机械

研究了面向转向架故障预警的轨边综合检测系统软硬件设计方法和样机实现，检测功能包含动态轮轨作用力、转向架运动稳定性和曲线通过性能三个模块。轮轨力检测子系统针对车轮踏面损伤和轮重均衡类故障，采用轨腰应变桥路与接近开关实现连续测量。失稳检测子系统针对转向架蛇行失稳和车轮偏磨等故障，采用非接触式电涡流传感器测量轮轨动态几何间隙。曲线通过性能检测子系统针对车辆曲线运行时脱轨系数、减载率以及轮轴横向力指标异常故障，通过测量轮轨横、垂作用力，基于多维数据和深度学习算法识别因轮轨异常磨耗或悬挂系统特性劣化导致的转向架曲线通过不良故障。主控软件采用设计模式满足功能柔性扩展需求，模块内部紧聚合而模块间松耦合。基于消息生产-消费、多数据队列模式实现异构硬件设备互联、传感器信号异速采集存储、数据实时计算以及数据库异步访问等集成机制。基于本地Web服务器和客户端浏览器实现用户界面，数据可视化与轨边数据采集分析功能完全解耦。

This study focuses on the wayside inspection system for bogie faults detection, including hardware and software design, and prototype implementation. The system is composed of three independent units, the wheel rail contact force detection module, the bogie stability detection module and the vehicle curving performance detection module. The contact force detection subsystem focuses on the wheel thread damages and wheel loads imbalance by strain measuring of rail waist and proximity switches. The bogie stability detection subsystem focuses on the high-speed running stability and asymmetrical wear of the wheelsets by measuring dynamic geometry gap with non-contact eddy current sensor. The curving performance detection subsystem focuses on the curving safety related faults such as the derailment factor, load reduction and abnormal lateral force of the axle. The design modes in software engineering were adopted for flexible requirement expansion. The function modules were designed with tight aggregation within modules and loose coupling between modules. Based on the producer-consumer and data queues design modes, the system integration mechanism was implemented, including heterogenic data acquisition hardware integration, multiple data sampling rates synchronization, data asynchronous processing algorithm and database access. The user interface was built by the local Web server and remote explorer. The data visualization in remote user interface is completely decoupled with the wayside data center.