Method and device for predictive operation of a device battery with a power supply network

Abstract

The invention relates to a computer-implemented method for operating a device battery (41) of a technical device (4), in particular at a charging station (3) with an electrical energy supply network (2) with varying energy availabilities, comprising the following steps: - determining (S4) a predicted temporal course of energy availability in the energy supply network (2) using an energy prediction model; - providing (S6) charging and discharging profiles for the device battery (41); - Providing an aging state model which is designed to determine an aging state change based on temporal profiles of operating variables of the device battery (41); - Providing an indication of a target state of charge and an optimization period after which the target state of charge is to be reached; - performing an optimization method based on a cost function that depends on the predicted energy availability in the energy supply network during the optimization period, the energy consumption and energy output of the device battery (41) during the optimization period and a change in the ageing state at the end of the optimization period, and the indication of the target state of charge in order to obtain a usage profile that indicates a temporal sequence of time periods of the use of charging profiles and / or discharging profiles; - Providing (S7) the usage profile for operating the device battery (41) as a buffer storage in order to provide no transmission of electrical energy, charging according to the respective charging profile or discharging according to the respective discharging profile in each time period according to the usage profile.

Description

Technical field

[0001] The invention relates to portable batteries that are regularly connected to an electrical power grid for charging. The invention further relates to an energy management system for cost-optimized operation of a portable battery at a charging station in conjunction with varying energy availabilities in the power grid. Technical background

[0002] Energy availability in electrical power grids varies considerably over time cycles, such as time of day, day of the week, seasonal fluctuations, and weather conditions (e.g., due to variable energy inputs from solar and wind power systems). This energy availability is typically reflected by variable prices on the energy spot market. Thus, during periods of high energy availability, negative prices may be generated for the consumed kilowatt-hour, while during periods of energy scarcity, very high prices per kilowatt-hour may occur.

[0003] To smooth the course of energy availability, energy storage systems are required, which are currently only available to a limited extent for an energy supply network and which make it possible to buffer or compensate for the temporal course of energy availability by releasing energy when energy demand is high and absorbing and storing energy when energy demand is low.

[0004] With the increase in electromobility, a variety of vehicle batteries are available as examples of portable batteries. However, the simple use of portable batteries as buffer storage, depending on energy availability, leads to repeated charging and discharging, sometimes under unfavorable conditions, such as low temperatures at high charging currents. The loss of value of the portable battery due to increased aging reduces or even negates its usefulness as an energy buffer for the energy supply grid. It is therefore advantageous to provide improved energy management for the operation of decentralized energy storage systems in an electrical power supply grid.In particular, predictive energy management, which involves temporarily storing energy from decentralized devices such as electric vehicles, can help to serve as an effective energy buffer, acting as a temporary sink in times of energy surplus and as a source of electrical energy in times of energy shortage. Disclosure of the invention

[0005] According to the invention, a method for operating a device battery at a charging station with an electrical energy supply network with varying energy availabilities according to claim 1 and a corresponding device according to the independent claim are provided.

[0006] Further embodiments are specified in the dependent claims.

[0007] According to a first aspect, a computer-implemented method for operating a device battery, in particular at a charging station with an electrical energy supply network with varying energy availabilities, is provided, comprising the following steps: - Determining a modeled or predicted temporal course of energy availability in the energy supply network using a data-based energy prediction model; - Providing charging and discharging profiles for the device battery; - Providing an aging state model which is designed to determine an aging state change based on temporal profiles of operating variables of the device battery; - Providing an indication of a target state of charge and an optimization period after which the target state of charge is to be reached; - Carrying out an optimization process based on a cost function that depends on the predicted energy availability in the energy supply network during the optimization period, the energy consumption and energy output of the device battery during the optimization period and a change in the ageing state at the end of the optimization period, and the specification of the target state of charge in order to obtain a usage profile that specifies a temporal sequence of time periods of use of charging profiles and / or discharging profiles; - Providing the usage profile for operating the device battery as a buffer storage in order to provide no transmission of electrical energy, charging according to the respective charging profile or discharging according to the respective discharging profile in each time period, according to the usage profile.

[0008] Furthermore, the device battery can be operated according to the usage profile.

[0009] As described above, energy availability in an energy supply grid can vary considerably over time. This may result in the need for energy storage for electrical energy during certain periods to effectively store excess electrical energy. Device batteries, such as vehicle batteries or household batteries, are ideal for this purpose. These batteries are regularly connected to the energy supply grid for charging or are permanently connected to it. These have relatively high storage capacities and, due to their large number, offer great potential for use as electrical buffer storage.

[0010] However, the simple demand-controlled supply or removal of electrical energy in or from these portable batteries leads to different aging behavior depending on the charging and discharging profiles used and the battery condition.

[0011] The state of health (SOH) is the key parameter for specifying the remaining battery capacity and remaining service life of portable batteries as energy storage devices. The state of health represents a measure of the aging of the portable battery. In the case of a portable battery, a battery module, or a battery cell, the state of health can be specified as the capacity retention rate (SOH-C). The capacity retention rate SOH-C, i.e. the capacity-related state of health, is specified as the ratio of the measured instantaneous capacity to the initial capacity of the fully charged battery and decreases with increasing aging. Alternatively, the state of health can be specified as the increase in internal resistance (SOH-R) relative to the internal resistance at the beginning of the service life of the portable battery. The relative change in internal resistance SOH-R increases with increasing battery age.

[0012] The aging state and the changes or progression of the aging state determine the remaining service life of the device battery and thus significantly determine its value. Therefore, when used as a buffer storage device for an energy supply network, it makes sense to predictively control the electrical energy input into the device batteries and the electrical energy output by the device batteries in order to enable cost-optimized use of the device batteries as electrical buffer storage devices for an energy supply network.

[0013] It can be provided that the energy prediction model is designed or trained to indicate energy availability in the energy supply network depending on a calendar time indication and in particular weather information, which may include solar radiation and wind speeds, and / or information on operating times of one or more connected energy sources.

[0014] First, an energy prediction model is provided that models the energy availability of an electrical power grid. The energy prediction model can be data-driven or hybrid and can enable a temporal prediction of energy availability. The energy prediction model can include an ARIMA model, a deep neural network, a recurrent neural network, a transformer, or a Gaussian process.

[0015] The energy prediction model is trained to model a measure of the energy availability of the energy supply grid at a specific time step, depending on a calendar time specification, the anniversary of the year, the day of the week, the time of day, public holidays, the geographical location, and, if applicable, weather forecasts such as solar radiation, wind speeds, wave movements, etc. By recurrently applying the energy prediction model, a prediction of energy availability can be made in one embodiment. This results in a future course of energy availability. The energy availability can be specified as an amount of energy or in the form of a cost value that reflects the energy availability. The energy availability can also be represented in reverse as energy demand. The prediction can be provided probabilistically and include, for example, a probability density function or a confidence interval.

[0016] Aging models and information on the respective battery status are available for each device battery. If a device battery is connected to the power grid via a charging station or is permanently connected, it can be used as a buffer storage for the power grid.

[0017] When used as a buffer storage system, charging and discharging processes are based on the energy availability in the power grid. This means that when energy availability is positive, electrical energy is stored in the buffer storage system, and when energy availability is negative, energy is released from the buffer storage system. The absorption and release of electrical energy can occur according to predefined charging and discharging profiles.

[0018] Using an aging state model, a change in the aging state—i.e., the degradation of the device battery—can be determined during the period in which the device battery is connected to a charging station or for a predetermined optimization period while the device battery is used as a buffer storage device. This degradation can be attributed to a loss of value in the form of a cost. These costs can be expressed in monetary terms or in the form of the remaining service life of the device battery. Alternatively, the SOH (State of Health) can be used, based on capacity or impedance, to technically quantify the battery's state of health.

[0019] The aging state and thus the corresponding change in the aging state can be simulated or determined using conventional and known models for determining the aging state. In particular, an electrochemical aging state model can be used, which is fundamentally based on an electrochemical battery model. Such an electrochemical battery model can comprise a system of differential equations that, based on differential equations parameterized via model parameters, models internal battery states, in particular equilibrium states and, where appropriate, kinetic states, using a time integration method and provides a relationship between temporal operating parameter profiles of the device battery, namely a battery current, a battery voltage, a battery temperature, and a state of charge of the device battery, and the internal battery state.Such electrochemical battery models are known, for example, from publications US20220179009A1, US20220334191A1, US20220099743A1, US 2016 / 023,566, US 2016 / 023,567, and US 2020 / 150,185. An aging state can be derived from the internal battery states. Data-based or hybrid models as aging state models can also be used to determine the change in aging state.

[0020] It can be provided that the charging profiles and / or discharging profiles are selected by the optimization method from a number of predetermined charging profiles and discharging profiles, wherein the predetermined charging profiles and discharging profiles are provided with the aid of an electrochemical battery model of the device batteries which is predetermined for the device batteries of the same type, by assuming different values for at least one electrochemical dimensioning parameter and in each case a charging profile or a discharging profile is determined which determines a charging current with regard to a state of charge by simulation in such a way that the anode overpotential always corresponds to the value for the at least one electrochemical dimensioning parameter.

[0021] With the help of simulations, an optimal sequence of application of charging and / or discharging profiles can now be determined depending on the energy availability, whereby the respective charging or discharging profile is selected from a large number of predefined charging and discharging profiles.

[0022] The optimization is based on a cost function that includes, on the one hand, a target charging to a certain state of charge within the specified optimization period, the loss of value or degradation of the device battery due to the aging effect of charging or discharging during the optimization period (i.e., the change in the aging state), and a benefit resulting from use as a buffer storage device (i.e., from the absorption or release of energy from the power grid depending on energy availability). The change in the aging state can be determined as the difference between the aging state at the beginning and the end of the optimization period.

[0023] It may be provided that the cost function further takes into account in particular: an indication of the grid stability of the energy supply network, e.g. as a type of volatility indication, which may in particular include a standard deviation of a historical time series, and an indication of the monetary charging process costs, because these are a measure of the technical availability of energy in the grid: positive energy availability at negative prices and negative energy availability at positive prices.

[0024] Since the power availability can change significantly during the optimization period, during which the device battery is to be charged to a certain level, several charging and discharging periods, each with different charging / discharging profiles, can be performed within the optimization period in order to optimize the system for the lowest possible degradation of the device battery, the achievement of a specific state of charge after the period has elapsed, and the utility as buffer storage by the energy supply grid. The simulation is carried out for the entire optimization period, by which the specified state of charge is to be achieved. The time periods can be determined by value ranges of different predicted energy availability or can be of the same duration.For example, the optimization period can be divided into a predetermined number of time periods of equal duration.

[0025] The charging and discharging profiles are determined by a maximum battery current dependent on the state of charge. Possible charging and discharging profiles can, for example, result from a Pareto optimization between an aging influence / change in the aging state and charging or discharging time information that indicates an aggregated or average charging or discharging time. The charging or discharging profiles can be determined by simulating a control system based on a specific dimensioning parameter, such as a limit value for an anode overpotential.

[0026] The charging curves or discharge profiles can be determined by simulation using an electrochemical battery model based on electrochemical and / or physical variables, such as the anode overpotential, and selected based on a Pareto-optimal relationship between the average / aggregated charging time information and the change in aging state. The Pareto-optimal relationship corresponds to a Pareto curve that can be determined by simulation using a predefined cost function that takes into account the charging time information or the discharging time information and the change in aging state. The charging profile can be generated in a conventional manner such that the anode overpotential does not fall below a limit value of a battery type-specific anode overpotential. In this context, several Pareto-optimal charging profiles orConfigurations of such charging profiles can be derived by specifying different limit values of the anode overpotential with the aid of control engineering approaches, such as simulation or control to a limit value.

[0027] The determination of suitable charging profiles, for example based on a battery model, is known from the state of the art, e.g. B. from US 9,153,991 B2, US 2017 / 033 866 A1, DE 10 2020 206 272 A1, DE 10 2020 210 132 A1, DE 10 2019 209 165 A1, DE 10 2020 212 579 A1, US 10,447,054 B2, US 9,153,991 B2, US 10,312,699 B2, DE 10 2019 216 015 A1 and from Doyle, M, Fuller, TF, & Newman, J., “Modeling of galvanostatic charge and discharge of the lithium / polymer / insertion cell”, United States, https: / / doi.org / 10.1149 / 1.2221597.

[0028] The charging and discharging profiles to be considered in the optimization process can be selected according to corresponding Pareto curves.

[0029] The determination of the corresponding usage profile, since the device battery's control system acts as a buffer storage device, can preferably be carried out in a central unit remote from the device, in which algorithms for predicting future energy availability, the aging model of the device battery in question, and the underlying cost function are implemented. Thus, after the device battery is connected to the power grid, a request can first be sent to the central unit to create an optimized usage profile for the device battery in question, subject to the conditions that a certain charge level must be reached within a predetermined period of time or by a predetermined time. The time period can then correspond to the optimization period.

[0030] The central unit determines the usage profile through the above optimization and transmits this back to the device control unit in order to operate the device battery during the period or until the predetermined time, according to the usage profile.

[0031] It may be provided that the device battery is used as a buffer storage if there is a smart contract for the device battery which includes a cost function for the optimization and / or specifies a dependency on a state of the device battery which is determined in real time and specifies a boundary condition, in particular by means of a limit value of an energy price or an efficiency factor, which specifies whether the device battery is to be used as an energy buffer.

[0032] When connecting the device battery to the power grid, it can also be determined by identifying the device battery whether its use as a buffer storage device is permitted. In particular, the use of the device battery as a buffer storage device with an associated quantification of an energy flow or energy quantity within a defined time frame is permitted if a smart contract exists that also allows billing with the energy grid operator for its use as a buffer storage device.

[0033] The smart contract can also be dependent on a state determined in real time and can also be dependent on a threshold value set by the user, e.g., the smart contract can specify that: - a user wants to fully charge his battery when the electricity price is cheaper than a defined limit; - the amount of energy to be removed should be selected depending on the current state of age and state of charge of the device battery, whereby the state of age can represent a boundary condition of the optimization problem because, for example, the battery management system (BMS) cannot implement certain high currents due to safety requirements, states and environmental conditions.

[0034] In addition to optimizing the cost function, the change in the state of the device battery when using certain charging profiles, current limits, operating strategies of the battery management system and other technical or economic variables can be taken into account. Brief description of the drawings

[0035] Embodiments are explained in more detail below with reference to the attached drawings. They show: Fig. 1 a schematic representation of a system with an energy supply network for providing electrical energy, a charging station for a vehicle battery and an electrically powered vehicle; Fig. 2 a flowchart illustrating a method for the optimized operation of a vehicle battery connected to a charging station depending on the energy availability in the energy supply network; Fig. 3 an illustration of the energy availability curve in the energy supply network; and Fig. 4 a Pareto curve for a charging or discharging profile of a vehicle battery. Description of embodiments

[0036] The method is described below using a vehicle battery as a device battery. Vehicle batteries are only temporarily located at a charging station and can serve as buffer storage. Alternative device batteries can also be house batteries in conjunction with a photovoltaic system, which are permanently connected to the energy supply grid but should preferably have a certain state of charge at a specific time that depends on the time of sunset. In addition, vehicle pools consisting of mobile fleet vehicles or stationary battery storage pools or parking garages for Vehicle2Grid applications can be connected to the energy supply grid and participate in the energy exchange. The method is described below using mobile vehicles, comprising a battery as an energy converter or energy storage system, but embodiments of the method with stationary energy converters orEnergy storage systems are possible.

[0037] Fig. Figure 1 schematically shows an energy supply system 1 with an energy supply network for providing electrical energy to consumers. The energy supply network 2 draws energy from a plurality of power generators 21. Furthermore, a charging station 3 is connected to the energy supply network 2, to which an electrically powered vehicle 4 with a vehicle battery 41 can be connected in order to charge the vehicle battery 41.

[0038] When a vehicle 4 arrives at the charging station 3, a charging end time can be specified according to a user specification or model prediction, by which the vehicle battery 41 should have reached a specified state of charge. In particular, a typical departure time can be learned and determined in a conventional manner using data-based models based on repeating patterns. This can be done probabilistically, in particular, and implemented with a hidden Markov model. The future charging end time thus lies a specified time period in the future and can be described probabilistically.

[0039] Furthermore, a central unit 5 remote from the vehicle is provided, which receives current data on energy availability from the energy supply network 2 or a supply network provider 22.

[0040] Furthermore, at the time of connecting the vehicle 4 to the charging station 3, the central unit 5 receives information about an identification of the vehicle battery 41, via which a current state of the vehicle battery 41, in particular the current state of charge and the current state of aging, can be recorded or retrieved.

[0041] A data-based energy prediction model is provided in the central unit 5, which predicts the energy availability of an electrical power grid. The energy prediction model can be data-based or implemented in a hybrid manner and can enable a temporal prediction of energy availability.

[0042] The energy prediction model is trained to model a measure of the energy availability of the energy supply grid 2 at a specific time step, depending on a calendar time specification, the anniversary, day of the week, time of day, holiday information, geographical location, other information about connected energy sources (power plant shutdowns, etc.), and, if applicable, weather forecasts. By recurrently applying the features of the energy prediction model, a prediction of energy availability can be made. This results in a future course of energy availability.

[0043] The energy prediction model can be supervised and trained using historical energy availability, possibly taking weather data into account. Because the label can be easily provided, the energy prediction model can be automated and regularly retrained, e.g., once a week, to incorporate the latest information, particularly inefficiencies in the energy market regarding energy availability.

[0044] Energy availability can be expressed as a quantity of energy or as a cost value that reflects energy availability. Energy availability can also be expressed in reverse as energy demand.

[0045] Furthermore, an aging state model can be implemented in the central unit. In particular, an electrochemical aging state model can be used, which is fundamentally based on an electrochemical battery model. Such an electrochemical battery model can comprise a system of differential equations that, based on differential equations parameterized via model parameters, models internal battery states, in particular equilibrium states and, if applicable, kinetic states, using a time integration method and provides a relationship between operating parameter profiles of the device battery, namely a battery current, a battery voltage, a battery temperature, and a state of charge of the device battery, and the internal battery state. Such electrochemical battery models are known, for example, from the publications US 2016 / 023,566, US 2016 / 023,567, and US 2020 / 150,185.An aging state can be derived from the internal battery states.

[0046] A possible aging state model can also be provided in the form of a hybrid aging state model, which corresponds to a combination of a physical aging model based on the evaluation of differential equations with a data-based correction model. In a hybrid model, a physical aging state can be determined using a physical or electrochemical aging model, and this can be applied with a correction value resulting from the data-based correction model, in particular by addition or multiplication.The physical aging model can be based on electrochemical model equations as described above, which characterize electrochemical states of a nonlinear differential equation system with regard to aging reactions, continuously calculate them according to a time integration method and map them to the physical aging state for output, as SOH-C and / or as SOH-R.

[0047] Furthermore, the correction model of the hybrid data-based aging state model can be designed with a probabilistic or artificial intelligence-based probabilistic regression model, in particular a Gaussian process model, and can be trained to correct the aging state obtained by the physical aging model. For this purpose, a data-based correction model can be provided to correct the capacity-related aging state and, if necessary, another data-based correction model can be provided to correct the resistance-change-related aging state. Possible alternatives to the Gaussian process are other supervised learning methods, such as those based on a random forest model, an AdaBoost model, a support vector machine, a transformer model, especially with multi-head attention, or a Bayesian neural network.

[0048] Furthermore, a cost function for an optimization procedure is implemented in the central unit 5.

[0049] A method for operating the energy supply system is shown in the flow chart of the Fig. 2 is described in more detail. The method is preferably carried out entirely with the aid of the remote central unit 5.

[0050] In step S1, it is first checked whether a vehicle battery 41 is connected to the charging station 3. If this is the case (alternative: yes), the method continues with step S2; otherwise, the method returns to step S1.

[0051] In step S2, identification information, a current state of charge SOC and a current state of aging SOH are retrieved from the vehicle battery 41.

[0052] In step S3, the identification information is used to check whether the vehicle battery may be used as a buffer storage device in the electrical power supply network.

[0053] Use as a buffer storage device may be permitted, for example, if a smart contract exists between the owner of the vehicle battery 41 and the supply network operator 22 regarding its use as a buffer storage device. The smart contract may consider the type of use as a buffer storage device, particularly defined as the maximum and minimum energy flow, the amount of energy within a defined period of time, and specified limits for use as a buffer storage device, particularly a price limit for the use of charging current, such as a maximum price per kilowatt hour. The smart contract may be stored using distributed ledger technology. In particular, this allows additional participants to be involved in the energy exchange, especially with regard to: - a high degree of automation, - Transparency, - Real-time capability, - Bi-directional contracts for peer-to-peer energy flows, - to optimise supply and demand of energy flows through a large number of decentralised participants in the energy market as best as possible and to allocate or assign energy flows as best as possible, particularly for buffering purposes.

[0054] If it is determined that the use of the vehicle battery as buffer storage is permissible (alternative: yes), the method continues with step S4; otherwise (alternative: no), the method returns to step S1.

[0055] In step S4, an energy availability in the energy supply network 2 is predicted based on the current time. For this purpose, the energy prediction model is used, which models the future predicted course of energy availability based on the current energy availability, the calendar time, possibly weather data, and domain-specific usage characteristics.

[0056] The model structure can be designed to model a state vector that describes energy availability in the form of an energy demand or a relative energy surplus. Alternatively or additionally, an absolute amount of energy can also comprise the target variables or target values of the model.

[0057] The energy prediction model can be hybrid or data-based and, for example, have a form xt=f(αt)+GP(αt,βt) where α t , β t , are features at time t that contain completely or partially identical features. A structured domain model f(α t ) is used, which is derived from a probabilistic, data-based machine learning model GP(α t , β t ) is corrected. The data-based machine learning model GP(α t , β t) is preferably carried out probabilistically and provides a confidence interval. Here, f(α t ) be implemented as a structured domain model or as a data-based domain model, e.g. in the form of an ARIMA model, which is empirically validated. This architecture has the advantage that in the interpolation case, where there is a sufficient data basis for the Gaussian process (GP), a precise correction can be made with low uncertainty or a narrow confidence interval. In the extrapolation case, however, the GP goes to its prior, which is close to 0, because the GP was trained on the residual of the structured domain model. In addition to the fact that the GP does not correct or only slightly in the extrapolation case because the prior = 0 or ~0, a high uncertainty, i.e. a wide confidence interval, is still signaled because the model is in extrapolation mode and no data was known in the training range.

[0058] The prediction model of energy availability x t may include a first function term f (domain model), which corresponds to a deterministic parameterized function term and which performs periodic or cyclical mappings of energy availability based on a usage pattern, and a data-based function term GP, particularly in the form of a Gaussian process model. It is thus possible to model energy availability in the supply grid depending on cyclical effects, such as time of day, day of the week, and seasonal energy consumption effects.

[0059] By recurrent modeling via a prediction model in the form xt=f(αt,xt−1)+GP(αt,βt,xt−1) a prediction of energy availability in the future can be made.

[0060] The prediction model can be trained based on historical energy availability data. This model can be trained sequentially, first fitting the domain model f and then fitting the data-based model GP to the residual.

[0061] In step S5, the energy prediction model is used to predict energy availability for the time period specified for the vehicle battery 41.

[0062] In Fig. Figure 3 shows an example of an energy demand EB (dashed curve) and a progression of energy availability V over time. The historical energy availabilities and the energy availability derived from the prediction model determined from the historical energy availability data are shown. Furthermore, the progression of the energy availability VP predicted by the energy prediction model is shown.

[0063] In step S6, an optimization is carried out based on a plurality of predetermined charging profiles and discharging profiles, each resulting from a Pareto optimization between charging and discharging time and aging state change, in order to enable cost-optimal use of the vehicle battery 41 as a buffer storage.

[0064] Using an electrochemical battery model, one or more charging profiles and / or one or more discharging profiles can be derived as Pareto-optimized state points (Pareto state points) from average / aggregated (dis)charging time information and aging state change based on electrochemical and / or physical parameters and one or more dimensioning parameters, such as the limit value of the anode overpotential, in the conflict of objectives between aging influence and required average / aggregated charging time information and provided as candidates for charging profiles and discharging profiles. Fig. 4 shows the course of the Pareto curve thus formed with the state points on it for selected charging and discharging profiles.

[0065] The determination of the charging and discharging profiles for the state points can be carried out in a conventional manner, taking into account the anode overpotential resulting from the battery chemistry in the central unit 5 and predefined dimensioning parameters, while specifying average / aggregated charging time information. The Pareto curve is obtained by minimizing a cost function that provides a cost value dependent on the change in the aging state and average / aggregated charging time information.

[0066] Fig. Figure 5 shows an example of a charging profile determined in this way. The charging profile for a vehicle battery 41 specifies a maximum charging current depending on the state of charge. Typical charging profiles have a maximum charging current that decreases with increasing state of charge, as shown, for example, in Fig.5, and each indicate a maximum charging current depending on a state of charge, ie, the stored usable charge of the vehicle battery 41, or on the charging time. The charging profiles can be determined such that the anode overpotential does not fall below a predetermined limit.

[0067] The optimization is performed for a specified optimization period, which corresponds to the time period by which a predetermined state of charge is to be achieved. A cost function is specified for the optimization that rewards the storage of electrical energy in the vehicle battery 41 when energy availability is high and penalizes the storage of electrical energy when energy availability is low or even negative in the energy supply grid. Conversely, discharging the vehicle battery when energy availability is high is penalized, and discharging when energy availability is low in the energy supply grid is rewarded.

[0068] Furthermore, there is an additional cost term that takes into account the change in the aging state during the optimization period. The lower the change in the aging state or degradation during the period, the lower the cost contribution in the cost function. Overall, the optimization period can be divided into one or more time periods, each of which uses one of the previously determined charging profiles or one of the previously determined discharging profiles to absorb or release energy, which was determined based on a model from a previous Pareto optimization regarding charging and discharging durations and changes in the aging state.

[0069] Furthermore, the optimization procedure can take into account the costs due to a change in state when applying the respective charging profiles or discharging profiles, the current market price for a unit of electricity (kWh), a maximum or minimum ampere-hour throughput, and / or a maximum or minimum charging or discharging current as additional constraints in the optimization.

[0070] The optimization method can be designed as a gradient-based method or as a black-box optimization method.

[0071] The optimization results in a usage profile consisting of a temporal sequence of time periods in which one of the charging profiles or discharging profiles is applied. This usage profile can be transmitted in step S7 to the vehicle battery 41 or to the battery management system and / or to the corresponding charging station 3 to which the vehicle battery 41 is connected, and can be taken into account there for the operation of the vehicle battery 41 during the optimization period. This enables cost-optimized use of the vehicle battery 41 as a buffer storage device during a period in which the vehicle battery 41 is connected to a charging station 3.

[0072] It can be provided that the optimization takes place in such a way that a desired or typical state of charge is probabilistically quantified at a defined point in time and is controlled accordingly, for example: - Driver usually leaves parking garage with 90% SOC + / - 5% at 17:30 at 95% confidence - It can be specified that the control is based on the 50% quantile for the target time of 5:30 p.m. However, it can also be specified that the control is based on a defined quantile, e.g., the 5% quantile or the 95% quantile. This allows for a risk-based interpretation and explainability of the probabilistic prediction in the decision-making process for energy allocation, which is carried out automatically.

[0073] It can be provided that in the event of global energy shortages or high volatility on the spot market, the quantile in particular is adaptively adjusted or adjusted as a control parameter in order to achieve local or global grid stability.

[0074] The optimization problem can be solved continuously and, in particular, implemented using a model predictive control.

[0075] In particular, sub-energy supply networks can be defined, for example, in a parking garage, in which the process described above is carried out in a subordinate control system so that available energy quantities can be allocated to vehicles in a distributed manner.

[0076] It may also be provided that a smart contract is concluded between the operator of the sub-energy supply network and the operator of the energy supply network, whereby the allocation of the consumption quantity of electrical energy to individual vehicles is carried out on the basis of individually determined usage profiles. QUOTES CONTAINED IN THE DESCRIPTION

[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited patent literature

[0000] US 20220179009A1

[0019] US 20220334191A1

[0019] US 20220099743A1

[0019] US 2016 / 023,566 [0019, 0045] US 2016 / 023,567 [0019, 0045] US 2020 / 150,185 [0019, 0045] US 9,153,991 B2

[0027] US 2017 / 033 866 A1

[0027] DE 10 2020 206 272 A1

[0027] DE 10 2020 210 132 A1

[0027] DE 10 2019 209 165 A1

[0027] DE 10 2020 212 579 A1

[0027] US 10,447,054 B2

[0027] US 10,312,699 B2

[0027] DE 10 2019 216 015 A1

[0027] Cited non-patent literature

[0000] Doyle, M, Fuller, T F, & Newman, J., „Modeling of galvanostatic charge and discharge of the lithium / polymer / insertion cell“, United States, https: / / doi.org / 10.1149 / 1.2221597

[0027]

Claims

[1] Computer-implemented method for operating a device battery (41) of a technical device (4), in particular at a charging station (3) with an electrical energy supply network (2) with varying energy availabilities, comprising the following steps: - determining (S4) a modeled or predicted temporal course of energy availability in the energy supply network (2) using an energy prediction model; - providing (S6) charging and discharging profiles for the device battery (41); - Providing an aging state model which is designed to determine an aging state change based on temporal profiles of operating variables of the device battery (41); - Providing an indication of a target state of charge and an optimization period after which the target state of charge is to be reached; - performing an optimization method based on a cost function that depends on the predicted energy availability in the energy supply network during the optimization period, the energy consumption and energy output of the device battery (41) during the optimization period and a change in the ageing state at the end of the optimization period, and the indication of the target state of charge in order to obtain a usage profile that indicates a temporal sequence of time periods of the use of charging profiles and / or discharging profiles; - Providing (S7) the usage profile for operating the device battery (41) as a buffer storage in order to provide no transmission of electrical energy, charging according to the respective charging profile or discharging according to the respective discharging profile in each time period according to the usage profile. [2] Method according to claim 1, wherein the energy prediction model is designed or trained to indicate energy availability in the energy supply network (2) depending on a calendar time indication and in particular weather information and / or information about operating times of one or more connected energy sources. [3] Method according to claim 1 or 2, wherein the energy prediction model is data-based or implemented as a hybrid model, wherein the energy prediction model comprises an ARIMA model, a deep neural network or a recurrent neural network or a transformer or a Gaussian process. [4] Method according to one of claims 1 to 3, wherein the charging profiles and / or discharging profiles are selected by the optimization method from a number of predetermined charging profiles and discharging profiles, wherein the predetermined charging profiles and discharging profiles are provided with the aid of an electrochemical battery model of the device batteries (41) predetermined for the similar device batteries (41) by assuming different values for at least one electrochemical dimensioning parameter and in each case a charging profile or a discharging profile is determined which determines a charging current with respect to a state of charge by simulation in such a way that the anode overpotential always corresponds to the value for the at least one electrochemical dimensioning parameter. [5] Method according to one of claims 1 to 4, wherein the change in the aging state is determined as the difference between the aging state at the beginning and at the end of the optimization period. [6] Method according to one of claims 1 to 5, wherein the time periods are determined by value ranges of different predicted energy availability or are of the same duration. [7] Method according to one of claims 1 to 6, wherein the usage profile is determined in a central unit (5) and wherein the usage profile is transmitted to the technical device (4) of the relevant device battery (41) in order to control its use as a buffer memory. [8] Method according to one of claims 1 to 7, wherein the device battery (41) is used as a buffer memory if there is a smart contract for the device battery (41), which smart contract comprises a cost function for the optimization and / or specifies a dependency on a state of the device battery, which is determined in real time, and specifies a boundary condition, in particular by a limit value of an energy price or an efficiency factor, which indicates whether the device battery should be used as an energy buffer. [9] Method according to one of claims 1 to 8, wherein the device battery (41) is operated as a buffer storage device according to the usage profile. [10] Method according to one of claims 1 to 9, wherein the optimization is carried out continuously in order to realize a model-predictive approach. [11] Method according to one of claims 1 to 10, wherein the energy availability is predicted probabilistically, wherein a quantile of a probabilistic prediction of the energy availability is used as an optimization parameter, wherein in particular the quantile is adaptively tracked or adjusted in order to carry out the optimization continuously to realize a model-predictive approach. [12] Apparatus for carrying out one of the methods according to one of claims 1 to 11. [13] Computer program product comprising instructions which, when the program is executed by at least one data processing device, cause the latter to carry out the steps of the method according to one of claims 1 to 11. [14] Machine-readable storage medium comprising instructions which, when executed by at least one data processing device, cause the device to carry out the steps of the method according to one of claims 1 to 11.

Images