of a Battery Module Type

The invention relates to a method for enhancing a battery module model of a battery module type.

The invention relates further to a device and a system carry ing out the method according to the invention.

The aging of a Lithium-Ion battery pack is a very important issue, which two factors contribute essentially. The first, calendar aging is due to the progression of time since the manufacturing of the cell or group of cells within a battery pack. The second is aging due to charge-discharge cycles. However, this second aging factor is a highly non-linear pro cess in which many factors such as temperature, charge/ dis charge rate, and changes in charge/ discharge rate play a role .

This lack of information regarding the battery aging can re sult in incorrect capacity estimations for state of charge algorithms .

In prior art battery aging and capacity was estimated by spe cific offline measurements to assess the remaining battery capacity. Such measurements include Open Circuit Volt age (OCV) measurement to deduce Internal resistance. Addi tionally, low C-rate charge-discharge measurements can be made to establish remaining capacity.

The objective of the invention is to provide an enhanced bat tery module model of a battery module type in order to allow accurately to estimate the effective age of a battery module of the battery module type without interrupting the continu ous usage of the module under investigation.

The objective of the invention is solved by the method ac cording to the preamble of claim 1, comprising the steps of: a) exposing a battery module of the battery module type including a multiplicity of battery cells to a first environment and measuring at the first environment at least one initial battery parameter based on at least one time parameter, b) setup and training of the battery module model of the battery module type with the at least one initial bat tery parameter based on the at least one time parame ter and based on machine learning techniques,

c) operating the battery module at a second environment with changing environmental conditions and capturing at least one operating battery parameter,

d) training the battery module model with the at least one operating battery parameter based on the at least one time parameter,

e) calculating an EOL parameter relating to an end-of-life prediction of the battery module. f) exposing the battery module to a third environment and measuring at the third environment at least one final battery parameter based on the at least one time param eter,

g) training of the battery module model with the at least one final battery parameter based on the at least one time parameter.

Thus, the estimation on active/in field batteries can be per formed without interfering with day-to-day operations.

In other words, the battery module of the battery module type including a multiplicity of battery cells is exposed during practicing the method subsequently to

• a first environment and measurements are performed there,

• a second environment and measurements are performed there,

• a third environment and measurements are performed there, whereas these three environments are different from each oth er .

The first environment can be for instance the production en vironment, where the battery module is manufactured or in a lab. There, preferably constant environmental conditions ap ply. The second environment can be for instance the operating en vironment, where the battery module acts as a power source, e.g. at an aircraft.

There, charge/ discharge measurements are performed during the operation of the battery module, e.g. at a car, an air craft or a mobile computer, delivering as in-field battery reporting the operating parameter for current and the operat ing parameter for voltage at the operating environment.

Consequently, non-constant environmental conditions can apply to the second environment. Thus, at the second environment the environmental conditions, like temperature or loads etc., can change during the duration of the exposure to the second environment .

Preferably, the second environment is applied during opera tion of the battery module, i.e. while providing power to an electric load. No controlled charge/discharge cycles are ap plied then and the charge/discharge "mode" of the electric load, e.g. a generator or a motor, connected to the battery module is used.

The third environment can be for instance a final environ ment, e.g. in the lab, where the module is removed from its operating environment, e.g. before battery recycling. There, preferably constant environmental conditions apply.

In prior art, subsequent measurements for building a model are only known at equal conditions, i.e. comparable environ ments .

According to the invention, the support of different environ ments during the measuring steps adds flexibility and lowers the costs of gathering the required measurement data.

Each environment can be characterized by individual tempera ture, air pressure, humidity, electrical load, mechanical load, etc.

This is achieved by the remote transmission of in use battery characteristics to a remote server for processing eliminating the constraint set by low power processors typically found within local battery management systems. Further, the method enhances safety by detecting battery packs which may be undergoing enhanced degradation due to manufacturing, environmental or abnormal usage patterns.

This is achieved by the aging prediction model which can de tect such packs as described.

Moreover, the down-time and costs are minimized due to unrec ognized battery packs which have decreased beyond useable life .

This is achieved as the battery Information is automatically made available and comparison to a reference pack made an op timized maintenance schedule can be derived.

Additionally, by predicting aging and capacity on a pre trained model means the battery pack under observation does not need to be removed for lab tests.

In one further development of the invention it is foreseen, that the at least one initial battery parameter and/or the at least one operating battery parameter and/or the at least one final battery parameter is related to a voltage, a current and/or a charge of the battery module or at least one cell of the multiplicity of battery cells.

In another further development of the invention it is fore seen, that the measuring step f) is performed by applying at least one controlled discharge on the battery module and/ or at least one cell of the multiplicity of battery cells.

In another further development of the invention it is fore seen, that after the measuring step f) the at least one final battery parameter is compared with the at least one initial battery parameter based on Bayesian classification, which classification results are used at the training step g) .

The objective of the invention is solved by a device for pre dicting the effective age of a battery module including a multiplicity of battery cells, wherein the device is config ured to

• receive at least one battery parameter,

• setup a battery module model based on machine learn ing techniques using the at least one battery parame ter, • train the battery module model with the at least one battery parameter set,

• output an end-of-life prediction parameter of the

battery module based on the battery module model,

• carry out the step after the measuring step f) ac

cording the method of the invention.

The objective of the invention is solved by a system for pre dicting the effective age of a battery module, wherein the system comprises a battery module including a multiplicity of battery cells and a prediction device according to the device according to the invention.

It is clear, that further not shown parts are necessary for the operation of a battery module e.g. electronic control components or measurement equipment. For the sake of better understanding these parts are not illustrated and described.

The invention is described by an embodiment in the accom plished figures in detail. The drawings show in

Fig. 1 a schematic illustration of a battery module of the battery module type including a multiplicity of battery cells,

Fig. 2 a schematic illustration of an embodiment of the steps of the method according to the invention,

Fig. 3 a further schematic illustration of the method according to Fig. 1,

Fig. 4 a schematic illustration of an embodiment of a device according to the invention.

Fig . 1 shows a schematic illustration of a battery module 1 of the battery module type including a multiplicity of bat tery cells 2.

Fig . 2 shows the method for enhancing a battery module model of a battery module type, comprising the steps of: a) exposing a battery module 1 of the battery module type including a multiplicity of battery cells 2 to a first environment 10 and measuring at the first environment 10 at least one initial battery parameter 11-13 based on at least one time parameter 3, b) setup and training of the battery module model of the battery module type with the at least one initial battery parameter 11-13 based on the at least one time parame ter 3 and based on machine learning techniques,

c) operating the battery module 1 at a second environment 20 with changing environmental conditions and capturing at least one operating battery parameter 21, 22,

d) training the battery module model with the at least one operating battery parameter 21, 22 based on the at least one time parameter 3,

e) calculating an EOL parameter 40 relating to an end-of- life prediction of the battery module 1, wherein, if the EOL parameter 40 exceeds a predetermined threshold value, continue with step f) , else continue with step c) , f) exposing the battery module 1 to a third environment 30 and measuring at the third environment 30 at least one final battery parameter 31, 32 based on the at least one time parameter 3,

g) training of the battery module model with the at least one final battery parameter 31, 32 based on the at least one time parameter 3.

Within step a) a set of battery modules can be used for providing initial data for the setup of the battery module model in step b) .

The at least one initial battery parameter 11-13 and the at least one operating battery parameter 21, 22 and the at least one final battery parameter 31, 32 are related to a voltage, a current and a charge of the battery module 1 and at least one cell of the multiplicity of battery cells 2.

The measuring step f) is performed by applying at least one controlled discharge on the battery module 1, i.e. at least one cell of the multiplicity of battery cells 2.

After the measuring step f) the at least one final battery parameter 31, 32 is compared with the at least one initial battery parameter 11-13 based on Bayesian classification, which classification results are used at the training

step g) .

The first environment 10 and the third environment 30 can be the same, e.g. at a lab.

As a result, an aged battery module at end-of-life can pro vide important data for the battery module model for another battery module of the same battery type during its ongoing operation .

Fig . 3 shows an alternative schematic illustration 100 of the embodiment of the steps of the method according to the inven tion .

Within step a) controlled charge/ discharge measurements 110 are performed, delivering the initial parameters for cur rent 11, the initial parameters for voltage 12 and the ini tial parameters for charge 13 at the initial environment 10 based on the time parameter 3.

The time parameter 3 can be a reference to all other measure ments, i.e. measurements can be normalized to the time param eter 3.

Within step b) the battery module model is setup and trained, i.e. an aging prediction model training 120 is performed, with the initial battery parameters 11-13 based on the time parameter 3 and based on machine learning techniques.

Within step c) charge/ discharge measurements 140 are per formed during the operation of the battery module, e.g. at a car, an aircraft or a mobile computer, delivering as in-field battery reporting 140 the operating parameter for current 21 and the operating parameter for voltage 22 at the operating environment 20. Thus, at the second environment the environ mental conditions can change during the duration of the expo sure to the second environment.

Subsequently, an aging model prediction 130 is performed, where the end-of-life parameter EOL, 40 is calculated and compared 150 to a pre-set value, e.g. 70%, 75% or 80% of the original, initial capacity of the battery module. The pre-set value depends on the application where the battery module is used and can vary over a wide range.

When end-of-life of the battery module is reached in

path 151, further controlled charge/ discharge measure ments 160 are performed in step f) , delivering the final pa rameters for current 31 and the final parameters for volt age 32 at the final environment 30 based on the time parame ter 3.

As long the end-of-life of the battery module in path 152 is not yet reached, the battery operation will continue in path 170. i.e. the "regular" battery operation within its ap plication is continued, and periodically step c) is repeated.

Fig . 4 depicts an embodiment of a device 5 according to the invention, which can be used in a system as well. The de vice 5 for enhancing a battery module model of a battery mod ule type is configured to:

• receive 200 at least one battery parameter 11-13, 21,

22, 31, 32 of a battery module 1 which includes a multi plicity of battery cells,

• setup 210 a battery module model based on machine learn ing techniques using the at least one battery parame ter 11-13,

• train 220 the battery module model with the at least one battery parameter set 11-13, 21, 22, 31, 32,

• output 230 an end-of-life prediction parameter 40 of the battery module 1 based on the battery module model,

• carry out 240 after the measuring step f) a comparison of the at least one final battery parameter 31, 32 with the at least one initial battery parameter 11-13 based on Bayesian classification, which classification results are used at the training step g) .

Thus, the device 5 provides an improved battery module model of a battery module 1 and/or its EOL parameter 40. Reference Numeral s :

1 battery module

2 multiplicity of cells of the battery module

3 time parameter

4 battery model

5 prediction device

10 initial environment

11 initial parameter for current

12 initial parameter for voltage

13 initial parameter for charge

20 operating environment

21 operating parameter for current

22 operating parameter for voltage

30 final environment

31 final parameter for current

32 final parameter for voltage

40 EOL parameter

100 method

110 Controlled charge/ discharge measurements

120 Aging prediction model training

130 Aging model prediction

14 0 In-field battery reporting

150 End of life?

151 Yes

152 No

1 60 Extract battery and perform lab measurement

17 0 actually no action of battery model adaption

200 receive measurement data as parameters

210 setup a battery module model

220 train the battery module model

230 output an end-of-life prediction parameter

24 0 carry out the method of the invention