Process monitoring and control using self-validating sensors

Details Process monitoring and control using self-validating sensors Process monitoring and control using self-validating sensors

2001-03-23

2004-11-09

Nghiem, Michael (Department: 2863)

Data processing: measuring, calibrating, or testing

Measurement system

Statistical measurement

Reexamination Certificate

active

06816810

ABSTRACT:

This invention relates to process monitoring and control using self-validating sensors.
The term ‘process’ is used herein in its broad control theory context to include controlled devices, plant and controlled systems generally.
The model for self-validating (SEVA) sensors was proposed by Henry and Clarke (1993). It specifies that every sensor should make use of all the information available to generate the following standard metrics for each measurement:
Validated Measurement Value (VMV): this is the best estimate of the measurement that can be provided by the sensor under the current circumstances. It is implied that the sensor should have self-diagnostic capability, and endeavours to correct for the effects of any faults on the ‘raw’ measurement as far as possible.
Validated Uncertainty (VU): if a reported measurement M has an uncertainty of &dgr;M, then by definition the true value of the measurement M
true
should be in the range
M−&dgr;M≦M
true
≦M+&dgr;M
with a certain probability (typically 95%). Methods for estimating the measurement uncertainty under a variety of conditions can be found in Yang (1994). Note that in the presence of a fault, a SEVA sensor is not only required to correct the VMV, but also to adjust the VU so as to cater for the reduction in the confidence of the corrected reading.
MV Status: this reflects the presence and the persistence of any sensor fault, and indicates how the VMV is generated. The MV Status can take any one of the following values:
Clear: the confidence in the ‘raw’ measurement is nominal, and the VMV is generated purely from the current ‘raw’ measurement.
Blurred: a fault has been diagnosed and it impairs measuring capability. The VMV is generated by compensating the current ‘raw’ measurement.
Dazzled: the ‘raw’ measurement is substantially abnormal and the confidence associated with it is zero, but the fault is judged to be temporary (eg, a spike). The VMV is generated from the device's past history.
Blind: a fault that destroys the measuring capability of the sensor has been diagnosed. There is no confidence in the ‘raw’ measurement. The VMV is generated from the device's past history.
Secure: the VMV is obtained by combining redundant SEVA measurements. The confidence in each SEVA measurement is nominal.
Unvalidated: validation is not taking place.
The implementation of a SEVA version of the Foxboro 871 Clark-type dissolved oxygen sensor is described by Clarke and Fraher (1995).
Control using self-validating sensors has been discussed by J. C.-Y. Yang and D. W. Clarke (1966). It was proposed that in the case of a simple feedback loop in which a SEVA sensor provides the feedback signal, it should be possible to make use of the metrics to select strategies to respond appropriately to sensor faults and unfavourable operating conditions. In the case that not only the nature of the fault, but also the bounds of the residual error, can be supplied by the SEVA sensor, then it may be possible for the controller to decide whether the fault effect is substantial enough to require controller re-tuning or reconfiguration.
In practice, most processes require for effective control the monitoring of a large number of plant variables involving the use of a large number of sensors, ie process variable transmitters. Practical control systems have not yet been developed which are able to fully utilise the metrics of the SEVA measurements. For example, a relatively sophisticated plant monitoring system is the CONNOISSEUR™ plant monitoring system version 14.00 sold by Simsci Limited of Stockport, England. The CONNOISSEUR™ 14.00 monitoring system, although capable of being used to monitor a relatively complex plant, such as a fluidised catalytic cracking unit, was not able to make full use of SEVA metrics from the, typical, thirty plus sensors required to monitor such a plant.
We consider that there is a need for a process monitor that is able to interact with one or more SEVA sensors so as to distinguish between actual changes in the process operation and fault conditions in one or more of the SEVA sensors in a hierarchical system.
According to one aspect of the present invention a multi-level (hierarchical) process monitoring system comprises a process monitoring unit, at a higher level of the system, and a plurality of sensors, at least one of the sensors having SEVA capability, at a lower level of the system, the sensors being adapted to provide respective measurement values of respective process variables to said monitoring unit, said monitoring unit being so arranged as to monitor the outputs of the sensors and to identify any significant apparent change in the process conditions as detected from an overview of said sensor outputs, and on detection of an apparent significant change, to request additional status information from at least one of the SEVA sensor/s to determine whether the apparent change is in reality due to a change in the characteristics of a particular SEVA sensor rather than an actual significant change in the process conditions.
The process monitoring system preferably comprises one or more actuators to effect process changes and the actuators are preferably arranged to provide the monitoring unit with actuator position signals.
Preferably most of the sensors are SEVA sensors.
The monitoring unit preferably implements a multivariate statistical analysis of the measurement values of the sensors, and compares the results of that analysis with reference information to identify any significant apparent change in process conditions, to determine whether or not to initiate interrogation of the SEVA sensor/s.
The reference information is preferably comprised of predictions of a model and historical data of stored statistical analyses. Thus, both history and model predictions are desirably used as the bases for characterising ‘normal process operation’.
The model preferably utilises actuator position information.
For convenience, we refer on occasions hereafter to ‘interrogation’ of a SEVA sensor for the process of requesting additional status information from the SEVA sensor.
Although the interrogation may often relate to the mere examination of at least some of the usual sensor outputs of the SEVA sensor, on other occasions, depending on the type of SEVA sensor, interrogation may involve the application of a non-routine test in or to the SEVA sensor. For example, when the SEVA sensor comprises a thermocouple for measuring temperature, interrogation may involve the application of current to the sensor to heat up the sensor to determine whether good thermal contact is present. Such a test would not normally be initiated by the internal software of the SEVA sensor, because the test directly affects the principal output of the sensor. Some tests reduce the life of a SEVA sensor and accordingly are not carried out on a routine basis or only relatively infrequently.
Thus the invention provides interactions between the higher level monitoring unit and the lower level SEVA sensor/s, in which the detection of an apparent significant change in process operation by the monitoring unit initiates interrogation of one or more of the SEVA sensors, and the resulting changes in the SEVA sensor outputs is then analysed by the monitoring unit to determine whether or not the perceived change is an actual process change. In the event that the monitoring unit determines that the perceived change is an actual change in the process conditions then this is preferably arranged to provide an alert signal which may be directed at a process operative or/and may result in attempted corrective action by a process control unit.
According to a second aspect of the invention we provide a process-monitoring unit adapted for use in a process monitoring system in accordance with the first aspect of the invention.
Although the invention is applicable to processes in which the various sensors measure different variables, the invention is also applicable to situations in which at least some SEVA sensors measure the same variable, and the SEVA m

Affiliated with

Also associated with

Rating

Process monitoring and control using self-validating sensors does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Process monitoring and control using self-validating sensors, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Process monitoring and control using self-validating sensors will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3280943