论文部分内容阅读
为利用工矿企业生产过程中已有的长期监测得到的元件故障数据,提出适合于处理这些数据的离散型空间故障树(DSFT).在DSFT基础上提出了确定元件故障概率空间分布的方法,即因素投影拟合法和基于神经网络(ANN)的方法.给出了这两种方法的具体实现过程,考虑了工作时间t和工作温度c对于元件故障的影响,研究了范围c:0~40℃,t:0~50天内的元件故障概率空间分布.并将这两种方法和CSFT得到的元件故障概率空间分布进行了对比分析.比较分析证明:使用DSFT的ANN预测法所得结果要比使用DSFT的因素投影拟合法更为接近CSFT的相对真实结果.在只有监测数据而不清楚系统结构的情况下这两种方法均适用,但ANN法更精确.
In order to make use of the long-term monitored fault information of the industrial and mining enterprises, a discrete Spatial Fault Tree (DSFT) suitable for processing these data is proposed. Based on the DSFT, a method of determining the spatial distribution of the fault probability is proposed Factor projection fitting method and neural network (ANN) -based method.The concrete realization processes of these two methods are given, and the influences of working time t and working temperature c on the component failure are considered. , t: 0 ~ 50 days of the component failure probability of spatial distribution, and these two methods and CSFT component failure probability of the spatial distribution of a comparative analysis of the comparative analysis showed that: ANN prediction method using DSFT results obtained than the use of DSFT The projection fitting method is closer to the relatively true result of CSFT, and both methods are suitable for monitoring data without knowing the structure of the system, but the ANN method is more accurate.