Research on Machine Fault Prediction Based on Machine Learning and Sensor Data

本文研究了利用机器学习算法预测机器故障。作者使用了和鲸社区的传感器数据，包含footfall，tempMode，AQ，USS，CS，VOC，RP，IP，Temperature和fail等特征。利用这些数据，分别构建了XGBoost、随机森林和KNN三种机器学习模型进行故障预测，并比较了它们的性能。实验结果表明，XGBoost模型在AUC和准确率等指标上表现最佳，AUC值为0.9721，准确率为0.9120。机器故障预测在工业生产中具有重要意义，可以减少设备停机时间，提高生产效率。本文的研究成果可以为企业提供有效的机器故障预测方法，具有一定的实际应用价值。This paper studies the prediction of machine failures using machine learning algorithms. The author utilized sensor data from the whale community, including features such as footfall, tempMode, AQ, USS, CS, VOC, RP, IP, Temperature, and fail. Using these data, three machine learning models, namely XGBoost, Random Forest and KNN, were respectively constructed for fault prediction, and their performances were compared. The experimental resultss show that the XGBoost model performs best in terms of indicators such as AUC and accuracy rate. The AUC value is 0.9721 and the accuracy rate is 0.9120. Machine failure prediction is of great significance in industrial production, which can reduce equipment downtime and improve production efficiency. The research results of this paper can provide enterprises with effective machine fault prediction methods and have certain practical application value.