Method for operating a virtual power plant

Abstract

The invention relates to a method (18) for operating a virtual power plant (2) which comprises a plurality of electrical consumers (4), at least one of which is formed by a heat pump. A current operating state of each consumer (4) is acquired and a control record for the consumers (4) is created. The consumers (4) are operated in accordance with the control record. The control record is created such that an auxiliary variable, which is created based on the current operating states and on a deviation between a setpoint value (34) for a power requirement and a forecast value (32) that corresponds to a power requirement of the consumers (44) during operation in accordance with the control record, is minimal. The invention further relates to a virtual power plant (2) and to a computer program product (16).

Description

[0001] Description

[0002] Procedure for operating a virtual power plant

[0003] The invention relates to a method for operating a virtual power plant. The virtual power plant has several electrical consumers, at least one of which is a heat pump. The virtual power plant is, in particular, in a power class between 50 kW and 200 kW. Furthermore, the invention relates to a virtual power plant and a computer program product.

[0004] Electricity is increasingly being generated using generators powered by renewable energy sources. Examples of such generators include photovoltaic systems and wind turbines. Due to the dependence on solar radiation and the prevailing wind, the electrical power provided by the generators is not constant over time. Thus, the electrical power supplied to a supply grid to which the generators are connected fluctuates over time, which can also lead to fluctuations in the electrical voltage supplied. In other words, grid stability is impaired. If the grid stability or the electrical voltage supplied by the supply grid fluctuates significantly, it is possible that at least some of the loads powered by them may malfunction.To prevent damage to these consumers, they are shut down, for example. If one of these consumers is part of an industrial plant, this results in production downtime.

[0005] Using theoretical models, it is possible to make a relatively accurate weather forecast for a limited future period, such as one day, and thus also for the solar radiation and prevailing wind. This also makes it possible to estimate the electrical output that will be provided by the power generators powered by these renewable energies over that period. To simplify organization, several such power generators are combined into a so-called virtual power plant. This takes advantage of the fact that individual power generators can complement each other. For example, when there is prevailing cloud cover, individual photovoltaic systems are exposed to different levels of solar radiation due to different geographical locations. This means that the electrical output provided by the individual photovoltaic systems does exhibit comparatively high fluctuations.However, when the photovoltaic systems are combined, these are at least partially balanced out, and a more uniform electrical output is provided by the virtual power plant.

[0006] Typically, (electrical) consumers are also added to the virtual power plant, allowing for flexible use of electrical power at specific times. One such consumer is an electrical energy storage device. This device stores at least a portion of the electrical power provided by the power generators if it exceeds a certain limit. If the electrical power provided by the power generators is then comparatively low, electrical energy is withdrawn from the energy storage device and fed into the grid, further balancing the supply. However, this requires an energy storage device.

[0007] Alternatively, or in combination, an electric vehicle is used as the consumer. This vehicle is charged when a comparatively high electrical output is provided by the power generator. In a further development, when a reduced electrical output is provided by the power generator, electrical energy is at least partially extracted from the electric vehicle and fed into the supply grid, so that the electric vehicle is used like an energy storage device. In a further development, the virtual power plant does not comprise a power generator, but only consumers which, however, have temporal flexibility for their operation. In this case, the consumers are only operated, or at least predominantly, when excessive electrical power is fed into the supply grid.In this way, the virtual power plant stabilizes the supply grid without requiring the reduction of the electrical power provided by the power generators. Furthermore, since the loads are operated as desired, the electrical energy drawn by the virtual power plant is actually used, thus increasing efficiency.

[0008] The requirements for the loads used are that comparatively large amounts of electrical energy can be consumed, possibly even within a comparatively short period of time. Furthermore, any loss of comfort due to the flexible operation should be either minimal or not noticeable to any potential user. For this reason, the only loads typically used are the aforementioned energy storage devices, which are explicitly designed for stabilization, or electric vehicles.

[0009] The invention is based on the object of specifying a particularly suitable method for operating a virtual power plant as well as a particularly suitable virtual power plant and a particularly suitable computer program product, wherein an area of ​​application is advantageously expanded, and wherein compatibility is expediently improved and / or robustness is increased.

[0010] With regard to the method, this object is achieved according to the invention by the features of claim 1, with regard to the virtual power plant by the features of claim 10 and with regard to the computer program product by the features of claim 11. Advantageous further developments and refinements are the subject of the respective subclaims. The method serves to operate a virtual power plant. The virtual power plant has a plurality of electrical consumers, which in particular are also referred to simply as consumers. Preferably, the virtual power plant has a supply connection for connection to a supply network. It is thus possible to feed the virtual power plant by means of the supply network and / or to transfer electrical energy between the supply network and the virtual power plant. In particular, an electrical alternating voltage, which is preferably three-phase, is carried by means of the supply network.In particular, the electrical voltage supplied to the supply network is 230 V or 110 V, and a frequency is in particular 50 Hz or 60 Hz.

[0011] The loads are interconnected in a suitable manner. For example, the loads are located within a specific area, or are conveniently decentralized. If all loads are located within a common area, they are conveniently electrically connected to the supply grid via the common supply connection. However, if the loads are decentralized, each load is assigned a portion of the supply connection. For example, some or all of the loads are assigned to the same building / structural unit, such as a private household or an industrial facility.

[0012] Electrical power is consumed by the consumers during operation. It is preferably possible to operate the consumers flexibly, at least for a limited period of time. The consumers, or at least some of them, are suitable, expediently provided and configured for this purpose. The virtual power plant suitably has between 10 consumers and 200 such consumers, preferably between 50 consumers and 150 consumers, and suitably essentially 100 consumers, with a tolerance of up to 20 consumers, 10 consumers, or 0 consumers, for example. In particular, the electrical power required by the virtual power plant due to operation of the consumers is between 50 kW and 200 kW. Suitably, the required power is greater than 60 kW or 80 kW. Preferably, the maximum required power is less than 150 kW or 120 kW.Conveniently, the required power is essentially equal to 100 kW, with a tolerance of up to 20 kW, 10 kW, or 0 kW being allowed in each case. Conveniently, the number and / or type of consumers is adapted to the corresponding required power or they are selected accordingly.

[0013] For example, one of the consumers is an energy storage device or an electric vehicle, which is electrically connected to other components of the virtual power plant, particularly via a so-called wall box or charging station. At least the electric vehicle is not used for transportation if it is a component of the virtual power plant and is electrically connected, for example, directly or via other components, to any supply connection. If the electric vehicle is used by a user, for example, it can no longer be charged, meaning that it is no longer part of the virtual power plant, at least temporarily.

[0014] At least one of the consumers is a heat pump. The heat pump is equipped, for example, in the manner of an air conditioning system. However, the heat pump is particularly preferably a component of a heating system, which is expediently a component of the virtual power plant. Preferably, several of the consumers are each formed by a heat pump, and, for example, each of the consumers is formed by a heat pump.

[0015] During operation, the heat pump is used to heat up a thermal buffer storage tank of the heating system, in particular a water tank. For this purpose, the heat pump suitably has an electric motor which drives a compressor. The heat pump can therefore be operated relatively flexibly because the thermal buffer storage tank has a certain inertia. In other words, the electrical flexibility during operation of the heat pump is provided by the inertia of the thermal buffer storage tank. In summary, it is particularly possible to operate the heating system for at least a certain period of time with a temperature of the thermal buffer storage tank that is (slightly) above or below a desired operating temperature, without causing any, at least excessive, loss of comfort for a user of the heat pump.

[0016] However, with a heat pump, particularly in comparison to an energy storage system, the electrical power drawn / required depends on the current operating parameters of the heat pump and / or environmental parameters, such as the current temperature of the thermal buffer storage and / or the ambient air temperature. In addition, there are boundary conditions for the operation of the heat pump, such as the thermal buffer storage not being heated above a certain critical temperature. It is also necessary for the thermal buffer storage to have a certain minimum temperature so that at least a slight heating / maintenance of the temperature of the structural unit to which the heat pump is assigned can occur. Thus, the boundary conditions to be observed with a heat pump are more stringent than, for example, with an energy storage system, and safety aspects must also be taken into account.

[0017] For example, the virtual power plant includes only the consumers, so that only electrical energy is supplied through it. Alternatively, the virtual power plant has one or more power generators, which are operated, in particular, using regenerative / renewable energy. For example, one of the power generators is a photovoltaic system or a wind turbine. Suitably, the virtual power plant includes several such power generators, at least some of which are photovoltaic systems and some of which are wind turbines.

[0018] The process records the current operating status of each consumer.

[0019] In other words, it is determined in particular whether the respective consumer is operating. For example, each or some of the operating states are determined only in binary form. In a further development, the currently required electrical power is also determined. For example, to record the current operating state, a query is carried out for each consumer, or the current current supply to the respective consumer is measured and the current operating state is determined based on this. Alternatively, the operating state is determined based on a previous control of the respective consumer. Thus, the operating state specified by a previous control is used as the current operating state.

[0020] Furthermore, a control set is created for the loads, and the loads are operated according to the control set. The control set specifically specifies the operating state each load should assume. For example, a specific power setting is assigned to each load, or the control set specifies how the load should be operated.

[0021] For example, the complete control set is transmitted to all consumers, which reduces the risk of errors and increases redundancy. Alternatively, only a control command corresponding to the control set is transmitted to the consumers. Preferably, a corresponding control command is created for each consumer. In this case, the control command is only transmitted to the consumer to which it corresponds, for example, so that the amount of data to be transmitted is reduced. In a preferred variant, the corresponding control command is only transmitted to those consumers for which a change in operating state is to occur. This further reduces the amount of data to be transmitted, which on the one hand reduces hardware requirements. On the other hand, the method can be implemented in a comparatively robust manner. For example, the control set orThe respective control command is transmitted to the consumers via a cable or, preferably, via radio. This reduces the manufacturing costs of the virtual power plant. The control set is created in such a way that an auxiliary variable is minimal. The auxiliary variable is created based on the current operating conditions and a deviation between a setpoint for a power requirement and a forecast value. The power requirement corresponds in particular to a consumption of electrical power. The setpoint is, for example, variable over time and is adapted in particular to current requirements. However, the setpoint is particularly preferably constant, which reduces effort and the required hardware resources.

[0022] The forecast value corresponds to the power demand of the consumers when operating at the control rate. In other words, the power demand that the consumers exhibit when operating at the control rate is assumed, and this assumption is used as the forecast value. The power demand is, in particular, a theoretical value and is determined, for example, using a theoretical model or a characteristic map. In particular, the forecast value thus corresponds to the assumed actual value for the power demand.

[0023] In summary, the auxiliary variable is functionally related to the current operating states of the consumers and the deviation between the setpoint and the forecast value. Consequently, the control set used to control the consumers fulfills a specific requirement with regard to the current operating states and the deviation. In other words, the control set is created in such a way that a specific condition for the deviation as well as for the current operating states is met, which is specified based on the design of the auxiliary variable. In other words, when the consumers are operated in accordance with the control set, the deviation between the assumed actual value for the power requirement and the setpoint and the current operating states are taken into account, in particular any change in the current operating states.This method makes it possible to add a heat pump to any existing virtual power plant or to create a virtual power plant using one or more heat pumps. This makes it possible to expand existing virtual power plants, thereby increasing their output. It is also possible to create virtual power plants based on previously unused consumers. This broadens the field of application. Since, depending on the configuration of the auxiliary variable, the target power requirement is usually met during operation of the virtual power plant, the compatibility of the virtual power plant is increased, particularly if the target power requirement is selected with a specific level of compatibility in mind.In other words, the virtual power plant's impact on the supply grid is reduced, and the virtual power plant provides system services, particularly voltage maintenance and / or grid frequency maintenance. At least, the virtual power plant expediently stabilizes an electrical voltage supplied by the supply grid, expediently by drawing electrical power corresponding to the setpoint.

[0024] Since the current operating conditions are taken into account when determining the control rate, it is possible to reduce the load on individual consumers, for example a mechanical and / or electrical load. In this case, the auxiliary variable is adjusted in such a way that the load on the heat pump is reduced, preferably on any mechanical components. This reduces the susceptibility to errors of the virtual power plant and increases robustness. This also reduces wear and tear on consumers, for example, and these have a comparatively long service life. Due to the reduced load, a heat pump user is therefore not prevented from making their heat pump available to the virtual power plant. This increases acceptance and provides the opportunity to create a large number of such virtual power plants.

[0025] The virtual power plant expediently has a control unit by means of which the method is at least partially implemented. The control unit is connected, in particular, to the consumers via signaling. For example, the method is only implemented once. However, it is particularly preferred that the method be implemented essentially continuously as long as the setpoint applies. As a result, the operation of the consumers is adjusted so that the condition specified by the auxiliary variable is met with regard to the current operating states and the deviation.

[0026] If the virtual power plant has one or more power generators, these are specifically taken into account, and the target value corresponds in particular to the power demand, i.e., the value of electrical power drawn from the supply grid. For this purpose, the electrical power provided by the power generators is appropriately determined, for example, by measurement.

[0027] For example, the variance or standard deviation is used as the deviation. However, the mean absolute error, i.e., in particular, the absolute value of the difference between the setpoint and the forecast value, is particularly preferred. This simplifies the determination and, in particular, ensures a comprehensible interaction with the consideration of current operating conditions when determining the auxiliary variable. The forecast value is, in particular, an assumption of the power requirement when the consumers are operated according to the control value. The setpoint, on the other hand, is, in particular, specified externally.

[0028] For example, a minimization algorithm is used to determine the control rate. The minimization algorithm can be a heuristic or an exact minimization algorithm. When executing the minimization algorithm, a preliminary control rate is first created and, preferably, the operating states resulting from it are determined. The power requirement resulting from the preliminary control rate is determined, and the deviations are calculated from this. The value of the auxiliary variable is determined based on this and the operating variables. This is repeated several times, with the preliminary control rate being varied until a minimum is found. The corresponding preliminary control rate is used as the new control rate. Alternatively, an MPC algorithm, i.e. a "model predictive control" algorithm, is used.This makes it possible to determine the control rate in a comparatively short period of time.

[0029] For example, the auxiliary variable comprises a case differentiation, wherein the individual cases are specified, in particular, as examples based on the current operating states and / or the deviation. However, the auxiliary variable particularly preferably comprises a weighted sum of the deviation of a variable. The auxiliary variable is expediently formed using the weighted sum. In other words, the weighted sum is thus used as the auxiliary variable. For example, the variable is adjusted such that the consumers are operated in a desired operating state. The variable preferably indicates a required change in the operating state for each consumer. In particular, the variable is increased if a change in the operating state is necessary due to the control rate used.

[0030] In other words, when determining the auxiliary variable, not only the current operating states are used, but also the change required due to the control rate used. In particular, the variable is adjusted in such a way that the number of required changes in the operating states during implementation of the method, i.e. as long as the setpoint applies, is optimal, preferably minimal. In particular, for each change, the variable is increased by a value which, for example, is a constant value or is adjusted to the respective consumer. As a result, the control rate used to operate the consumers is that for which the change and / or number of changes in the operating states is comparatively small. In particular, this results in an essentially constant operation of the respective consumers, so that the load on them is reduced.This also allows users to see that the loads are operating essentially consistently, increasing acceptance. Using the weighted sum reduces the effort involved. It is also possible to adapt the weights to the needs of the users and / or the loads used, for example, significantly reducing the deviation or the number of changes in operating states.

[0031] For example, a certain electrical power to be consumed by the heat pump is specified using the control kit. However, it is particularly preferred if only a binary setting is stored for the heat pump in the control kit. The control kit therefore only has a binary setting for the heat pump. In particular, the setting specifies whether the heat pump is to be operated or not. If the setting is set to one value, the heat pump should operate and otherwise not. In this way, it is possible, on the one hand, to use an existing so-called “Smart Grid Ready” connection or “Smart Grid Ready” interface of the heat pump, which is why there are no additional requirements for the heat pump to be added to the virtual power plant.On the other hand, the operation of a heat pump depends on a comparatively large number of parameters and environmental influences that do not need to be taken into account in this design of the process. It is only necessary to either start the heat pump or stop / pause its operation, i.e., switch off the heat pump. In this way, the heat pump user can still adjust the individual settings of the heat pump themselves, such as hysteresis, heating temperature, etc., so that it is operated in a way that is adapted to the actual installation situation. Furthermore, the heat pump settings remain within the control of the heat pump user / owner. This reduces the required data exchange between the heat pump and, in particular, a control unit used to carry out the process.In this way, the individual operating data of the heat pump also remains within the user's sphere of influence, thus improving data protection and increasing data security. Furthermore, when implementing the process, it is not necessary to take any safety regulations and / or restrictions into account, and the heat pump itself ensures that these are met. Thus, no certification or similar is required for the operation of the virtual power plant. For example, the forecast value is created using an artificial intelligence algorithm and / or by extrapolating a current actual value for the power demand. For this purpose, the actual value is first determined, for which purpose the actual power demand required by the consumers is expediently measured. Alternatively, the complete (aggregated) power demand is used.However, the forecast value is particularly preferably created based on consumption profiles, with each consumer being assigned one of the consumption profiles. Using the respective consumption profile, a required electrical power is plotted over time, i.e., in particular, the temporal progression of the required / used electrical power. If the consumer is an electric vehicle, for example, the consumption profile corresponds, for example, to a power limit or any value between the power limit and a consumer demand of 0 W. With an electric vehicle, the electrical power used for charging is essentially freely selectable, at least as long as the electric vehicle is not yet fully charged.Preferably, the consumption profile takes into account whether the electric vehicle is available for the entire period for which the target value applies, and / or whether a trip is pending is taken into account as a boundary condition, and thus whether the electric vehicle should be charged at a specific time. In particular, the consumption profile corresponds to the current operating state of the consumer, i.e., in particular, whether the electric vehicle is being charged. The consumption profile for the heat pump, for example, is determined theoretically or empirically.

[0032] Alternatively or in combination with this, machine learning, in particular an "artificial intelligence" algorithm, is used to determine the consumption profiles. The algorithm is used to determine in particular whether the electric vehicle is actually present, or how likely it is that it will be disconnected by a user and thus no longer forms part of the virtual power plant while the target value applies. Particularly preferably, the actual value of the power demand is determined, expediently recorded, and compared with the forecast value. The consumption profiles are then adjusted depending on the comparison. For example, this is carried out only for one or more of the consumers, preferably the heat pump. Preferably, this is carried out for all consumers. For example, the actual value of the power demand of the entire virtual power plant is determined.However, it is particularly preferred to determine the actual value for each consumer and compare it with the forecast value, namely preferably a part thereof, suitably the respective consumption profile used. For example, the respective actual value is used as the new consumption profile. Alternatively, the actual value is determined several times, expediently for a specific period of time. For example, the mean of these actual values ​​is used as the new consumption profile, whereby, for example, a standard deviation (A / variance) is also taken into account. This is expediently done if the consumer corresponds to an electric vehicle. Machine learning is particularly preferably used to adapt the consumption profiles. For example, an "artificial intelligence" algorithm (Kl) is used. For example, a neural network is used for this purpose.However, Gaussian Process Regression is particularly preferred for fitting.

[0033] The adjusted consumption profiles are expediently used to recreate the forecast value. This occurs, for example, when the process is carried out again, for example by first stopping and then restarting it. Preferably, however, the forecast value is continuously recalculated during the process, at least as long as the target value applies, so that in particular the control rate is also recreated. Due to the now adjusted consumption profiles, the deviation is thus particularly reduced. In particular, the consumption profiles are adjusted in certain intervals, with the length of such a step being between 1 second and 20 minutes, preferably between 30 seconds and 5 minutes. Suitably, the consumption profiles are recreated every minute, and thus also the forecast value. Consequently, a new control rate is also created then, in particular.Thus, there is a comparatively precise match between the actual value and the target value, with comparatively little effort.

[0034] For example, the setpoint is specified for the entire time the virtual power plant is operating. However, it is particularly preferred that the setpoint is used only for a predefined time window. In other words, the setpoint applies only to the time window. In particular, predefined time periods are provided, with only one such time window per time period, during which operation occurs according to the setpoint. The virtual power plant expediently also operates outside of the time window, but no setpoint is specified. In particular, the method is terminated after the end of the time period and preferably restarted for the next time period.

[0035] In particular, the time window is at most half the length of the period. Preferably, 24 hours, i.e. one day, is used as the predefined period. Thus, in particular, a setpoint is specified once for each day. The time window is in particular between 15 minutes and 5 hours, between 30 minutes and 2 hours and, for example, essentially equal to 1 hour, with a deviation of 25%, 10%, 5% or 0%, for example. During the time window, the virtual power plant serves in particular to stabilize the supply network, whereas otherwise the operation of the consumers is not restricted by the setpoint. The time window is, for example, divided into several time periods that are spaced apart from one another. Preferably, however, the time window is contiguous, which makes it easier to determine the control rate.

[0036] This makes it possible to operate the consumers outside of the time window according to the wishes of the respective user, thus increasing their comfort. Because the system is operated using the setpoint, comfort is only slightly reduced during the time window. Because the time window is comparatively short, there is at most only a slight reduction in comfort, especially with heat pumps, since any thermal buffer storage is heated slightly more or less than desired during the time window, for example. Subsequently, however, operation takes place again, for example according to a heating curve. However, using the setpoint during the time window improves the stability of the supply network.

[0037] For example, the target value is selected essentially arbitrarily or according to certain specifications. However, it is particularly preferred that a maximum value and a minimum value are determined for the power requirement during the predefined time window. The minimum value and / or the maximum value are expediently variable or, for example, constant. The target value is then selected between these. This ensures that the target value can actually be achieved. The maximum value and the minimum value apply in particular to the entire virtual power plant. In particular, the maximum value and the minimum value are determined beforehand, in particular with a time delay from the predetermined time window. It is particularly preferred that after these have been determined, this is communicated, for example to a grid operator, in particular in the form of an offer.The operator or another person then expediently specifies and selects the target value. The method is suitably part of a business model, with the minimum and maximum values ​​being specified by an operator / owner of the virtual power plant, in particular to the grid operator. The grid operator then selects the target value between the minimum and maximum values. The virtual power plant is then operated for the specific time window using the target value. In return for consuming the electrical power corresponding to the target value during the time window, the operator of the virtual power plant expediently receives a compensation payment from the grid operator.

[0038] For example, any value can be used as the minimum value. Preferably, 0 W is used as the minimum value, or each consumer is assigned a minimum demand, and the minimum value corresponds to the sum of all minimum demands. The maximum value is, for example, determined theoretically or always corresponds to the same value. This is preferably smaller than the sum of all power demands of the consumers at maximum power. This ensures that the maximum value can be realized using the virtual power plant.

[0039] Particularly preferably, however, the maximum value corresponds to operation of the consumers according to a second control set. In other words, it is predicted / assumed how great the power requirement of the consumers would be if they were operated with the second control set. In other words, the maximum value is therefore an assumed value. In particular, any consumption profiles are used to determine the maximum value. The second control set is created in such a way that a second auxiliary variable is minimal. For example, the second auxiliary variable is created in the same way as the auxiliary variable. Particularly preferably, the second auxiliary variable is created based on predicted operating states, i.e. assumed operating states, and based on a deviation between the maximum value and the predicted value. Furthermore, the second auxiliary variable is also created based on the maximum value itself.Preferably, the second auxiliary variable is thus created in the same way as the auxiliary variable, but additionally takes the maximum value into account. For example, the same predicted operating states are always used to determine the maximum value, or these can also change over the predefined period. Preferably, the second auxiliary variable comprises a weighted sum of the deviation between the maximum value and a second variable that indicates a required change in the predicted operating state for each consumer, and the reciprocal of the maximum value or the negated maximum value. The second auxiliary variable is expediently formed using this sum.In summary, the second control set is created such that the maximum value is as large as possible, while the deviation between the maximum value and the forecast value is as small as possible, and the number of changes to the operating states required for this purpose for the time window is as small as possible. If the virtual power plant has one or more power generators, a forecast is preferably created for the electrical power provided by the power generators for the time window. The forecast is taken into account in particular when determining the maximum value, and this is reduced by this amount compared to using only the (assumed) power demand of the consumers.

[0040] For example, the predefined time window is always the same or different for each period. Particularly preferably, the maximum and minimum values ​​are determined for each period for several such time windows. These values ​​can differ, in particular depending on the respective forecast operating states. Alternatively, or in combination, these values ​​differ because a different number of consumers is present, in particular if at least some of them are electric vehicles. For example, the period also includes times that are not assigned to any of the time windows. Particularly preferably, however, the period is divided among the time windows, which preferably have the same length.

[0041] A time window is selected from the time windows. The setpoint then applies to this time window. For example, the time window is selected depending on the selected setpoint, or the setpoint is selected for the selected time window. Once the time window has begun, the control rate is generated using the auxiliary variable, and the consumers are operated accordingly. In particular, the potential business model is adapted in such a way that the operator of the virtual power plant offers the minimum and maximum values ​​for each of the time windows, with the grid operator selecting one of the time windows and specifying the applicable setpoint, which lies between the minimum and maximum values ​​applicable to the time window.During operation, the operator of the virtual power plant receives compensation or a price is reduced for the electrical power consumed during the time window. The virtual power plant has a number of electrical consumers. These are, for example, located locally or distributed. In the assembled state, the consumers are electrically connected to a supply network, for example, via a common supply connection or each consumer is individually connected. One of the consumers is formed by a heat pump. For example, all of the consumers are heat pumps, or at least several of the consumers are heat pumps. For example, one of the consumers is an electric vehicle, an air conditioner, a refrigerator or a server system. For example, the virtual power plant has one or more electricity generators, at least some of which are powered by renewable energy.Alternatively, the virtual power plant, for example, is without a power generator and thus formed solely by the consumers. At the very least, the virtual power plant has a power demand during operation. The virtual power plant is operated according to a method in which the current operating state of each consumer is recorded. A control rate for the consumers is created, and the consumers are operated according to the control rate. The control rate is created in such a way that an auxiliary variable is minimal. This auxiliary variable is created based on the current operating states and a deviation between a target value for a power demand and a forecast value that corresponds to a power demand of the consumers when operating according to the control rate.

[0042] The virtual power plant has, in particular, a control unit that is provided and configured to carry out the method. The control unit comprises, for example, an application-specific integrated circuit (ASIC) or, particularly preferably, a computer that is suitably designed to be programmable. In particular, the control unit comprises a storage medium on which a computer program product, also referred to as a computer program, is stored. Upon execution of this computer program product, i.e. the program, the computer is prompted to carry out the method. The control unit is expediently connected to all consumers and / or the supply connection via signaling, preferably via a corresponding connection of the control unit. For example, the control unit is connected directly to each individual consumer separately. Alternatively, a bus system is formed, in particular.For example, the signaling connection is established via a cable or, suitably, at least partially via a radio connection that complies, in particular, with a Bluetooth, mobile radio, or WLAN standard. In particular, the control signal, or at least part of it, is transmitted to the respective consumers via the signaling connection.

[0043] The invention further relates to such a control unit. The control unit is provided and configured to carry out a method for operating a virtual power plant having a plurality of electrical consumers, at least one of which is formed by a heat pump. In the method, a current operating state of each consumer is recorded, and a control set for the consumers is created. The consumers are then operated according to the control set. The control set is created in such a way that an auxiliary variable is minimal, which is created based on the current operating states and a deviation between a target value for a power requirement and a forecast value. The forecast value corresponds to a power requirement of the consumers when operating according to the control set.

[0044] The control unit has, for example, an application-specific integrated circuit (AS IC) and / or a microprocessor, by means of which the method is at least partially carried out. In particular, the control unit comprises a computer program product that is stored in a memory and, when the program is executed by a computer, such as the microprocessor, causes the computer to carry out the method. Preferably, the control unit, in the assembled state, is a component of one of the consumers, preferably the heat pump, or of a higher-level control system of the virtual power plant. For example, the higher-level control system is formed by the control unit. Thus, additional functions are also performed by the control unit.

[0045] The computer program product comprises a number of instructions which, when the program (computer program product) is executed by a computer, cause the computer to carry out a method for operating a virtual power plant having a plurality of electrical consumers, at least one of which is formed by a heat pump. In the method, a current operating state of each consumer is recorded, and a control set for the consumers is created. The consumers are then operated according to the control set. The control set is created in such a way that an auxiliary variable is minimal, which is created based on the current operating states and a deviation between a target value for a power requirement and a forecast value. The forecast value corresponds to a power requirement of the consumers when operated according to the control set.The computer is expediently a component of a control unit and is formed, for example, by means of the control unit. The computer preferably comprises a microprocessor or is formed by means of the microprocessor. The computer program product is, for example, a file or a data carrier containing an executable program that, when installed on a computer, automatically executes the method.

[0046] The invention further relates to a storage medium on which the computer program product is stored. Such a storage medium is, for example, a CD-ROM, a DVD, or a Blu-ray disc. Alternatively, the storage medium is a USB stick or other storage device that is, for example, rewritable or only writable once. Such a storage device is, for example, a flash memory, a RAM, or a ROM.

[0047] The further developments and advantages explained in connection with the method are also to be transferred mutatis mutandis to the virtual power plant / the control unit / the computer program product / the storage medium as well as to each other and vice versa.

[0048] An embodiment of the invention is explained in more detail below with reference to a drawing. In the drawings:

[0049] Fig. 1 schematically shows a virtual power plant,

[0050] Fig. 2 shows a method for operating the virtual power plant, Fig. 3 shows a period of time comprising several time windows,

[0051] Fig. 4 shows several consumption profiles and the resulting time course of a maximum value for one of the time windows, and

[0052] Fig. 5 shows several consumption profiles and the resulting temporal course of a forecast value as well as a target value and an actual value for the time window.

[0053] Corresponding parts are provided with the same reference numerals in all figures.

[0054] Figure 1 shows a simplified schematic of a virtual power plant 2 which has a plurality of electrical consumers 4, which are also simply referred to as consumers 4. The virtual power plant 2 is electrically connected to a supply grid via a supply connection 6. A three-phase alternating electrical voltage is carried via the supply grid. In the example shown, the virtual power plant 2 does not comprise a power generator and is therefore generator-free. Each of the consumers 4, if it is in operation, has a power requirement. This power is covered from a supply grid. For this purpose, the consumers 4 are suitably electrically contacted with one another and with the supply connection 6. Thus, during operation, the virtual power plant 2 only draws electrical power from the supply grid, namely when the consumers 4 are operated accordingly.

[0055] Some of the consumers 4 are each formed by an electric vehicle, which is charged during the operation of the virtual power plant 2. When the respective electric vehicle is used, i.e., driven, by a user, it is removed from the network of the virtual power plant 2 and thus no longer forms part of the virtual power plant 2. Thus, some of the consumers 4 are only temporarily assigned to the virtual power plant 2.

[0056] Several of the consumers 4 are each formed by a heat pump. Each heat pump is a component of a heating system, which also has a thermal buffer storage. The heat pumps are assigned to different structural units and are operated in different ways, depending on the settings of the respective user. It is therefore possible that so-called heating curves and / or domestic water temperatures differ between the individual heat pumps / heating systems, so that the consumers 4 are operated differently. It is also possible that the individual heat pumps have different hardware.

[0057] However, the heat pumps each have an "SG Ready" ("Smart Grid Ready") interface and, like the other consumers 4, are signal-connected to a control unit 8. For this purpose, the control unit 8 has a communication device 10 capable of wireless communication. The communication device 10 complies with a WLAN, mobile radio, and / or Bluetooth standard. The virtual power plant 2 is operated by means of the control unit 8, and the control unit 8 is suitable, intended, and configured for this purpose.

[0058] The control unit 8 has a computer 12 in the form of a programmable microprocessor and a storage medium in the form of a memory 14. A computer program product 16 is stored on the memory 14. The computer program product 16 comprises a plurality of instructions which, when executed by the computer 12, cause the computer 12 to execute a method 18, shown in Figure 2, for operating the virtual power plant 2. In other words, the virtual power plant 2 is operated according to the method 18, and the control unit 8 is provided and configured to carry out the method 18.

[0059] In a first work step 20, a minimum value 26 and a maximum value 28 are determined for each of three predefined time windows 22 of a predefined period 24 shown in Figure 3. The length of the predefined period 24, which is also simply referred to as a period, is 24 hours, and the method 18 is carried out at the beginning of the period 24, but before one of the time windows 20 has begun. The length of each predefined time window 22, which is also simply referred to as a time window 22, is 1 hour. In the example shown, the position of the time windows 22 in the period 24 is always the same, i.e., also when the method 18 is subsequently carried out. The minimum values ​​26 and maximum values ​​28 are constant during the respective time window 22, but can differ between the individual time windows 22. In the example shown, the minimum value 26 for the first two time windows 22 is 0 W.In the subsequent, i.e. last, time window 22 of the period 24, the minimum value 26 is increased and corresponds to the sum of the power requirements when all consumers 4 are operated at minimum power.

[0060] The maximum value 28 corresponds to operation of the consumers 4 during the respective time window 22 according to a second control set, wherein the maximum value 28 is constant and in particular corresponds to a second target value. To determine the maximum value 28, the power requirement of the individual consumers 4 when operated according to the second control set is first determined. For this purpose, a consumption profile 30 assigned to the respective consumer 4 is used, three of which are shown as examples in Figure 4. The consumption profiles 30 are stored in the memory 14 and correspond to an assumed power requirement of the respective consumers 4. A specific operating state is initially assumed for each consumer 4, which results from the second control set used. The respective consumption profile 30 is derived from the operating states predicted in this way.

[0061] The consumption profile 30 for each electric vehicle is constant and corresponds to the maximum power consumption when the respective electric vehicle is being charged. In other words, the consumption profile 30 of the electric vehicle corresponds to operation at full charging power.

[0062] The consumption profiles 30 of the heat pumps, on the other hand, are modified and have different forms that depend on the respective setting, such as a desired domestic water temperature. This is not specified by the second control set, but rather it only specifies in binary whether the respective heat pump is operated or not. In other words, only one binary setting is stored for each heat pump in the second control set. However, it is possible to flexibly select the times at which a change in the respective operating state occurs, i.e. whether the heat pump is operated or not, within the time window 22. The consumption profiles 30 are not always different from 0 W during the entire time window 22 in order to avoid overheating of the associated thermal buffer storage.

[0063] The sum of the consumption profiles 30 corresponds to a forecast value 32. Since the consumption profiles 30 vary over time, the forecast value 32 also changes during the time window 22. The deviation, namely the mean absolute error, between the forecast value 32 and the (assumed) maximum value 28 for the time window 22 is created. A value is also determined that corresponds to the number of changes in the operating states of the consumers 4 during the time window 22. The deviation, the value, and the negation of the maximum value 28, i.e. the maximum value 28 multiplied by "-1" ("minus one"), are summed with different weights to form a second auxiliary value. The second control set is changed until the second auxiliary value is at a minimum. For this purpose, a minimization algorithm is used, for example, in particular a "model predictive control" (MPC) algorithm, whereby a "mixed integer linear problem solver" is suitably used.

[0064] Since the forecast operating states of the consumers 4 can differ for the different time windows 22, the maximum values ​​28 of the time windows 22 are also different: For example, it is possible that the number of electric vehicles 4 present has changed in one of the time windows 22, or that, for example, one of the heaters to which one of the heat pumps is assigned previously had a comparatively high demand for heating power, so that now only a comparatively low power consumption is possible without overheating occurring.

[0065] In summary, the maximum value of 28 corresponds to an operation of

[0066] Consumer 4 according to the second control set, whereby the second control set is created in such a way that the second auxiliary variable is minimal. The second auxiliary variable is in turn created based on the forecast operating states, based on the deviation between the maximum value 28 and the forecast value 32 and based on the maximum value 28 itself. The forecast value 32 corresponds to the sum of the power requirements of consumers 4 when operating according to the second control set, and the maximum value 28 is constant for the time window 22. For the predefined period 24, the maximum value 28 and the minimum value 26 are determined for each of the three time windows 22. In a variant not shown in detail, the entire period 24 is divided into the time windows 22. Thus, there are a total of 24 such time windows 22, and the minimum value 26 and the maximum value 28 are determined for each.

[0067] In a second step 34, the temporal position of each time window 22, as well as the minimum values ​​26 and maximum values ​​28 applicable to each of these, are transmitted from the operator of the virtual power plant 2 to the operator of the supply grid. One of the predefined time windows 22 is selected from the operator. Furthermore, a target value 36, shown in Figure 5, is selected for the time window 22. The target value 36 is located between the minimum value 26 and the maximum value 28 of the selected time window 22. The target value 36 is selected only for one of the time windows 22 of the period 24.

[0068] As long as the selected time window 22 has not yet begun, the consumers 44 are only operated according to the specified settings / organizations of the respective user. The power requirement of the virtual power plant 2 in the west is thus predetermined based on the needs of the users and can exhibit comparatively high fluctuations. As soon as the selected time window 22 has begun, a third work step 38 is carried out. In this step, the current operating state of each consumer 4 is first recorded. For the consumers 4 configured as heat pumps, it is checked whether they are currently operating or not. For the consumers 4 configured as electric vehicles, a current power consumption is recorded. In the third work step 38, a control set for the consumers 4 is also created.For this purpose, the forecast value 32 is first determined, which corresponds to the power requirement of the consumers when operated according to the control rate. However, here only a binary setting is stored for the heat pumps in the control rate. The forecast value 32 is determined for the entire time window 22, whereby the operating state of each consumer 4 is changed during the time window 22 depending on the control rate. The deviation between the forecast value 32 and the setpoint 34, which is constant for the time window 22, is determined, with the mean absolute error also being used as the deviation. The forecast value 32 is created here, just as in the first step 22, based on the consumption profiles 30, with one assigned to each consumer 4. The current operating state of the respective consumer 4 is taken into account for each of the consumption profiles 30.

[0069] Furthermore, the number of changes in the operating state required to achieve this forecast value 32 is determined, and the sum of these changes is used as a variable. This variable and the deviation are weighted and summed to form an auxiliary variable. In other words, the auxiliary variable is formed using the weighted sum of the deviation between the target value 34 and the forecast value 32, as well as the variable that characterizes the required change in the operating states for each consumer 4.

[0070] The control rate is varied until the auxiliary variable is at a minimum. Here, too, a model predictive control (MPC) algorithm is used, suitably using a mixed-integer linear problem solver. In summary, the control rate is determined in the same way as the second control rate, but the negation of the maximum value 28 is not used, and the setpoint value 34 is used instead of the maximum value 28. Also, instead of the forecast operating states, the current operating states of the consumers 4 are used. In the example shown in Figure 5, one of the consumption profiles 30 is divided into two disjoint time periods because the setpoint value 34 is lower than the maximum value 28. The operating states are also changed, for example, compared to the forecast operating states.

[0071] The control rate created in this way is used to operate the loads 4. For this purpose, the control rate is transmitted to them so that the loads 4 are operated according to the control rate. Only the portion of the control rate applicable to each load 4 is transmitted, thus reducing the amount of data transmitted.

[0072] In a subsequent fourth work step 40, an actual value 42 of the power demand is measured and compared with the forecast value 32. For each of the consumers 4, the actual power demand, i.e., the respective actual value, is measured and compared with the respective consumption profiles 30, based on the sum of which the forecast value 32 is determined. In the example shown, a comparatively large discrepancy exists.

[0073] In a subsequent fifth step 44, the consumption profiles 30 are adjusted based on the comparison. For the consumers configured as electric vehicles, the actual power consumption is compared with the predicted power consumption, i.e., the used consumption profile 30. If these differ by more than a tolerance value, the mean of the (actual / realized) power consumption is used as the new consumption profile 30 for the electric vehicle, with a standard deviation also being taken into account. In a further development, machine learning is used for the adjustment. For the consumers 4 configured as heat pumps, machine learning is also used to adjust the respective consumption profile 30, namely a Gaussian process regression.

[0074] As soon as the consumption profiles 30 have been adjusted, which expediently occurs every minute, the third work step 38 is performed again, now using the adjusted consumption profiles 30. In other words, the control rate is now created based on the adjusted consumption profiles 30. As a result, the deviation between the forecast value 32 and the actual value 42 is reduced, and this essentially corresponds to the target value 34. Following this, the fourth work step 40 and the fifth work step 44 are performed again, and the consumption profiles 30 are adjusted again if necessary. The third, fourth, and fifth work steps 38, 40, 44 are thus repeated several times during the time window 22.

[0075] Due to the use of the control set, the behavior of consumers 4 during time window 22 does not fully match the wishes of the respective users. However, the power demand of virtual power plant 2 during time window 22, namely the actual value 42, is comparatively constant, which is why the load on the supply grid is reduced, or why virtual power plant 2 contributes to stabilizing supply grid 2. When time window 22 ends, a sixth work step 46 is performed, and consumers 4 are again operated according to the user specifications. After the end of time period 24, method 18 is terminated and immediately restarted for the subsequent time period 24.

[0076] The invention is not limited to the exemplary embodiment described above. Rather, other variants of the invention can also be derived therefrom by those skilled in the art without departing from the scope of the invention. In particular, all individual features described in connection with the exemplary embodiment can also be combined with one another in other ways without departing from the scope of the invention.

[0077] List of reference symbols

[0078] 2 virtual power plant

[0079] 4 consumers

[0080] 6 Supply connection

[0081] 8 Control unit

[0082] 10 Communication device

[0083] 12 computers

[0084] 14 storage

[0085] 16 Computer program product

[0086] 18 procedures

[0087] 20 first step

[0088] 22 time slots

[0089] 24 period

[0090] 26 Minimum value

[0091] 28 Maximum value

[0092] 30 Consumption profile

[0093] 32 Forecast value

[0094] 34 second step

[0095] 36 Setpoint

[0096] 38 third step

[0097] 40 fourth step

[0098] 42 Actual value

[0099] 44 fifth step

[0100] 46 sixth step

Claims

Claims 1. A method (18) for operating a virtual power plant (2) having a plurality of electrical consumers (4), at least one of which is formed by a heat pump, in which - a current operating status of each consumer (4) is recorded, - a control rate for the consumers (4) is created, and - the consumers (4) are operated according to the control rate, wherein the control rate is created in such a way that an auxiliary variable is minimal, which is created on the basis of the current operating conditions and a deviation between a setpoint value (34) for a power requirement and a forecast value (32) which corresponds to a power requirement of the consumers (44) when operating according to the control rate.

2. Method (18) according to claim 1, characterized in that the auxiliary variable comprises a weighted sum of the deviation and a variable which characterizes a required change in the operating state for each consumer (4).

3. Method (18) according to claim 1 or 2, characterized in that only a binary setting is stored for the heat pump in the control set.

4. Method (18) according to one of claims 1 to 3, characterized in that the forecast value (32) is created on the basis of consumption profiles (30), wherein each consumer (4) is assigned one of the consumption profiles (30).

5. Method (18) according to claim 4, characterized in that the forecast value (32) is compared with an actual value of the power requirement, wherein the consumption profiles (30) are adapted depending on the comparison.

6. Method (18) according to one of claims 1 to 5, characterized in that the setpoint value (36) is used only for one predefined time window (22) per predefined period (24).

7. Method (18) according to claim 6, characterized in that a maximum value (28) and a minimum value (26) for the power requirement during the predefined time window (22) are determined, the target value (36) being selected between these.

8. Method (18) according to claim 7, characterized in that the maximum value (28) corresponds to an operation of the consumers (4) according to a second control set, wherein the second control set is created in such a way that a second auxiliary variable is minimal, which is created on the basis of predicted operating states and on the basis of a deviation between the maximum value (28) and the forecast value (32) and on the basis of the maximum value (28) itself.

9. Method (18) according to claim 7 or 8, characterized in that for the predefined period (28) for several such time windows (22) the maximum value (28) and the minimum value (26) are determined, and that one of the time windows (22) is selected.

10. Virtual power plant (2) which has a plurality of electrical consumers (4), at least one of which is formed by a heat pump, and which is operated according to a method (18) according to one of claims 1 to 9.

11. A computer program product (16) comprising instructions which, when executed by a computer (12), cause the computer (12) to execute a method (18) according to any one of claims 1 to 9.

Images