Modular Energy Storage System

Abstract

An energy storage system has at least one string of N modules, with each module including an energy storage device and a switching unit configured to for either serially connect the energy storage device into the string or to provide a short circuit. The energy storage system additionally includes a controller configured to perform (during on-load operation of the ESS) the steps of: changing the state of at least one switching unit of a module; measuring a current and a voltage at the energy storage device of the module, and determining characteristics of the energy storage device on a basis of at least a current through the string and change over time of the voltage measured before and after change of the state of the switching unit.

Claims

1. An energy storage system comprising: at least one string of N modules with an integer N>1, the at least one string comprising at least one first end and at least one a second end, wherein each module comprises: at least one input and at least one output, wherein the at least one output of the (n)-th module is connected to the at least one input of the (n+1)-th module for each integer n with 0<n<N, the input of the first module is connected to the at least one first end and the output of the (n+1)-th module is connected to the at least one second end; an energy storage device; a switching unit configured to switch between at least two states of operation including connecting the energy storage device between the at least one input and the at least one output, and providing a short circuit between the at least one input and the at least one output; and a controller configured to perform the following steps during on-load operation of the ESS: to change a state of at least one switching unit of the P-th module P.sub.m with 0<P<=N, to measure a current I and a voltage V.sub.mP at an energy storage device P of the P-th module P.sub.m, and to determine characteristics of the energy storage device P on a basis of at least the current I and a change over time of said voltage V.sub.mP measured before and after a change of the state of the at least one switching unit of the P-th module P.sub.m, wherein at least one sample of the voltage V.sub.mP is taken before the change of said state and a plurality of samples of the voltage V.sub.mP is taken after the change of said state, and at least one sample of current I is taken before and/or after the change of said state.

2. An energy storage system of claim 1, wherein said controller is configured to take at least one sample of current I before the state of the at least one switching unit is changed from said connecting the energy storage device between the at least one input and the at least one output to said providing a short circuit between the at least one input and the at least one output, and to take at least one sample of current I after the state of the at least one switching unit is changed from said providing a short circuit between the at least one input and the at least one output to said connecting the energy storage device between the at least one input and the at least one output.

3. An energy storage system of claim 1, wherein said controller is further configured to determine said characteristics of the energy storage device P on a basis of an estimated state of charge (SOC) of such energy storage device.

4. An energy storage system of claim 1, wherein said controller is configured to average measured values of the voltage V.sub.mP and current I over multiple changes of the state of the at least one switching unit of the P-th module P.sub.m, and/or to calculate multiple equivalent circuit parameters based on the measured values of the voltage V.sub.mP and current I.

5. An energy storage system of claim 1, wherein said controller is configured to repeatedly change the state of at the least one switching unit of the P-th module P.sub.m and/or wherein the controller is further configured to control the energy storage device P to charge or discharge to a predefined state of charge (SOC) level or to a predefined voltage.

6. An energy storage system of claim 1, wherein the switching unit is configured to select, in the state of said connecting the energy storage device between the at least one input and at least one output a polarity of the energy storage device.

7. An energy storage system of claim 1, wherein the controller is further configured to change states of corresponding switching units of a subset of M modules with M<=N, wherein such changes of the states are made such that all modules are used over time in a balanced manner to achieve a balanced state of charge (SOC) for all energy storage devices with an exception of at least the module P.sub.m of an energy storage device P, said at least the module P.sub.m being comparatively unbalanced to achieve charging or discharging that is faster than that of the remaining modules.

8. An energy storage system of claim 1, wherein determined characteristics of the energy storage device P comprise at least one or more parameters of an equivalent circuit including an internal resistance of the energy storage device P at one or more state of charge (SOC) levels.

9. An energy storage system of claim 1, wherein each energy storage device includes at least one of: a battery, a battery cell, a battery pack, a fuel cell, a stack of fuel cells, a solid state battery, and a high-energy capacitor.

10. An energy storage system of claim 1, wherein at least one switching unit is configured to switch at least two energy storage devices in series and/or in parallel, and wherein the at least one switching unit comprises at least one of: a three pole switch, a half bridge, wherein a half-bridge comprises two switches; two half-bridges; and one or two full bridges, wherein each full bridge comprises four switches and/or a battery switch to bypass a corresponding energy storage device.

11. An energy storage system of claim 1, wherein the controller comprises: a plurality of controller units, each controller unit being associated with one or more modules, and one or more measurement units configured to measure at least one of the current through and the voltage at a corresponding energy storage device.

12. A method for determining characteristics of energy storage devices of an energy storage system (ESS) during an on-load operation of said ESS, said ESS comprising: at least one string of N modules with an integer N>1, the string comprising at least one first end and at least one second end, each module comprising: at least one input and at least one output, wherein the at least one output of the (n)-th module is connected to the at least one input of the (n+1)-th module for each integer n with 0<n<N, the at least one input of the first module is connected to the at least one first end and the at least one output of the (n+1)-th module is connected to the at least one second end; an energy storage device; a switching unit, a controller, wherein the method comprises the steps of: changing a state of a switching unit of the P-th module P.sub.m with 0<P<=N either by connecting a corresponding energy storage device P of the P-th module Pm between the at least one input and the at least one output, or by providing a short circuit between the at least one input and the at least one output; measuring a current I and a voltage V.sub.mP at the energy storage device P of the P-th module Pm, and determining characteristics of the energy storage device P on a basis of at least the current I and a change over time of said voltage V.sub.mP measured before and after a change of the state of the switching unit of the P-th module Pm, wherein at least one sample of the voltage V.sub.mP is taken before the change of said state and a plurality of samples of the voltage V.sub.mP is taken after the change of said state and at least one sample of current I is taken before and/or after the change of said state.

13. A method of claim 12, wherein the characteristics of the energy storage device P comprise at least one of: one or more parameters of an electric equivalent circuit diagram including an internal resistance, and state of health (SOH) of the energy storage device P; and/or wherein the determining characteristics of the energy storage device P is further based on at least one of: an estimated state of charge (SOC) of the energy storage device P, wherein the SOC is estimated by integrating the current through the energy storage device P and dividing the integrated current by an available capacity Cx of the energy storage device P, and/or an assessed temperature of the energy storage device P.

14. A method of claim 12, wherein said determining characteristics of the energy storage device P further comprises determining the available capacity C.sub.x of the energy storage device P by applying at least one substantially fully discharge and/or charge cycle to the energy storage device P, thereby integrating current I to obtain a total charge transfer during the at least one substantially fully discharge or charge cycle.

15. A method of claim 12, wherein the state of health is estimated by at least one of: dividing the available capacity C.sub.x by a nominal capacity CN of a new energy storage device P, and dividing an actual internal resistance by a nominal internal resistance of the new energy storage device P.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In the following, the invention will be described by way of example, without limitation of the general inventive concept, on examples of embodiment and with reference to the drawings.

[0057] FIG. 1A shows a basic structure of an energy storage system, ESS, in an embodiment.

[0058] FIG. 1B shows a basic structure of an energy storage system, ESS, according to another embodiment.

[0059] FIG. 2A to 2C illustrates different types of switching units allowing a serial connection possibility of modules and energy storage devices.

[0060] FIG. 3A shows a module structure with parallel connection possibility in an embodiment.

[0061] FIG. 3B to 3E illustrates different types of switching units allowing a serial and/or parallel connection possibility of modules and energy storage devices.

[0062] FIG. 4 shows different module configuration sets and their effects to the string output voltage.

[0063] FIG. 5A illustrates an exemplary sequence of different module configurations over time to generate a sinusoidal string output voltage.

[0064] FIG. 5B shows the impact to the SOC of the exemplary sequence of the different module configurations shown in FIG. 5A over a plurality of periods.

[0065] FIG. 5C shows pulse pattern shared between two modules and the result on the output waveform.

[0066] FIG. 6A shows a time sequence of a measured voltage response V.sub.mP at an energy storage device triggered by two consecutive current transitions of it with different polarities.

[0067] FIG. 6B shows a time sequence of measurement cycles at different SOC levels of energy storage device P.

[0068] FIG. 7 illustrates examples of equivalent circuit diagrams of battery cells.

[0069] FIG. 8 shows the SOC behave over time for during a determination of the available capacity in an embodiment.

[0070] FIG. 9 shows a flow diagram of a method for determining characteristics of an energy storage device.

[0071] FIGS. 10A to 10D illustrate exemplary sequences of different module configurations over time to generate a sinusoidal string output voltage.

[0072] FIGS. 11A to 11D illustrate exemplary sequences of different module configurations over several sine wave periods to generate a sinusoidal string output voltage.

[0073] FIG. 12 shows in more detail in diagram 910 a current transition and the resulting change of battery voltage over time.

[0074] FIGS. 13 and 14 show different transition patterns.

[0075] Generally, the drawings are not to scale. Like elements and components are referred to by like labels and numerals. For the simplicity of illustrations, not all elements and components depicted and labeled in one drawing are necessarily labels in another drawing even if these elements and components appear in such other drawing.

[0076] While various modifications and alternative forms, of implementation of the idea of the invention are within the scope of the invention, specific embodiments thereof are shown by way of example in the drawings and are described below in detail. It should be understood, however, that the drawings and related detailed description are not intended to limit the implementation of the idea of the invention to the particular form disclosed in this application, but on the contrary, the intention is to cover all modifications, equivalents and alternatives falling within the spirit and scope of the present invention as defined by the appended claims.

DETAILED DESCRIPTION

[0077] FIG. 1A shows a string of N modules, with an integer N>1, of an energy storage system, ESS, which, in an embodiment, generates a stepwise output voltage V.sub.AB. The ESS is constructed according to a modular multilevel converter, MMC, while being equipped with a plurality of integrated energy storage devices (131-134). Throughout the description, the term energy storage device includes preferably a li-ion based battery cell or a battery module, wherein the battery module consisting of at least two or more parallel or serial hard wired battery cells.

[0078] The ESS includes a string (100) including a plurality of modules N (111-114) connected in series. Each module (111-114) includes a switching unit (121-124) configured to selectively put the respective energy storage device (131-134) in or out the current path, which generates string output voltage V.sub.AB whereas the subset of modules in the current path will be denominated as M in the following. Each module further includes a respective module controller unit (141-144) configured to control the switching unit (121-124) of the respective module (111-114). Furthermore, each of the modules (111-114) includes a measurement unit (151-154) to measure at least a voltage at the respective module. Preferably, the module measurement unit may also measure the current through the energy storing device locally on a module level. In an embodiment, each measurement unit further includes one or more temperature sensors to measure the temperature at the respective energy storage device (131-134). While the measurement unit (151-154) is shown as a separate component, a skilled person would understand that the measurement unit could be integrated into the respective module controller unit (141-144).

[0079] FIG. 1B shows a structure of an ESS. In an embodiment which contains a central controller (160) comprising a plurality of module controller units (142, 144, 145). The system further includes a string measurement unit (180) associated with the central controller (160). While the string measurement unit (180) is shown as a separate component, a skilled person would understand that the string measurement unit (180) could be integrated into the central controller (160). The string measurement unit (180) measures the string (100) current I.sub.AB and string output V.sub.AB through and at the string (100). In some aspects, a module controller unit (145) may be associated with more than one switching units and energy storage devices. The ESS may further include a cloud server (170), which may run some calculations and may store data associated with the ESS. The central controller (160) may exchange information with the module controller unit (141-145), with the string measurement unit (180) and/or with the cloud server (170).

[0080] In an embodiment, the central controller (160) may be located in the ESS, or alternatively may be located at a remote location. In some aspects, the ESS may include a communication interface to communicate via a communication network, such as LAN, WLAN, Bluetooth, etc., with a remote server (174) or cloud (170). In an embodiment, the central controller (160) may collect the measured data and provide them to the remote server (174) or to the cloud (170) via the communication network.

[0081] In some aspects, a remote user (172) may remotely control the operations performed on the central controller (160), for example by employing a software routine on it. Hence, the operation of the ESS may be changed during operation of the ESS, for example, in an electric vehicle, such, that a particular energy storage device (131-134) may be characterized as described herein. The results and/or measurement data may then be sent back to the remote user (172) via the communication network. Alternatively, the results and/or measurement data may be sent, via the communication network, to a particular remote server (174) being associated with a service provider having interest in the current state of the ESS via the communication network, such as an original equipment manufacturer, a supplier, a consumer, an insurance company, etc.

[0082] FIG. 2A to 2C illustrates different types of switching units (121-124), allowing to achieve a serial connection of a plurality of energy storage devices (131-134) to generate the current path of the string (100). FIGS. 2A and 2B each show a two-quadrant module to either bypass the energy storage device or to put same into the current path, thereby increasing the string output voltage by the voltage of the energy storage device V.sub.bat. These modules have one input and one output. FIG. 2B shows a simple embodiment of a switching unit. It has an input (222) and an output (223). A battery (221) may be connected between input and output if the series switch (225) is closed and the parallel switch (224) is open. The battery is disconnected, if the series switch (225) is open. Further, the parallel switch (224) is closed to provide a direct connection (short circuit) between input and output. It should be avoided to close both switches at the same time as this may lead to a short circuit of the battery. FIG. 2C illustrates a four-quadrant module represented by a full bridge providing the function of a two-quadrant module, but additionally allowing the energy storage device (210) to be switched inversely into the current path of the string (100), thereby decreasing the string output voltage V.sub.AB by V.sub.bat., whereas V.sub.bat is the voltage of one energy storage device of a module. FIG. 2C shows by the dashed lines the current path taken through the module in case the energy storage device (210) is put serially into the current path and shows by the dotted lines the current path in case of an inverse serial connection. Hence, a string output voltage V.sub.AB in a range between −M.Math.V.sub.bat to M.Math.V.sub.bat may be generated, whereby M is the number of serially connected energy storage devices (131-134).

[0083] FIG. 3A illustrates a string configuration according to which the plurality of modules (301, 302, 303) have two inputs (306, 307) and two outputs (304, 305) to achieve not only serial but also parallel connectivity of the modules. Such multiple input multiple out, MIMO, modules may replace the single input single output modules (111-114) as shown in FIG. 1A.

[0084] FIGS. 3B to 3E illustrate different types of switching units (121-124) that may be employed in the MIMO modules shown in FIG. 3A. In particular, FIG. 3B shows by means of the dashed lines a two-quadrant MIMO module in a state where the energy storage device is put serially into the current path. The energy storage device may be put by closing switches 1 (310) and 2 (320) in a parallel manner into the current path. FIG. 3C shows another type of a two-quadrant module which may be employed in the MIMO module. FIGS. 3D and 3E, each show a four-quadrant MIMO module represented by two full bridges providing the function of a two-quadrant MIMO module, but additionally allowing the energy storage device to be switched inverse into the current path in case of an inverse parallel connection. An illustration of the current path taken through the modules of FIGS. 3C to 3D depending on the switching states is omitted for the sake of clarity. The switching units illustrated in FIGS. 3C and 3E may additionally include a battery switch (350, 351) to separate the energy storage device from the current path independently of the switching state of the other switches within the respective switching unit. These modules have two inputs and two outputs.

[0085] FIG. 4 illustrates by some examples the effect of different module configurations on the string output voltage V.sub.AB. In these examples it is assumed that four modules are available and that each module may be switched between four different states, namely serial, parallel, bypass and inverse.

[0086] According to configuration 1, module 1 and module 2 are serially connected and generate an output voltage 2.Math.V.sub.bat. Modules 3 and 4 are bypassed.

[0087] While the module voltage is assumed to be positive, an inversely connected module may be illustrated with a negative voltage contribution, as discussed in the following.

[0088] In configuration 2, modules 1 to 3 are serially connected and generate an output voltage 3.Math.V.sub.bat. Module 4 is inverse serially connected and reduces the output voltage by V.sub.bat to 2.Math.V.sub.bat. As a result, an output voltage equal to the output voltage of configuration 1 is generated albeit using a different module configuration.

[0089] According to configuration 3, modules 1 and 2 are connected in a parallel and share the same absolute but halved string current. In addition, module 3 is serially connected to modules 1 and 2 and, together, generate an output voltage 2.Math.V.sub.bat. Module 4 is bypassed.

[0090] According to configuration 4, module 1 and module 2 are serially connected and generate an output voltage 2.Math.V.sub.bat, whereby the voltage of module 1 is less than the voltage of module 2. Modules 3 and 4 are bypassed.

[0091] FIG. 5A illustrates an exemplary sequence of different module configurations over time to generate an approximately sinusoidal string output voltage V.sub.AB (t) by a stepwise/staircase output voltage typically needed for grid or motor applications.

[0092] On top of FIG. 5A, the string output voltage V.sub.AB in units of V.sub.bat (511) is illustrated over time by a solid stepped line and the resultant current through the string (512) is schematically illustrated by means of dash-dotted lines. The current through the string (512) is typically smoothed by an inductive load or filter and may have a phase shift towards the string output voltage depending on the used current controller, filter and the load. The three lower diagrams show by means of a solid line the changing module switching states s.sub.min (521, 531, 541) of the respective three modules, which may be serial (=1), bypass (=0) and inverse (−1), of three particular modules together synthesizing the string output voltage (511) of FIG. 5A. Moreover, the diagrams illustrate by dashed-dotted lines the respective currents I.sub.mn (522, 532, 542) seen by the three modules. Each current through a respective module corresponds to a fragment of the sinusoidal string current (512) in case the respective module is not bypassed. The current I.sub.mn indicates the current through module n and switching state s.sub.mn indicates the switching state of module n, with n being and integer with 0<n<N.

[0093] Each time a module is put into or out the current path, a respective positive or negative current transition (543-546) is effected at the module. If the switching occurs in two consecutive steps (545, 546) with reversed polarity, a current pulse (547) is generated at the module. In case of MIMO modules, a current transition may be effected by switching modules in parallel. The current amplitude of the positive or negative current transition may be set by measuring the string current and determine, by comparing the string current with a predefined current amplitude, a particular point in time the positive or negative transition is to be applied.

[0094] The module configurations may be selected in a manner that a subset of the modules or all modules within the string (100) are loaded substantially even, such, that their SOC is maintained at a very similar level to each other, hereinafter referred to as balancing. For example, FIG. 5B shows by solid lines the substantially uniform SOC of modules 2 and 3 (550) over time caused by the uniform and balanced loading of the modules over several sine wave periods.

[0095] In an embodiment, at least one is loaded intentionally unbalanced to determine certain characteristics of its respective energy storage device during operation of the ESS at one or more predefined SOC levels. For example, module 1 is more heavily loaded during the shown single sine period compared to module 2 and 3 in FIG. 5A. Simplified speaking, if the particular module is determined to be more heavily loaded, said module may be switched on first and switched off last or it may be switched on more frequently compared to other modules. In addition, the module may be connected in serial, but not be connected in parallel. On the contrary, if the particular module is determined to be less heavily loaded, said module may be switched on last and switched off first or it may be switched on less frequently compared to other modules. In addition, the module may be connected in parallel. Hence, the SOC of module 1 will change faster than the SOC of modules 2 and 3. FIG. 5B illustrates by dashed lines the SOC over time caused by unbalancing of module 1 (560) over several sine wave periods.

[0096] In an embodiment, the module configurations may be selected in manner that at least two modules may be loaded in the opposite direction to accelerate the time needed to unbalance at least one module to a predefined SOC level. For example, if there are more modules available then needed for holding the ESS operational, a module may also work against the other, balanced, modules. So when the balanced modules are being discharged, they may not only be discharged into the grid or load but also into said module. For charging the inverse applies. As another example, if there are two modules more available then needed for holding the ESS operational, one module may be intentionally discharged while the other module is intentionally charged by the load removed from said first module. FIG. 5B illustrates by dashed-dotted lines the SOC over time of a module 4 (570) caused by unbalancing of module 4 inverse to module 1 (560) over several sine wave periods. This is in particular advantageous, in case, the balanced modules have a higher SOC (550) and consequently also a higher voltage, such, that the probability that the unbalanced modules are needed for operation of the ESS being low. Hence, loading a module to an unbalanced SOC level may be accelerated.

[0097] According to another embodiment, a module may be pulsed in a much shorter period to generate a pulse-width modulation, PWM, signal, which smoothens the staircase approach of the desired sine wave voltage of the string output (511). In some embodiments, two or more modules may pulse against or with each other to generate a desired output voltage. For example, FIG. 5C shows an example where two or more modules pulse (575, 580) with each other, without changing the output function of the string or likewise the sum signal (571) of both modules. According to another embodiment, interleaved pulsing may be used to generate a PWM signal without the need for each module to pulse in the full PWM frequency. Hence, from a system point of view, the PWM frequency (572) seems to be higher than the PWM frequency of each module. In an embodiment, the current amplitude may be set based on the right timing in relation to the load current on the string, but may also by setting the number of modules working in parallel. The averaged voltage resultant from the PWM signals within a PWM period for each, module 1, 2 and the sum signal, are illustrated over several PWM periods in dashed lines in FIG. 5C.

[0098] In an embodiment, the abbreviation P is used to refer to a module, which is intentionally unbalanced to characterize the respective energy storage device P thereof.

[0099] It was shown that positive or negative current transitions can be generated by putting an energy storage devices into or out of the current path of the string. Further, it was shown that a particular energy storage device within a module can be intentionally unbalanced to an SOC different to the SOC of the other modules.

[0100] In an embodiment, said two control options are combined to characterize the energy storage device P. Simplified speaking, the energy storage device P is characterized by the voltage response triggered from a current transition or current pulse reflecting a certain load change. Based on said voltage response, one or more characteristics including the elements of an equivalent circuit diagram, the internal resistance, and the SOH may be determined. Since, the elements of an equivalent circuit diagram and the SOH typically depend on the SOC level, the energy storage device P is “unbalanced” to reach several SOC levels. Respective measurements may thus be carried out at different SOC levels of the energy storage device P, thereby generating SOC level dependent parameters for the equivalent circuit diagram.

[0101] Parameters carried out by the respective measurements may be used to update models describing the energy storage devices, such as a digital twin, an SOC estimation model, or modules for estimating the actual aging of the energy storage devices in an expanded parameter space. Parameters may include values of the elements of an equivalent circuit diagram and/or values describing the functional dependency of the elements of an equivalent circuit diagram on the SOC, temperature, and/or current intensity. The parameters and models may be used to predict maintenance, e.g., replacement requirements, warranty cases or the lifetime of an energy storage device.

[0102] In more detail, FIG. 6A shows a time sequence of the measured voltage response V.sub.mP (620) at an energy storage device P triggered by two consecutive current transitions (612, 614) of it with different polarities at time t0 and t1. The current (610) is a fragment of the sinusoidal string current I.sub.AB (615) represented by the solid line during the period t0 to t1.

[0103] Based on the voltage response shown in the lower part of FIG. 6A, one or more elements of an equivalent circuit diagram of energy storage device P may be determined.

[0104] FIG. 7 illustrates examples of equivalent circuit diagrams of battery cells. A first equivalent circuit diagram (710) consists of a voltage source (711), the so called open circuit voltage, OCV, and an internal resistance (712). Alternatively, a battery cell is modeled more precisely with additional serially connected RC elements (723, 734) as shown by the equivalent circuit diagrams 2 to 4 (720, 730, 740). Equivalent circuit diagram 4 (720) additionally includes two Warburg elements (745, 746) representing an even more precise battery model. The elements of the equivalent circuit diagrams may dependent on at least the SOC, the temperature, and the current intensity of the battery.

[0105] Referring back to FIG. 6A, internal resistance (712) may be determined by dividing the voltage drop (623) triggered by current transition (612) at t0 through the measured magnitude of said current transition. The voltage drop (623) itself may be measured the voltage at the energy storage device just before (t0−) and just after (t0+) the current transition (612) is applied.

[0106] Alternatively or additionally, the internal resistance (712) may be determined based on the voltage rise (627) between t1− (626) and t1+ (628) triggered by current transition (614) with reversed polarity. In an embodiment, the internal resistance (712) is determined by obtaining the mean value of both said determinations to achieve a higher parameter accuracy.

[0107] The different and overlaid gradients of voltage drop (625) between t0+ and t1− may be used to identify the RC (723, 734) and Warburg (745, 746) elements of one of the equivalent circuit diagrams 2 to 4. The different and overlaid gradients in the relaxation process (629) may, instead or additionally be used to identify said RC (723, 734) and Warburg (745, 746) elements.

[0108] A new OCV voltage (711) may be assessed after the negative current transition (614) is applied and the capacities of the energy storage device have been substantially discharged. Such state is typically reached at the end of relaxation process at time t2. The relaxation process may take up more than 24 hours.

[0109] In an embodiment, mathematical methods, such as at least one of: curve fitting, neuronal networks, machine learning or support vector machines may be used to determine the elements of the equivalent circuit diagrams (710, 720, 730, 740) based on these measurements. The multiple execution of the measurements allows measurement errors to be minimized by averaging the measured values.

[0110] In the above it has been shown how one or more parameters of the equivalent circuit diagram may be determined based on the measured voltage response triggered by a positive and/or negative current transition. As previously discussed, the one or more values of the equivalent circuit diagrams are SOC dependent.

[0111] In some embodiments, not only the current and voltage through the energy storage device may be taken into account to determine one or more parameters, but also the temperature at which the respective measurement has been carried out.

[0112] In some embodiments, not only the current and voltage through the energy storage device may be taken into account to determine one or more parameters, but also the SOC at which the respective measurement has been carried out. In an embodiment, the measurements may be carried out at 5% SOC intervals to increase the model accuracy of the SOC-dependent parameters. FIG. 6B shows how the energy storage device P is charged to predefined SOC levels (55%, 60%, etc.). After a predefined SOC level is reached, the respective measurements or likewise measurement cycles may be carried out. In particular, FIG. 6B shows by dashed lines an example of different current transitions or likewise current pulses (690, 692) with different current amplitudes and pulse durations applied to the energy storage device at 55% and 60% SOC to parameterize the equivalent circuit of energy storage device P. The switching of module P may be performed more than once and repeatedly for each measuring cycle. It is further possible to load energy storage device P with current pulses of different polarity and different time periods. These pulse patterns may be repeatedly applied to energy storage device P in order to compensate for measurement errors through averaging or curve fitting methods. By varying the pulse patterns the amount of information which may be extracted from the time series measurements increases. Most information may be extracted if the pulse patterns resemble a quasi-noise pattern, where the pulse duration, polarity of pulses and amplitude of pulses at least seem to not have a regularity or dependency on each other. Since the SOC may be not directly measurable, it may be estimated as shown below.

[0113] The SOC represents the remaining capacity related to the available capacity C.sub.x of the energy storage device. The SOC may be estimated by means of an ampere-hour meter according to

[00001]$\mathrm{SOC}\left(t\right)=SOC\left(t-1\right)+{\int }_{t-1}^t\frac{i\left(t\right)}{C_x}\mathrm{dt}.$

[0114] Methods to determine the available capacity are shown in FIG. 8. Alternatively, the rated or nominal capacity of a new energy storage device C.sub.N may be used.

[0115] A suitable starting point for the estimation may be reached by unbalancing the energy storage device P until it reaches its charge cutoff or discharge cut-off voltage, respectively corresponding to its fully (SOC=100%) or discharged state (SOC=0%).

[0116] An alternative estimation of the SOC uses the measured voltage during the relaxation process (629) on a SOC/OCV mathematical model previously estimated or provided by the battery cell manufacturer for a new energy storage device.

[0117] In an embodiment, the charge estimator may be designed as a classical ampere-hour counter with extensions like lookup tables or more complex estimation methods like a Kalman filter (extended, unscented, etc.), as a generalized Kalman filter in form of a particle filter, via neural networks etc. Indeed, the estimations will be more precise if the underlying parameter, the available capacity C.sub.x of the energy storage device, is determined as precisely as possible. Depending on the needed calculation power of the SOC estimator, the estimation may be calculated on the module controller unit (141-145). Alternatively, the estimation may be calculated on the central controller (160) or in the cloud (170).

[0118] In the above, SOC estimation models have been described. Additionally, it has been shown that the SOC estimation may depend on the available capacity of an energy storage device. The available capacity decreases as the energy storage system ages over time. Hence, the SOC estimation model may be updated by replacing the value of the available capacity with the new determined available capacity to improve the SOC estimate.

[0119] FIG. 8 illustrates an example of a full charging and discharging cycle of energy storage device P to determine the available capacity C.sub.x of the energy storage device P. In an embodiment, the energy storage device P is charged (810) from its current SOC state (SOC=60%) (805) to a substantially fully charged state (SOC=100%) (815) corresponding to the charge cut-off voltage. After the fully charged state (815) is reached, the energy storage device is discharged to its fully discharged state (SOC=0%) (825) associated with the discharge cut-off voltage. The available capacity C.sub.x is determined by measuring and integrating the measured current flow through the energy storage device P during discharge cycle (820) between t1 and t2.

[0120] Alternatively, the available capacity C.sub.x may be determined by applying and using a full cycle (850) for the measurement and determination. A full cycle may be applied by charging (830) the energy storage device from discharged state (825) back to its fully charged state (835), thereby measuring and integrating the measured current flow through the energy storage device P during the charge cycle (830) between t2 to t3. The available capacity C.sub.x is thereby determined by calculating an average value of the available discharge and charge capacity measurements. Alternatively, the smaller value of the available discharge and charge capacity may be used to indicate the available capacity C.sub.x. The energy storage device P may be charged or discharged to another SOC or being assigned to the balancing algorithm and strategy again, which brings it back to the SOC of the other balanced modules.

[0121] In the above, it has been shown that the SOC estimation model may be updated by measuring the total charge transferred during one of a discharge, charge or full cycle. Further, it has been indicated that the available capacity C.sub.x decreases as the energy storage systems age over time.

[0122] The ageing of an energy storage system is preferably represented by the state of health, SOH, and may be estimated based on a ratio of the available capacity C.sub.x to the nominal capacity C.sub.N of a new energy storage device according to:

[00002]$\mathrm{SOH}=\frac{C_x\left(t\right)}{C_n}$

[0123] Alternatively, the SOH may be estimated based on the rise of the internal resistance (712) in relation to the internal resistance of a new state. Alternatively, the SOH may take into account both of the mentioned ratios above and/or also include further embodiments.

[0124] In an embodiment, the available capacity C.sub.x may be determined based on the internal resistance (712) over time by equalizing the two SOH equations.

[0125] FIG. 9 shows a flow diagram of the method for determining characteristics of at least an energy storage device contained in an energy storage system, ESS, during operation of said ESS. The ESS comprising a plurality of modules, wherein each module comprising an energy storage device and a switching unit. In some embodiments, one or more of the steps may omitted, repeated, and/or performed in different order.

[0126] Initially, a subset of a plurality of modules may be connected by means of the switching units into a module configuration according to which the respective energy storage devices of the subset of modules M are serially connected into a current path to provide an output voltage of the ESS (901). Next, the module configuration is changed by switching an energy storage device P of at least one module into or out of the current path (902). In this way, a current change or rather a current transition is applied to a particular energy storage device P, resulting in a respective voltage response. Next, the current I and a voltage V.sub.mP at the energy storage device are measured (903). Subsequently, characteristics of the energy storage device P are determined on basis of at least the measured current I and the change over time of the voltage V.sub.mP measured before and after switching at the energy storage device P (904). Said characteristics represent the current status of said energy storage device P, such as parameters of an equivalent circuit diagram, internal resistance, SOH, etc.

[0127] In an embodiment, the current through a particular energy storage device may be measured by the respective module measurement unit (151-154). It is advantageous to measure the current at module level instead of string level to reduce measurement inaccuracies caused by interferences, the not always appropriate time sample coverage of the current measurement timing and changes in the switching state and impedances of the modules, cabling and filters.

[0128] Alternatively, the current through an energy storage device and module may be determined based on the measured current at string level by the string measurement unit (180) and the known switching state of the modules (111-114). The current I.sub.mn through module n (n being and integer with 0<n<N) may be determined according to:

[00003]$I_{mn}=\frac{s_{mn}}{p}{I}_{\mathrm{AB}},$

[0129] wherein s.sub.mn is the switching status (1=active; 0=bypassed, −1=active with inversed polarity) of module n; p is the number of parallel modules with an integer p≥1, wherein for p=1 no parallel connections are used, and wherein I.sub.AB is the current flowing through the string. Accordingly, I.sub.mP describes the current at module P through energy storage device P.

[0130] In an embodiment, the current measurement may be performed several times to statistically determine measurement errors and/or reduce same.

[0131] In an embodiment, at least one of the current or voltage measurement may work with a sampling frequency greater than 10.Math.f.sub.0, with f.sub.0 being the mains frequency, to ensure a sufficiently high measurement accuracy. This is advantageous, since the amount of energy that has been flown per time unit may be balanced based on the current measurement and the sampling rate. Thus, the charge quantities may be added up per discharging and charging direction, which may allow an estimation of the total available capacity of the at least one energy storage device P in a more precise manner.

[0132] In an embodiment, the voltage and current measurements may be measured at a high temporal resolution more than 10 kHz. Alternatively, the voltage and current measurements may be measured at a low temporal resolution less than 10 kHz.

[0133] As previously discussed, the one or more values of the equivalent circuit diagrams may be temperature dependent. The measured values of an EIS at imaginary part=0 are a reliable measurement for the internal temperature of the energy storage device, not depending on aging or SOC. In an embodiment, a temperature determination similar to that with an EIS at imaginary part=0 may be carried out without additional measuring circuits. In more detail, depending on the load current and the respective timing, pulses may be generated to enable a temperature determination. Thus, on the one hand, temperature sensors may be eliminated and on the other hand, the parameter determination may be stored based on a more precise temperature.

[0134] In an embodiment, the ESS is composed of different energy storage devices at least in terms of mixed battery modules regarding voltage, SOC, SOH, used cell chemistry and number of cells. In an embodiment, batteries described herein may be li-ion based batteries. In an embodiment, cathode material such as LiCoO2, LiMn2O4, Li(NiCoMn)O2, LiFePO4, LiNiCoAlO2 may be used within the li-ion batteries.

[0135] In an embodiment, the module controller units (141-145) may process the measurement data and may perform the necessary mathematical functions.

[0136] In an embodiment, a logging may be carried out on the temporal course of the determined parameters. This is useful in particular, to determine how the SOH changes over time and between measurements. Depending on the available memory of the module controller units (141-145), this logging may also be performed on the higher-level central controller, externally in the cloud (170) or on a server belonging to a user.

[0137] In an embodiment, depending on their complexity and memory requirements, calculations may be performed on the central controller (160) or in the cloud (170). In this case, the task of the module controller units (141-145) may be restricted to data acquisition, aggregation and transmission.

[0138] FIGS. 10A to 10D illustrate exemplary sequences of different module configurations or likewise pulse patterns over a sine wave period to generate a step shaped output voltage, which may approximate a sinusoidal string output voltage V.sub.AB resulting in an approximate sinusoidal current I.sub.AB. The x-axis is over time any the y-axis gives output voltage level in units of V.sub.bat. The figures are based on a simplified embodiment with a string having three modules. For grid voltages the frequency in Europe usually may be 50 Hz depending on the country, so a sine wave period has a duration of 20 ms. Also other frequencies are possible, e.g. railway (16.66 Hz) or aircraft supply voltages (400 Hz). For cars the frequency is variable from 1 Hz to 400 Hz and depending on the motor even higher, up to 1 kHz. Each pulse pattern shown in FIGS. 10A to 10D generates the same output voltage as shown in FIG. 5A. In the respective three lower subplots the y-axis illustrates the polarity and the activation of the respective module [−1, 0, 1] are shown (Module 1: S.sub.m1, Module 2: S.sub.m2, Module 3: S.sub.m3). FIG. 10A illustrates a pulse pattern which may be used to substantially evenly load modules 1 to 3 to keep them balanced at a substantially same SOC. FIG. 10B shows a pulse pattern wherein the modules 1 has the largest load as its positive on-time where it provides power is larger than the negative on-time where it is charged. Module 3 is even charged, as the charging time is larger than the power delivery time. FIG. 10C illustrates a pulse pattern with four modules. FIG. 10D illustrates a pulse pattern with very short pulses corresponding to high-frequency and almost noise-like pulses. An equivalent circuit may also be parametrized using such noisy and high-frequency pulses. The dotted lines in FIGS. 10A to 10D illustrate schematically the resultant string current I.sub.AB.

[0139] FIGS. 11A to 11D illustrate exemplary sequences of different module configurations over several sine wave periods to generate a sinusoidal string output voltage (x-axis: time, y-axis amplitude). In more detail, the lower diagrams of FIGS. 11A to 11D show the switching state of the particular module P (y-axis is the polarity and the activation of the module [−1, 0, 1]) from a plurality of modules used to generate the string output voltage illustrated in the respective upper diagram over several sine periods. According to FIG. 11A, the switching pattern shown in the lower diagram generates pulses with positive polarity and a frequency of 100 Hz (for a 50 Hz sine wave period) at the respective module. According to FIG. 11B, the switching pattern shown in the lower diagram generates pulses with positive polarity and a frequency of 200 Hz (for a 50 Hz sine wave period) at the respective module. According to FIG. 11C, the switching pattern shown in the lower diagram generates pulses with a frequency of 200 Hz (for a 50 Hz sine wave period) and a polarity change at every third pulse at the respective module. According to FIG. 11D, the switching pattern shown in the lower diagram generates triple pulses with alternating polarity [+1;−1;+1] at the respective module. The different pulse patterns illustrated in FIGS. 11A to 11D may be used to stimulate different chemical reactions of the energy storage device, which may result in diagnostic benefits.

[0140] FIG. 12 shows in more detail in diagram 910 a current transition and the resulting change of battery voltage over time (horizontal axis). When the battery 221 is switched on, for example, by closing the series switch 225 and opening the parallel switch 224, this may result in a current rise as shown in curve 911 from a low current 913 to a high current 914. The voltage at the battery may drop as shown in curve 912 from a first voltage 916, which may be an idle voltage, to a voltage approximating a value 915. The function over time of the battery voltage is explained by the circuit diagram 730 of FIG. 7. The first voltage 916 may correspond to the voltage U.sub.OCV of the diagram 730. The voltage drop in the first section 917 is proportional to the current rise and is caused by the inner resistance R.sub.i of the battery. The second section 918 of the curve 912 is determined by the polarization which can be described by the first RC combination RC.sub.1. The third section 919 of the curve 912 is determined by the diffusion in the battery which can be described by the second RC combination RC.sub.2, and which normally has a longer time constant than the first RC combination.

[0141] The parameters of an equivalent circuit, e.g. as given in the circuit diagram 730 cannot be determined by a sampling before and another sampling after the current rise. Instead multiple samples have to be made to measure the waveforms.

[0142] In an embodiment, at least one sample of the battery voltage is measured before the current transition (which is when the switches change state) and a plurality of measurements are made after the current transition. The current transition coincides with a change of state, which is a change between a state where the energy storage device is connected between the at least one input and at least one output and another state having a short circuit between the at least one input and the at least one output. In the first state the battery may be connected to the string and in the second state the battery may be disconnected from the string.

[0143] There may be 10 to 100 samples, 20 to 200 samples or more than 100 samples measured after the change of state. The measurement of a sample before the change of state may be immediately before the change of state. It may be determined by the time resolution of the measuring devices employed, such that this measurement is clearly made before the transition. It may be made less than 100 microseconds before the transition to suppress low frequency deviations of the voltage. Measurement after the transition may start immediately after the transition. It may be determined by the time resolution of the measuring devices employed.

[0144] Further, at least one sample of current I is taken before and/or after the change of state. In an embodiment, the controller is configured to take at least one sample of current I before the state is changed from connecting the energy storage device between the at least one input and the at least one output to providing a short circuit between the at least one input and the at least one output, and

to take at least one sample of current I after the state is changed from providing a short circuit between the at least one input and the at least one output to connecting the energy storage device between the at least one input and the at least one output. This improves efficiency in sampling and data processing, as no current measurements are made, when the battery is disconnected, which may result in a current close to zero.

[0145] In a perfect system with perfect measurement equipment a single measurement (including multiple samples) may be sufficient to specify the parameters of the equivalent circuit model. Normally an energy storage system may operate on a power grid while doing the measurement. Therefore, the environment is noisy and the currents are not rectangular but fragments of sine waves. Additionally, the measurement equipment is very simple and may include microcontrollers and simple integrated sensors.

[0146] In order to increase the quality of the measurement data, measurements have to be repeated multiple times. The measurement results may be fitted by mathematical methods (e.g. recursion, machine learning, support vector machines) to the battery model. It is beneficial to have a plurality of measurements (and therefore datapoints) in order to have meaningful battery model parameters.

[0147] One issue which will be taken into account by multiple measurements is the sample time error. Normally, a microcontroller has a distinct sample time. But with this distinct sample time it won't be able to directly measure e.g. the inner resistance since it will be represented as an instantaneous drop in battery voltage when a current is applied. Multiple measurements make it possible to more precisely determine the real instantaneous voltage drop. The inner resistance may be calculated as R_i=|(V1−V0)/(I1−I0)|. One has to keep in mind, that the resistance is dependent on temperature, SOC and SOH.

[0148] Measuring transient processes in a noisy environment gives distorted (=noisy) measurement results. In order to decrease the noise multiple measurements may to be taken. The noise may be reduced by the square root of the number of measurements.

[0149] Relevant information can be obtained faster if the system does not wait for full relaxation but if it includes a new current pulse more frequently. The fast processes are harder to measure, so they have to be measured more often in order to increase data quality and validity. So the system may start new pulses before the end of third section 919 or even at the end or during the second section 918. This is shown in FIG. 13 with diagram 920 and FIG. 14 with diagram 930.

[0150] In order to correctly fit the model and to obtain a reliable SOH state, it may also necessary to repeat these measurements for different SOCs and temperatures. The basic idea is to gather relevant measurement information in order to have sufficient (low quality compared to lab measurements) data for mathematical methods of curve fitting for equivalent circuit models.

[0151] Also changing amplitude or direction of the current increases the data quality since new behaviors are being measured which have not been measured before. Ideally the fitting algorithm has an indicator of the data quality supplied and an indicator for “blind spots”, e.g. measured behaviors where there is no or too little data material.

[0152] It will be appreciated to those skilled in the art having the benefit of this disclosure that this invention is believed to provide an energy storage system. Further modifications and alternative embodiments of various aspects of the invention will be apparent to those skilled in the art in view of this description. Accordingly, this description is to be construed as illustrative only and is provided for the purpose of teaching those skilled in the art the general manner of carrying out the invention. It is to be understood that the forms of the invention shown and described herein are to be taken as the presently preferred embodiments. Elements and materials may be substituted for those illustrated and described herein, parts and processes may be reversed, and certain features of the invention may be utilized independently, all as would be apparent to one skilled in the art after having the benefit of this description of the invention. Changes may be made in the elements described herein without departing from the spirit and scope of the invention as described in the following claims.