范国栋  李博涵

（重庆交通大学机电与车辆工程学院，重庆 400074）

摘要：为提高工业生产效率和安全性，研究基于机器学习的工业机械设备故障预测方法。首先，利用斯皮尔曼等级相关系数分析工业机械设备故障特征之间的相关性，并过滤冗余特征；然后，采用随机森林算法筛选影响工业机械设备故障的3个核心特征；最后，基于逻辑回归、朴素贝叶斯、XGBoost、决策树等机器学习算法分别建立工业机械设备的故障预测模型和故障类型预测模型。经实验验证，基于XGBoost算法构建的工业机械设备故障预测模型和决策树训练出来的工业机械设备故障类型预测模型具有较高的准确性。该方法具有实际的应用价值，可有效地预测不同工业机械设备的故障类型，为工业安全生产提供技术支持。

关键词：机器学习；工业机械设备；故障预测；斯皮尔曼相关性分析；随机森林算法；预测模型

中图分类号：TP399             文献标志码：A           文章编号：1674-2605(2023)04-0003-07

DOI：10.3969/j.issn.1674-2605.2023.04.003

Fault Prediction Method of Industrial Machinery Equipment             Based on Machine Learning 

FAN Guodong  LI Bohan

(School of Electromechanical and Vehicle Engineering, Chongqing Traffic University,              Chongqing 400074, China)

Abstract: To improve industrial production efficiency and safety, a machine learning based fault prediction method for industrial machinery and equipment is studied. Firstly, the Spearman rank correlation coefficient is used to analyze the correlation between fault features of industrial machinery equipment, and redundant features are filtered; Then, the random forest algorithm is used to screen the three core features that affect the faults of industrial machinery and equipment; Finally, based on machine learning algorithms such as logistic regression, naive Bayes, XGBoost, and decision tree, a fault prediction model and a fault type prediction model for industrial machinery equipment are established. Through experimental verification, the industrial machinery equipment fault prediction model constructed based on XGBoost algorithm and the industrial machinery equipment fault type prediction model trained from decision trees have high accuracy. This method has practical application value and can effectively predict the fault types of different industrial machinery and equipment, providing technical support for industrial safety production.

Keywords: machine learning; industrial machinery and equipment; fault prediction; Spearman correlation analysis; random forest algorithm; prediction model