Research On Cable Layout Design Method Of More Electric Aircraft Electronic System Based On Machine Learning

Cables are important channels for transmitting energy and signals in electronic system.There are many kinds of cables in more electric aircraft(MEA),and the cable wiring path and its electromagnetic environment are very complex.The quality of cable layout directly affects the overall performance of the system.Aiming at the problems of low efficiency and difficult to obtain the optimal cable layout meeting various constraints in the existing cable layout design methods,this thesis takes the interconnection systems composed of various types of single line,coaxial and other cables and several typical electronic and electrical equipment in MEA as the research object,some key problems in the automatic cable layout design of MEA are studied.The combination of machine learning,path planning algorithm and intelligent optimization algorithm is used in the automatic cable layout technology.According to different design scenarios and requirements,on the basis of considering the complex constraints such as electromagnetic compatibility(EMC)performance,mechanical performance,installation convenience,spatial redundancy and weight,a series of automatic cable layout design methods based on machine learning are proposed,and a set of automatic cable layout system is developed.The research results have been applied to the product development and design of a related project of an electronics research institute,which is of great significance in ensuring the comprehensive performance of the interconnected system of electrical and electronic equipment related to MEA,such as electromagnetic compatibility performance,mechanical performance and installation convenience,as well as reducing space redundancy and the total cost of cable layout,and establishing a basis for further improving the quality and efficiency of the overall system design of MEA.The research content and results of this thesis mainly include the following aspects:(1)A cable harness layout method based on Quasi-Physical and Quasi-Human algorithm and improved A*algorithm is proposed to solve the problem that the existing cable harness layout methods cannot be used for the cable harness layout of MEA airborne equipment considering complex engineering rules constraints.Firstly,the evaluation function of traditional A*algorithm is improved.Then,the equivalent radius of the cable harness is calculated based on the Quasi-Physical and Quasi-Human algorithm,and the search space automatic processing algorithm and the corner node reasonableness judgment algorithm considering the constraints of engineering rules are proposed.On this basis,the process of cable harness path planning using improved A*algorithm is given.An example verification of cable harness laying for an airborne equipment shows that compared with the traditional A*algorithm,the cable harness layout scheme obtained using the proposed improved A*algorithm can not only satisfy the complex constraints but also effectively reduce the total wiring cost.(2)Aiming at the problem of cable layout with cable-carbon fiber reinforced composite(CFRP)integrated components in the electronic system of MEA,a cable layout method based on improved response surface method(RSM)and machine learning is proposed,the cable layout is divided into two steps.The first step is to determine the embedded position of the cables in the cable-CFRP integrated component:firstly,the finite element analysis model and electromagnetic compatibility analysis model of the integrated component are established;then,according to the characteristics of the cable layout problem studied,the traditional response surface method is improved,and a method to solve the parameter optimization problem of cable embedding position is proposed;finally,using this method,a set of high-quality cable embedding position parameters are predicted by comprehensively considering the mechanical properties,EMC and manufacturing constraints of the cables.The experimental results show that the integrated components obtained by this method can meet all constraints,and the average error between the predicted and actual values of all responses is very small.The second step is to determine the overall layout scheme of these cables:a cable layout method based on machine learning algorithm,improved RRT algorithm and genetic algorithm is proposed.Using this method,on the premise of meeting all constraints,the globally optimal routing parameters are calculated and a high-quality cable layout scheme is obtained.(3)Aiming at the layout problem of electronic systems with multiple cables and cable bundles without branches,a cable layout method based on stacking-LSRM multi-stage ensemble learning method and improved differential evolution algorithm is proposed.Firstly,a multi-stage ensemble learning(EL)method of stacking-LSRM is proposed,and the process of establishing the cable crosstalk prediction model using this method is given.The correctness of the cable crosstalk simulation modeling method is verified through the comparison between simulation and experiment.Through two examples of crosstalk prediction of electronic subsystems of MEA,the effectiveness of the proposed method and its superiority over the existing methods are verified;then,the initial cable layout is obtained by using the path planning algorithm based on the improved RRT algorithm.According to the characteristics of the cable layout problem in this study,the traditional differential evolution(DE)algorithm is improved.On this basis,a method for solving the cable layout optimization problem of electronic system based on EL and improved DE algorithm is proposed.The effectiveness and superiority of this method over the existing methods are verified by two cable layout examples of MEA electronic system with different wiring modes;finally,the EL models established by using the stacking-LSRM multi-stage EL method are combined with the improved DE algorithm to solve the optimization problem in this study,and a better actual optimization effect is obtained,which further verifies the effectiveness of the stacking-LSRM multi-stage EL method in solving the cable layout problem in this study.(4)Aiming at the layout problem of complex electronic systems with multi branch cable bundles,a layout method of cable harness with branch based on fast linear weighted machine learning fusion model and NSGA-Ⅱ algorithm is proposed.There are many cables in complex electronic systems.These cables are usually arranged to cable harness with multiple branches.In the cable harness layout problem,it is hoped that the total length and total weight of all cables are the smallest,the cables are arranged in harness as much as possible,and The EMC performance of the system is required to be as good as possible.To solve this problem,firstly,a path planning algoritlun for cable harness with branches is proposed to quickly obtain the initial layout of the cable harness;then,a fast weighted linear fusion method based on prior information and genetic algorithm is proposed to establish the machine learning fusion prediction model;finally,for the single-target cable harness layout design problem,an optimization algorithm based on fast linear weighted machine learning fusion model and genetic algorithm is proposed,and the comparison shows the advantages of the fusion model established by the proposed method over the individual machine learning model;for the multi-target cable harness layout design problem,a multi-objective optimization algorithm based on fast linear weighted machine learning fusion model and NSGA-Ⅱ algorithm is proposed.This algorithm is used to optimize the routing parameters,and a high-quality layout scheme is obtained.Finally,the effectiveness of the proposed method is verified.(5)A set of automatic cable layout system based on UG secondary development is developed.Firstly,the system can read the geometric model of wiring space and automatically match the cable connection relationship,obtain,fit and optimize the path,and obtain the optimal path;then,it can automatically check the cable wiring rules,and output the inspection results and relevant inspection data;finally,the deployment drawing and detailed list of cables can be obtained automatically,which provides convenience for subsequent cable production,processing and assembly,so as to effectively improve the wiring quality and wiring efficiency.