基于BIM-QNN的机电设备故障自动化检测系统

Automated Fault Detection System for Electromechanical Equipment Based on BIM-QNN

摘要:

针对传统机电设备故障检测多依赖单一信号维度、难以表征多物理场耦合特征且实时性受限的问题,本文设计了一种基于BIM-QNN的机电设备故障自动化检测系统。硬件层采用模块化架构,由电源、驱动、多通道数据采集与隔离抗干扰输出等模块构成,并引入高精度时钟实现多源数据的时序同步；软件层以BIM三维模型承载设备几何/拓扑与运维信息,通过量子特征映射将振动、温度、电流等多源序列编码为高维量子态,构建可训练参数化量子线路提取非线性耦合特征,并设计量子-经典混合注意力机制实现特征自适应融合；进一步采用多任务学习同步完成故障类型识别、严重度评估与剩余寿命预测。以工业级轴流式主通风机为对象开展实验验证,结果表明：与对比方法相比,本系统误报率长期处于较低水平,对微弱振动异常具有更高灵敏度。综上,所提系统通过BIM与QNN的深度融合提升了弱故障可检性与诊断维度,可为复杂机电设备的智能运维提供支撑。

Abstract:

To address the limitations of traditional electromechanical equipment fault detection—namely, its reliance on a single signal dimension, difficulty in characterizing multi-physics coupling features, and constrained real-time performance—this paper designs an automated fault detection system for electromechanical equipment based on BIM-QNN. At the hardware level, a modular architecture is adopted, consisting of power supply, drive, multi-channel data acquisition, and isolated anti-interference output modules; a high-precision clock is also introduced to achieve temporal synchronization of multi-source data. At the software level, a BIM-based 3D model is used to host the equipment’s geometric/topological information and O M data. Via quantum feature mapping, multi-source time-series signals such as vibration, temperature, and current are encoded into high-dimensional quantum states. A trainable parameterized quantum circuit is then constructed to extract nonlinear coupling features, and a quantum–classical hybrid attention mechanism is designed for adaptive feature fusion. Furthermore, a multi-task learning strategy is employed to jointly perform fault type identification, severity assessment, and remaining useful life prediction. Experiments are conducted on an industrial-grade axial main ventilation fan, and the results show that, compared with baseline methods, the proposed system maintains a consistently low false-alarm rate over long-term operation and exhibits higher sensitivity to subtle vibration anomalies. In summary, by deeply integrating BIM and QNN, the proposed system improves the detectability of incipient faults and expands diagnostic dimensions, providing support for intelligent operation and maintenance of complex electromechanical equipment.