基于神经网络的卷烟包装机运行监测与故障分析

Operation Monitoring and Fault Analysis of Cigarette Packaging Machines Based on Neural Networks

摘要

卷烟包装机属于多工位连续节拍设备，理烟、成束、上膜、热封、切刀与输送环节相互牵连，单点报警往往难以直接对应真实故障源。现场数据具有采样频率不一致、噪声冲击强、工况切换频繁、同类故障表现分散等特点，导致传统阈值与经验判别在跨班次、跨配方与跨速度条件下稳定性不足。神经网络在该类设备中的应用关键不在于替代控制逻辑，而在于把多源信号按工序语义重新组织，使动作质量、负载变化与耦合关系以可学习的形式被表达，并将模型输出转化为检修可复核的证据链。基于此，本文将基于神经网络的卷烟包装机运行监测与故障展开深入探讨，以供参考。

Abstract

Cigarette packaging machines are multi-station continuous rhythm processing systems where processes such as cigarette handling, bundling, film application, heat sealing, cutting, and material conveying are interdependent. Single-point alarms often fail to directly identify the actual fault source. Field data exhibit characteristics including inconsistent sampling frequencies, strong noise interference, frequent operational transitions, and dispersed manifestations of similar faults, rendering traditional threshold-based and empirical diagnostic methods unstable across shifts, formulations, and speed variations. The key application of neural networks in such equipment lies not in replacing control logic, but in reorganizing multi-source signals according to process semantics to represent operational quality, load variations, and coupling relationships in a learnable format, thereby transforming model outputs into verifiable evidence chains for maintenance purposes. Based on this, this paper conducts an in-depth investigation into operational monitoring and fault diagnosis for cigarette packaging machines utilizing neural networks as a reference.

关键词

神经网络 / 卷烟包装机 / 运行监测 / 故障

Key words

neural network / cigarette packaging machine / operation monitoring / fault diagnosis

参考文献

[1] 张廷龙. 卷烟厂包装机智能化监控系统设计与实现[J]. 中文科技期刊数据库(文摘版)工程技术, 2023(011):000.

[2] 樊柄宏, 刘清, 周国彬, 等. 基于模糊神经网络的一体化监测设备可靠性评价[J]. 现代矿业, 2024, 40(3):230-234.

[3] 贺雯静. 基于卷积神经网络的IT设备运行故障自动化监测技术[J]. 电子设计工程, 2025, 33(17):80-84.