Battery Health Quantification for TDRS Spacecraft by Using ... › archive › nasa ›...

EXPLORATION AND SPACE COMMUNICATIONS PROJECTS DIVISION NASA GODDARD SPACE FLIGHT CENTEREXPLORATION AND SPACE COMMUNICATIONS PROJECTS DIVISION NASA GODDARD SPACE FLIGHT CENTER

Battery Health Quantification for TDRS Spacecraft by Using Signature Discriminability Measurement

Presenter: Kenneth Y. Ma

https://ntrs.nasa.gov/search.jsp?R=20200001625 2020-07-24T11:54:10+00:00Z

EXPLORATION AND SPACE COMMUNICATIONS PROJECTS DIVISION NASA GODDARD SPACE FLIGHT CENTER

Introduction: Tracking and Data Relay Satellites (TDRS)• NASA’s Goddard Space Flight Center (GSFC) operates a constellation of ten active

geosynchronous Tracking and Data Relay Satellites (TDRS). The primary mission of TDRS constellation is to provide relay communications from low-Earth orbiting spacecraft to the primary ground stations.

• The Space Network project office at GSFC has managed the constellation of spacecraft over the past 30 years.

• This paper presents a new concept of battery analysis called telemetry quality quantification (TQQ) and discusses the progress that has been made in battery performance estimation for the second-generation TDRS spacecraft.

• This activity is important because many of the TDRS fleet of spacecraft have exceeded their on-orbit design lifetime and, therefore, NASA must carefully manage the spacecraft to continue operations while avoiding an end-of-mission scenario that leaves a non-functioning spacecraft in geosynchronous orbit.


Tracking and Data Relay Satellites (TDRS)

Sensor Extract Telemetry

DataModulation

Transmission

Demodulation

On-Orbit

Ground Station

Raw Telemetry

Data

DataStorage

DataProcessing


Telemetry Data Quantification Analysis (TDQA)• TDQA is quite different from traditional analysis currently being used, such as expert model and limit

checking.• TDQA makes use of hundreds of different telemetry parameters to characterize spacecraft’s health

situation and function of behavior by using quantification algorithm.• It provides many advantages and benefits that current analysis method cannot. Such as, fast, easy to

implement and real time processing.

174:12:34 174:16:07 174:19:41 174:23:14 175:02:47 175:06:21 175:09:54 175:13:27 175:17:01 175:20:34

Time(Day:Hour:Minute)

0

500

1000

1500

Anom

ily S

core

2009Day174__10:49:01 to Day175__22:07:22

RRXD result

174:12:34 174:16:07 174:19:41 174:23:14 175:02:47 175:06:21 175:09:54 175:13:27 175:17:01 175:20:34

Time(Day:Hour:Minute)

0

10

20

30

40

50

Cur

rent

(I)

50

50.5

51

51.5

52

Volta

ge

2009Day174__10:49:01 to Day175__22:07:22

BVL1

BVL2

SOLPANI

DRVTOTSHI

DRVTOTSCI

BUSVOL

TDQA


Application in TDQA• Application in TDQA

o Subsystem anomalies detectiono Spacecraft behavior classificationo Subsystem anomalies quantificationo Battery quality quantification

• Advantage of TDQAo Fast analyzingo Soft decision rather than hard decisiono Real time processing


Battery Performance evaluation and discriminability measurement (1)• The second generation of TDRS spacecraft contain a single nickel hydrogen battery consisting of four

battery packs with a total of 29 cells with 110 Amp-hour (Ah) total capacity per spacecraft and provide power to the spacecraft during each eclipse and for contingency support.

• In recent years, several battery cells have exhibited divergence indicating cell capacity loss, which can affect the long-term spacecraft reliability.

• The NASA has a great deal of experience with battery reconditioning on the 1st gen satellites NiCad batteries. That design was 3 for 2 redundant allowing one battery to be reconditioned while the other two were available to support the bus.


Example of TDRS Battery during eclipse condition• Nickel hydrogen battery provide reliable discharge processes during

each eclipse season to TDRS spacecraft. • Battery during discharge and recharge process provide considerable

information in regrading to battery performance and remaining cell capacity.

• The analysis of battery performance should be executed by using the telemetry data that battery running discharge or recharge rather than trickle charge.

Day

62 0

2:31

Day

62 0

3:23

Day

62 0

4:14

Day

62 0

5:05

Day

62 0

5:56

Day

62 0

6:47

Day

62 0

7:39

Day

62 0

8:30

Day

62 0

9:21

Day

62 1

0:12

Day

62 1

1:03

Day

62 1

1:55

Day

62 1

2:46

XLabel

3000

3500

4000

4500

5000

5500

kPA

P1PressureA

P1PressureB

-8

-7

-6

-5

-4

-3

-2

-1

0

1

2

3

C°

P1TempA

P1TempB

0

1

2

3

4

5

6

7

Amp

Flight9 Year2011 Pack1

Battery Charge current

Battery Trickle charge

Battery discharge process

Run into Eclipse

End of Eclipse

Battery Charging

Go Back to Trickle Charge

End of Recharge

Day

62 0

2:31

Day

62 0

3:23

Day

62 0

4:14

Day

62 0

5:05

Day

62 0

5:56

Day

62 0

6:47

Day

62 0

7:39

Day

62 0

8:30

Day

62 0

9:21

Day

62 1

0:12

Day

62 1

1:03

Day

62 1

1:55

Day

62 1

2:46

Day

62 1

3:37

Day

62 1

4:28

1.2

1.25

1.3

1.35

1.4

1.45

1.5

1.55

1.6

Volta

ge

0

5

10

15

20

25

30

35

40

45

Amp

Flight9 Year2011 Pack1

Cell1

Cell2

Cell3

Cell4

Cell5

Cell6

Cell7

Cell8

Discharge Current


Battery Performance evaluation and discriminability measurement (2)• One critical issue of TDRS batteries is the challenge of battery cell capacity estimation when the

spacecraft is operating on orbit. Since the batteries never fully discharge in each eclipse condition, telemetry can only observed a partial V/Ah curve. Consequently, the current quality of the battery cells is hard to predict until the failure of a cell appears.

0 10 20 30 40 50 60

AH

1.1

1.15

1.2

1.25

1.3

1.35

1.4

1.45

1.5

Volta

ge

2016 Day 255

BCVMV01

BCVMV04

BCVMV08

BCVMV10

BCVMV12

BCVMV16

BCVMV20

0 10 20 30 40 50 60

AH

1.1

1.15

1.2

1.25

1.3

1.35

1.4

1.45

1.5

Volta

ge

2015 Day 071

BCVMV01

BCVMV04

BCVMV08

BCVMV10

BCVMV12

BCVMV16

BCVMV20

0 10 20 30 40 50 60

AH

1.1

1.15

1.2

1.25

1.3

1.35

1.4

1.45

1.5

Volta

ge

2018 Day 067

BCVMV01

BCVMV04

BCVMV08

BCVMV10

BCVMV12

BCVMV16

BCVMV20


Signature Discriminability Measurement (SDM)• SDM is primarily based on measuring the divergence between two input signals. This can be significantly useful

to compute discriminability between health subsystem signal and failed substance signal.• Euclidean Distance (ED):

• Signature Angle Mapper (SAM):

1 11 12 1( , ,..., )Ls s s= Ts 2 21 22 2( , ,..., )Ls s s= Ts

21 2 1 2 1 22 1

ED( , ) ( )Ll ll

s s=

= − = −∑s s s s

1 211 2

1 2

,SAM( , )=cos−

s ss s

s s

)ED(𝐬𝐬1, 𝐬𝐬2

1s

1s

2s

SAM(𝐬𝐬1,𝐬𝐬2)Geometric illustration for SAM and ED


Signature Discriminability Measurement (SDM)

• Signature Information Divergence (SID) method using relative entropy in information theory to compute discriminability between two signals.

1 1 11/ L

l l llp s s

== ∑ 2 2 21

/ Ll l ll

p s s=

= ∑

1 2 1 1 21( ) log( / )L

l l llD p p p

==∑s s

2 1 2 2 11( ) log( / )L

l l llD p p p

==∑s s

1 2 1 2 2 1SID( , ) ( ) ( )D D= +s s s s s s


SDM Battery Cell Performance Analysis

• In this paper, we focus on the V vs Ah curve analysis to estimate the health of the battery cells during battery discharge process.

• The healthy cell data that we selected as our reference is extracted from TDRS 11 at the launched year. Even though TDRS 11 from a different generation than TDRS 8-10, the batteries that are installed in both spacecraft are similar and can be used as the reference data.

0 10 20 30 40 50

AH

1.2

1.25

1.3

1.35

1.4

1.45

1.5

Volta

ge

F11__f__2013

F8__f__2011

F8__f__2012

F8__f__2013

F8__f__2014

F8__f__2015

F8__f__2016

DCHI( 1) DCHI( )Ah( )2

t tt t− −= ∆


Non-Linear Regression Approach in Estimation of True Cell Signal• Statistically, linear regression is an approach to modeling the relationship between a scalar response

and one or more variables. • Linear or nonlinear regression function:

1 for 1

Nc

i i oi

y x cα α=

= + ≥∑

0 10 20 30 40 50 60

AH

1.2

1.25

1.3

1.35

1.4

1.45

Volta

ge

0 10 20 30 40 50 60

AH

1.2

1.25

1.3

1.35

1.4

1.45

Volta

ge

Observed Cell Voltage Data Estimated Cell Voltage Data


SDM Battery Cell Performance Analysis Results (1)

• SDM applied to quantify current performance of battery cell. The cell with abnormal discriminability when compared to the rest of the cells in same pack indicates the anomalous event. These anomalies could be the failure of a scanner, the leaking of chemical substance or other unknown events.

2008 2010 2012 2014 2016 2018 2020

year

0

0.002

0.004

0.006

0.008

0.01

0.012

SAM

Spring Season F8 Pack 3 SAM

Cell17

Cell18

Cell19

Cell20

Cell21

2008 2010 2012 2014 2016 2018 2020

year

0

0.5

1

1.5

2

2.5

3

SID

10-3 Spring Season F8 Pack 3 SID

Cell17

Cell18

Cell19

Cell20

Cell21

SDM results showing that we can observe some signs of failure

since spring 2014

F8 cell 20 failed at fall 2015


SDM Battery Cell Performance Analysis (2)• Flight 9 results shows that 2 cells in Pack 1 shows significant divergence from rest of the cells in same pack,

which indicate these 2 cells have loss of capacity. Both SAM and SID shown similar results for cell 1 and cell 3.

2008 2010 2012 2014 2016 2018 2020

year

0

0.5

1

1.5

2

2.5

3

SID

10-3 Spring Season F9 Pack 1 SID

Cell1

Cell2

Cell3

Cell4

Cell5

Cell6

Cell7

Cell8

2008 2010 2012 2014 2016 2018 2020

year

0

0.002

0.004

0.006

0.008

0.01

0.012

SAM

Spring Season F9 Pack 1 SAM

Cell1

Cell2

Cell3

Cell4

Cell5

Cell6

Cell7

Cell8

F9 Cell 1 and cell 3 started to diverge

from rest cells since 2016. We can say

these two cells have potential failure.


SDM Battery Cell Performance Analysis (3)• Similar examples can also be found in analysis of TDRS 10 data as shown in Figure. Cell 1 and cell 2 in TDRS

10 also diverge from the rest of the cells beginning at 2010. However, these two cells do not seem to become worse as we can see in 2018. Moreover, we can observe that the ED is not as sensitive as SID. This is because ED only considers the distance between two signals and does not take the signal shape into account.

2008 2010 2012 2014 2016 2018 2020

year

0

0.2

0.4

0.6

0.8

1

ED

F10 fall ED P1

Cell1

Cell2

Cell3

Cell4

Cell5

Cell6

Cell7

Cell8

2008 2010 2012 2014 2016 2018 2020

year

0

0.5

1

1.5

2

2.5

3

SID

10-3 F10 fall SID P1

Cell1

Cell2

Cell3

Cell4

Cell5

Cell6

Cell7

Cell8

F10 Cell 2 and cell 8 show divergence form

other cells since 2010.


Conclusion• Our experiment indicates that the SDM algorithm can be helpful for managing spacecraft batteries

and for quantifying and trending the performance of cells.• A yearly or monthly trending of the data is an important methodology to analyze the performance

of the spacecraft• By using the SDM algorithm to quantify the performance of each individual spacecraft component,

we can easily analyze the spacecraft performance by yearly or monthly trending of the SDM• Continue develop algorithms for analyzing each subsystems to improve the efficiency in

monitoring spacecraft quality.• Future Work:

• Reaction wheel anomaly detection and analysis• Bus voltage limiter (BVL) anomaly detection and analysis.• Thruster anomaly and failure prediction.


Thanks for your attention

Battery Health Quantification for TDRS Spacecraft by Using ... › archive › nasa ›...

Documents

Transcript of Battery Health Quantification for TDRS Spacecraft by Using ... › archive › nasa ›...