EP4205050A1

EP4205050A1 - Method for controlling energy exchange between a set of assets and an energy network, computer program, computer system and virtual power plant

Info

Publication number: EP4205050A1
Application number: EP20761814.1A
Authority: EP
Inventors: Juan Bernabe MORENO; Martin CLEVEN; Giorgio Cortiana; Corey O'MEARA
Original assignee: EOn Digital Technology GmbH
Current assignee: EOn Digital Technology GmbH
Priority date: 2020-08-25
Filing date: 2020-08-25
Publication date: 2023-07-05
Also published as: WO2022042828A1

Abstract

The invention relates to a method for controlling energy exchange between a set of assets (16) and an energy network (19), a computer program, a computer system (10) comprising at least one classical computer (11), a quantum computer (12) and an asset controller (13) and a virtual power plant (1).

Description

Method for controlling energy exchange between a set of assets and an energy network, computer program, computer system and virtual power plant

D e s c r i p t i o n

The invention relates to a method for controlling energy exchange between a set of assets and an energy network, a computer program, a computer system and a virtual power plant.

A virtual power plant is commonly understood as a network of decentralized assets such as power generating units, flexible power consumers and/or storage systems. The power generation and power consumption of the assets are controlled by a central computer system of the virtual power plant. Nonetheless the assets may remain independent in their operation and ownership. One primary target of a virtual power plant is to relieve the load on the grid by smartly distributing the power generated by the individual assets during periods of peak load. To achieve the above target and/or other targets, prior to importing power into the electrical grid from the assets or exporting power from the electrical grid to the assets during the peak load, the computer system of the virtual power plant has to predict its capability of power to be exchanged with the electrical grid during the period of peak load and optimize it for the purpose of achieving the set target. For this purpose, the computer system has to take several constraints of each one of the assets into account and to determine for each asset a quantity of power and time to be imported to the electrical grid or exported from the electrical grid to which they are connected. This optimization step is performed prior to the actual period of peak load, which may also be referred to as the delivery period, in which the power is imported into the electrical grid or exported therefrom into the assets. The power to be imported or exported in the delivery period is then contracted for that period. In other words, the virtual power plant promises the operator of the electrical grid to deliver the power to be imported or exported in the delivery period according to the optimization performed by the computer system.

The objective of the invention is to provide an improved method for controlling energy exchange between a set of assets and an energy network, a corresponding computer program, computer system and virtual power plant. In particular, the objective of the invention is to utilize the assets to optimally achieve at least one set target.

This object is solved by the subject-matter of the claims. In particular, the object is solved by a method for controlling energy exchange between a set of assets and an energy network according to claim 1 , a computer program according to claim 27, a computer system according to claim 28 and a virtual power plant according to claim 31. Further details of the invention unfold from the other claims as well as the description and the drawings. Thereby, the features and details described in connection with the method of the invention apply in connection with the computer program of the invention, the computer system according to the invention and the virtual power plant according to the invention, so that regarding the disclosure of the individual aspects of the invention it is or can be referred to one another.

According to a first aspect of the invention, the objective is solved by a method for controlling energy exchange between a set of assets and an energy network, whereby the method comprises the steps of: (a) predicting an energy exchange flexibility of each asset of the set of assets with the energy network for a specified time period, and

(b) solving a first optimization function, whereby

(i) the first optimization function is configured such that its solution determines for each asset of the set of assets an energy exchange schedule for exchanging energy between the respective asset and the energy network in the specified time period to achieve at least one set target, and

(ii) the first optimization function contains at least the predicted energy exchange flexibility of each asset of the set of assets with the energy network as a constraint,

(c) contracting the energy exchange as contracted energy exchange according to the determined energy exchange schedules of the assets,

(d) controlling each asset of the set of assets to provide the energy network with the contracted energy exchange according to the energy exchange schedules of each asset of the set of assets in the specified time period,

(e) solving a second optimization function in the specified time period, whereby the second optimization function is configured to minimize a difference between the contracted energy exchange according to the energy exchange schedule and an actually provided energy exchange at the energy network by optimizing the energy exchange schedules of assets of the set of assets, whereby the second optimization function is solved by a quantum computer, and

(f) steering of the controlling of the set of assets to provide the energy network with energy exchange according to the optimized energy exchange schedules optimized by the solution of the second optimization function solved by the quantum computer in the specified time period.

Accordingly, the method provides for utilizing the assets to achieve a set target by optimizing a second optimization function for enabling optimal steering of the controlling of the assets to fulfil the contracted energy exchange schedule as contracted in order to achieve the least one set target or, in other words, at least one predetermined target optimized by the solution of the first optimization function. This may be optimally performed by the method of the invention in particular due to solving the second optimization function by means of the quantum computer. That is because the steering of the controlling of each asset is performed in real-time in the specified time period, i.e. when the energy network is provided with the contracted energy exchange according to the energy exchange schedule and the difference between the contracted energy exchange according to the energy exchange schedules and the actually provided energy exchange at the energy network occurs. Therefore, the second optimization function must be solved within that specified time period and in real-time to deliver an optimal result. A solution to the second optimization function however should not only be found quickly, that is in as little calculation time as possible, but also the found solution should ideally be optimal, such that the steering according to the solution can fulfil the contracted energy exchange in the specified time period.

Compared to known algorithms, such as genetic algorithms, executed on a classical computer, it has been found that, despite the errors a quantum computer can produce when solving the particular function, the solution delivered by the quantum computer for the specific second optimization function enables optimized steering of the controlling of the assets, such that the contracted energy exchange is fulfilled to a larger or even full degree compared to other solving methods, such as the genetic algorithm executed on the classical computer.

In one specific embodiment of this invention, however, the method according to the first aspect of the invention may be provided with the difference, that the second optimization function is solved by a classical computer and/or a quantum computer. Accordingly, the classical computer alone or in combination may be used in order to solve the second optimization function.

The method according to the first aspect of the invention and in particular its steps are further described in the following, whereby some features thereof are further explained by non-limiting examples or definitions and some features thereof are further specified by additional optional features.

The prediction of the energy exchange flexibility of each asset for the specified time period is performed in particular before or in advance of the specified time period. The prediction may be based on information gathered from previous data based on the control of energy exchange or in other words, based on experience, and/or information gathered from the set of assets or each of the assets individually by means of a communication of the set of assets or each of the assets with an asset controller, for example. Accordingly, the method according to the first aspect of the invention may comprise a further step prior to the predicting of the energy exchange flexibility, wherein the further step is gathering data from the assets for predicting of the energy exchange flexibility relevant for the specified time period. The target or multiple targets may as well be set in particular before or in advance of the specified time period.

The energy may be in particular in the form of electrical energy or power. Accordingly, the energy network may be referred to as an electrical energy network or electrical grid. Also, the energy exchange may be in particular referred to as power flow.

The specified time period may be a time period in which the energy network typically experiences peak load during the day or due to special events, for example. However, the specified time period may alternatively be any other time period, such as a time period in which the energy network experiences low or medium load.

Also, the first optimization function in particular is solved before or in advance of the specified time period, such that the energy exchange can be provided according to the energy exchange schedules of the assets determined by the solution of the first optimization function. The first optimization function is formulated such that it optimizes the energy exchange schedules of the assets given at least the constraint of the predicted energy exchange flexibility to achieve the at least one set target, in particular optimize, i.e. minimize or maximize, at least one target value of the at least one target. The energy exchange schedule of each asset contains the information of the planned energy exchange, that is a quantity of energy, for the respective asset of the set of assets with the energy network in the specified time period. The energy exchange schedules can specify the energy exchange for each asset of the set of assets in specified time blocks within the specified time period. Whether there are specified time blocks within the specified time period or not may depend on the duration of the specified time period, which may depend upon the at least one set target, in particular conditions of the at least one target. When the energy exchange according to the energy exchange schedules is bidirectional in the specified time period, i.e. energy can be exported to the energy network from the assets and energy can be imported from the energy network to the assets, the energy exchange schedules may contain the information of the quantity of energy to be imported in specified time blocks or the specified time period and/or to be exported in the specified time period in specified time blocks or the specified time period. Accordingly, for example, each asset of the set of assets according to the energy exchange schedule can import energy within a first specified time block and export energy within a second specified time block after the first specified time block and in the specified time period.

The contracting step in the method is a step in which the energy exchange according to the energy exchange schedules determined by the solution of the first optimization function is contracted to a contractor, which may be the operator of the energy network, a supplier of energy or a consumer of energy connected to the energy network, for example. In other words, the contracting step is a binding plan of energy exchange, which is being contracted according to the energy exchange schedule. Thus, this step is performed in particular before or in advance of the controlling step in the specified time period. The energy exchange being contracted according to the determined energy exchange schedules of the assets does not necessarily mean that the specific energy exchange schedules of the assets become part of the contract. The overall energy exchange according to the energy exchange schedules are contracted and it is usually not important to the contractor which specific assets do supply that energy exchange according to the determined energy exchange schedules. Further, the contracted energy exchange does not necessarily equal the available energy exchange according to the energy exchange schedules. Rather, the contracted energy exchange may be different from, in particular less than, the energy exchange available according to the determined energy exchange schedule. This is because in the contracting process not necessarily all the energy exchange available according to the energy exchange schedule gets contracted. However, normally, no more energy exchange can be contracted than what is available according to the determined energy exchange schedule.

The controlling of each asset of the set of assets to provide the energy network with the contracted energy exchange occurs in the specified time period and therefore after the prediction of the energy exchange flexibility, the determination of the energy exchange schedules for each of the assets and the contracting of the energy exchange. Here, each of the assets is controlled according to the determined energy exchange schedules such that the contracted energy exchange is provided. The second optimization function delivers a solution of optimized energy exchange schedules of assets of the set of assets. Here, at least one, some or all energy exchange schedules of the assets of the set of assets previously determined by the first optimization function may be optimized, whereby one, some or all energy exchange schedules are amended such that the difference between the contracted energy exchange and the actually provided energy exchange is minimized. The actually provided energy exchange may be determined by the step of controlling each of the asset of the set of assets. When, for example, the predicted energy exchange flexibility of an asset is not available at the step of controlling that asset due to it not being connected to the energy network or not being able to provide the per control requested energy exchange, it is determined that there is an absence of energy exchange, which will need to be compensated for in order to provide the contracted energy exchange. Thereby, the difference between the contracted energy exchange for the specified time period and the actually provided energy exchange in the specified time period may occur due to false predictions. The prediction cannot be accurate, especially in the case where the assets are not being owned and operated by the operator of the virtual power plant and the owners of the assets cannot be bound to the predicted energy exchange flexibility in the specified time period or shall be provided with flexibility in terms of their energy exchange flexibility in the specified time period.

The steering of the controlling the set of assets is consecutively based on the energy exchange schedules determined by the first optimization function and on the energy exchange schedules that have been optimized by the solution of the second optimization function solved by the quantum computer in the specified time period. Obviously, in a case where all energy exchange schedules are optimized by the solution of the second optimization function, there is no control of the assets based on the energy exchange schedules determined by the first optimization function. However, usually the prediction of the energy exchange flexibility should be rather accurate and there should be no need of optimizing all energy exchange schedules of the assets by means of the solution of the second optimization function. Rather, in the step of steering of the controlling of each asset only some of the energy exchange schedules should need to be optimized to minimize the difference between the contracted energy exchange and the actually provided energy exchange. However, locating the assets of which the energy exchange schedules are to be optimized among all assets and determining how these are to be optimized is a very complex problem to be solved, which, according to the first aspect of the invention, is being solved by solving the second optimization function and therefore by the quantum computer.

Accordingly, the second optimization function may be formulated as a binary combinatorical problem. This formulation of the second optimization function may be solved very effectively by the quantum computer. In particular, the second optimization function may be formulated as a Hamiltonian formulation. Specifically, the Hamiltonian formulation may be used in QAOA (Quantum Approximate Optimization Algorithm), Variational Quantum Eigensolver algorithm (VQE) or Quadratic Unconstrained Binary Optimization problem (QUBO)/lsing formulation and/or any generalization of these algorithms. Accordingly, the solving of the second optimization function may be performed by means of a quantum computer using QAOA, i.e. a back and forth between quantum computing and classical computing, and does not necessarily be solved entirely but only partially by means of the quantum computer itself.

Further, the second optimization function may be quantum computer technology agnostic. Thereby, different types of quantum computers may be used in order to solve the second optimization function. In particular, when the second optimization function is solved by a quantum computer embedded in a cloud infrastructure this may be very advantageous, because the second optimization function may be solved on any of the quantum computers in the cloud infrastructure and does need to queue less for availability of a specific quantum computer of that cloud infrastructure. This effectively reduces the time that the solution of the second optimization function requires.

Moreover, the first optimization function may be solved by a quantum computer. The same or a different quantum computer from the one used to solve the second optimization function may be used. This is particularly important when the quantum computer to be used is embedded in the mentioned cloud infrastructure. It has been found that the solving of the first optimization function benefits from the fast and optimized solution found by the quantum computer as well despite the possible errors a quantum computer may generate.

Accordingly, the first optimization function may be formulated as a binary combinatorical problem. Thereby, the same advantages as previously explained with respect to the second optimization function being formulated as a binary combinatorical problem may be achieved. In particular, the first optimization function may be formulated as a Hamiltonian formulation. Specifically, the Hamiltonian formulation may be used in QAOA (Quantum Approximate Optimization Algorithm), Variational Quantum Eigensolver algorithm (VQE) or Quadratic Unconstrained Binary Optimization problem (QUBO)/lsing formulation and/or any generalization of these algorithms. Accordingly, the solving of the first optimization function may be performed by means of a quantum computer using QAOA, i.e. a back and forth between quantum computing and classical computing, and does not necessarily be solved entirely but only partially by means of the quantum computer itself.

Also, accordingly, the first optimization function may be quantum computer technology agnostic. Thereby, the same advantages as previously explained with respect to the second optimization function being quantum computer technology agnostic may be achieved.

The method may further include the steps of solving after a predetermined time interval within the specified time period again the second optimization function in the specified time period, whereby the second optimization function is configured to minimize a difference between the contracted energy exchange according to the energy exchange schedules and an actually provided energy exchange at the energy network by optimizing the optimized energy exchange schedules optimized by the solution of the previously solved second optimization function, whereby the second optimization function is again solved by the quantum computer, and steering of the controlling of the set of assets to provide the energy network with the contracted energy exchange according to the optimized energy exchange schedules optimized by the solution of the again solved second optimization function solved by the quantum computer in the specified time period.

Accordingly, the second optimization function is being solved again with updated data of the optimized energy exchange schedule optimized by the second optimization function and the actually provided energy exchange according to the controlling of the assets after the predetermined time interval. This process relating to these two steps may be run iteratively for several predetermined time intervals within the specified time period with as much and short or long time intervals as necessary in order to minimize the difference between the contracted energy exchange and the actually provided energy exchanged such that the contract can be fulfilled and possible penalties due to not importing or exporting contracted energy from or into the energy network are avoided. The predetermined intervals may be within every 3 to 60 minutes, in particular within every 5 to 30 minutes, such as every 15 minutes, for example.

The controlling and steering of the controlling of each asset of the set of assets may be performed by an asset controller. The asset controller may be a program executed on a classical computer. The classical computer may be used for this purpose in particular due to its wide availability, low cost and better performance for handling the operation of controlling the assets.

Herein, a classical computer is understood as a computer which is not a quantum computer. A classical computer solely uses classical mechanics, such as dedicated circuit diagram for flowing electrical voltage, gates etc., whereas a quantum computer requires quantum mechanical features to perform its operations on data. The quantum computer uses qubits as its main information unit.

Further, the second optimization function may be submitted to being solved by the quantum computer by a classical computer. Accordingly, the second optimization function, the contracted energy exchange, the energy exchange schedules determined by the first optimization function and the actually provided energy exchange according to the controlling of the assets may be submitted to the quantum computer as data by the classical computer. The classical computer is used for this purpose in particular due to its availability and better performance for handling this sort of operation. Especially in a cloud infrastructure rather cost- expensive time of usage of the quantum computer is thereby reduced to a minimum and the data is gathered and submitted with high reliability to the quantum computer in form of, or in other words, together with the second optimization function to be solved. Accordingly, the method according to the first aspect of the invention may than be referred to as a classical- quantum hybrid method as both, a classical computer and a quantum computer, are used in a hybrid application.

Therein, the second optimization function is submitted to being solved by the quantum computer in predetermined time intervals within the specified time period. These may be the time intervals according to which the steps of again solving the second optimization function and the steering of the controlling of each asset are iteratively performed.

Therein, the energy exchange schedules optimized by the solution of the second optimization function may be forwarded to a classical computer. This may be the same classical computer or a classical computer different from the one by which the second optimization function to being solved by the quantum computer was submitted, which again becomes relevant with respect to usage of the cloud infrastructure. That classical computer may forward the solution of the second optimization function, that is the optimized energy exchange schedules, to the asset controller. The asset controller may be executed on the same classical computer or a different classical computer.

The second optimization function may be configured such that it optimizes the energy exchange schedules of a subset of the set of assets. In other words, the quantum computer may solve the second optimization function to optimize the energy exchange schedule for a subset of the set of assets. Accordingly, the second optimization function includes only the data of the subset of the set of assets for which the energy exchange schedule is to be optimized. The subset contains only some assets of the set of assets. The subset may be preselected and submitted accordingly with the second optimization function to the quantum computer. The solving of the second optimization function by the quantum computer for the selected subset of the set of assets can significantly reduce calculation time and thereby provide a quick and good solution for a selected subset of the set of assets.

Further, the quantum computer may be configured such that it optimizes the energy exchange schedules of the set of assets by consecutively solving the second optimization function for subsets of the set of assets. Consequently, the entire set of assets may be split into the subsets of which the energy exchange schedules are optimized consecutively until the energy exchange schedules of the set of assets are optimized according to current data, in particular the actually provided energy exchange.

The energy exchange provided by the set of assets in the specified time period may be bidirectional. Accordingly, energy may be exported to the energy network from at least some of the assets of the set of assets in a specified time block within the specified time period and energy may be imported from the energy network to at least some assets of the set of assets in another specified time block within the specified time period.

The energy exchange schedules may contain, for different time blocks within the specified time period, a quantity of energy to be exported to the energy network from the respective assets and/or to be imported to the respective assets from the energy network.

The predicting of the energy exchange flexibility considers an energy exchange buffer for minimizing the difference between the contracted energy exchange according to the energy exchange schedules and the actually provided energy exchange at the energy network. The energy exchange buffer acts as a safety buffer, which may be used in the second optimization function to fulfil the contracted energy exchange despite false predictions of the energy exchange flexibility. The energy exchange buffer may, for example, be provided by assuming less energy exchange flexibility of some or all assets in the set of assets or by not considering some assets of the set of assets for the prediction of the energy exchange flexibility, whereby the predicted energy exchange flexibility is in any case less than what could actually be predicted based on the given data.

The prediction of the energy exchange flexibility of each asset of the set of assets with the energy network for the specified time period and/or the contracting of the determined energy exchange schedules of the assets of the set of assets may be performed by at least one classical computer. The at least one classical computer may be one classical computer and thus for both steps, i.e. predicting and contracting, the same. The at least one classical computer may be the same classical computer as previously mentioned or a different classical computer or different classical computers, which again becomes relevant to a cloud infrastructure. The prediction may be performed not only by using classical computing on the classical computer but also using data analysis of the data mentioned before gathered for the prediction step and/or machine learning models, in particular a mixture of both may be used.

The set target may be based on at least one target value related to the energy exchange between the set of assets and the energy network in the specified time period. The solution of the first optimization function is then to optimize the at least one target value, that is minimize or maximize, for example, based at least on the predicted energy exchange flexibility as a constraint. The target value may thus be included, in particular together with other variables and/or the at least one constraint, in the first optimization function or submitted along with it for solving the first optimization function.

Therein, the at least one target value related to the energy exchange may be one of a quantity of carbon emissions related to the generation of the exchanged energy, a portion of a renewable energy origin of the exchanged energy and a cost of the exchanged energy. The aforementioned are to be understood as non-limiting examples of target values to be optimized by solving the first optimization function. Accordingly, in the case where the target value is a quantity of carbon emissions related to the generation of the exchanged energy, the first optimization function may be configured such that its solution minimizes the quantity of carbon emissions of a predetermined exchanged energy contained in the set of assets or the energy network after the specified time period, for example. Further, in the case where the target value is a portion of a renewable energy origin of the exchanged energy, the first optimization function may be configured such that its solution maximizes the portion of renewable energy origin of a predetermined energy contained in the set of assets or the energy network after the specified time period, for example. And, in the case where the target value is a cost of the exchanged energy, the first optimization function may be configured to minimize the cost of a predetermined energy contained in the set of assets after the specified time period, for example. Alternatively, the first optimization function may be configured such that its solution maximizes the cost of the energy imported from the set of assets into the energy network in the specified time period and/or minimizes the cost of energy exported from the energy network into the set of assets, for example. These costs may be based on a price or prices determined by an energy market. Any type and number of energy markets may be used for determining the price such as day-ahead markets, intraday markets and frequency restoration, where essentially the energy exchange flexibility is marketed.

Further, the assets in the set of assets may be electric vehicles. The electric vehicles may be plug-in electric vehicles and/or battery electric vehicles in particular. The energy storage systems, in particular the batteries, in the electric vehicles can be utilized to provide the energy exchange flexibility with the energy network due to their electrical connection with the energy network when being coupled to an electrical charger, in particular when being charged or discharged. The electric vehicles have particularly good availability in the night time, when the electric vehicles are being coupled to an electrical charger connected to the energy network and possibly charged for usage through the following day. Accordingly, the specified time period may in particular be in the night time. However, the electric vehicles are also charged during the day and at that time can provide a high energy exchange flexibility for a long period, for example when the electric vehicle is a company vehicle and is coupled to an electrical charger at the company’s site during the work time of an employee driving that electric vehicle. Alternatively, or additionally, the assets in the set of assets may be households and/or power plants, for example.

The electric vehicles in the set of assets may be decentralized. Decentralized means that it is not necessary for the electric vehicles as assets to be at one location but they may be located at different locations. Of course, if the electric vehicles are centralized, for example when they are all charged at one site, for example at a company site and are part of a company fleet, then the prediction of the energy exchange flexibility is rather accurate compared to when the assets are decentralized. However, with decentralized assets in the set of assets a comparatively large energy exchange flexibility may be provided because typically centralized electric vehicles are rare, meaning that according electric vehicle fleets are typically small.

At least an arrival state of charge and a departure state of charge of each of the electric vehicles may be determined for the predicting of the energy exchange flexibility. For this purpose, information may be gathered directly from the electric vehicles or owners of the electric vehicles based on direct feedback via the navigation system of the electric vehicles, via apps installed on the owner’s smartphones or via previous experience based on the energy exchange availability of the electric vehicles, for example. The arrival state of charge determines the energy available to the electric vehicle at the start of the predetermined period. The departure state of charge of each of the electric vehicle determines the energy that must be contained in the electric vehicle upon its departure. In this sense, the owner of the electric vehicle provides their electric vehicle with an arrival state of charge and demands that the electric vehicle is charged to the departure state of charge at their departure. In between the arrival time and departure time, the predicted energy exchange flexibility may be used by the method according to the first aspect of the invention to achieve the set target. For the arrival state of charge, the arrival state of charge and the arrival time may be forecasted. For the departure state of charge, the departure state of charge and departure time may be set by the owner of the electric vehicle, e.g. via the app, or per contract terms.

Additionally, or alternatively, at least a predetermined maximum state of charge and a predetermined minimum state of charge of each of the electric vehicles may be provided for the prediction of the energy exchange flexibility. The energy exchange flexibility is restricted accordingly to the maximum state of charge, for example because the owner of the electric vehicle does not want an entire charge of his battery, and a minimum state of charge, for example because the owner of the electric vehicle wants the possibility to depart prior to his planned departure.

Further, additionally, or alternatively, at least a predetermined battery capacity of each of the electric vehicles is provided for the prediction of the energy exchange flexibility. This information is provided because the energy exchange flexibility may obviously be based only on the actually available battery capacity of the electric vehicles.

Also, the first optimization function may contain a minimum charging power and/or a maximum charging power for each electric vehicle as further constraints. These constraints may be for technical reasons, because the electric vehicles may not be charged with less or more power, or may be due to other technically feasible restrictions or charger specifications, such as provided by the owner of the vehicle for example, to mitigate battery aging caused by very slow or very fast charging, for example.

Moreover, the first optimization function may contain the site load at the set of assets as a further constraint. The site load may be restricting the energy exchange that can actually be provided at the energy network, e.g. a maximum power. A prediction of the site load may be performed in order to include it as a further constraint. For example, when the assets are electric vehicles, the load provided by the electric vehicle charging must at all times be within the energy network capacity constraints of the given site.

According to a second aspect of the invention, the objective is solved by a computer program configured to be executed on a computer system to carry out the method according to the first aspect of the invention. According to a third aspect of the invention, the objective is solved by a computer system comprising at least one classical computer, a quantum computer and an asset controller, whereby the computer system is configured to carry out the method according to the first aspect of the invention.

Therein, the asset controller may be a program to be executed on the at least one classical computer or another classical computer. The at least one classical computer may store the first optimization function and the second optimization function, and receive the corresponding data for solving these, such as the energy exchange flexibility and the at least one constraint. The classical computer may then transmit these to the quantum computer for solving these functions. Of course, the at least one classical computer may also be configured to predict the energy exchange flexibility and to contract the energy exchange according to the determined energy exchange schedules.

At least one of the at least one classical computer, the quantum computer and the asset controller may be embedded in a cloud infrastructure. In particular, the entire computer system may be embedded in the cloud infrastructure. Alternatively, the classical computer, quantum computer and asset controller may be embedded in an on-premise architecture.

Further, the asset controller may be configured for communication with each asset of the set of assets via a communication protocol for predicting the energy exchange flexibility of each asset and/or for controlling each asset of the set of assets. Any technically feasible communication protocol may be used. In particular, communication over the internet may be facilitated.

According to a fourth aspect of the invention, the objective is solved by means of a virtual power plant comprising the system according to the third aspect of the invention and the set of assets, wherein the assets of the set of assets are configured for communication with the asset controller of the computer system.

Further advantages, features and details of the invention unfold from the following description, in which by reference to drawings FIGs. 1 to 3 embodiments of the present invention are described in detail. Thereby, the features from the claims as well as the features mentioned in the description can be essential for the invention as taken alone or in an arbitrary combination. In the drawings, there is schematically shown:

FIG. 1 a schematic representation of a virtual power plant according to an embodiment of the invention;

FIG. 2 a schematic representation of a method for controlling energy exchange between a set of assets and an energy network according to an embodiment of the invention in the virtual power plant of FIG. 1 ; and

FIG. 3 a further schematic representation of the method of FIG. 2.

FIG. 1 depicts a computer system 10 embedded in a cloud infrastructure 14 and used in a virtual power plant 1 according to an embodiment of the invention.

Therein, the computer system 10 comprises a classical computer 11 , which is not a quantum computer, a quantum computer 12 and an asset controller 13. The asset controller 13 may be a separate unit, in particular a further classical computer, equipped with a software for controlling a set of assets 16.1 , 16.2, 16.3, 16.4, 16.5 or it may be alternatively executed on the classical computer 11 .

In any case, the asset controller 13 is connected by means of a communication network 15 over the internet with each asset 16.1 , 16.2, 16.3, 16.4, 16.5 of the set of assets 16.1 , 16.2, 16.3, 16.4, 16.5. In FIG. 1 , five assets 16.1 , 16.2, 16.3, 16.4, 16.5 (in the following referenced as 16) are depicted. However, the number of assets 16 may be less or more, in particular typically much more, for example at least 100 or at least 1.000 assets 16 may be used in the virtual power plant 1. Accordingly, the herein exemplarily described virtual power plant 1 and therewith associated method are to be understood as very particular and exemplary embodiments for better comprehension of the invention but not in any way for limiting the invention. These assets 16 may be decentralized and operated and owned by different entities. Each of the assets 16 of this particular embodiment has an energy storage unit 17.1 , 17.2, 17.3, 17.4, 17.5 (in the following referenced as 17). In this particular embodiment, all of the multiple assets 16 of the virtual power plant 1 are of the same type. Further, in this particular embodiment, the assets 16 are electric vehicles 16, such as battery electric vehicles or plug-in electric vehicles, and their energy storage units 17 are traction batteries. However, the virtual power plant 1 is not limited to having same types of assets 16 and may comprise further assets 16 being of a different type, such as households with energy storage units or power plants, for example.

Each of the energy storage units 17 of the assets 16 is electrically connected or connectable to an electric charging station 18.1 , 18.2, 18.3, 18.4, 18.5, whereby the electric charging stations 18.1 , 18.2, 18.3, 18.4, 18.5 are connected via an energy network 19 with each other. The energy network 19 distributes the energy inside of the energy network 19. An electric mast 20 is exemplarily shown connected to the energy network 19 for illustrating this in an illustrative way.

The energy network 19, which may be referred to as an electrical grid as well, is obviously larger than what is shown as a small extract thereof in FIG. 1 and comprises energy generators and energy consumers attached thereto (not shown). The goal of the virtual power plant 1 is to achieve a set target or predetermined target. The set target may be to charge the energy storage units 17 of the assets 16 to individual and predetermined levels of charge within a predetermined time period, for example over the night time, and having as little carbon emissions as possible associated with the energy stored inside of the energy storage units 17 after the predetermined time period. The set target is thereby based on the target value of carbon emissions associated with the energy inside of the energy storage units 17, which are to be minimized. This is only one example of a set target and according target value. The set target may be a different one or further comprise other targets or target values, such as, for example, providing the energy stored inside of the energy storage units 17 at the lowest cost while ensuring as little carbon emissions as possible associated with the stored energy or a mix thereof, for example.

Generally, an energy exchange between the assets 16, i.e. their energy storage units 17 (in the following the assets 16 will be referred to as that the energy storage units 17 are part of the assets 16), may be bidirectional, meaning that the assets 16 can export energy into the energy network 19 from their respective energy storage units 17 and import energy from the energy network 19 by storing it in the energy storage units 17. Of course, the ability to import and export energy depends on constraints such as the individual level of charge of the assets 16 and further constraints given by the owners of the assets 16, such as a maximum level of charge and a minimum level of charge, but also a minimum charging power and a maximum charging power and further the predetermined levels of charge and grid capacity constraints on site.

According to the method of the invention, which will be explained with reference to FIGS. 2 and 3 in the following, there is a first optimization function 31 formulated which is at least temporarily stored in the classical computer 11 . This first optimization function 31 is formulated so as to determine an energy exchange schedule 34 (note that there are individual energy exchange schedules 34 for all assets 16, for the sake of simplicity however this is illustrated in FIG. 3 as one schematic box referred to as 34) for each asset 16 of the set of assets 16 for exchanging energy between the respective asset 16 and the energy network 19 in the specified time period to achieve the above-set target. However, the first optimization function 31 may be formulated to achieve any desired target, such as the ones explained in this description.

FIGs. 2 and 3 depict representations of the method of the invention. Herein, the computer system 10 is provided with certain sets of data 30.1 , 30.2, 30.3, 30.4. These sets of data 30.1 , 30.2, 30.3, 30.4 may, in the current example according to this embodiment, relate to the assets 16, to the energy network 19, to the set target, to information relating to a contracting process as further described below and to other data, such as holidays, weather and so on, which may be relevant with respect to the energy network 19 and availability of energy, for example. These sets of data 30.1 , 30.2, 30.3, 30.4, as far as they are important for solving the first optimization function 31 , are included therein or, in other words, submitted together with it.

Upon receipt of the sets of data 30.1 , 30.2, 30.3, 30.4, the computer system 10 executes the method for controlling energy exchange between the set of assets 16 and the energy network 19 according to the steps 40 to 47 described in detail in the following with reference to FIGs. 2 and 3. In the first step 40, an energy exchange flexibility of each asset 16 of the set of assets 16 with the energy network 19 for a specified time period is determined. The specified time period in this particular embodiment is the night time, during which the assets 16 in form of the electric vehicles 16 are typically not moved and need to be recharged for the next day.

The owners of the electric vehicles 16 according to the set target in this particular embodiment would like to use their electric vehicles 16 with as little carbon emissions as possible. Therefore, the set target is to recharge the electric vehicles 16 to a predetermined state of charge at departure in the morning. This predetermined state of charge is collected prior to determining the energy exchange flexibility and transmitted to the classical computer 11 as one of the data 30.1 , 30.2, 30.3, 30.4.

As exemplarily shown in FIG. 3, one set of data 30.1 comes from outside of the computer system 10 and one set of data 30.2 comes from the asset controller 13. The asset controller 13 is a software executed on the classical computer 11 in this particular embodiment. The energy exchange flexibility therefore in this particular embodiment is determined by the amount of energy that has to be imported into the electric vehicles 16 based on a state of charge at the beginning of the night time, at which the electric vehicles 16 are plugged into their respective charging stations 18, and by the end of the predetermined period.

In the second step 41 , the above-mentioned first optimization function 31 is solved. The first optimization function 31 is configured such that its solution determines for each asset 16 of the set of assets 16 its individual energy exchange schedule 34 for exchanging energy between the respective asset 16 and the energy network 19 in the night time to achieve the set target transmitted as one of the sets of data 30.1 , 30.2, 30.3, 30.4. In this second step 41 , the first optimization function 31 is transmitted from the classical computer 11 to the quantum computer 12 as a quadratic unconstrained binary optimization problem. The first optimization function 31 contains at least the predicted energy exchange flexibility of each asset 16 of the set of assets 16 with the energy network 19 as one of the sets of data 30.1 , 30.2, 30.3, 30.4 and a constraint. Based on the determined energy exchange schedules 34 of the assets, in the third step 42, an energy exchange is contracted with a supplier of energy via the energy network 19 or the operator of the energy network 19. This third step 42 may be executed by the classical computer 11 as shown in FIG. 3. According to the contracted energy exchange 35, the virtual power plant 1 guarantees to the supplier of the energy or the operator of the energy network 19 to import or export the quantity of the contracted energy exchange 35 from the energy network 19 in the therein specified time period, i.e. the night time in this particular embodiment.

Accordingly, in the thereon following fourth step 43, the asset controller 13 is controlling each asset 16 of the set of assets 16 to provide the energy network 19 with the contracted energy exchange 35, in this particular embodiment to import a corresponding quantity of energy, according to the energy exchange schedules 34 of each asset 16 of the set of assets 16 in the specified time period.

However, although the predicted energy exchange schedules 34 are rather accurate, it may be that there is a difference between the contracted energy exchange 35 contracted according to the energy exchange schedules 34 and an actually provided energy exchange 36 at the energy network 19. This may in particular be because the electric vehicles 16 are not owned by the operator of the virtual power plant 1 but by other entities. For example, there may be a change of plans of the owner of the electric vehicle 16.5 and they do not plug in their electric vehicle 16.5 because they actually do not arrive at the respective charging station 18.5 within the specified time period or for some reason forget to do so.

Now, to keep the energy network 19 stable and to avoid penalties of the operator of the virtual power plant 1 because the contracted energy exchange 35 cannot be actually provided as per the priorly determined energy exchange schedules 34, i.e. the quantity of contracted energy exchange 35 cannot be imported or exported, the energy exchange schedules 35 of assets 16 of the set of assets 16 are optimized by solving a second optimization function 32 in a fifth step 44.

The second optimization function 32 is configured to minimize the difference between the contracted energy exchange 35 and the actually provided energy exchange 36 at the energy network 19 by optimizing the energy exchange schedules 34 of one, some or all assets 16 of the set of assets 16, whereby the energy exchange schedule 34 of one, some or all assets 16 is changed as determined by the second optimization function 32.

For this purpose, an energy exchange buffer is considered, which was taken into account when predicting the energy exchange flexibility. For example, the energy exchange buffer may have been a sixth electric vehicle 36.6, which had originally not been considered for being recharged in the specified time period. The second optimization function 32 is also submitted from the classical computer 11 to the quantum computer 12 as a problem and consequently solved in the fifth step 44.

The solution of the second optimization function 32, i.e. the optimized energy exchange schedules 37 of the assets 16, are then sent from the quantum computer 12 via the classical computer 11 to the asset controller 13, which accordingly in a sixth step 45 steers the ongoing control of the assets 16 to provide the energy network 19 with the contracted energy exchange

34 according to the optimized energy exchange schedules 37 optimized by the solving of the second optimization function 32 solved by the quantum computer 12 in the specified time period.

Then, after predetermined time intervals, e.g. 15 minutes, within the specified time period, the steps 44 and 45 are essentially repeated as seventh step 46 and eighth step 47 with current data at that time to enable steering of the control in the case that the prediction yet again deviates from reality such that there is a difference between the contracted energy exchange

35 and actually delivered energy exchange 36 at that new time, so that the contracted energy exchange 35 may nonetheless be fulfilled.

Referen ce s i g n l i st

1 virtual power plant

10 computer system

11 classical computer

12 quantum computer

13 asset controller

14 cloud infrastructure

15 communication network

16 asset, electric vehicle

17 energy storage unit

18 charging station

19 energy network

20 electric mast

30 set of data

31 first optimization function

32 second optimization function

34 energy exchange schedules

35 contracted energy exchange

36 actually provided energy exchange

37 optimized energy exchange schedules

40 first step

41 second step

42 third step

43 fourth step

44 fifth step

45 sixth step

46 seventh step

47 eighth step

Claims

C l a i m s Method for controlling energy exchange between a set of assets (16) and an energy network (19), whereby the method comprises the steps of:

(a) predicting an energy exchange flexibility of each asset (16) of the set of assets (16) with the energy network (19) for a specified time period, and

(b) solving a first optimization function (31), whereby

(i) the first optimization function (31) is configured such that its solution determines for each asset (16) of the set of assets (16) an energy exchange schedule (34) for exchanging energy between the respective asset (16) and the energy network (19) in the specified time period to achieve at least one set target, and

(ii) the first optimization function (31) contains at least the predicted energy exchange flexibility of each asset (16) of the set of assets (16) with the energy network (19) as a constraint,

(c) contracting the energy exchange as contracted energy exchange (35) according to the determined energy exchange schedules (34) of the assets (16),

(d) controlling each asset (16) of the set of assets (16) to provide the energy network (19) with the contracted energy exchange (35) according to the energy exchange schedules (34) of each asset (16) of the set of assets (16) in the specified time period,

(e) solving a second optimization function (32) in the specified time period, whereby the second optimization function (32) is configured to minimize a difference between the contracted energy exchange (35) according to the energy exchange schedules (34) and an actually provided energy exchange (36) at the energy network (19) by optimizing the energy exchange schedules (34) of assets (16) of the set of assets (16), whereby the second optimization function (32) is solved by a quantum computer (12), and

(f) steering of the controlling of the set of assets (16) to provide the energy network (19) with the contracted energy exchange (35) according to the optimized energy exchange schedules (37) optimized by the solution of the second optimization function (32) solved by the quantum computer (12) in the specified time period. Method according to claim 1 , whereby the second optimization function (32) is formulated as a binary combinatorical problem.

- 24 -

3. Method according to claim 1 or 2, whereby the second optimization function (32) is quantum computer technology agnostic.

4. Method according to any of the previous claims, whereby the first optimization function

(31) is solved by a quantum computer (12).

5. Method according to claim 4, whereby the first optimization function (31) is formulated as a binary combinatorical problem.

6. Method according to claim 4 or 5, whereby the first optimization function (31) is quantum computer technology agnostic.

7. Method according to any of the previous claims, whereby the method further includes the steps of: solving after a predetermined time interval within the specified time period again the second optimization function (32) in the specified time period, whereby the second optimization function (32) is configured to minimize a difference between the contracted energy exchange (35) according to the energy exchange schedules (34) and an actually provided energy exchange (36) at the energy network (19) by optimizing the optimized energy exchange schedules (37) optimized by the solution of the previously solved second optimization function (32), whereby the second optimization function (32) is again solved by the quantum computer (12), and steering of the controlling of the set of assets (16) to provide the energy network (19) with the contracted energy exchange (35) according to the optimized energy exchange schedules (37) optimized by the solution of the again solved second optimization function

(32) solved by the quantum computer (12) in the specified time period.

8. Method according to any of the previous claims, whereby the controlling and steering of the controlling of each asset (16) of the set of assets (16) are performed by an asset controller (13), whereby the asset controller (13) is a program executed on a classical computer (11).

9. Method according to any of the previous claims, whereby the second optimization function (32) is submitted to being solved by the quantum computer (12) by a classical computer (11).

10. Method according to claim 9, whereby the second optimization function (32) is submitted to being solved by the quantum computer (12) in predetermined time intervals within the specified time period.

11. Method according to claim 9 or 10, whereby the optimized energy exchange schedules (37) optimized by the solution of the second optimization function (32) are forwarded to a classical computer (11).

12. Method according to any of the previous claims, whereby the second optimization function (32) is configured such that it optimizes the energy exchange schedules (34) of a subset of the set of assets (16).

13. Method according to claim 12, whereby the quantum computer (12) optimizes the energy exchange schedules (34) of the set of assets (16) by consecutively solving the second optimization function (32) for subsets of the set of assets (16).

14. Method according to any of the previous claims, whereby the energy exchange provided by the set of assets (16) in the specified time period is bidirectional.

15. Method according to any of the previous claims, whereby the energy exchange schedules (34) contain a quantity of energy for different times or time frames within the specified time period to be exported to the energy network (19) from the respective assets (16) and/or to be imported to the respective assets (16) from the energy network (19).

16. Method according to any of the previous claims, whereby the predicting of the energy exchange flexibility considers an energy exchange buffer for minimizing the difference between the contracted energy exchange (35) according to the energy exchange schedules (34) and an actually provided energy exchange (36) at the energy network (19).

17. Method according to any of the previous claims, whereby the predicting of the energy exchange flexibility of each asset (16) of the set of assets (16) with the energy network (19) for the specified time period and/or the contracting of the determined energy exchange schedules (34) of the assets (16) of the set of assets (16) are performed by at least one classical computer (11).

18. Method according to any of the previous claims, whereby the set target is based on at least one target value related to the energy exchange between the set of assets (16) and the energy network (19) in the specified time period.

19. Method according to claim 17, whereby the at least one target value related to the energy exchange is one of a quantity of carbon emissions related to the generation of the exchanged energy, a portion of a renewable energy origin of the exchanged energy and a cost of the exchanged energy.

20. Method according to any of the previous claims, whereby the assets (16) in the set of assets (16) are electric vehicles (16).

21. Method according to claim 20, whereby the electric vehicles (16) in the set of assets (16) are decentralized.

22. Method according to any of claims 20 or 21, whereby at least an arrival state of charge and a departure state of charge of each of the electric vehicles (16) are determined for the predicting of the energy exchange flexibility.

23. Method according to any of claims 20 to 22, whereby at least a predetermined maximum state of charge and a predetermined minimum state of charge of each of the electric vehicles are provided for the prediction of the energy exchange flexibility.

24. Method according to any of claims 20 to 23, whereby at least a predetermined battery capacity of each of the electric vehicles (16) are provided for the prediction of the energy exchange flexibility.

- 27 -

25. Method according to any of claims 20 to 24, whereby the first optimization function (31) contains a minimum charging power and/or a maximum charging power for each electric vehicle as further constraints.

26. Method according to any of the previous claims, whereby the first optimization function (31) contains the site load at the set of assets (16) as a further constraint

27. Computer program configured to be executed on a computer system (10) to carry out the method according to any of the previous claims.

28. Computer system (10) comprising at least one classical computer (11), a quantum computer (12) and an asset controller (13), whereby the computer system (10) is configured to carry out the method according to any of the claims 1 to 26.

29. Computer system (10) according to claim 28, wherein at least one of the at least one classical computer (11), the quantum computer (12) and the asset controller (13) is embedded in a cloud infrastructure (14).

30. Computer system (10) according to claim 28 or 29, wherein the asset controller (13) is configured for communication with each asset (16) of the set of assets (16) via a communication protocol for predicting the energy exchange flexibility of each asset (16) of the set of assets (16) and/or for controlling each asset (16) of the set of assets (16).

31. Virtual power plant (1) comprising the computer system (10) according to claim 30 and a set of assets (16), wherein the assets (16) of the set of assets (16) are configured for communication with the asset controller (13) of the computer system (10).

- 28 -