Application of mechanical product design parameter optimization based on machine learning in identification
提出基于Stacking集成学习的参数设计优化识别模型，用于早期识别机械产品设计参数缺陷，以浆叶模型为例验证了模型对空化比例的预测精度（绝对误差<0.02），并指导参数选择以减少结构缺陷。
The selection of mechanical product design parameters is a key factor in determining product quality. The defects caused by unreasonable product parameter design are one of the main reasons for the extension of the product development cycle and the impact on product market competitiveness. A slurry blade is one of the most common mechanical structures, which is widely used in wind power, ship, hydropower, and other fields. To identify the parameter defects of mechanical product design in the early stage of the design stage, this paper proposes the parameter design optimization identification model based on Stacking integrated learning and studies the parameter selection and results in the analysis of the slurry blade model. Space Claim software in Ansys software establishes three-dimensional slurry blade models with different scales. Combined with the NSGA II optimization algorithm, the multi-objective optimization design of the aerofoil structure is carried out. The CFD software is used to simulate and verify the slurry blade model after parameter design optimization. The results show that the optimization identification model for the design of slurry blade parameters based on the Stacking integrated learning algorithm can realize the influence of the unexpected change of design parameters on the design results. The prediction accuracy of cavitation proportion is high under different working conditions, and the absolute error is <0.02. The simulation of the optimized blade shows that the cavitation proportion e reaches a minimum of 0.005. The purpose of guiding the selection of design parameters, machining parameters, and assembly accuracy parameters is achieved, the structural defects of the slurry blade are reduced, and the working efficiency of the edge is improved.