Automatic rotating-machine fault diagnosis with confidence level indication
Automatic fault diagnosis is performed on vibration data sensed from a machine. A set of faults to screen for is identified from the machine configuration. For each fault there are characteristic symptoms. For each characteristic symptom, there is a corresponding indication used to diagnose the symptom. The indications are based on analyses of the current vibration data. The diagnosed symptoms have weights assigned according to a Bayesian network, and are used to derive a Bayesian probability for the fault. A fault having a Bayesian probability exceeding a threshold value is identified as being present in the machine. For each fault a confidence level is derived. The confidence level for a first fault is based on a similarity between characteristic symptoms for the first fault and characteristic symptoms for each one of the other faults being screened.
1. A rotating-machine fault diagnostic system comprising:
a rotating machine comprising a housing and a plurality of select parts, the plurality of select parts including a shaft, a rotor and a bearing, the rotating machine having a plurality of test points;
a sensor unit that collects diagnostic data of the rotating machine when mounted at any test point of said plurality of test points;
a portable data collection and analysis device comprising a processor, a memory, an embedded camera, and a display;
a photographic image stored in memory for each one test point of the plurality of test points, said photographic image depicting an appropriate sensor unit position and an appropriate sensor unit orientation for data collection by said sensor unit at said one test point;
wherein said portable data collection and analysis device is configured to perform the steps of:
determining a machine configuration corresponding to the plurality of select parts present on the rotating machine by identifying the rotating machine based on one of either an operator input or by receiving a signal from a scan of a code embedded on the rotating machine;
for each one test point of the plurality of test points to be tested with said sensor unit,
loading a test point prescription for said one test point, said one test point being a location on the rotating machine for diagnosing a corresponding select part among the plurality of select parts of the rotating machine,
displaying setup instructions from the test point prescription for installing said sensor unit at said one test point,
the processor loading and the display displaying a corresponding photographic image from the photographic images stored in memory that shows said appropriate position and orientation for positioning and orienting at said one test point said sensor unit, wherein said sensor unit at said one test point is removably mounted by an operator so as to be in accordance with said appropriate position and orientation shown in said corresponding photographic image and said setup instructions;
receiving diagnostic data from said sensor unit for said one test point while mounted at said one test point;
configuring the portable data collection and analysis device to represent the machine configuration of the identified rotating machine as a naive bayesian network which is independent of the received diagnostic data;
for each one part of said plurality of select parts to be evaluated using said sensor unit,
deriving a probability of presence of a potential fault for said one part based on the received diagnostic data and the naive bayesian network, said potential fault having one or more characteristic symptoms, and each one of said one or more characteristic symptoms having a symptom weight, and
determining a confidence level C(i) for presence of said potential fault i for said one part as follows:
C
(
i
)
=
1
-
∑
j
≠
i
P
j
×
Similarity
(
i
,
j
)
∑
j
≠
i
P
j
where P j is the derived probability of presence of potential fault j; and
Similarity
(
i
,
j
)
=
InterCount
(
i
,
j
)
UnionCount
(
i
,
j
)
where
InterCount
(
i
,
j
)
=
∑
k
∈
(
S
i
⋂
S
j
)
1
-
w
(
i
,
k
)
-
w
(
j
,
k
)
and
UnionCount
(
i
,
j
)
=
S
i
+
S
j
-
InterCount
(
i
,
j
)
and
where w(i, k)∈[0,1] is the probability that characteristic symptom k of potential fault i is present;
∥S i ∥ denotes the number of characteristic symptoms in the set S i of characteristic symptoms for the potential fault i; and
∥S j ∥ denotes the number of characteristic symptoms in the set S j of characteristic symptoms for potential fault j,
wherein a greater dis-similarity between sets of characteristic symptoms, as weighted, for said potential fault i and said potential faults for each other part of said select parts corresponds to a higher confidence level; and
diagnosing that said one part has a corresponding deteriorating condition, when the probability of presence P of said potential fault i for said one part exceeds a first prescribed threshold and the confidence level for the probability of presence P of said potential fault i for said one part exceeds a second prescribed threshold.
2. The system of claim 1 , wherein said portable data collection and analysis device is configured to perform image capturing, including for each one test point of said plurality of test points the steps of:
capturing with said embedded camera said corresponding photographic image by imaging said one test point on the machine as configured to show said appropriate sensor unit position and said appropriate sensor unit orientation for the sensor unit; and
storing said photographic image in memory.
3. The system of claim 1 , wherein the portable data collection and analysis device is configured to perform the further step of:
displaying a recommendation to perform maintenance on said one part based on the diagnosis that said one part has a corresponding deteriorating condition.
4. The system of claim 1 , wherein the portable data collection and analysis device is configured to perform the further step of:
displaying a warning recommending that complementary analysis be performed when the probability of presence of said potential fault for said one part exceeds the first prescribed threshold and the corresponding confidence level does not exceed the second prescribed threshold.
5. The system of claim 1 , wherein the sensor unit is a sensor and further comprising a wire that connects the sensor to the portable data collection and analysis device.
6. The system of claim 1 , wherein the portable data collection and analysis device further comprises a pyrometer;
wherein said portable data collection and analysis device is configured to further perform receiving diagnostic data from said pyrometer; and
wherein among the characteristic symptoms of said potential fault i of said one part is a first characteristic symptom based on diagnostic data received from the pyrometer.
7. The system of claim 1 , wherein the portable data collection and analysis device further comprises a stroboscope;
wherein said portable data collection and analysis device is configured to further perform receiving diagnostic data from said stroboscope; and
wherein among the characteristic symptoms of said potential fault i of said one part is a first characteristic symptom based on diagnostic data received from the stroboscope.
8. The system of claim 1 , wherein said diagnositc data comprises machine vibration data that represent sample measurements of vibration of the rotating machine.
9. The system of claim 1 , wherein said diagnositc data comprises machine temperature data that represent sample measurements of temperature of the rotating machine.
10. A rotating-machine fault diagnostic system comprising:
a rotating machine comprising a housing and a plurality of select parts, the plurality of select parts including a shaft, a rotor and a bearing, the rotating machine having a plurality of test points;
one or more sensors that collect diagnostic data of the rotating machine from among the plurality of test points;
a portable data collection and analysis device comprising a processor, a memory, an embedded camera, and a display;
wherein said portable data collection and analysis device is configured to perform the steps of:
for each one test point of the plurality of test points to be tested, receiving diagnostic data from a corresponding sensor unit among said plurality of sensor units into said memory;
configuring the portable data collection and analysis device to represent the machine configuration of the identified rotating machine as a naive bayesian network which is independent of the received diagnostic data;
for each one part of said plurality of select parts to be evaluated using said received diagnostic data,
deriving a probability of presence of a potential fault for said one part based on the received diagnostic data and the naive bayesian network, said potential fault having one or more characteristic symptoms, and each one of said one or more characteristic symptoms having a symptom weight, and
determining a confidence level C(i) for presence of said potential fault i for said one part as follows:
C
(
i
)
=
1
-
∑
j
≠
i
P
j
×
Similarity
(
i
,
j
)
∑
j
≠
i
P
j
where P j is the derived probability of presence of potential fault j; and
Similarity
(
i
,
j
)
=
InterCount
(
i
,
j
)
UnionCount
(
i
,
j
)
where
InterCount
(
i
,
j
)
=
∑
k
∈
(
S
i
⋂
S
j
)
1
-
w
(
i
,
k
)
-
w
(
j
,
k
)
and
UnionCount
(
i
,
j
)
=
S
i
+
S
j
-
InterCount
(
i
,
j
)
and
where w(i, k)∈[0,1] is the probability that characteristic symptom k of potential fault i is present;
∥S i ∥ denotes the number of characteristic symptoms in the set S i of characteristic symptoms for the potential fault i; and
∥S j ∥ denotes the number of characteristic symptoms in the set S j of characteristic symptoms for potential fault j,
wherein a greater dis-similarity between sets of characteristic symptoms, as weighted, for said potential fault i and said potential faults for each other part of said select parts corresponds to a higher confidence level; and
diagnosing that said one part has a corresponding deteriorating condition, when the probability of presence P j of said potential fault i for said one part exceeds a first prescribed threshold and the confidence level for the probability of presence P j of said potential fault i for said one part exceeds a second prescribed threshold.
11. The system of claim 10 , wherein said one or more sensors comprise a first sensor unit removably mounted to the rotating machine so as to collect diagnostic data from a test point among the plurality of test points, the first sensor unit removably mounted at different locations on the machine to collect diagnostic data from differing test points among said plurality of test points.
12. The system of claim 10 , further comprising a wire that connects a first sensor of said one or more sensors to the portable data collection and analysis device.
13. The system of claim 10 , wherein the portable data collection and analysis device further comprises a pyrometer, and the pyrometer is a first sensor among said one or more sensors.
14. The system of claim 10 , wherein the portable data collection and analysis device further comprises a stroboscope, and the stroboscope is a first sensor among said one or more sensors.
15. The system of claim 10 , wherein the portable data collection and analysis device is configured to perform the further step of:
displaying a recommendation to perform maintenance on said one part based on the diagnosis that said one part has a corresponding deteriorating condition.
16. The system of claim 10 , wherein the portable data collection and analysis device is configured to perform the further step of:
displaying a warning recommending that complementary analysis be performed when the probability of presence of said potential fault for said one part exceeds the first prescribed threshold and the corresponding confidence level does not exceed the second prescribed threshold.
17. The system of claim 10 , wherein said diagnositc data comprises machine vibration data that represent sample measurements of vibration of the rotating machine.
18. The system of claim 10 , wherein said diagnositc data comprises machine temperature data that represent sample measurements of temperature of the rotating machine.