Surveillance of industrial processes with correlated parameters
First Claim
1. 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.
4 Assignments
0 Petitions
Accused Products
Abstract
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.
-
Citations
27 Claims
-
1. 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. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17)
-
-
18. A method for performing surveillance of an industrial process using industrial sensors to assess at least one of an operational state of the industrial process and viability of the sensors, comprising the steps of:
-
sensing raw data information from the industrial process using a plurality of sensors; converting the raw data information to computer data and outputting the computer data to a computer; generating analyzable sensor data for the computer; executing computer software using the computer to remove serial correlation information present for each sensor alone between the raw data information accumulated by each of the sensors at different times; accumulating training data for the industrial process and the sensors characteristic of raw data information known to be representative of a properly functioning sensor and industrial process; further executing the computer software to calculate a statistical deviation from the training data by determining Mahalanobis distribution data for both the raw data information and the training data; and applying a probability ratio test to determine alarm conditions for abnormal states of the industrial process and the sensors. - View Dependent Claims (19, 20, 21, 22, 23, 24, 25, 26, 27)
-
Specification