基于机器学习的模具故障预测与维护系统研究

打开文本图片集

中图分类号：TP713 文献标志码：A

Abstract：To improve the reliability of mold production lines， it is essential to incorporate mold failure prediction and maintenance into production scheduling，in addition to routine machine maintenance. This study employs principal component analysis （PCA） and Isomap for dimensionality reduction of the dataset，followed by predictive analysis using support vector machine（SVM）），neural network （NN），and logistic regression（LR） models. Subsequently，genetic algorithms are utilized for multi-objective optimization. Experimental results demonstrate that dimensionality reduction methods significantly enhance the accuracy of predictive models，with the PCA combined with NN achieving an overall accuracy of 98.04% . Moreover， this study integrates genetic algorithms with simulated annealing to improve optimization strategies， exhibiting excellent global search capabilities in solving optimization problems. These findings further validate the practical value of the proposed approach in mold maintenance and failure prediction.

Key words： machine learning；mold failure prediction; proactive maintenance；model algorithm; intelligent fault diagnosis

0 引言

由于塑料具有高强度、低成本、重量轻，以及易用性等独特优势，目前已经被广泛应用于日常生活中[1]。（剩余11185字）