Controlling and maintaining operational status during component failures

Patent metadata

Field	Value
Publication number	US-11048249-B2
Application number	US-201816048273-A
Country	US
Kind code	B2
Filing date	Jul 28, 2018
Priority date	Jul 28, 2017
Publication date	Jun 29, 2021
Grant date	Jun 29, 2021

Abstract

A system, a control unit, and a method for controlling operation of a technical system are provided. The technical system includes a plurality of sensors. The method includes receiving first sensor data from a first sensor of the plurality of sensors. The method includes detecting a first sensor anomaly based on failure of the first sensor to generate the first sensor data. The failure of the first sensor includes generation of anomalous first sensor data. The method also includes validating the first sensor anomaly based on a comparison between the first sensor data and a virtual first sensor data. Thereafter, a control command is generated to the technical system by replacing the virtual first sensor data in lieu of the first sensor data when the first sensor anomaly is validated.

First claim

The invention claimed is:

1. A method for controlling operation of a technical system comprising a plurality of sensors, the method comprising:
generating a system model of the technical system based on a multi-physics probability model in a pre-operation phase of the technical system, wherein the system model is a high fidelity simulation model of the technical system generated based on Bayesian calibration, wherein the system model comprises virtual sensor data for each sensor of a plurality of sensors associated with the technical system, and wherein the virtual sensor data comprises virtual first sensor data;
receiving first sensor data from a first sensor of the plurality of sensors, in an operation phase of the technical system;
detecting a first sensor anomaly based on failure of the first sensor to generate the first sensor data, wherein failure of the first sensor comprises generation of anomalous first sensor data;
validating the first sensor anomaly based on a comparison between the first sensor data and the virtual first sensor data;
validating the first sensor anomaly based on a sensor relationship model, when sensor data from other sensors of the plurality of sensors surrounding the first sensor is aligned with the virtual sensor data, except for the first sensor data; and
generating a control command to the technical system, the generating of the control command comprising replacing the virtual first sensor data in lieu of the first sensor data when the first sensor anomaly is validated.

2. The method of claim 1 , further comprising:
updating the system model with sensor data from the plurality of sensors to reflect a current state of the technical system; and
effecting change in the sensor data, wherein the effecting of change in the sensor data comprises updating input parameters associated with operation of the technical system based on the system model.

3. The method of claim 2 , further comprising:
continuing operation of the technical system based on the virtual first sensor data when the first sensor anomaly is validated.

4. The method of claim 2 , further comprising:
generating the virtual sensor data at a time instant when the plurality of sensors fail to generate the sensor data.

5. The method of claim 1 , further comprising:
determining sensor limits based on operation limits of the technical system using a supervised learning model; and
determining a tolerance deviation for each sensor of the plurality of sensors based on the supervised learning model, wherein the tolerance deviation is an acceptable deviation from the sensor limits.

6. The method of claim 5 , wherein the detecting of the first sensor anomaly based on failure of the first sensor to generate the first sensor data comprises:
comparing a deviation between the first sensor data and the virtual first sensor data with the tolerance deviation; and
detecting the first sensor anomaly when the deviation exceeds the tolerance deviation.

7. The method of claim 1 , further comprising:
detecting a virtual first sensor anomaly when the first sensor anomaly validation is false; and
updating a system model of the technical system with a degradation model associated with the technical system when the virtual first sensor anomaly detection is continuous.

8. The method of claim 1 , further comprising:
determining a sensor sensitivity for each sensor of the plurality of sensors, the determining of the sensor sensitivity comprising performing a perturbation analysis on each sensor of the plurality of sensors; and
generating the sensor relationship model between the plurality of sensors based on the sensor sensitivity using a neural network.

9. A controller for controlling operation of a technical system including a plurality of sensors, the controller comprising:
a receiver configured to receive sensor data from the plurality of sensors, wherein the sensor data includes first sensor data from a first sensor of the plurality of sensors;
at least one processor; and
a memory communicatively coupled to the at least one processor, the memory comprising:
a model generator module configured to generate a system model of the technical system based on a multi-physics probability model in a pre-operation phase of the technical system, wherein the system model comprises virtual sensor data for each sensor of the plurality of sensors, wherein the system model is a high fidelity simulation model of the technical system generated based on Bayesian calibration, and wherein the virtual sensor data comprises virtual first sensor data, and wherein the model generator module is operable to update the system model with sensor data from the plurality of sensors to reflect a current state of the technical system;
an anomaly detection module configured to detect a first sensor anomaly based on failure of the first sensor to generate the first sensor data;
a validation module configured to validate the first sensor anomaly based on a comparison between the first sensor data and virtual first sensor data and to validate the first sensor anomaly based on a sensor relationship model, when the sensor data from other sensors of the plurality of surrounding the first sensor is aligned with the virtual sensor data, except for the first sensor data; and
a sensor selection module configured to output the virtual first sensor data in lieu of the first sensor data when the first sensor anomaly is validated,

wherein the at least one processor is configured to generate a control command to the technical system based on the virtual first sensor data, the generation of the control command comprising replacement of the virtual first sensor data in lieu of the first sensor data when the first sensor anomaly is validated.

10. The controller of claim 9 , wherein the anomaly detection module comprises:
a virtual anomaly detection module configured to detect a virtual first sensor anomaly when the first sensor anomaly validation is false,
wherein a system model of the technical system is updated with a degradation model associated with the technical system when the virtual first sensor anomaly is detected.

11. The controller of claim 9 , wherein the memory further comprises:
a model generator module configured to generate a system model of the technical system based on a multi-physics probability model, wherein the system model comprises virtual sensor data for each sensor of the plurality of sensors, and wherein the virtual sensor data comprises the virtual first sensor data, and
wherein the model generator module is operable to update the system model with sensor data from the plurality of sensors to reflect a current state of the technical system.

12. The controller of claim 9 , wherein the memory further comprises:
a sensor limit module configured to determine sensor limits based on operation limits of the technical system using a supervised learning model; and
a tolerance deviation module configured to determine a tolerance deviation for each sensor of the plurality of sensors based on the supervised learning model,
wherein the tolerance deviation is acceptable deviation from the sensor limits.

13. The controller of claim 9 , wherein the memory further comprises:
a sensor sensitivity module configured to perform a perturbation analysis on each sensor of the plurality of sensors to determine sensor sensitivity for each sensor of the plurality of sensors; and
a sensor relationship module configured to generate the sensor relationship model between the plurality of sensors based on the sensor sensitivity using a neural network.

14. The controller of claim

Assignees

• B R Chethan Ravi
• Mathew Bony
• Obst Birgit
• Siemens Ag

Inventors

• B R Chethan Ravi
• Mathew Bony
• Obst Birgit

Classifications

• G05B23/0262 Primary

  Confirmation of fault detection, e.g. extra checks to confirm that a failure has indeed occurred · CPC title

• G05B23/0221 Primary

  Preprocessing measurements, e.g. data collection rate adjustment; Standardization of measurements; Time series or signal analysis, e.g. frequency analysis or wavelets; Trustworthiness of measurements; Indexes therefor; Measurements using easily measured parameters to estimate parameters difficult to measure; Virtual sensor creation; De-noising; Sensor fusion; Unconventional preprocessing inherently present in specific fault detection methods like PCA-based methods · CPC title

• G05B9/02

  electric · CPC title

• G05B23/0213

  Modular or universal configuration of the monitoring system, e.g. monitoring system having modules that may be combined to build monitoring program; monitoring system that can be applied to legacy systems; adaptable monitoring system; using different communication protocols · CPC title

• G01D5/24457

  Failure detection · CPC title