提高传感器故障检测能力的研究

Research on Improving Fault Detectability of Sensor

提高传感器故障检测系统检测小故障的能力,对提高工业过程的安全和经济性有重要意义.通过推导证明:对过程数据进行指数加权滑动平均(EWMA)滤波后,(T2H)指标的阈值保持不变,而T2H指标相对于滤波前得到放大,因此滤波后T2H指标的故障检测精度得到提高,有利于及时检测到更小的传感器故障,提高了T2H指标检测小故障的能力,降低误报率,更易区分传感器的偏差故障和漂移故障.由于EWMA滤波会滤除一定的过程噪声,因此不推荐采用滤波方法检测传感器精度下降的故障.

To improve the detectability of small faults in sensor fault detection systems is of great importance for improving security and economy of industry processes. The derivationproves that, after process data have been treated by an exponentially weighted moving average （EWMA） filter, the threshold value of Hawkins statistics （TH^2） can be kept constant, while TH^2 is being amplified by the filtering. Therefore the accuracy of fault detection can be improved by using the filtered TH^2, which makes even smaller sensor faults to be detectable in time. Moreover, improving the detectability of small sensor fault based on T~ is helpful for reducing false alarm rate and facility＇s identification of bias and drift faults. Because the EWMA filter is able to remove certain process noises, this method is not recommended for sensor fault detection in case it affects the accuracy of sensor.