# US 12,001,199 B2 · Method, system and computer program product for evaluation of energy consumption in industrial environments

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 12,001,199 B2
- Document ID: US12001199B2
- Status: Granted
- Kind: B2
- Application number: 17/420,265
- Filed: 2019-12-30
- Published: 2022-03-24
- Granted: 2024-06-04
- Earliest claimed: 2019-01-02
- Assignees: Gestamp Servicios, S.A.
- Applicants: Gestamp Servicios, S.A.; FundacióN Tecnalia Research and Innovation
- Inventors: Iñaki Grau Unda; Alicia Rubio Gaitán; Sergio Gil López; Eugenio Perea Olavarría; Izaskun Mendia Tellería; Adelaida Lejarazu Marquina
- Web page: https://patentgrep.com/patent/US12001199B2

## CPC

- G05B 23/0224: {Process history based detection method, e.g. whereby history implies the availability of large amounts of data}
- G05B 23/027: {Alarm generation, e.g. communication protocol; Forms of alarm}
- Y02P 90/82: Energy audits or management systems therefor
- G05B: CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS (systems for controlling or regulating non-electric variables G05D; systems for regulating electric or magnetic variables G05F; control devices or systems insofar as characterised by mechanical features only G05G)
- Y02P: CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS

## Abstract

A method for evaluating energy consumption in an industrial plant, includes capturing sensor data from at least one level of the industrial plant, from the sensor data, digitally obtaining energy consumption curves, an energy consumption curve representing, along a certain time period, discrete values of energy consumption corresponding to time intervals Δt. The method also includes applying a clustering algorithm for digitally computing K energy consumption patterns, wherein each pattern represents a set of energy consumption curves grouped according to a similarity metric and includes discrete values of energy consumption corresponding to time intervals Δt; capturing data of production achieved during the time period at at least one level; calculating the aggregated sum of the discrete values, thus obtaining the aggregated energy consumption for each curve; and digitally establishing a relationship between aggregated energy consumption for each energy consumption curve during the time period and captured data of production.

## Claims

### Claim 1

A computer-implemented method for evaluating the energy consumption in an industrial plant, the method including the following steps:
at a plurality of time instants, capturing with a plurality of sensors, sensor data from at least one level of the industrial plant, wherein said plurality of sensors comprises energy measuring sensors,
from said sensor data, digitally obtaining a plurality of energy consumption curves xj, wherein j=0, 1, 2, . . . , J-1, J being a natural number, an energy consumption curve representing, along a certain time period T, discrete values of energy consumption corresponding to time intervals Δt into which said time period T is divided,
from said energy consumption curves xj, applying a clustering algorithm for digitally computing a plurality of K energy consumption patterns Ck, wherein k=0, 1, 2, . . . , K-1, K being a natural number, K<J, wherein each energy consumption pattern Ck represents a set of energy consumption curves xj grouped together according to a similarity metric, wherein each energy consumption pattern Ck comprises discrete values of energy consumption corresponding to said time intervals Δt into which said time period T is divided,
capturing data of the production achieved during said time period T at said at least one level of the industrial plant,
calculating, for each energy consumption curve belonging to each pattern, the aggregated sum of the discrete values of energy consumption during said time period T, thus obtaining the aggregated energy consumption for each energy consumption curve during said time period T, and
digitally establishing a relationship between aggregated energy consumption for each energy consumption curve during said time period T and said captured data of production.

### Claim 2 (depends on 1)

The method of claim 1, wherein stage of capturing with a plurality of sensors, sensor data from at least one level of the industrial plant, is done at periodic time instants separated at time intervals Δt.

### Claim 3 (depends on 1)

The method of claim 1, wherein said at least one level of the industrial plant is at least one of industrial plant, line, process, machine or component of a machine.

### Claim 4 (depends on 1)

The method of claim 1, wherein each component of energy consumption pattern Ck is calculated as the mean value of corresponding components xij of each energy consumption curve xj belonging to said pattern.

### Claim 5 (depends on 1)

The method of claim 1, wherein said clustering algorithm applies the following iterative process for minimizing the distance between elements forming a cluster and its corresponding pattern:
arg min{Σi=1I∥xij−Cikt∥∀k=1, . . . ,K}
wherein xij is the i-th component of the j-th energy consumption curve xj, Cikt is the i-th component of the k-th energy consumption pattern at time instant t, K is the total number of energy consumption patterns and I is the number of measurements within the whole time period,
wherein the optimum number K of patterns for the whole set of curves xj ∀ j=1, . . . , J is given by the point of maximum deflection in the representation of the variance between groups divided by the total variance of the selection from the set of curves, that is to say:
argmax
⁢
{
(
VAR
in
k
+
1
VAR
)
-
(
VAR
in
k
VAR
)
}
,
∀
k
=
1
,
…
⁢
,
K
wherein VARin is the variance between the elements of a group or cluster, and
wherein VARin is defined:
VARin=Σk=1Knk(Ckt−Ct)/(K−1),
wherein K is the total number of energy consumption patterns, that is to say, the total number of groups of energy consumption curves related by a similarity metric defined for a group, nk is the number of elements forming the k-th group or curves, Ckt is the characteristic pattern for the k-th group of curves and Ct is the mean pattern of the whole selection of energy consumption curves.

### Claim 6 (depends on 1)

The method of claim 1, wherein for a new energy consumption curve obtained from sensor data at a new time period T, the degree of similarity/dissimilarity with respect to current patterns is evaluated by comparing the new energy consumption curve with the K patterns already defined and, if the new energy consumption curve mathematically fits in a previously defined pattern in terms of a similarity metric, it is determined that said new energy consumption curve belongs to said pattern in which the new energy consumption curve fits.

### Claim 7 (depends on 6)

The method of claim 6, wherein if the new energy consumption curve mathematically fits in a previously defined pattern in terms of a similarity metric, the new energy consumption curve is digitally compared with all the energy consumption curves that belong to the cluster represented by the pattern, thus comparing the new curve with the curve of maximum energy consumption and with the curve of minimum energy consumption.

### Claim 8 (depends on 6)

The method of claim 6, wherein if the new energy consumption curve belongs to an existing energy consumption pattern, the new energy consumption curve is associated to the group of curves represented by said pattern and the pattern is updated taking into account the new energy consumption curve.

### Claim 9 (depends on 6)

The method of claim 6, wherein if the new energy consumption curve does not fit in a previously defined pattern in terms of a similarity metric, determining if the new energy consumption curve represents an anomalous consumption or if the new energy consumption curve represents the appearance of a new type of energy consumption pattern.

### Claim 10 (depends on 9)

The method of claim 9, wherein if the new energy consumption curve is determined to represent an anomalous consumption, an alarm associated to the energy consumption is triggered.

### Claim 11 (depends on 10)

The method of claim 10, further including the step of using sub-metering measurements to interpret the difference in energy consumption causing the anomalous consumption.

### Claim 12 (depends on 6)

The method of claim 6, further including the following steps:
capturing data of the production achieved during the same time period T of the new energy consumption curve, and
digitally comparing the energy consumption and associated production during said time period of the new energy consumption curve, with the relationship energy versus production inferred from historic curves.

### Claim 13 (depends on 12)

The method of claim 12, wherein if as a result of the digital comparison energy inefficiency is determined, an alarm associated to the production is triggered.

### Claim 14 (depends on 1)

A system comprising processing means, the system being configured to perform the steps of the method of claim 1.