Power generation systems with monitoring for anomaly detection via nonlinear relationship modeling

Patent metadata

Field	Value
Publication number	US-10340734-B2
Application number	US-201715680796-A
Country	US
Kind code	B2
Filing date	Aug 18, 2017
Priority date	Aug 22, 2016
Publication date	Jul 2, 2019
Grant date	Jul 2, 2019

Abstract

A power generator system with anomaly detection and methods for detecting anomalies include a power generator that includes one or more physical components configured to provide electrical power. Sensors are configured to make measurements of a state of respective physical components, outputting respective time series of said measurements. A monitoring system includes a fitting module configured to determine a predictive model for each pair of a set of time series, an anomaly detection module configured to compare new values of each pair of time series to values predicted by the respective predictive model to determine if the respective predictive model is broken and to determine a number of broken predictive model, and an alert module configured to generate an anomaly alert if the number of broken predictive models exceeds a threshold.

First claim

What is claimed is:

1. A power generator system with anomaly detection, comprising:
a power generator that includes one or more physical components configured to provide electrical power;
a plurality sensors configured to make measurements of a state of respective physical components, outputting respective time series of said measurements; and
a monitoring system, comprising:
a fitting module configured to determine a predictive model for each pair of a set of time series, to determine polynomial bases for modeling a polynomial relationship between the time series, to solve a corresponding Sparse Group Lasso problem for the set of time series, and to correct coefficients of a solution of the corresponding Sparse Group Lasso problem by linear regression;
an anomaly detection module configured to compare new values of each pair of time series to values predicted by the respective predictive model to determine if the respective predictive model is broken and to determine a number of broken predictive models; and
an alert module configured to generate an anomaly alert if the number of broken predictive models exceeds a threshold.

2. The power generator system of claim 1 , wherein the fitting module is further configured to determine a periodic relationship between the time series and a non-periodic relationship between the time series.

3. The power generator system of claim 1 , wherein the fitting module is further configured to perform a Fourier transform on the time series to separate periodic components from non-periodic components.

4. The power generator system of claim 1 , wherein the fitting module is further configured to determine a time delay between time series that have a periodic relationship.

5. The power generator system of claim 1 , wherein the anomaly detection module is further configured to use only predictive models that have a fitness score higher than a threshold.

6. The power generator system of claim 5 , wherein the fitting module is further configured to calculate a fitness score for each predictive model based on an r-squared error.

7. A method for detecting anomalies in a power generation system, comprising:
measuring a state of one or more physical components of a power generator using a plurality of sensors, outputting respective time series of said measurements;
determining a predictive model for each pair of a set of time series, each time series being associated with a component of a system, including determining a period relationship between the time series and a non-periodic relationship between the time series, wherein determining the non-period relationship between the time series includes determining polynomial bases for modeling a polynomial relationship between the time series, solving a corresponding Sparse Group Lasso problem for the time series, and correcting coefficients of a solution of the corresponding Sparse Group Lasso problem by linear regression;
comparing new values of each pair of time series to values predicted by the respective predictive model to determine if the respective predictive model is broken;
determining a number of broken predictive models; and
generating an anomaly alert if the number of broken predictive models exceeds a threshold.

8. The method of claim 7 , wherein determining the predictive model for a pair of time series further comprises performing a Fourier transform on the time series to separate periodic components from non-periodic components.

9. The method of claim 7 , wherein determining the predictive model for a pair of time series further comprises determining a time delay between time series that have a periodic relationship.

10. The method of claim 7 , wherein comparing values of each pair of time series only uses predictive models that have a fitness score higher than a threshold.

11. The method of claim 10 , further comprising calculating a fitness score for each predictive model based on an r-squared error.

Assignees

• Nec Lab America Inc
• Nec Corp

Inventors

• Yan Tan
• Song Dongjin
• Chen Haifeng
• Jiang Guofei
• Xu Tingyang

Classifications

• G06F2218/20

  by applying autoregressive analysis · CPC title

• G06F18/24

  Classification techniques · CPC title

• G06F18/21345

  enforcing sparsity or involving a domain transformation · CPC title

• G06F17/142

  Fast Fourier transforms, e.g. using a Cooley-Tukey type algorithm · CPC title

• G06N7/08

  using chaos models or non-linear system models · CPC title