| 摘要 |
PURPOSE:To detect that abnormality other than noise was generated in a measuring signal without delay, by smoothing a measuring signal by an adaptive type filter and further calculating a differential signal showing steepness of the wave form of the smoothed value and the estimation error of low frequency noise and a present value by a differential filter, a noise estimation device and a square deviation calculator. CONSTITUTION:The high frequency noise of a measuring signal is removed by an adaptive type filter constituted of a combination of Haar conversion and inverse Haar conversion and a differential signal showing the steepness of a wave form and a square deviation signal showing the estimation error of low frequency noise and a present value are further calculated by estimation calculation and square deviation calculation due to differentiation and a self- regression model and, by using both of them, it is detected that abnormality other than noise was generated in the measuring signal. At the same time, a pulse like signal with a specific shape is extracted by a Haar signal to judge whether abnormality is generated and, therefore, the presence of an abnormal point is detected without delay to issue and alarm and detection capacity can be enhanced. |
