Research On Layered Optimization Method Of Mechanical Structure Based On Parameter Sensitivity

There are a lot of complex mechanical structures in the field of power engineering.The complexity of them is mainly reflected in two aspects: one is that the structure will be subjected to various loads when working,and the other is that these structures are usually composed of multiple substructures through different connection modes.Therefore,the optimization design of complex mechanical structure is usually multi-objective optimization problem with multi variables and constraints.There is a contradiction between the sub-objective functions.The influence of different design variables on the optimization objectives is also very different.It is very difficult to solve the optimization design of these complex structures by traditional direct optimization method.The layered optimization method provides an effective way for the multi-objective optimization design of complex mechanical structures.However,at present,the layered optimization technology is more focused on the application of large-scale complex truss structure,but it is of great significance for the research and application of the structural components of engineering machinery.Firstly,the paper summarizes and analyzes different layered optimization strategies and model building methods from the theoretical perspective.Aiming at the multi-objective optimization of member structure,a layered optimization method based on parameter sensitivity analysis is proposed.The sensitivity analysis of the input and output parameters of the mechanical structure optimization problem is carried out to get the influence degree of each input parameter on the output parameters.Then the overall multi-objective optimization problem is divided into multi-layer optimization problem.The design variables of each sub-layer optimization problem are the design variables which have a great influence on the sub-level optimization objectives.Through layered optimization,the optimization scale of each sub-layer problem is small,and each sub-layer problem can choose the optimization algorithm suitable for the sub-layer problem,so as to reduce the dimension of the overall optimization problem and easy to calculate.In order to solve the problem of multi-objective optimization with the minimum mass and the highest natural frequency of a diesel engine,the direct global multi-objective optimization method and the layered optimization method based on parameter sensitivity are used to optimize the design.Among them,AMO optimization algorithm and MOGA algorithm are used to solve the whole optimization problem directly.The layered optimization divides the whole problem into two sub-layer problems according to the influence of design variables on the optimization objectives.Each sub-layer problem is solved according to one of the original multi-objective optimization problems.The results of the optimization are obtained by 311 times of total iteration calculation,and the quality of the optimization is16.8% lower than that of the original structure.The lower order natural frequency is increased by 15.4%.The multi-objective algorithm of AMO is calculated 536 times iteratively,and the structure quality is reduced by 13.2% after optimization.The natural frequency of the lower order is increased by 18.8%.The MOGA multi-objective algorithm is calculated 682 times iteratively,and the structure quality is reduced by 13.1% after optimization.The lower order natural frequency is increased by 15.2%.The results show that the optimization of layered optimization method is more accurate and efficient than direct optimization method.The research results in this paper provide new ideas for multi-objective optimization design of mechanical structure,and also provide reference and useful reference for other multi constraint and multi variable optimization design of complex mechanical structure.