标题： 基于人工智能的工业泵故障诊断机器学习方法 显示英文标题

标题： AI-Powered Machine Learning Approaches for Fault Diagnosis in Industrial Pumps

摘要： 本研究提出了一种实用的方法，利用来自大型垂直离心泵在严苛海洋环境中的实际传感器数据，实现工业泵系统的早期故障检测。 监测了五个关键运行参数：振动、温度、流量、压力和电流。 应用了双阈值标注方法，结合固定的工程限值和自适应阈值，自适应阈值计算为历史传感器值的第95百分位数。 为了解决已记录故障的稀有性，使用领域特定规则将合成故障信号注入数据中，模拟合理操作范围内的关键警报。 训练了三种机器学习分类器——随机森林、极端梯度提升（XGBoost）和支持向量机（SVM），以区分正常运行、早期警告和关键警报。 结果表明，随机森林和XGBoost模型在所有类别中均表现出高准确性，包括代表罕见或新兴故障的少数情况，而SVM模型对异常的敏感性较低。 视觉分析，包括分组的混淆矩阵和时间序列图，表明所提出的混合方法具有强大的检测能力。 该框架可扩展、可解释，并适用于实时工业部署，在故障发生前支持主动维护决策。 此外，它可以适应具有类似传感器架构的其他机械，突显了其作为复杂系统预测性维护可扩展解决方案的潜力。 显示英文摘要

摘要： This study presents a practical approach for early fault detection in industrial pump systems using real-world sensor data from a large-scale vertical centrifugal pump operating in a demanding marine environment. Five key operational parameters were monitored: vibration, temperature, flow rate, pressure, and electrical current. A dual-threshold labeling method was applied, combining fixed engineering limits with adaptive thresholds calculated as the 95th percentile of historical sensor values. To address the rarity of documented failures, synthetic fault signals were injected into the data using domain-specific rules, simulating critical alerts within plausible operating ranges. Three machine learning classifiers - Random Forest, Extreme Gradient Boosting (XGBoost), and Support Vector Machine (SVM) - were trained to distinguish between normal operation, early warnings, and critical alerts. Results showed that Random Forest and XGBoost models achieved high accuracy across all classes, including minority cases representing rare or emerging faults, while the SVM model exhibited lower sensitivity to anomalies. Visual analyses, including grouped confusion matrices and time-series plots, indicated that the proposed hybrid method provides robust detection capabilities. The framework is scalable, interpretable, and suitable for real-time industrial deployment, supporting proactive maintenance decisions before failures occur. Furthermore, it can be adapted to other machinery with similar sensor architectures, highlighting its potential as a scalable solution for predictive maintenance in complex systems.