EP0436312A2
Expired / Lapsed
EP
⏱️ 0.0 years remaining
Diagnostic expert system monitor
Assignee: Westinghouse Electric Corp
•
Filing: 1990-12-06
•
Granted: —
•
Expires: 2010-12-06
Abstract
Abstract An online valve monitoring system senses operational characteristics of a valve and provides a diagnosis of aberrations in the characteristics. Sensor signals from the valve are transmitted to a diagnostic processor which extracts characteristic features from the sensor signals as deviations from baseline data for the valve or operational values for predetermined reference operations. An expert system applies a rule base to the characteristic features or operational values; the expert system being updated in a largely automatic process by collecting operational and maintenance data which is linked together to produce data-derived rules. Criteria are defined for comparing the data-derived rules and pre-existing rules in the rule base. All data-derived rules meeting the criteria and, if only single-valued conclusion parameters are in the rules, the pre-existing rules meeting the criteria are placed on the agenda. Each of the rules on the agenda are compared with all of the other rules or, when a multi-valued conclusion parameter is used in the criteria, with all of the pre-existing rules in a pool of rules meeting the criteria. For each pair of rules, if the conclusions conflict and the premises are not mutually exclusive (or independent if a multi-valued conclusion parameter is used in the criteria), the pair of rules are placed on a conflict list which is output to a human expert for evaluation to produce an updated rule base.
Inventors
- 👤 Jan Jozef Schreurs
- 👤 James Moore Forker
- 👤 Fred Harold Bednar
- 👤 Frank Gerald Arcella
- 👤 Neil Pessall
Bibliographic Details
Application Number
—
Jurisdiction
EP
Data Source
google_patents
Last Synchronized
2026-10-26 09:15
CPC Classifications
G; G06; G06N; G06N5/00; G06N5/02; G06N5/022; G06F; G06F17/00; G05; G05B; G05B23/00; G05B23/02; G05B23/0205; G05B23/0218; G05B23/0224; G05B23/0227; G05B23/0229
Claims Analysis
Total Claims: 28
Claim 1
Independent Claim
93 words
An online plant operation monitoring system for monitoring operational units (20) in a plant, comprising: sensor means (24, 28, 31) for continuously sensing conditions of the operational units (20) in the plant to produce sensor signals; data acquisition means (40, 42, 45) for transmitting the sensor signals; and processing means (44, 46, 48, 50) for evaluating the sensor signals by application of an expert system rule base, characterized by: modification means (46) in said processing means (44, 46, 48, 50) for modifying the expert system rule base in dependence upon the sensor signals.
Claim 2
Dependent Claim
91 words
An online plant operation monitoring system as recited in claim 1, further characterized by said data acquisition means comprising: at least one radio transmitter (45), operatively connected to at least one of said sensor means (42), for receiving the sensor signals from the at least one of said sensor means (42) and transmitting radio signals containing the sensor signals; and a radio receiver (45), operatively connected to said processing means (44, 46, 48, 50), for receiving the radio signals and supplying the sensor signals to said processing means (44, 46,48, 50).
Claim 3
Dependent Claim
51 words
An online plant operation monitoring system as recited in claim 2, wherein the plant includes a containment structure substantially enclosing the at least one sensor means (42) and said radio transmitter (45), and wherein said radio receiver (45) and said processing means (44, 46, 48, 50) are outside the containment structure.
Claim 4
Dependent Claim
33 words
An online plant operation monitoring system as recited in claim 2, wherein said radio transmitter (45) comprises multiplexer means (45) for multiplexing the sensor signals prior to the generation of the radio signals.
Claim 5
Dependent Claim
23 words
An online valve monitoring system as recited in claim 2, wherein said radio transmitter (45) and said radio receiver (45) use frequency modulation.
Claim 6
Dependent Claim
117 words
An online plant operation monitoring system as recited in claim 1, wherein said processing means (44, 46, 48, 50) is further characterized by: valuation means (44) for determining operational values corresponding to each of the sensor signals for predetermined reference operations of the operational units (20); and expert system means (50) for applying operational variation rules, containing design basis values, in the expert system rule base to the operational value of each of the sensor signals to provide a first diagnosis of a first malfunction capable of causing said valuation means (44) to generate the operational values, and said modification means (46) modifies the expert system rules in dependence upon the operational values and operational variation rules.
Claim 7
Dependent Claim
204 words
An online plant operation monitoring system as recited in claim 6, wherein said processing means (44, 46, 48, 50) further comprises characteristic feature means (46) for extracting characteristic features of the sensor signals from each of the operational units (20) by comparing the sensors signals with baseline data to determine deviations therefrom in proportion to the baseline data, whereby the characteristic features are dimensionless, and for providing a significance indication for each of the characteristic features in dependence upon the magnitude of the deviations between the sensor signals and the baseline data corresponding thereto, wherein said expert system means (50) further evaluates the characteristic features by application of characteristic feature rules in the expert system rule base to provide a second diagnosis of a second malfunction capable of causing the deviations in the sensor signals for each operational unit (20) corresponding thereto and compares the deviations corresponding to each operational unit (20) to determine a confidence factor for the second diagnosis thereof, and wherein said online plant operation monitoring system is further characterized by selection means (52) for selecting between application of the operational variation rules and the characteristic feature rules in said expert system means (50) for a set of the sensor signals.
Claim 8
Dependent Claim
124 words
An online plant operation monitoring system as recited in claim 7, wherein said operational units (20) comprise valves (20) in a piping system, said valves (20) having a valve stem (30) actuated by a motor (26), and wherein said sensor means (24, 28, 31) associated with said valves includes at least three sensors (24, 28, 31), comprising: a motor current sensor (24) for detecting current drawn by the motor (26) operating a corresponding valve (20); a strain sensor (28), directly coupled to the valve stem (30), for sensing strain of the valve stem (30) of the corresponding valve (20); and a position sensor (31), coupled to the valve stem (30), for sensing a position of the valve stem (30) of the corresponding valve (20).
Claim 9
Dependent Claim
19 words
An online plant operation monitoring system as recited in claim 6, wherein the operational units (20) are valves (20).
Claim 10
Independent Claim
69 words
A method of monitoring operational units in a plant, comprising the steps of: (a) continuously sensing (72) conditions of the operational units in the plant to produce sensor signals; (b) transmitting the sensor signals; and (c) evaluating (74, 82) the sensor signals by application of an expert system rule base, said method characterized by: (d) modifying (108, 112) the expert system rule base in dependence upon the sensor signals.
Claim 11
Dependent Claim
104 words
A method as recited in claim 10, further characterized by: step (c) comprising the steps of: (c1) determining (74) an operational value corresponding to each of the sensor signals for predetermined operations of the operational units; and (c2) applying (82) operational variation rules, containing design basis values, in the expert system rule base to the operational value of each of the sensor signals to provide a first diagnosis of a first malfunction capable of causing generation of the operational values in step (c1), and step (d) comprising modifying (108, 112) the expert system rules in dependence upon the operational values and operational variation rules.
Claim 12
Dependent Claim
185 words
A method as recited in claim 11, further characterized by: step (c) further comprising the steps of: (c3) extracting (74) characteristic features of the sensor signals from each of the operational units by comparing the sensor signals with baseline data to determine deviations therefrom in proportion to the baseline data, whereby the characteristic features are dimensionless; (c4) providing (74) a significance indication for each of the characteristic features in dependence upon the magnitude of the deviations between the sensor signals and the baseline data corresponding thereto; (c5) evaluating (82) the characteristic features by application of characteristic feature rules in the expert system rule base to provide a second diagnosis of a second malfunction capable of causing the deviations in the sensor signals for each operational unit corresponding thereto; and (c6) comparing (82) the deviations corresponding to each operational unit to determine a confidence factor for the second diagnosis thereof, and said method further comprising the step of (e) selecting (56) between application of the operational variation rules and the characteristic feature rules in the expert system rule base for a set of the sensor signals.
Claim 13
Independent Claim
107 words
A method of comparing rules in first and second rule bases of an expert system, each of the rules containing a premise and a conclusion, said method characterized by: (a) storing the first and second rule bases; (b) defining (114) a criteria for conclusions to be used for conflict evaluation; (c) forming (116) an agenda containing all of the rules in the first and second rule bases having conclusions meeting the criteria defined in step (b); (d) comparing (120, 128) each of the rules on the agenda with all of the other rules on the agenda to identify conflicting rules; and (e) outputting (132) the conflicting rules.
Claim 14
Dependent Claim
123 words
A method as recited in claim 13, wherein step (b) comprises the step of inputting (114) a single-valued parameter used in at least one of the conclusions of the rules in the first rule base, wherein step (c) comprises the step of automatically placing (116) all of the rules in the first and second rule bases, having the single-valued parameter in the conclusion thereof, on the agenda, and wherein step (d) comprises the step of automatically comparing (120, 128) each of the rules on the agenda with all of the other rules on the agenda to detect all conflicting pairs of rules having overlapping premises, where the single-valued parameter in the conclusion of each rule in the pairs of rules has different values.
Claim 15
Dependent Claim
53 words
A method as recited in claim 13, wherein step (d) comprises the step of identifying (128) a nonconflicting pair of rules when a condition in the premise of one of the nonconflicting pair of rules is contrary to a condition in the premise of the other rule in the nonconflicting pair of rules.
Claim 16
Independent Claim
180 words
A method of updating pre-existing rules in an expert system rule base used to analyze a domain, each of the pre-existing rules having a premise and a conclusion, said method characterized by: (a) storing (102) domain characteristic data capable of being input, if necessary after conversion, to the expert system to produce a diagnosis; (b) inputting (102) actual condition data, corresponding to the domain characteristic data, indicating actual conditions existing in the domain when the domain characteristic data was generated; (c) automatically generating (104) data-derived rules from the domain characteristic data and the actual condition data, each data-derived rule having a premise and a conclusion; (d) defining (114) a set of conclusions for conflict evaluation; (e) forming (116) an agenda containing all of the pre-existing rules and the data-derived rules having conclusions in the set of conclusions defined in step (d); (f) comparing (118, 120, 122, 124, 126, 130) each of the pre-existing and data-derived rules on the agenda with all other pre-existing and data-derived rules on the agenda to identify conflicting rules; and (g) outputting (132) the conflicting rules.
Claim 17
Dependent Claim
74 words
A method as recited in claim 16, wherein step (f) comprises the steps of: (f1) popping (118) a top rule from the agenda for use as a current rule, thereby making an immediately following rule the top rule; (f2) comparing (120) the current rule with all of the pre-existing and data-derived rules remaining on the agenda; and (f3) repeating (124, 130) steps (f1) and (f2) until only the top rule remains on the agenda.
Claim 18
Dependent Claim
122 words
A method as recited in claim 16, wherein step (d) comprises the step of inputting (114) a single-valued parameter used in at least one of the conclusions of the pre-existing rules, wherein step (e) comprises the step of automatically placing (116) all of the pre-existing and data-derived rules, having the single-valued parameter in the conclusion thereof, on the agenda, and wherein step (f) comprises the step of automatically comparing (120) each of the pre-existing and data-derived rules on the agenda with all of the other pre-existing and data-derived rules on the agenda to detect all conflicting pairs of rules having overlapping premises, where the single-valued parameter has different values in the conclusion of each rule in each of the pairs of rules.
Claim 19
Dependent Claim
48 words
A method as recited in claim 16, wherein step (f) comprises the step of identifying (126) a non-conflicting pair of rules when a condition in the premise of one of the non-conflicting rules is contrary to a condition in the premise of the other of the non-conflicting rules.
Claim 20
Independent Claim
131 words
A method of comparing rules in first and second rule bases of an expert system, each of the rules containing a premise and a conclusion, said method characterized by: (a) storing the first and second rule bases; (b) defining (114) a criteria for conclusions to be used for conflict evaluation; (c) forming (116a) an agenda containing all of the rules in the first rule base having conclusions meeting the criteria defined in step (b); (d) forming (116b) a pool of all of the rules in the second rule base having conclusions meeting the criteria defined in step (b); (e) comparing (120', 122, 124', 126', 128, 130') each of the rules on the agenda with all of the rules in the pool to identify conflicting rules; and (f) outputting the conflicting rules.
Claim 21
Dependent Claim
129 words
A method as recited in claim 20, wherein step (b) comprises the step of inputting (114) a multi-valued parameter used in at least one of the conclusions of the rules in the second rule base, wherein steps (c) and (d) comprises the step of automatically placing (116a, 116b) all of the rules in the first and second rule bases, having the multi-valued parameter in the conclusion thereof, on the agenda and in the pool, respectively, and wherein step (e) comprises the step of automatically comparing (120') each of the rules on the agenda with all of the rules in the pool to detect all conflicting pairs of rules having overlapping premises, where the multi-valued parameter in the conclusion of each rule in the pairs of rules has contradictory values.
Claim 22
Independent Claim
195 words
A method of updating pre-existing rules in an expert system rule base used to analyze a domain, each of the pre-existing rules having a premise and a conclusion, said method characterized by: (a) storing (102) domain characteristic data capable of being input, if necessary after conversion, to the expert system to produce a diagnosis; (b) inputting (102) actual condition data, corresponding to the domain characteristic data, indicating actual conditions existing in the domain when the domain characteristic data was generated; (c) automatically generating (104) data-derived rules from the domain characteristic data and the actual condition data, each data-derived rule having a premise and a conclusion; (d) defining (114) a set of conclusions for conflict evaluation; (e) forming (116b) a pool containing all of the pre-existing rules having conclusions in the set of conclusions defined in step (d); (f) forming (116a) an agenda containing all of the data-derived rules having conclusions in the set of conclusions defined in step (d); (g) comparing (118, 120', 122, 124', 126', 130') each of the data-derived rules on the agenda with all of the pre-existing rules in the pool to identify conflicting rules; and (h) outputting (132) the conflicting rules.
Claim 23
Dependent Claim
76 words
A method as recited in claim 22, wherein step (g) comprises the steps of: (g1) popping (118) a top rule from the agenda for use as a current agenda rule, thereby making an immediately following rule the top rule on the agenda; (g2) comparing (120', 122, 124', 126') the current agenda rule with all of the pre-existing rules in the pool; and (g3) repeating (130') steps (g1) and (g2) until no rules remain on the agenda.
Claim 24
Dependent Claim
75 words
A method as recited in claim 23, wherein step (g2) comprises the step of identifying (126') a non-conflicting pair of rules when a condition in the premise of the current agenda rule is contrary to a condition in the premise of one of the rules in the pool and when a condition in the premise of the current agenda rule is independent of a condition in the premise of the other of the non-conflicting rules.
Claim 25
Dependent Claim
126 words
A method as recited in claim 22, wherein step (d) comprises the step of inputting (114) a multi-valued parameter used in at least one of the conclusions of the pre-existing rules, wherein steps (e) and (f) comprise the steps of automatically placing (116a, 116b) all of the pre-existing and data-derived rules, having the multi-valued parameter in the conclusion thereof, in the pool and on the agenda, and wherein step (g) comprises the step of automatically comparing (118, 120', 122, 124', 126', 130') each of the data-derived rules on the agenda with all of the pre-existing rules in the pool to detect all conflicting pairs of rules having overlapping premises, wherein the multi-valued parameter is contradictory in the conclusion of each rule in the pairs of rules.
Claim 26
Dependent Claim
129 words
A method as recited in claim 22, wherein the conclusion in each of the pre-existing and data-derived rules includes at least one diagnosis, wherein step (d) comprises the step of inputting (114) a specified diagnosis, wherein steps (e) and (f) comprise the steps of automatically placing (116a, 116b) all of the pre-existing and data-derived rules, containing a reference to the specified diagnosis, on the agenda and in the pool, respectively, and wherein step (g) comprises the step of comparing (118, 120', 122, 124', 126', 130') each of the data-derived rules on the agenda with all of the pre-existing rules in the pool to detect all conflicting pairs of rules having overlapping premises, when the specified diagnosis is inconsistent in the rules in each of the conflicting pairs of rules.
Claim 27
Dependent Claim
96 words
A method as recited in claim 26, wherein steps (e) and (f) comprise storing (116a, 116b) an array for each of the pre-existing and data-derived rules, respectively, each element of the array having a value representing a probability that a corresponding diagnosis is true, wherein step (d) comprises the step of specifying (114) one of the elements in each array created in steps (e) and (f), and wherein step (g) comprises comparing (120') the values of the one of the elements in the array specified in step (d) in determining whether the specified diagnosis is inconsistent.
Claim 28
Dependent Claim
68 words
A method as recited in claim 27, wherein the value in each element of each array created in steps (e) and (f) is an integer between -100 and 100, where -100 represents that the corresponding diagnosis is certainly false, 100 represents that the corresponding diagnosis is certainly true, and 0 indicates that the corresponding diagnosis is unrelated to the premise of a corresponding rule represented by the array.
References Cited (Backward Citations)
No citation records indexed.
USPTO Annuity & Maintenance Windows
Maintenance schedule is calculated for granted US utility patents.