基于深度学习的故障预警诊断平台设计与开发

Fault Early Warning and Diagnosis Platform Based on Deep Learning

现代电站系统日益庞大且复杂,且火电机组设备大多处于高温、高尘、高速等恶劣环境中运行,极易出现故障。故障的发生会影响机组正常运行脱离最优工况降低经济效益,重大故障甚至会造成人员伤亡。因此,对这些复杂设备进行在线实时过程监测与故障预警,是保障电站设备运行可靠性、经济性的重要手段。对此采用了基于深度学习算法对DCS系统中海量过程数据进行数据驱动建模,挖掘过程数据中所代表的设备运行信息,开发了电站设备在线实时过程监测与故障预警系统。试验结果表明,该系统提供了简单高效的建模方法与精准迅速的故障预警,为现场运行人员给予可靠指导分析,提高了全场管理水平与效率。

Modern power station systems are becoming increasingly large and complex,and most of the thermal power plant equipment is operating in harsh environments such as high temperature,high dust,and high speed,which is extremely prone to failure.The occurrence of faults will affect the normal operation of the unit and deviate from the optimal working conditions and reduce economic benefits.Major faults may even cause casualties.Therefore,online real-time process monitoring and fault early warning of these complex equipment are important means to ensure the reliability and economy of power station equipment.Based on the deep learning algorithm,this paper conducts data-driven modeling of the massive process data in the DCS,mines the equipment operation information represented in the process data,and develops an online realtime process monitoring and fault warning system for power plant equipment.The test results show that the system provides a simple and efficient modeling method and accurate and rapid fault warning,provides reliable guidance and analysis for on-site operators,and improves the level and efficiency of the entire site management.