制造实验平台故障诊断方法研究

Fault Diagnosis Method for a Manufacturing Experimental Platform

摘要: 制造实验平台集PLC/CNC控制、主轴驱动、坐标轴进给系统、夹具与换刀机构及现场传感系统于一体，运行过程同时包含离散事件和连续监测变量。针对监测信息来源多样、故障类型复杂以及传统阈值报警难以描述故障因果层级和传播过程的问题，提出一种故障树分析与模糊Petri网相结合的诊断方法。首先依据PLC/CNC状态、S7通讯、RS485数据、主轴电流、坐标轴位置与误差及夹具反馈等信息建立“监测数据–异常表现–故障类型”映射；其次构建故障树并按逻辑门规则转换为FTA-Petri网；最后引入模糊隶属度、规则权重、变迁阈值和故障置信度进行不确定异常推理。所建模型可描述底层异常向中间故障、子系统故障和系统级故障的传播关系，为故障报警、定位和处理建议提供模型基础。

Abstract: A manufacturing experimental platform integrates PLC/CNC control, spindle drive, coordinate-axis feed systems, fixtures and tool-changing mechanisms, and field sensor systems, and its operation involves both discrete events and continuous monitoring variables. To address the problems of diverse monitoring information sources, complex fault types, and the limited capability of conventional threshold-based alarms to describe fault causal hierarchies and propagation processes, a fault diagnosis method combining fault tree analysis and fuzzy Petri nets is proposed. First, a mapping relationship among monitoring data, abnormal manifestations, and fault types is established based on PLC/CNC status, S7 communication, RS485 data, spindle current, coordinate-axis position and error information, and fixture feedback. Then, a fault tree model is constructed and converted into an FTA-Petri net according to logic-gate conversion rules. Finally, fuzzy membership degrees, rule weights, transition thresholds, and fault confidence levels are introduced to perform reasoning for uncertain abnormal information. The proposed model can describe the propagation of underlying abnormal events to intermediate faults, subsystem faults, and system-level faults, providing a modeling basis for fault alarm, fault localization, and maintenance recommendations.