论文部分内容阅读
为了充分利用机组运行的实时信息以提高机组运行的安全性,提出了基于概率计算的故障检测方法。该故障检测方法模拟运行人员的故障检测过程,首先根据运行规程和运行经验得到基本故障检测知识,然后通过对历史数据的学习得到精确故障检测知识,再在此基础上,根据相关实时数据实现运行故障检测。为减少故障检测的漏报和误报,提出了该方法故障检测知识的自适应学习策略。理论分析和实验结果表明:该故障检测方法具有鲁棒性强、检测灵敏、知识获取容易和易于模块化等优点,可以有效地构成基于厂级监控信息系统的通用故障检测软件包。
In order to make full use of the real-time information of unit operation to improve the safety of unit operation, a fault detection method based on probability calculation is proposed. The fault detection method simulates the fault detection process of the operator. First, the basic fault detection knowledge is obtained according to the operation rules and operation experience. Then the knowledge of the fault detection is obtained by learning the historical data. Based on this, the operation is realized according to the related real-time data Fault detection. To reduce the omission and false positive of fault detection, an adaptive learning strategy of fault detection knowledge is proposed. Theoretical analysis and experimental results show that the proposed fault detection method has the advantages of robustness, sensitive detection, easy access to knowledge and easy modularity. It can effectively constitute a general fault detection software package based on plant-level monitoring information system.