US 10,884,383 B2Granted

Advanced control systems for machines

AssigneeSAS Institute Inc.

InventorsMichael James Leonard, David Bruce Elsheimer, Yuelei Sui

Life of this patent

Estimating the term…

Loading patent…

Abstract

Machines can be controlled using advanced control systems that implement an automated version of singular spectrum analysis (SSA). For example, a control system can perform SSA on a time series having one or more time-dependent variables by: generating a trajectory matrix from the time series, performing singular value decomposition on the trajectory matrix to determine elementary matrices; and categorizing the elementary matrices into groups. The elementary matrices can be automatically categorized into the groups by: generating one or more w-correlation matrices based on spectral components associated with the time series, determining w-correlation values based on the one or more w-correlation matrices; categorizing the w-correlation values into a predefined number of w-correlation sets, and forming the groups based on the predefined number of w-correlation sets. The control system can then generate a predictive forecast using the groups and control operation of a machine using the predictive forecast.

Claims27

Dependent claims

A method comprising:

receiving, by a processing device, a time series having one or more time-dependent variables, the time series being associated with a machine;
performing, by the processing device, singular spectrum analysis on the time series at least partially by:
generating a trajectory matrix from the time series, the trajectory matrix being a multi-dimensional representation of the time series, wherein the multi-dimensional representation of the time series has at least one dimension that is dependent on how many time-dependent variables are present in the time series;
performing singular value decomposition on the trajectory matrix to generate elementary matrices;
generating spectral components associated with the time series by performing diagonal averaging on the elementary matrices;
and
automatically categorizing the elementary matrices into a plurality of groups by:
generating one or more w-correlation matrices based on the spectral components, each w-correlation matrix being generated by determining weighted correlations between a respective pair of the spectral components associated with a respective time-dependent variable present in the time series;
determining w-correlation values based on the one or more w-correlation matrices;
categorizing the w-correlation values into a predefined number of w-correlation sets such that, for each w-correlation set in the predefined number of w-correlation sets, all of the w-correlation values in the w-correlation set are above a predefined threshold value;
and
forming the plurality of groups based on the predefined number of w-correlation sets, each respective group in the plurality of groups including a respective subset of the elementary matrices corresponding to the w-correlation values in a respective w-correlation set among the predefined number of w-correlation sets;
and
determining a plurality of component time-series based on the plurality of groups;
generating, by the processing device, a predictive forecast using the plurality of component time-series;
predicting, by the processing device, an event related to the machine using the predictive forecast;
and
controlling, by the processing device, the machine by modifying an operational setting of the machine in response to the predicted event.

15 more elements

A system comprising:

a processing device;
and
a memory device including instructions that are executable by the processing device for causing the processing device to perform operations comprising:
receiving a time series having one or more time-dependent variables, the time series being associated with a machine;
performing singular spectrum analysis on the time series at least partially by:
generating a trajectory matrix from the time series, the trajectory matrix being a multi-dimensional representation of the time series, wherein the multi-dimensional representation of the time series has at least one dimension that is dependent on how many time-dependent variables are present in the time series;
performing singular value decomposition on the trajectory matrix to generate elementary matrices;
generating spectral components associated with the time series by performing diagonal averaging on the elementary matrices;
and
automatically categorizing the elementary matrices into a plurality of groups by:
generating one or more w-correlation matrices based on the spectral components, each w-correlation matrix being generated by determining weighted correlations between a respective pair of the spectral components associated with a respective time-dependent variable present in the time series;
determining w-correlation values based on the one or more w-correlation matrices;
categorizing the w-correlation values into a predefined number of w-correlation sets such that, for each w-correlation set in the predefined number of w-correlation sets, all of the w-correlation values in the w-correlation set are above a predefined threshold value;
and
forming the plurality of groups based on the predefined number of w-correlation sets, each respective group in the plurality of groups including a respective subset of the elementary matrices corresponding to the w-correlation values in a respective w-correlation set among the predefined number of w-correlation sets;
and
determining a plurality of component time-series based on the plurality of groups;
generating a predictive forecast using the plurality of component time-series;
predicting an event related to the machine using the predictive forecast;
and
controlling the machine by modifying an operational setting of the machine in response to the predicted event.

18 more elements

A non-transitory computer-readable medium comprising instructions that are executable by a processing device for causing the processing device to perform operations including:

receiving a time series having one or more time-dependent variables, the time series being associated with a machine;
performing singular spectrum analysis on the time series at least partially by:
generating a trajectory matrix from the time series, the trajectory matrix being a multi-dimensional representation of the time series, wherein the multi-dimensional representation of the time series has at least one dimension that is dependent on how many time-dependent variables are present in the time series;
performing singular value decomposition on the trajectory matrix to generate elementary matrices;
generating spectral components associated with the time series by performing diagonal averaging on the elementary matrices;
and
automatically categorizing the elementary matrices into a plurality of groups by:
generating one or more w-correlation matrices based on the spectral components, each w-correlation matrix being generated by determining weighted correlations between a respective pair of the spectral components associated with a respective time-dependent variable present in the time series;
determining w-correlation values based on the one or more w-correlation matrices;
categorizing the w-correlation values into a predefined number of w-correlation sets such that, for each w-correlation set in the predefined number of w-correlation sets, all of the w-correlation values in the w-correlation set are above a predefined threshold value;
and
forming the plurality of groups based on the predefined number of w-correlation sets, each respective group in the plurality of groups including a respective subset of the elementary matrices corresponding to the w-correlation values in a respective w-correlation set among the predefined number of w-correlation sets;
and
determining a plurality of component time-series based on the plurality of groups;
generating a predictive forecast using the plurality of component time-series;
predicting an event related to the machine using the predictive forecast;
and
controlling the machine by modifying an operational setting of the machine in response to the predicted event.

15 more elements

Dependencies are inferred from the claim text.

Facts

Application: 16/388,119Filed 18 Apr 2019 · continuation in part
Priority: 19 Oct 2016
Published: US 2019/0250569 A115 Aug 2019
Granted: 5 Jan 2021
Term ends: Not estimatedThe source records don’t support an estimate.

Classification15 codes

G05B 13/0255
Not using a model or a simulator of the controlled system: the criterion being a time-optimal performance criterion
G05B 11/01
Automatic controllers: electric
G05B 13/0265
Electric: the criterion being a learning criterion
G05B 13/048
Involving the use of models or simulators: using a predictor

Family

Application62/409,964Us provisional application · filed 19 Oct 2016
Application62/433,002Us provisional application · filed 12 Dec 2016
PCT/US2017/056777Parent of this continuation-in-part · filed 19 Oct 2017
US 2019/0250569 A1Pre-grant publication · published 15 Aug 2019
US 10,884,383 B2This patent

From the US filing history. Foreign family members aren’t listed.