Whole vehicle size matching deviation prediction method based on multi-model fusion

Abstract

The invention discloses a whole vehicle size matching deviation prediction method based on multi-model fusion, belongs to the field of vehicle production. The method comprises steps that deviation analysis is applied based on a Monte Carlo simulation method to establish a mechanism model M1 of a whole vehicle size matching deviation prediction position; a model learning data set D is established, wherein the data set D is composed of j samples, each sample is composed of i feature items and one target item, and the i feature items come from influence factors in the mechanism model M1; a multivariate linear regression model M2 based on a normal equation, a multivariate linear regression model M3 based on gradient descent and a ridge regression model M4 based on L2 regularization are established by using the data set D; the mechanism model M1, the regression model M2, the regression model M3 and the ridge regression model M4 are fused to obtain a prediction model M, and the prediction model M is used for carrying out prediction analysis on the whole vehicle size matching related position. According to the invention, the effectiveness and accuracy of vehicle size matching deviation prediction can be effectively improved.

Description

Technical field [0001] The invention belongs to the field of vehicle production, and more specifically, relates to a vehicle size matching deviation prediction method based on multi-model fusion. Background technique [0002] The current vehicle size matching deviation prediction methods include: [0003] ①Extreme value method. The calculation method of the extreme value method is: the maximum limit size of a closed ring is obtained when all increasing rings are at the maximum limit size and all decreasing rings are at the minimum limit size; the minimum limit size is when all increasing rings are at the minimum limit size. is the minimum limit size and all reduced rings are the maximum limit size. The extreme value method is based on 100% interchangeability of parts, so the predicted results are often much different than the actual results. [0004] ②The square root method is based on a certain confidence level (usually it is assumed that the tolerances of each component ...

AI Technical Summary

Problems solved by technology

The extreme value method is based on 100% interchangeability of parts, so the predicted results are often very different from the actual results

[0004] ②The square and root method is based on a certain level of confidence (it is usually assumed that the tolerances of each component ring and closed ring obey a normal distribution, and the assembly function is a linear relationship, and the confidence level P=99.73%), does not require 100% interchange , only large numbers are required to be swapped, but the influence of actual production factors is not considered, so the predicted results differ greatly from the actual results

However, this method does not consider the influence of other external factors on the assembly process, and does not consider factors such as assembly and positioning force deformation, thermal expansion, etc., so there is still a large difference between the predicted results and the actual results

Images