Detecting anomalous sensors

1. An apparatus comprising:

a plurality of sensors;

a processor coupled to the plurality of sensors; and

one or more computer readable mediums collectively including instructions that, when executed by the processor, cause the processor to

monitor the plurality of sensors,

obtain a plurality of healthy sensor data, each of the healthy sensor data includes a plurality of sensed values of a corresponding sensor among the plurality of sensors in normal operation and does not include data from anomalous sensors,

generate a healthy data distribution of at least two sensors among the plurality of sensors corresponding to the plurality of healthy sensor data, and

generate a function of a parameter probability distribution of sensor parameters of the plurality of sensors under a condition of sensor data based on the healthy data distribution, each parameter indicating whether the corresponding sensor is healthy or anomalous.

2. The apparatus of claim 1 , wherein the instructions further cause the processor to:

estimate a score of each parameter from a new sensor data.

3. The apparatus of claim 2 , wherein the instructions further cause the processor to estimate a parameter probability distribution from the new sensor data.

4. The apparatus of claim 3 , wherein the instructions further cause the processor to approximate the parameter probability distribution with a Bernoulli distribution.

5. The apparatus of claim 4 , wherein the instructions further cause the processor to estimate the parameter probability distribution from the new sensor data by using a Bernoulli distribution as a prior distribution of the plurality of parameters.

6. The apparatus of claim 5 , wherein the instructions further cause the processor to estimate the parameter probability distribution from the new sensor data by using a coordinate descent method.

7. The apparatus of claim 2 , wherein each score represents at least one probability that the corresponding sensor is anomalous.

8. The apparatus of claim 2 , wherein the instructions further cause the processor to detect an anomalous sensor by comparing the score of each parameter with a threshold value.

9. The apparatus of claim 1 , wherein the instructions further cause the processor to generate the function of a parameter probability distribution further based on an anomalous data distribution, wherein the anomalous data distribution is approximated with a uniform distribution.

10. The apparatus of claim 1 , wherein the instructions further cause the processor to generate the healthy data distribution of a pair of sensors according to a partition of the plurality of sensors wherein the distribution is two dimensional.

11. The apparatus of claim 1 , wherein each sensor of the plurality of sensors is configured to sense a physical quantity.

12. A computer-implemented method comprising:

monitoring a plurality of sensors to obtain a plurality of healthy sensor data, each of the healthy sensor data includes a plurality of sensed values of a corresponding sensor among a plurality of sensors in normal operation and does not include data from anomalous sensors;

generating a healthy data distribution of at least two sensors among the plurality of sensors corresponding to the plurality of healthy sensor data; and

generating a function of a parameter probability distribution of sensor parameters of the plurality of sensors under a condition of sensor data of the plurality of sensors based on the healthy data distribution, each parameter indicating whether the corresponding sensor is healthy or anomalous.

13. The computer-implemented method of claim 12 , further comprising:

estimating a score of each parameter from a new sensor data.

14. The computer-implemented method of claim 13 , further comprising:

estimating a parameter probability distribution from the new sensor data.

15. The computer-implemented method of claim 14 , further comprising:

approximating the parameter probability distribution with a Bernoulli distribution.

16. The computer-implemented method of claim 15 , further comprising:

estimating the parameter probability distribution from the new sensor data by using a Bernoulli distribution as a prior distribution of the plurality of parameters.

17. A computer program product comprising including one or more computer readable mediums collectively including instructions that, when executed by the processor, cause the processor to:

monitor a plurality of sensors to obtain a plurality of healthy sensor data, each of the healthy sensor data includes a plurality of sensed values of a corresponding sensor among a plurality of sensors in normal operation and does not include data from anomalous sensors;

generate a healthy data distribution of at least two sensors among the plurality of sensors corresponding to the plurality of healthy sensor data; and

generate a function of a parameter probability distribution of sensor parameters of the plurality of sensors under a condition of sensor data of the plurality of sensors based on the healthy data distribution, each parameter indicating whether the corresponding sensor is healthy or anomalous.

18. The computer program product of claim 17 , wherein the instructions further cause the processor to:

estimate a score of each parameter from a new sensor data.

19. The computer program product of claim 18 , wherein the instructions further cause the processor to estimate a parameter probability distribution from the new sensor data.

20. The computer program product of claim 19 , wherein the instructions further cause the processor to approximate the parameter probability distribution with a Bernoulli distribution.