基于数据驱动的烟机主轴箱机械故障预测研究

Research on Data-Driven Mechanical Fault Prediction of Tobacco Machine Spindle Box

摘要: 烟草加工行业中，主轴箱作为核心传动部件，其故障易导致生产中断、成本增加，而传统维护模式难以适配现代化生产的精准运维需求。文章以烟机主轴箱为研究对象，结合其运行工况与故障特点，构建数据驱动的故障预测体系，通过信号预处理、时频特征提取及特征降维优化数据质量，设计传统机器学习与混合深度学习预测模型并进行对比优选，最终通过实验验证模型有效性。研究表明，优选的混合深度学习模型预测精度高、鲁棒性强，可实现故障提前预警，为烟机主轴箱这一特定关键设备提供并验证了一套完整且高效的数据驱动故障预测方案，为烟机主轴箱精准运维提供技术支撑，助力烟草生产连续稳定运行。

Abstract: In the tobacco processing industry, the spindle box is a core transmission component. Its failure can easily lead to production interruptions and increased costs. Traditional maintenance methods are difficult to adapt to the precise operation and maintenance needs of modern production. This paper takes the tobacco machine spindle box as the research object, and constructs a data-driven fault prediction system based on its operating conditions and fault characteristics. Through signal preprocessing, time-frequency feature extraction, and feature dimensionality reduction to optimize data quality, traditional machine learning and hybrid deep learning prediction models are designed, compared, and optimized. Finally, the effectiveness of the model is verified through experiments. The study shows that the optimized hybrid deep learning model has high prediction accuracy and strong robustness, and can realize early warning of faults. It provides and verifies a complete and efficient data-driven fault prediction scheme for the tobacco machine spindle box, a specific key piece of equipment, providing technical support for the precise operation and maintenance of the tobacco machine spindle box and helping to ensure the continuous and stable operation of tobacco production.