Research on Quality Control and Optimization of Automobile Parts Production Line Based on Statistical Modeling and Bootstrap Inference

Aiming at the fluctuation problem of bolt diameter—a critical quality parameter—in the production process of automobile parts, this study establishes an optimized model for parts quality control based on linear regression, statistical inference, and Bootstrap methods. By simulating production data of 150 batches (including categorical variables such as pressure, temperature, feed rate, tool wear, machine number, and material batch), the influence law of each factor on bolt diameter is analyzed.

(1) The ggpairs function was used for exploratory analysis to obtain correlation relationships, and a multiple linear regression model was established. The broom package was applied to calculate correlation coefficients and confidence intervals, revealing that pressure and tool wear are the influencing factors—i.e., pressure and tool wear have a significant impact on bolt diameter, while temperature and feed rate show no obvious effects.

(2) ANOVA (Analysis of Variance) was employed to verify the interaction effect between machine batches and material batches. Meanwhile, the Boot package was used to perform Bootstrap sampling with R=1000 iterations, constructing confidence intervals for key characteristic coefficients and verifying the stability of the pressure coefficient.

(3) Based on statistical results, reasonable process improvement suggestions are proposed: appropriately increasing pressure, reasonably reducing tool wear, switching to Machine M1, and monitoring/improving production quality via the Process Capability Index (Cpk).

This study fully demonstrates the application potential of the Tidyverse ecosystem in the field of statistical quality control for industrial engineering, providing a reference method for the digital and intelligent development of industrial enterprises.