective and traditional multi-objective evolutionary algorithms will fall into trouble.Therefore,a new multi-objective cooperative co-evolutionary method is proposed to decompose the problem based on divide and rule strategy,and get the Pareto optimal solutions.This dissertation takes the delegated related satellite design projects and subsidiary subjects in aerospace "863" project as the engineering application background,studies a new multi-objective cooperative co-evolutionary method aiming at a class of multi-cabin satellite equipment layout problems under the funding of National Natural Science Foundation of China.The main work of this dissertation is shown as following:(1)Based on the single objective optimization model of single cabin satellite equipment layout problem,this thesis presents a multi-objective optimization model in system level.Compared with the single cabin satellite,the multi-cabin satellite equipment layout optimization problem has more design variables(more than 200)and complex structures(Multiple cabins),which is difficult to solve by typical multi-objective optimization method,generally needs to be resolved after the system decomposition.This study proposes a multi-objective optimization model for multi-cabin satellite equipment layout optimization problem and gives a multi-objective constraint treatment method.It laid the foundation for the proposed follow-up method.(2)A non-dominated sorting multi-objective cooperative differential evolution algorithm(NSCCDE for short)is proposed for multi-cabin satellite equipment layout optimization problem and gets the Pareto optimal solution set.NSCCDE adopts an external archive to store the current obtained elitism Pareto solution set and provides the collaborations for sub-system to constitute the complete solutions.This operation supplies the high-quality resources for non-dominated sorting.This archive executes self-renewal operation by spatial dispersal and mutation operation in the iterative process of the algorithm,so as to increase the Spacing(SP)and Maximum Spread(MS)of the Pareto optimal solution set in achieve.NSCCDE is used to solves a multi-cabin satellite equipment layout case,and obtain the Pareto optimal solution set.Experimental resultes demonstrate the proposed algorithm outperforms NSCCGA and NSGA-?.(3)This dissertation also upgrades the multi-objective cooperative co-evolutionary algorithm to the strategy layer and presents a multi-objective cooperative co-evolutionary multi-agent system(CCMO-EMAS for short).Each agent in CCMO-EMAS is used to represent a multi-objective co-evolution strategy(including the algorithm and its setting).The problem is optimized while evolving the strategies.The result shows that CCMO-EMAS can improve all the properties of multi-objective cooperative co-evolutionary(Maximum Spread MS,Spacing SP and Generation Distance GD).Also,CCMO-EMAS is used to analyze the evolutionary process of agents and study the multi-objective cooperative co-evolutionary strategy,in order to build a targeted multi-objective cooperative co-evolutionary algorithm.(4)For sufficiently utilizing the engineer's participation and previous experiences,the human-computer cooperation method is herein introduced into the multi-objective cooperative co-evolutionary algorithm,and a human-computer cooperative multi-objective co-evolutionary differential evolution algorithm(HNSCCDE for short)is proposed.In HNSCCDE,human-computer cooperation in sub-system level is adopted to increase the Maximum Spread(MS)and Spacing(SP)of the algorithm in the early iteration;human-computer cooperation in system level is taken to increase the Spacing(SP)and Generation Distance(GD)in later iteration.The case of multi-cabin satellite equipment layout shows that the multi-object human-computer cooperation method can improve the calculation efficiency while ensuring the calculation accuracy.Finally,we give the system integration and realization of a multi-objective cooperative co-evolutionary design platform based on the method in this dissertation.In summary,this study try to effectively solve the MLSELP,which has theoretical significance and application value.Its direct application background is the layout design for the space laboratory and resource module layout of the multi-cabin satellite in our country.This study is hoped to apply to a class of engineering system's layout optimization designs in national defense and industrial equipment.