EP3698261A1 - Method for determining parameters of a simplified model of an energy storage system, control method using such a model and associated device - Google Patents

Abstract

One aspect of the invention relates to a method for determining parameters of a simplified model of an energy storage system, said system comprising an energy storage device and a conversion device, said system being able to be modelled by way of a complex model (MC) including a model of the energy storage device and a model of the conversion device; said complex model receiving a setpoint power Pac_sp and a state of charge SOCp at input and providing the state of charge SOC of the storage device and the power Pac at the output of the storage device at output; said method comprising a first step of implementing a plurality of simulations of the energy storage system using the complex model; a second step of calculating, on the basis of the results obtained in the first step: - a table of the variation in the state of charge of the system as a function of the setpoint power Pac_sp and of a state of charge SOC; - a table of maximum power as a function of the state of charge SOC; - a table of minimum power as a function of the state of charge SOC. The simplified model thus obtained makes it possible to assign a power Pac and a state of charge of the system SOC as a function of a setpoint power Pac_sp and of a state of charge SOCp that are provided at input, and on the basis of the tables determined in the second step.

Description

 METHOD FOR DETERMINING THE PARAMETERS OF A SIMPLIFIED MODEL OF AN ENERGY STORAGE SYSTEM, A STEERING METHOD USING SUCH A MODEL AND ASSOCIATED DEVICE

TECHNICAL FIELD OF THE INVENTION The technical field of the invention is that of energy management. The present invention relates to a method for determining the parameters of a physical model and in particular to a method for determining the parameters of a simplified model of an energy storage system.

BACKGROUND OF THE INVENTION In the context of renewable energies, it is essential to optimize the storage of energy and the use of this stored energy to reduce the intermittent nature of certain energy sources such as wind turbines. or photovoltaic panels. The storage is, in general, provided by an energy storage system comprising a storage device in charge of storing the energy itself and a conversion device which will ensure the charging or discharging of the device. storing and converting the stored energy into a form suitable for domestic or industrial use, for example by converting a direct current to alternating current. It should be noted that a storage system often includes auxiliaries that may be associated with the storage device or the conversion device. In the following text, when we speak of storage system, these auxiliary systems are considered included in said storage system.

In order to drive such a storage system efficiently, it is essential to predict the behavior of the latter so as to anticipate the load, the discharge and the power that the device can provide or absorb. This control is performed by updating operating instructions, said updating being performed at regular intervals, generally of the order of a minute or even the hour. In order to determine the instructions adapted to the control of the installation, it is it is therefore necessary to simulate the behavior of the storage system before re-establishing each new setpoint. For this, a model representing the storage device and a model representing the conversion device are assembled so as to constitute a complex model. In order to control the energy storage system, it may seem reasonable to carry out a simulation with a time step of the same order of magnitude as the piloting time step (ie of a few minutes or even a few hours), the simulation as well performed to generate a setpoint. However, it is generally necessary to adopt a time step well below the pilot time step (ie of the order of one second or even one millisecond) in order to obtain sufficient accuracy to generate a reliable setpoint. Thus, for each new setpoint, it is necessary to perform a greedy simulation in computing resource and in memory.

There is therefore a need for a control method in which the instructions are generated using a simulation whose time step is of the same order of magnitude as the time step of piloting itself, and which thus requires a less computing power (or less computing time for a given computing power) while ensuring sufficient accuracy. There is therefore also a need for a simplified model for implementing such a control method. SUMMARY OF THE INVENTION

The invention offers a solution to the problems mentioned above, by making it possible to obtain a simplified model whose simulation time is limited. Indeed, the simplified model obtained using a method according to the invention makes it possible to perform a simulation of a system for storing energy with a time step of the same order of magnitude as the time step of piloting said system, while guaranteeing a precision simulated results similar to that obtained with a simulation having a time step much smaller than the time step of the pilot itself. A first aspect of the invention relates to a method for determining the parameters of a simplified model of an energy storage system, said system comprising an energy storage device and a conversion device, said system being able to be modeled using a complex model including a model of the energy storage device and a model of the conversion device; said complex model receiving as input a nominal power Pac sp and a charge state SOC _P , and outputting the state of charge SOC of the storage device and the power P _ac at the output of the storage device; said method being characterized in that it comprises:

 a first step of implementing a plurality of simulations of the energy storage system using the complex model, each simulation being for example carried out with a time step Δΐ;

 a second calculation step based on the results obtained during the first step:

■ a table of the time variation of the load state of the system based on set power Pac_ _sp and the state of charge SOC;

^■ of a maximal power table provided depending on the state of charge SOC;

 ■ a minimum power table provided according to the state of charge SOC.

The simplified model obtained makes it possible to allocate, as a function of a reference power Pac sp and a state of charge SOC _P supplied as input, and from the tables determined during the second stage, a power P _ac and a state SOC system load.

In the following, by convention, the power Pac during charging is considered negative and the power P _ac during the discharge is considered positive. Thanks to the invention, it is no longer necessary to choose between a simulation time step of the same order of magnitude as the pilot time step (ie the time separating two updates of the piloting instructions) leading to a unsatisfactory accuracy, and no more simulation time low to obtain a simulation certainly accurate, but greedy computing resources and memory. Indeed, the simplified model obtained using a method according to a first aspect of the invention makes it possible to adopt a simulation time step of the same order of magnitude as the pilot time step while maintaining a precision sufficient for said piloting.

In addition to the features that have just been mentioned in the preceding paragraph, the method according to a first aspect of the invention may have one or more additional characteristics among the following, considered individually or in any technically possible combination. Advantageously, each value in the table of the change in the system state of charge is obtained with a simulation of a duration iess made for a state of charge SOC (j) and a desired power Pac_ _sp (i) given and belonging to a first subset of the plurality of simulations, said value being:

- equal to the value of the variation of ASOC charge state obtained in said simulation time iess if the average power <P _ac> during said simulation is equal to the desired power Pac_ _sp.

- otherwise, equal to: interp {[P _ac _ _sp (i - l), <P _ac {i, j)>}, [* ≡ ^, * ≡ "], P _{ac sp} (i)} with:

■ ^d50 ^ ^{,; ~ 1),} the load variation obtained or calculated in the simulation corresponding to the state of charge SOC (j) and a desired power Pac_ _sp (i-1);

^■ ASOC (i, j), the variation of charge obtained in the simulation corresponding to the state of charge SOC (j) and a desired power Pac_ _sp (i);

! ^{■ <i>} ac (i, _^j)>> a mean value of the power P _ac in the simulation corresponding to the state of charge SOC (j) and a desired power Pac_ _s (i);

^■ interp _{[x0, xl], [y _0, y], x] is the function that determines the value of y corresponding to the value of x by interpolation from the values of Thus, it is possible to establish the temporal variation of the state of charge corresponding to a given simulation even when the latter is not constant during said simulation. Mean that equal power P _AC is equal to the desired power Pac_ _sp plus or minus 5% or plus or minus 2%, preferably roughly 1%.

Alternatively, each value of the table of the variation of the state of charge of the system is obtained by means of a simulation of a duration ÎESS carried out for a state of charge SOC (j) and a nominal power Pac_ _sp (i) given and belonging to a first subset of the plurality of simulations, said value being equal to the value of the variation of ASOC charge state obtained in said simulation time iess.

Thus, it is possible to establish the temporal variation of the state of charge corresponding to a given simulation even when the latter is not constant during said simulation by resorting to an approximation that is not very demanding in computing resources. Indeed, the inventors have demonstrated that, surprisingly, this approximation allowed to obtain very good results.

Still alternatively, each value of the table of the variation of the state of charge of the system is obtained by means of a simulation of a duration SESS performed for a state of charge SOC (j) and a nominal power Pac_ _sp (i) given and belonging to a first subset of the plurality of simulations, said value being:

- equal to the value of the variation of ASOC charge state obtained in said simulation time iess if the average power <P _ac> during said simulation is equal to the desired power Pac_ _sp.

- otherwise, equal to ^DSOC (™ 'II _or ^{dS0C (™>} D _is | _a time variation of charge obtained in the simulation corresponding to the state of charge SOC (j) and the target power Pac_ _s ( m), Pac_ _s (m) being the power the closest to the target power set Pac_ _sp (i) for which the average power <P _ac> during said simulation is equal to the desired power Pac_s _P (m).

Thus, it is possible to establish the temporal variation of the state of charge corresponding to a given simulation even when the latter is not constant during said simulation by resorting to an approximation that is not very demanding in computing resources.

Advantageously, each simulation the first sub-set of simulations is performed to a state of charge SOC (j) and a desired power Pac_ _sp (i) data and in that each state of charge SOC (j) is separated from the previous SOC (j-1) and / or the next SOC (j + 1) by a step of adaptive state of charge and / or each desired power Pac_ _sp (i) is separated from the previous P _ac _s (i-1) and / or the following Pac_ _s (i + 1) a step of adaptive target power.

Thus, the number of simulations performed to constitute the table of the temporal variation of the state of charge is reduced.

Advantageously, the simulation step is repeated for a plurality of ISES durations. Thus, it is possible to integrate a temporal variable with the simplified model.

Advantageously, the plurality of simulations comprises a simulation carried out with an initial state of charge SOCini equal to the maximum state of charge SOCmax, a duration equal to the time necessary for the complete unloading of the storage system tDch and a reference power P _ac _ _s infinite positive, and the calculation of the maximum power table provided as a function of SOC state of charge comprises:

 a step of determining periods of duration ISESS within said simulation, each period being identified by means of a positive integer k;

for each of these periods, a step of determining the state of charge SOCk at the beginning of the period k and the power average <P _ac > k during period k.

Thus, the plurality of pairs (SOCk, <P _ac > k) constitute the maximum power table provided as a function of the charge state SOC. The term power _sp positive infinity Pac_ set a desired power Pac_ _sp much higher in absolute value to the power that the system can provide. As a reminder, by convention, a positive value of the power P _ac is associated with the discharge of the storage system.

Advantageously, the plurality of simulations includes a simulation carried out with an initial load state of Socini equal to the condition of minimum load SOCmin, a duration equal to the duration necessary to complete loading of tch storage system and a target power Pac_ _sp infinite negative, and the calculation of the minimum power table supplied as a function of the state of charge SOC comprises:

 a step of determining periods of duration ISESS within said simulation, each period being identified by means of a positive integer k ';

for each of these periods, a step of determining the state of charge SOCk 'at the beginning of the period k' and the average power <P _ac > k 'during the period k';

Thus, the plurality of pairs (SOCk ', <P _ac > k) constitutes the minimum power table supplied as a function of the state of charge SOC. Mean negative infinity _sp Pac_ target power a target power P _ac _ _s much greater, in absolute value than the power that the system may receive during charging. As a reminder, by convention, a negative value of the power P _ac is associated with the load of the storage system. A second aspect of the invention relates to a method of controlling an energy storage system calculating operating instructions of said system from a model of said system, said model being obtained using a method according to a first aspect of the invention. A third aspect of the invention relates to a device for controlling an energy storage system comprising means for sending operating instructions to the energy storage system, means for receiving data concerning the operation of the energy storage system, energy storage system and means for implementing a control method according to a second aspect of the invention.

A fourth aspect of the invention relates to a computer program product comprising instructions which, when the program is executed by a computer, lead it to implement the steps of the method according to a first aspect of the invention.

A fifth aspect of the invention relates to a computer program product comprising instructions that drive the driver according to a third aspect of the invention to perform the steps of the method according to a second aspect of the invention. A sixth aspect of the invention relates to a computer readable medium on which the computer program according to a fourth or fifth aspect of the invention is stored.

The invention and its various applications will be better understood by reading the following description and examining the figures that accompany it. BRIEF DESCRIPTION OF THE FIGURES

The figures are presented as an indication and in no way limit the invention.

 - Figure 1 shows a flow chart of an embodiment of a method according to a first aspect of the invention.

 - Figure 2 shows a schematic representation of an energy storage system.

- Figure 3 shows a block diagram of a method according to a first aspect of the invention. FIGS. 4A and 4B show a simulation involved in a method according to a first aspect of the invention.

 FIGS. 5A and 5B show a simulation involved in a method according to a first aspect of the invention.

 FIGS. 6A and 6B show a 3D illustration of a table of the temporal variation of the state of charge of the system according to a first aspect of the invention.

 FIGS. 7A and 7B show a simulation involved in a method according to a first aspect of the invention.

 FIGS. 8A and 8B show a simulation involved in a method according to a first aspect of the invention.

 FIG. 9 shows a graph illustrating the time and precision performances of a model obtained using a method according to a first aspect of the invention. DETAILED DESCRIPTION OF AT LEAST ONE EMBODIMENT OF THE INVENTION

Unless otherwise specified, the same element appearing in different figures has a unique reference.

A first embodiment of a method according to a first aspect of the invention illustrated in FIG. 1 concerns a method 100 for determining the parameters of a simplified model MS of an energy storage system ESS. The ESS energy storage system illustrated in FIG. 2 comprises an energy storage device DSE, for example a battery, and a DC conversion device, and can be modeled by means of a complex model MC illustrated in FIG. Figure 3 including a model of the MCS energy storage device and a model of the MCC converter device. The complex model MC receives as input a reference power Pac sp and a state of charge SOC _P , and outputs the state of charge SOC of the storage device and the power P _ac at the output of the storage system of the energy. The method 100 according to the invention comprises a first step 101 of implementing a plurality of simulations of the storage system SSE energy using the complex model MC, each simulation being performed for example with a time step Δΐ. During these simulations, the state of charge SOC _P supplied at the input of the complex model is the initial state of charge of the system for the first iteration then, for the following iterations, the state of charge calculated during the previous iteration .

The method according to a first aspect of the invention also comprises a second calculation step 102 based on the results obtained during the first step 101:

- a SOCV_TC table of the time variation of the load state of the system depending on the desired power Pac_ _sp and the state of charge

SOC;

 a PAC_MAX_TC table of maximum power as a function of the state of charge SOC;

 - a PAC MIN TC table with minimum power depending on the state of charge SOC.

As a reminder, by convention, the power P _ac during charging is considered negative because it is absorbed and the power P _ac during the discharge is considered positive because it is supplied outside the considered system. Thus, for a given state of charge, the minimum power is the minimum Pac_min power (negative) that can be absorbed system for said state of charge and the maximum power output is the maximum P _ac _max power (positive) that can provide the system for said state of charge.

The simplified model obtained makes it possible to allocate as a function of a reference power ac sp and a charge state SOC _P provided as input, and from the tables SOCV_TC, PAC_MAX_TC, PAC_MIN_TC determined during the second step 102, a Pac power and a state of charge of the SOC system. Indeed, from a target power Pac_ _sp SOCP and a state of charge provided as input to the model, it is possible to determine the minimum power with the minimum power PACJVIIN table TC and maximum power using the maximum power table PAC_MAX_TC. If the target power Pac_s _P is in this range, then the power delivered by the system is equal to said Pac_ _sp set, otherwise it is equal to the limit value closest to said target. It is also possible, from the saturated Pac sp once saturated power and SOC _P charge state input, to determine the temporal variation of the state of charge using the SOCV TC table. the temporal variation of the state of charge of the ESS energy storage system. The state of charge SOCp provided input may for example correspond to a state of charge of a previous iteration when the simplified model is used to perform a simulation. Thus, the model obtained makes it possible to model an ESS energy storage system in a rapid manner without calling into question the accuracy of the simulations making it possible to model said system.

In one embodiment, each value of the SOCV table TC of the temporal variation of the state of charge of the system is obtained by means of a simulation performed for a state of charge SOC (j) chosen as state of charge. Socini initial and a target power Pac_ _sp (i) given and belonging to a first subset of the plurality of simulations, said simulation being performed on a iess duration. This simulation will make it possible to associate with each reference power ac sp (i) and with each state of charge SOC (j) a temporal variation of the state of charge ^ (ij ^' ). In one embodiment, the time step At used for the simulations using the complex model is chosen such that 1 0 ² At <tEss, preferably 1 0 ³ At <tEss.

FIGS. 4A-4B and 5A-5B illustrate two simulations of the first subset of the plurality of simulations performed for two different SOC load states and for the same ISE simulation time. FIGS. 4A and 5A reproduce the power P _ac at the output of the energy management system as a function of time, and FIGS. 4B and 5B reproduce the state of charge SOC of the energy management system as a function of time. The simulation illustrated in Figures 4A and 4B describes a simulation in which the power P _ac output of the energy storage system on the whole iess simulation time is equal to the power setpoint Pac_ _sp. Equals means that the power P _ac at the output of the system ESS is equal to the reference power Pac_p _sp at plus or minus 5%, or more or less 2%, preferably plus or minus 1%. During this simulation, the state of charge SOC decreases steadily so that it is easy, from this simulation, to extract a value of the temporal variation of the state of charge FIGS. 5A and 5B describe for their part, a simulation in which the power Pac at the output of the energy storage system over the duration ISESS of the simulation is not constant, but varies during the simulation. This variation can for example be explained by the fact that the storage device DS of the energy storage system ESS is almost fully loaded (or discharged). During this simulation, the temporal variation of the state of charge ^ p of the storage system ESS is no longer constant, the state of charge SOC having a tendency to form a plateau on the end of the simulation for the reasons that come to be mentioned. In this case, it may be difficult to assign a value properly representing the time variation of the state of charge pp of the energy storage system associated with the desired power Pac_ _sp for a given state of charge SOC. In order to take these two cases into account, in an embodiment called "with interpolation", the method according to a first aspect of the invention provides that each value of the correspondence table SOCV TC is:

- equal to the value of the variation of ASOC charge state obtained in said simulation time iess if the average power <P _ac> during said simulation is equal to the power _sp Pac_ setpoint (position illustrated in Figures 2A and 2B);

in the opposite case (situation illustrated in FIGS. 3A and 3B), equal to: interp {[P _{ac sp} {i - 1), <P _ac (i)>]>

with DSOC (it, j)

 -, the temporal variation of load obtained during the dt

simulation or calculated corresponding to the state of charge SOC (j) and a desired power Pac_ _sp (i-1);

ASOC (i, j), the variation of charge obtained in the simulation corresponding to the state of charge SOC (j) and a desired power Pac_ _sp (i);

^<I> ac (i, _^j)>> the average value of the power P _ac in the simulation corresponding to the state of charge SOC (j) and a desired power Pac_ _sp (i);

interp {[x ₀ , xl], [y ₀ , yi], x] is the function that determines the value of y corresponding to the value of x by interpolation from the values of

In other words, an interpolation is performed using simulations previously performed. To realize this interpolation, it is for example conceivable to carry out, from simulations of the first subset, a first table in which the columns represent different states of SOC fillers, rows represent different target powers Pac_ _sp and whose cells contain the average power <P _ac> during said simulation (see table 1) and a second table wherein the columns represent different states of SOC fillers, lines represent target powers Pac_ _s different and which the boxes contain the total variation of the ASOC load state during the simulation (see Table 2). Specifically, for each charge state SOC, a simulation is performed for a nominal power P _ac _ _s zero then the following simulations are performed for powers instructions _ac P _ _s growing up. Thus, when one reaches a power P _ac _ _s which can not be maintained during the simulation and which therefore requires interpolation, the temporal variation of the state of charge dSOC (il, j)

 obtained at the previous simulation can be used for the said

interpolation. The same is true of the negative values of the desired power P _ac _ _s . In other words, if we want to calculate the table temporal variation of the charge state for the set powers Pac_ _sp from Ps to Ps, simulations are carried out starting with a zero target power Pac_ that _sp is grown between each simulation to a value equal to Ps setpoint power. then, then performs the simulation by starting with a target power Pac_ _sp zero that is decreased between each simulation up to a desired power value equal to Ps.

Table 1

SOC (O) SOC (d-1) SOC (d) SOCO + 1) SOC (k ')

Pac sp (k) ASOC (k, 0) ASOC (k, j-1) ASOC (k) ASOC (k, j + 1) ASOC (k, k ')

Pac_sp (i + 1) ASOC (i + 1, 0) ASOC (i + 1, j-1) ASOC (i + 1) ASOC (i + 1 j + 1) ASOC (i + 1, k ')

Pac sp (1) ASOC (i, 0) ASOC (i, j-1) ASOC (ij) ASOC (i, j + 1) ASOC (i, k ')

Pac_sp (i-1) ASOC (i-1, 0) ASOC (i-1, j-1) ASOC (i-1) ASOC (i-1, j + 1) ASOC (i-1, k ')

Pac sp (O) ASOC (0.0) ASOC (0, j-1) ASOC (O) ASOC (0, j + 1) ASOC (0, k ')

 Table 2

6A and 6B illustrate 3D table the change in TC SOCV load state of the system (along the z axis) as a function of the desired power Pac_ _sp and the state of charge SOC. These two figures clearly show two zones separated by two black lines which correspond to the limits imposed by the minimum power and the maximum power, these limit values varying according to SOC state of charge as will be described later. This representation allows to reveal an important aspect on the table of the change in SOCV TC charge state of the system: it can assign a value to the time variation of the state of charge for setpoint powers Pac_ _sp higher (in absolute value) than the limit values which are the maximum power and the minimum power for a given state of charge. This is made possible only by the use of interpolation as just described or by using two other variants called "without interpolation" or "with the previous value" that we will describe now. It should be noted that the purpose of this interpolation is not to inform the table for values which in all cases will not be used because they are not attainable, but to make the points at the edge of the previously described zones more precise. In an alternative embodiment called "without interpolation", each value of the SOCV TC match table is equal to ^{A50 (t ,;)} , with ASOC (i, j) the tESS

load variation obtained in the simulation corresponding to the state of charge SOC (j) and a desired power Pac_ _sp (i). In fact, the inventors have demonstrated that, surprisingly, this approximation makes it possible to obtain very good results at a lower cost in terms of computing power (this aspect of the invention will be illustrated hereinafter).

In an alternative embodiment called "with the previous value", each value of the table of the variation of the state of charge SOCV TC of the system is equal to:

- the value of the variation of the state of charge ASOC obtained in said simulation time iess if the average power <P _ac> during said simulation is equal to the desired power Pac_ _sp.

in the opposite case, where is the variation of the charge obtained dt dt ³ during the simulation corresponding to the state of charge SOC (j) and the reference power Pac_s _P (m), Pac_s _P (m) being the power from nearest set the desired power Pac_ _s (i) for which the average power <P _ac> during said simulation is equal to the desired power Pac_ _s (m). Thus, when the target power P _ac _ _s can not be maintained during the simulation, the results of a previous simulation are used to assign a value to the time variation of the state of charge of the storage system ESS energy. This aspect makes it possible to limit the calculations necessary to obtain the SOCV TC table of the temporal variation of the state of charge of the system with however some limitations discussed in the following and illustrated in FIG. 9.

In one embodiment illustrated in FIGS. 7A and 7B, in order to obtain the table PAC_MAX_TC of maximum power as a function of the state of charge SOC, the plurality of simulations comprises a simulation carried out with a state of charge. initial SOCini equal to the state of maximum load SOCmax, a duration equal to the time necessary for the complete unloading of the storage system tDch (that is to say until the state of charge reaches the value of the minimum charge state SOCmin) and a power Pac_ _sp positive infinite set. The term power _sp positive infinity Pac_ set a desired power Pac_ _sp much greater than the power that the system can provide. Thus, the average power <Pac> k over a given interval can be considered as the maximum power for the state of charge SOCk corresponding to said interval. In addition, the calculation of the table PAC_MAX_TC of maximum power as a function of the state of charge SOC comprises:

 a step of determining periods of duration ISESS within said simulation, each period being identified by means of a positive integer k;

for each of these periods, a step of determining the state of charge SOCk at the beginning of the period k and the average power <P _ac > k during the period k.

The plurality of pairs (SOCk, <P _ac > k) then constitutes the table PAC_MAX_TC of maximum power as a function of the state of charge SOC so that for each state of charge SOCk given, it is possible to attribute a power maximum (equal to <P _ac > k). In one embodiment illustrated in FIGS. 8A and 8B, in order to obtain the PACJVIIN table TC of minimum power as a function of the state of charge SOC, the plurality of simulations comprises a simulation carried out with an initial state of charge SOCini equal to the minimum state of charge SOCmin, a duration equal to the time necessary for the complete charging of the storage system tcn (that is to say until the state of charge reaches the value of the maximum state of charge SOCmax) and a negative power of infinite _sp Pac_ set. Shall mean a set of _ac power P _ _s infinite negative a nominal power P _ac _ _s much greater than the power that the system can accept when charging. Thus, the average power <P _ac > k over a given interval can be considered as the minimum power for the state of charge SOCk 'corresponding to said interval. In addition, the calculation of the table PACJVIIN TC of minimum power according to the state SOC charge includes:

 a step of determining periods of duration ISESS within said simulation, each period being identified by means of a positive integer k ';

for each of these periods, a step of determining the state of charge SOCk at the beginning of the period k 'and the average power <P _ac >k' during the period k '.

The plurality of pairs (SOCk, <P _ac > k) then constitutes the table PAC_MIN_TC of minimum power as a function of the state of charge SOC so that for each state of charge SOCk 'given, it is possible to assign a minimum power (equal to <Pac> k).

In one embodiment, each simulation of a first subset of simulations is performed for a charge state SOC (j) and a given ac power sp (i) and each state of charge SOC (j) is separated. SOC of the previous (j-1) and / or the next SOC (j + 1) by a step of adaptive state of charge and / or each desired power Pac_ _sp (i) is separated from the previous Pac_ _sp (i- 1) and / or the next power Pac_s _P (i + 1) of an adaptive nominal power step.

In one embodiment, the simulation step is repeated for a plurality of ISES durations. Thus, one establishes a correspondence table simulation duration ÎESS, the correspondence table used by the control system being chosen according to the required accuracy.

FIG. 9 makes it possible to highlight the advantages of the model obtained by means of a method according to a first aspect of the present invention. The first HMC histogram corresponds to the simulation time when the simulation is performed using a complex system and with a time step Δΐ equal to 1 second. The graph also comprises five groups H1, H2, H3, H4, H5 of four histograms, each group corresponding to a simulation identical to that performed with the complex model, but carried out using a simplified model obtained by a method according to a first aspect of the invention and for a given simulation time. In other words, each grouping corresponds to a simulation performed with a simplified model and with a simulation time step equal to the simulation time ÎESS used during the method for determining the parameters of said simplified model.

In each grouping, the first histogram concerns the simulation time and the second, third and fourth histograms relate to:

 for the second histogram, the RMSE error between the simulation using the complex model and a simulation using a model obtained by means of a method according to a first aspect of the invention in the so-called "interpolation" embodiment ;

 for the third histogram, the RMSE error between the simulation using the complex model and a simulation using a model obtained by means of a method according to a first aspect of the invention in the so-called "no interpolation" embodiment ;

 for the fourth histogram, the RMSE error between the simulation using the complex model and a simulation using a model obtained using a method according to a first aspect of the invention in the embodiment with the value former ".

The grouping H1 relates to a case where the time ISESS is equal to 926 seconds. In this case, the precision obtained with the methods "with interpolation" and "without interpolation" is substantially identical while the precision obtained with the method "with the previous value" is lower. In addition, the simulation time using a model obtained by a method according to the invention is significantly less than the simulation time using a complex model according to the state of the prior art. The group H2 relates to a case where the time ISESS is equal to 463 seconds. It can be seen that the accuracy of the "interpolation" and "non-interpolation" methods is significantly improved while the precision with the "with the previous value" method has deteriorated significantly.

H3 groupings for a case where the time ISESS is equal to 93 seconds, H4 for a case where the time ISESS is equal to 47 seconds and H5 for a time the case where the time ISE is equal to 10 seconds only confirms the trends observed in the two preceding groupings, namely that the error decreases when the time ISESS decreases and the methods "with interpolation" and "without interpolation" make it possible to get similar results while the "with the previous value" method is always less efficient.

In order to illustrate the flow of the method according to a first aspect of the invention, we will now present a complex model comprising a model relating to a conversion device and a model relating to a storage device. It is important to note that the method according to a first aspect of the invention does not depend on the type of complex model used and the following complex model is given for purely illustrative purposes. It will make it possible to show the advantage in terms of simplification of the modeling (and therefore of the piloting) of an energy storage system that the model obtained by using a method according to a first aspect of the invention. In the following, the input values are initialized during the first iteration and then updated during the following iterations. The output values are calculated according to the input values as well as the model parameters. The model parameters are usually provided by the manufacturer of the conversion device or storage device, but can also be determined from experimental tests and measurements. On the other hand, we note X the value of the magnitude X at the iteration t. In addition, the function related to a correspondence table Y will be noted fv. It is also assumed that when a value is not directly available in a correspondence table, the latter is obtained by interpolation, for example linear interpolation, from the values available in said correspondence table (it is a standard method of using a lookup table).

The use of the model relating to a conversion device will now be detailed. The inputs, parameters and outputs relating to the model will be introduced as and when. The model allows firstly to calculate the φ phase angle between the active part Pac_ _sp and _sp Qac_ the reactive part of the target power. The active part Pac_ _sp and the reactive part Qac_s _P of the power set points are inputs of the model. The angle of departure is simply obtained using the following relation: However, to simplify the description that follows, the reactive power setpoint Qac_ _sp, is null the entire simulation and thus the value of the phase angle φ is independent of iteration.

Once the known phase angle is known, the model makes it possible to calculate the apparent maximum power Smax_LUT by means of a correspondence table Smax_LUT providing the maximum apparent power Smax as a function of the phase shift angle φ so that:

Smax = fs _max LUT (Φ)

The model then makes it possible to calculate the active powers Pac sat and reactive Qac_stat maximum that the conversion system can absorb. This calculation is carried out using a saturation function fsat which models the saturation applied to the active and reactive power setpoints. In other words :

^> ac_sat> Pac_sp> Qac_sp)

As mentioned above, the reactive power Qac_ _sp set is null throughout the simulation and thus, it is the same Q _ac _sat. The previous relationship can be rewritten as follows:

This condition ensures that even if the _sp Pac_ power setpoint is greater than the maximum S max power that can be supplied by the conversion device, the power actually delivered is limited by the maximum power conversion device can provide. The model then calculates the DC power Pdc corresponding to the target power Pac_ _sp using the correspondence table giving Pac_LUT alternative _ac power P as a function of the DC power Pdc, reactive power Ckc, tension Udc received at the input and the voltage U _ac ; in other words :

Pac ⁼ fp _ac _LUT (Pdc> Qao U do U ac)

The previous relationship should then be used to determine Pdc. For this, the values of P _ac and Q _ac are chosen so that P _has c = Pec sp and Q _a c = Ckc_sp, the voltage U _ac being known and equal to the nominal voltage of the conversion system U input name input , and the method of Newton-Raphson is used to solve the following minimization: m ^m (| ^ ac_sp ^- fp _ac _LUT (Pdc>Qac_sp> ^ do ^ ac) \ where P ^* dc is the initial value taken equal to P _ac _ _sp from which the minimization is carried out, said minimization making it possible to obtain the continuous power Pdc. The model then makes it possible to send the target current of the battery ldc_ _sp to the model of the storage device after the latter has been calculated using the following relationship:

The model also makes it possible to calculate the active power P _ac using the correspondence table P _ac _LUT introduced previously so that:

Pac ⁼ fp _ac _LUT ISPdo Qao U do U ac)

As mentioned above, the set reactive power Q _ac _ _sp is considered to be zero throughout the simulation and therefore the same is true of Q _ac . It is also important to note that the output voltage U _ac is imposed by the system so that it does not have a fluctuating voltage and its value is given by the nominal AC voltage of the Unom conversion system, the latter being a parameter of the model.

In this same embodiment, we consider a model on a storage device that inputs the reference current ldc_ _sp (supplied by the conversion device) and temperature. This model is coupled to the previously detailed model to form the complex model of the ESS energy storage system. In the following, we consider that the temperature of the storage device is provided by a model external to the complex model and provided at the input of the latter. In addition, the assumption is made that the temperature does not influence the behavior of the battery or that the temperature is maintained by an external device. In this case the inlet temperature remains constant and equal to the initial temperature. However, this is only an example of implementation and it is also conceivable to couple the complex model with a thermal model when the previous assumptions are no longer respected. The model initially used to calculate the power lost in the auxiliary system using a corresponding table Pdc_aux_LUT providing power lost Pdc_aux in the auxiliary storage device system according to the setpoint current ldc_ _sp received by the storage device so that:

^ dc_aux ~ fPdc_aux ^LUT 0dc_s) Considering that the elements of the auxiliary system (the fans for example) operate at their optimum voltage known to the user and equal to Udc_aux, the current consumed by said auxiliary system, the model makes it possible to obtain the auxiliary current I of to using the following relation:

 dcjiux

Once the current consumed by said auxiliary system ldc_aux known, the model is used to evaluate the influence of the latter on the setpoint current ldc_ _{sp in} order to obtain a modified setpoint current Idc 'using the following relation : dc ^l dc_sp ^l dc_aux

In other words, when the storage device is charged at a current ldc_ _sp, the current actually used for said filler is equal to Idc. In the same way, when the storage device is discharged by a current ldc_ _sp , the current that the storage system can actually provide is equal to Idc '.

The model also makes it possible to take into account the fact that the current Idc 'is limited by a maximum current IchMax during the charging of the storage system and a maximum current i DchMax during the discharge of said system. For this, the maximum current IchMax during charging is determined using a correspondence table lchMax_LUT providing the maximum current I chMax during charging according to the state of charge SOC, the state of health SOH and temperature T: lchliax ⁼ 0H ^t _j r ^t )

Similarly, the maximum current I DchMax during the discharge is obtained using correspondence table l DchMax_LUT providing the maximum current I DchMax during the discharge according to the state of charge SOC, the state of health SOH and temperature T:

chMax ~ flDchMax-L T ÎSOC ^ SOH ^ T ¹ )

In other words, the current Idc can be determined using the following relation: ax

The model also makes it possible to calculate the variation of the state of charge of the storage device. This variation can be obtained using a SOCs _P eed_LUT correspondence table providing the temporal variation of the state of charge as a function of SOC state of charge, SOH state of health, Idc current. and the temperature T so that:

The initial state of charge being known, the model is able to calculate the states of charges corresponding to the following iterations using the following relations: dSOC *

SOC ^{t + 1} = SOC ¹ + - ^■ x At

 dt

where Δΐ is the time step between two successive iterations.

The model also makes it possible to calculate the state of health of the storage means. This state of health is estimated through aging that corresponds to a change in health status. This aging has two components (which are negative):

aging due to time (called calendar) Α50Η _α15ρββά ; and

aging due to ASOH _CycSpeed cycles.

The model makes it possible to determine aging due to time ASOHj _alSpeed by means of a mapping table ASOHcais _P eed_LUT providing aging over time as a function of state of charge SOC, state of health and the temperature so that:

Similarly, the model makes it possible to determine the aging due to the cyclization ASOHcycspee _d using a correspondence table ASOHcycs _P eed_LUT providing the aging due to the cycling as a function of the state of charge SOC, the state of health SOH , the current Idc and the temperature T so that:

ASOH ^ _Speed = / ASOH _cycSpeed SSOH ^t , T ^t , SOC ^t , I _ac )

The state of health is then calculated using the following relationship

A5OH ^f - ASOH _çalSpped + ASOH _çycspeed The model then makes it possible to determine the resistance Rch of the system during charging by means of a correspondence table Rch_LUT providing the resistor Rch of the system during charging as a function of the current Idc during charging, the state of SOC load, SOH health status and T temperature so that:

RCH ¹ = fR _ch _LUT (SOH ^T , T ^T , SOC ^T , I ^ ¹ )

In the same way, the model allows to determine the system resistance Rûch during the discharge using a correspondence table RDch_LUT providing the resistance Rûch of the system during the discharge according to the current Idc during the discharge, the state of charge SOC, state of health SOH and temperature T so that:

On the other hand, the model makes it possible to know the open circuit voltage OCVch during charging by means of a correspondence table OCVch_LUT providing the open circuit voltage OCVch during charging as a function of the charge state SOC, of the state of health SOH and current Idc so that:

Similarly, the model makes it possible to know the open circuit voltage OCVuch during the discharge by means of a correspondence table OCVDch_LUT supplying the open circuit voltage OCVuch during the discharge as a function of the state of charge SOC, of the state of health SOH and current Idc so that:

In addition, the model makes it possible to determine the DC voltage Udc at the terminals of the storage system using Ohm's law by means of the following relation: U _dc = OCVc _h / _Dch - l _dc x Rch / och where I is negative when charging and positive when discharging. It should be noted that when the value obtained for DC voltage Udc at the terminals of the storage system exceeds the maximum voltage U ^ _axch , an indicator for exceeding the maximum voltage Umax_limit_status is updated. This maximum voltage can be calculated using the following relation:

U ChMax ⁼ OCV ^t - l _C hMax ^x RCh

Similarly, when the value obtained for DC voltage Udc at the terminals of the storage system is lower than the minimum voltage ItachMax, an overflow indicator and the minimum voltage Umin_limit_status is updated. This maximum discharge voltage can be calculated using the following relation:

^ DchMax ⁼ OCV ^t - IochMax ^x RDch

The model then makes it possible to calculate the power supplied by the battery Pdc by means of the following relation:

? dc ⁼ Ud.c ^X ld.c Of course, this power value is limited by the voltages and the limit currents so that:

PfnaxCh ~ ^ maxCh ^x ^ maxCh

 -

maxDch 'maxDch ^{Λ u} maxDch

As already mentioned, the conversion device model is combined with the storage device model to obtain a complex model capable of modeling an energy management device. Thus, the energy storage device can be simulated. However, as has just been detailed, the complex model comprises a large number of parameters and variables to be calculated and is difficult to use for efficient control of such an installation, especially since the pilot's time step of such a system is of a few minutes, even a few hours, whereas the step of simulation time in the case of a complex model must be of the order of one second or even less than the second so as not to lose precision in the results obtained. With the solutions of art previous and having only the complex model that has just been detailed, the user is therefore placed in front of two options both unsatisfactory:

 or it chooses a simulation time step of the same order of magnitude as the pilot time step in order to limit the computing power required, but the accuracy of the results of the simulation (and therefore of the piloting) will be unsatisfactory;

 - Or it maintains a low simulation time step to maintain good accuracy, but the resources used to perform all calculations are then much larger and may not be available.

Thus, no solution to retain the advantage of each of these options is available, and this absence is filled by the present invention. The latter proposes for this, as has been described above, a plurality of simulations is performed using a complex model as just described so as to obtain three tables of correspondence:

- a table of the variation of the state of charge (SOCV TC) system depending on the desired power Pac_ _sp and the state of charge SOC;

a maximum power table (PAC_MAX_TC) according to the state of charge SOC;

 - a minimum power table (PAC_MIN_TC) according to the state of charge SOC.

Using the model thus obtained and to part of a target power Pac_ _sp, and a charge state SOCP inputted (SOC ¹ noted here because its value is given by the iteration t), it is It is possible to calculate the minimum and maximum output power of the system using the maximum power table PAC_MAX_TC provided according to the state of charge SOC and the maximum power table PACJVIIN TC accepted according to the state of SOC load. More particularly, the maximum power is given by:

PACM X ⁼ fpAC_MAx_Tc (SOC ^T )

Similarly, the minimum power is given by: PlcMin - PAC_MI _TC ( _' 5OC ^Î :)

If the power set point is in the range [ΡΑΟΜΙΠ, ΡΑΟΜΒΧ], then the Pac power is equal to the power setpoint Pac_ _sp. Otherwise, the Pac power is equal to the power limit PACMin / PACmax closest to the target power Pac_ _sp.

In addition, the time variation of the state of charge can be calculated using the table of the variation of the state of charge (SOCV TC) of the system according to the target power and a Pac_ _sp state of charge SOC using the following relation: dSOC * _ _tt

~ ^~ dt ^{~ ~} fsOCV _TC (.SOC> Pac_sp)

It is interesting to note that only the initial value of the SOCini state of charge is necessary, the following values of the state of charge being determined using the following relation:

, dSOC '

soc ^{t + 1} = so + - xt _ESS

^t dt It is important to note here that when the system is simulated using a model according to the invention, the time step between two iterations is not equal to Δΐ like, but the term used when iess implementation of the method according to a first aspect of the invention. However, since the model's correspondence tables were calculated from simulations performed with a time step Δΐ, the precision of the results obtained with the simplified model remains very close (see Figure 9) of those obtained with a complex model, without however, require as much computing resource. In other words, the simplified model obtained using a method according to a first aspect of the invention makes it possible to go from a simulation time step equal to Δΐ to a simulation step equal to ÎESS without significant loss in accuracy. results obtained. We understand then all the interest that can have such a model, especially in the control of a storage system energy. For this, an embodiment of a second aspect of the invention relates to a method for controlling an energy storage system ESS calculating operating instructions of said ESS system from a model of said system characterized by said model is obtained using a method 100 according to one of the preceding claims.

For example, the method may comprise a first initialization phase during which a method 100 according to a first aspect of the invention is implemented. At the end of this first phase, the method comprises a second phase during which instructions to the energy storage system are generated at regular intervals using the model obtained during the initialization phase of the system. to drive said ESS system. In general, said setpoints take the form of an optimal flight path whose calculation is performed for a given time horizon, typically of the order of several hours, for example 12h. This trajectory is updated regularly according to a period of PEI update, usually several times per hour, for example every 15 minutes. For each update (every 15 min for example), a simulation covering the horizon of the trajectory (12h for example) must be carried out within the control system. This update takes into account the evolution of the state of charge of the storage system and, possibly, information concerning parameters likely to influence the evolution of the system (such as weather forecast information if the system of storage is connected to a renewable energy source for example). This update is performed using a simulation whose time step is equal to the duration ISESS, for example 1 minute, said simulation being carried out using a simplified model obtained by a method according to a first aspect of the invention. It is interesting to note that with a complex model according to the prior art, a simulation time step of one minute would not make it possible to obtain sufficient accuracy to update the flight path. It would therefore be necessary to reduce the simulation time step and thus increase the calculation time and the resources required for this calculation. The simplified model obtained by means of a method according to a first aspect of the invention makes it possible to solve this technical problem by allowing precisely to adopt a much higher time step. than that of the prior art without this entailing significant losses in the precision of the flight path.

During the first initialization phase, it is therefore preferable to take into account the duration ÎPIL separating two updates of the trajectory. For this purpose, in one embodiment, the simulation duration ESSESS is chosen such that ÎESS <ÎPIL, preferably such that nxtEss = ÎPIL, with n a positive integer. Thus, knowing the duration ÎPIL separating two updates of the pilot trajectory, it is possible to choose the most appropriate simulation simulation time step ÎESS.

In order to implement such a method, an embodiment of a third aspect of the invention relates to a control device of an energy storage system ESS comprising means for sending operating instructions to the system of storage. storing the ESS energy and means for receiving data relating to the operation of the ESS energy storage system. For example, the control device and the ESS energy storage system communicate via an Ethernet network and the control device comprises an Ethernet type network card. The control device also comprises means for implementing a control method 100 according to the preceding claim. The control device may in particular comprise data acquisition means relating to the complex model relating to the energy storage device DSE and / or the conversion device DC, such as a keyboard associated with a screen or even a touch screen. Alternatively or additionally, the control device comprises means of connection to a network, for example the Internet, and the data relating to the complex model associated with the energy storage system ESS are retrieved from a server, for example the server of the manufacturer of the DC conversion and energy storage devices DE that comprises said ESS system. The control device also comprises calculation means, for example a processor or an ASIC card, said calculation means making it possible to carry out the steps of the method according to a first aspect of the invention in order to obtain a simplified model of the storage system. energy, but also to generate operating instructions from said simplified model.

Claims

A method (100) for determining the parameters of a simplified model of a computer-implemented energy storage system (ESS), said system (ESS) including an energy storage device (ESD) and a conversion device (DC), said system (ESS) being modelable by means of a complex model (MC) including a model of the energy storage device (MCS) and a model of the conversion device (MCC) ; said complex model (MC) receiving as input a desired power Pac_ _sp and a state of SOC load _P, and outputting the state of charge SOC of the storage device and the Pac output power of the storage device; said method being characterized in that it comprises:

 a first step (101) of implementing a plurality of simulations of the energy storage system (ESS) using the complex model (MC);

 a second calculation step (102) based on the results obtained during the first step (101):

^■ a table of the time variation of the state of charge (SOCV TC) of the system according to the ac power setpoint SP and the state of charge SOC;

^■ a maximum power table (PAC MAX TC) according to the state of charge SOC;

^■ a minimum power table (PAC_MIN_TC) according to SOC state of charge;

the simplified model obtained for assigning a function of a desired power Pac_ _sp and a state of charge SOC _P inputted, and from the tables (SOCV_TC, PAC_MAX_TC, PAC_MIN_TC) determined during the second step ( 102), a power P _ac and a state of charge of the SOC system.

Method (100) according to the preceding claim characterized in that each value of the table of the variation of the state of charge (SOCV TC) of the system is obtained with a simulation of a duration iess made for a state of charge SOC (j) and a desired power Pac_ _sp (i) given and belonging to a first subset of the plurality of simulations , said value being:

equal to the value of the variation of the state of charge ASOC obtained during said simulation on the time ISESS given by ^ 2 £. _S j \ _Tess average power <P _ac> during said simulation is equal to the desired power Pac_ _sp.

 - otherwise, equal to:

Γ Γ _Γ . _f -, _r .

interp {[P _{ac sp} (i - 1), <P _ac (i,

 with:

^{■ d50} ^ ^{,; ~ 1),} the variation of charge obtained in the simulation corresponding to the state of charge SOC (j) and a desired power Pac_ _sp (i-1);

^■ ASOC (i, j), the variation of charge obtained in the simulation corresponding to the state of charge SOC (j) and a desired power Pac_ _sp (i);

^■ <Pac _{-> i)>>} a mean value of the power P _ac in the simulation corresponding to the state of charge SOC (j) and a desired power Pac_ _s (i)!

^■ interp {[x ₀ , xl], x} is the function that determines the value of y corresponding to the interpolation value from the values of x ₀ , x ₁ , y ₀ and y ₁ .

A method (100) according to claim 1, characterized in that each value of the state of charge variation table (SOCV TC) of the system is obtained by means of a simulation of a duration ISESS performed for a state of charge SOC (j) and a desired power Pac_ _s (i) given and belonging to a first subset of the plurality of simulations, said value being equal to the value of the variation of the state of charge ASOC obtained during said simulation on the time ISESS. Method (100) according to claim 1, characterized in that each value of the table of the temporal variation of the state of charge (SOCV TC) of the system is obtained by means of a simulation of a duration ISESS carried out for a charge state SOC (j) and a desired power Pac_ _sp (i) given and belonging to a first subset of the plurality of simulations, said value being:

- equal to the value of the variation of ASOC charge state obtained in said simulation time iess given by if the TESS average power <P _ac> during said simulation is equal to the desired power Pac_ _sp.

 ,, dSOCCm) dSOC (m, i). . . . . .

- in the opposite case, equal to - ^ - or - ^ - is the variation of

³ dt dt

charge obtained in the simulation corresponding to the state of charge SOC (j) and the target power Pac_s _P (m), Pac_s _P (m) being the nearest target power of the desired power Pac_ _s (i ) for which the average power <P _ac > during said simulation is equal to the reference power Pac_ _s (m).

Method (100) according to one of the three preceding claims, characterized in that each state of charge SOC (j) is separated from the preceding SOC (j-1) and / or the following SOC (j + 1) by a step of state adaptive load and / or each desired power Pac_ _s (i) is separated from the previous P _ac _s (i-1) and / or the following Pac_ _s (i + 1) of a desired power of not adaptive.

Method (100) according to one of Claims 3 to 5, characterized in that the simulation step is repeated for a plurality of ISES durations.

Method (100) according to one of the preceding claims, characterized in that the plurality of simulations comprises a simulation performed with an initial state of charge SOCini equal to the maximum state of charge SOC max, a duration equal to the duration necessary for unloading TDCH complete the storage system and a target power P _ac _ _s infinite positive and in that the calculation of the table PAC_MAX_TC maximum power depending on the state of charge SOC comprises: a step of determining periods of duration ISESS within said simulation, each period being identified by means of a positive integer k;

for each of these periods, a step of determining the state of charge SOCk at the beginning of the period k and the average power <P _ac > k during the period k;

the plurality of couples (SOCk, <P _ac > k) constituting the table PAC_MAX_TC.

8. Method (100) according to one of the preceding claims characterized in that the plurality of simulations comprises a simulation carried out with an initial load state SOCini equal to the state of minimum load SOCmin, a duration equal to the time required to fully loaded tch storage system and negative infinity _sp Pac_ desired power and in that the calculation of the minimum power PACJVIIN table TC depending on the state of charge SOC comprises:

 a step of determining periods of duration ISESS within said simulation, each period being identified by means of a positive integer k ';

for each of these periods, a step of determining the state of charge SOCk 'at the beginning of the period k' and the average power <P _ac > k 'during the period k';

the plurality of couples (SOCk, <P _ac > k) constituting the table PAC_MIN_TC.

9. A method for controlling an energy storage system (ESS) calculating operating instructions of said system (ESS) from a model of said system (ESS), characterized in that said model is obtained from using a method (100) according to one of the preceding claims.

10. A device for controlling an energy storage system (ESS) comprising means for sending operating instructions to the energy storage system (ESS), means for receiving data relating to the operation of the system storage of energy (ESS) characterized in that it comprises means for implementing a control method according to the preceding claim.