초록 ▼

A system and method for surveillance of an industrial process. The system and method includes a plurality of sensors monitoring industrial process parameters, devices to convert the sensed data to computer compatible information and a computer which executes computer software directed to analyzing t

A system and method for surveillance of an industrial process. The system and method includes a plurality of sensors monitoring industrial process parameters, devices to convert the sensed data to computer compatible information and a computer which executes computer software directed to analyzing the sensor data to discern statistically reliable alarm conditions. The computer software is executed to remove serial correlation information and then calculate Mahalanobis distribution data to carry out a probability ratio test to determine alarm conditions.

대표청구항 ▼

A system for surveillance of an industrial process to assess at least one of an operational state of the industrial process and the operational state of sensors monitoring the process, comprising: at least one sensor means for sensing raw data information from the industrial process corresponding to

A system for surveillance of an industrial process to assess at least one of an operational state of the industrial process and the operational state of sensors monitoring the process, comprising: at least one sensor means for sensing raw data information from the industrial process corresponding to the operational state of the industrial process; means for converting the raw data information to computer data; computer means for receiving the computer data and generating sensor data analyzable by the computer means; computer software means executable by the computer means for carrying out manipulation of the sensor data, said computer software means including, (a) means for removing serial correlation information between any of the means for sensing raw data information and (b) means for calculating Mahalanobis distribution data and determining a statistical deviation from previously accumulated training data for the industrial process, wherein the training data is characteristic of raw data information with the means for sensing operating normally and the industrial process properly functioning; and means for applying a probability ratio test to the Mahalanobis distribution data to determine alarm conditions for said industrial process and said sensor means.