Research On Structural Modeling Technology Of Mechanical Products Based On Machine Learning

Aiming at the problem of excessive time consuming in the process of modeling and simulation of mechanical structure,taking the static deformation quantity of grinder bed as the goal,the overall structure parameter of grinder bed is the design variable,the finite element statics model of grinder bed is established,the sample point of sampling and extraction is studied,This paper explores and synthetically compares the modeling methods and prediction accuracy of various algorithms models,and selects the optimal algorithm models to fuse them,which has high prediction accuracy and good effect,provides reference and guidance for the static analysis of mechanical product structure and the construction of machine learning algorithm data model.Firstly,the fine mesh model of grinder bed is established,the load and boundary conditions of grinder bed in normal operation are discussed and modeled,the finite element statics model of grinder bed is established,the results are calculated,the post-processing and calculation analysis are carried out.Secondly,aiming at the strategy of establishing the data set,the Latin hypercube design method is used to extract the training samples and test samples,to ensure the uniform distribution of the sample points and the spatial distribution of the stochastic intensity,and on this basis,the finite element statics analysis of the grinder bed is carried out,the grinder bed body variables of the corresponding structural parameters are obtained,according to the possible correlation between each output value,a priori estimation scheme of three kinds models is proposed.And then,according to three different model schemes,the modeling techniques of machine learning algorithms such as support vector machine,back propagation neural algorithm,decision tree and simple Bayesian algorithm are studied respectively,the method of parameter feature selection,and the influence of parameter feature selection on the prediction accuracy of algorithm model,and the data is re-sampled by 10 percent cross-verification method.An optimization Model index is introduced to evaluate the performance of the model and optimize it,which effectively improves the generalization ability of the model.Based on the comprehensive consideration of the stability and prediction accuracy of the model,the optimal machine learning algorithm is selected to construct the model.Finally,the finite element analysis results of the test samples are compared by the optimal model,and the model constructed by the machine learning algorithm is verified,which has the validity and good accuracy in solving the structural design problems of the mechanical products.