CONTROL METHOD OF ELECTRIC ENERGY FLOWS WITHIN A RADIO ACCESS SYSTEM TO A COMMUNICATION NETWORK, AND CORRESPONDING CONTROL DEVICE

Description

Technical Scope of the Invention

[0001] The invention relates to the field of electrical energy management in a radio access system to a communications network. It relates more particularly to a method for controlling electrical energy flows within a radio access system to a communications network comprising a radio transceiver, a device for connection to an electrical network, a device for producing electrical energy and a battery for storing electrical energy produced by the device for producing and supplying electricity to the radio transceiver. State of the art

[0002] In the context of global warming, telecommunications operators are seeking to reduce the ecological footprint of mobile telecommunications networks. Among the various components of a mobile telecommunications network, the radio access network, which includes base stations, is the largest consumer of energy. These base stations consume 50% to 90% of a mobile network's energy.

[0003] A known solution for reducing the ecological footprint of a mobile telecommunications network is to power the network's base stations, at least partially, from green or renewable energy such as solar or wind power. To this end, the radio access network integrates photovoltaic panels and / or wind turbines to supply the base stations with electrical energy.

[0004] Photovoltaic panels and wind turbines are intermittent sources of electrical energy production because green or renewable energies are not available permanently. Their availability is highly variable over time and uncontrollable. In order to mitigate the intermittent nature of such electrical energy production, it is known to add an electrical energy storage battery between the intermittent electrical energy source (photovoltaic panel or wind turbine) and the base station. The battery temporarily stores the electrical energy produced by the intermittent source and provides this electrical energy in a smoothed manner to the base station. Ultimately, it acts as an energy storage and power supply buffer.

[0005] On the figure 1, a system 1 is shown comprising a base station 2 covering a small cell 3, or "small cell", a photovoltaic panel 4 and a battery 5. This system 1 is called "Green Small Cell" or "GSC" (i.e. "small green cell"). It is connected to a smart grid 6 by means of a bidirectional connection allowing a bidirectional exchange of electrical energy between the electrical network and the GSC 1. Thus, the GSC system 1 can buy or sell electrical energy to the network 6. An energy monitoring system 7, interposed between the GSC system 1 and the electrical network 6, controls the energy flows (represented by the arrows F1 on the figure 1 ) between the GSC system 1 and the electrical network 6, from data and / or information coming from them (represented by the arrows F2 on the figure 1 ).

[0006] Using green energy to power base stations significantly reduces CO2 emissions from the mobile access network and therefore reduces its carbon footprint.

[0007] The following documents relate to energy management for GSC systems similar to the GSC 1 system shown in the figure 1 : [1] J. Leithon, S. Sun, and T. J. Lim, "Energy Management Strategies for Base Stations Powered by the Smart Grid," in IEEE GLOBECOM, 2013. [2] D. Niyato, X. Lu, and P. Wang, "Adaptive Power Management for Wireless Base Stations in a Smart Grid Environment," IEEE Wireless Communications, vol. 19, no. 6, pp. 44-51, 2012. [3] R. Kaewpuang, D. Niyato, and P. Wang, "Decomposition of Stochastic Power Management for Wireless Base Station in Smart Grid," IEEE Wireless Communications Letters, vol. 1, no. 2, pp. 97-100, 2012. [4] J. Leithon, T. J. Lim, and S. Sun, "Online Energy Management Strategies for Base Stations Powered by the Smart Grid," in IEEE SmartGridComm, 2013, pp. 199-204.

[0008] The aim of these various documents is to minimize the operational expenses related to the telecommunications operator's electrical energy consumption by developing a strategy for purchasing and selling electricity between the GSC system and the smart grid, over a given period. Taking into account the variation in the purchase and sale prices of electricity in the electricity grid 6, the electrical consumption of the base station and the production of electrical energy by the photovoltaic module, a strategy can be defined, at the GSC system level, to: consume electrical energy from the electricity grid when the purchase price of electricity from the grid 6 is low and provide excess electrical energy produced by the photovoltaic module 4 and temporarily stored in the battery when the sale price of electricity to the grid is high.

[0009] The approaches proposed in documents [1] to [4] make it possible to control energy exchanges between the electricity grid and the GSC with the aim of reducing energy costs. These approaches implement a high-level control of the energy flows globally exchanged between the system comprising the base station, the battery and the photovoltaic module, and the smart electricity grid and require prior knowledge of the evolution of the random variables of interest relating to the production, consumption and price of electrical energy. However, they are based on an approximate knowledge of the variable price of electricity, photovoltaic production and energy consumption of the base station, for example on average profiles of solar irradiation "IS", load or electricity consumption by the GSC "CH GSC" and electricity price "Px E", as represented on the figure 2as an illustrative example. This results in models that are either very complex, requiring frequent updates and incurring high maintenance costs, or insufficient to realistically reflect changes in the environment, resulting in a loss of performance. Furthermore, these energy management methods are only aimed at reducing the telecommunications operator's electricity bill.

[0010] The present invention improves the situation, also seeking to preserve battery life.

[0011] Document US2009 / 319090A1 describes a method for optimizing the use of renewable and conventional energy sources in the context of a building's energy consumption in order to obtain an advantage aimed at reducing the costs and carbon footprint associated with this consumption. Document WO2015 / 107299A1 describes a method for managing the state of charge of a battery capable of being connected to a distribution network, the method aiming to minimize the calendar degradation or aging of such a battery and thus to avoid a deterioration in its performance. Document WO 2015 / 031581 describes a method for calibrating and optimally distributing energy resources.WO 2016 / 013089 and US 2017 / 186108 describe a method and a supply-demand regulation device, which is connected via a communication network to charge-discharge regulation devices which regulate the charging and discharging of accumulator devices connected to distribution lines of a distribution system managed by a locality. Object of the invention

[0012] To this end, the invention relates to a method for managing electrical energy flows within a radio access system to a communication network comprising a radio transceiver, a device for connection to an electrical network, an electrical energy production device and a battery for storing electrical energy produced by said device for producing and supplying electricity to the radio transceiver, characterized in that a control device determines an operating strategy for the battery by optimizing an objective function which associates a cost with a possible operating strategy for the battery during a period of determined duration T and by respecting at least one constraint for protecting the battery against accelerated aging and controls the incoming and outgoing electrical energy flows from the battery according to the determined operating strategy,and characterized in that a battery protection constraint is to limit the duration of non-use of the battery to a predefined maximum duration, and in that the maximum duration of non-use of the battery is determined as a function of the temperature of the battery and / or the current state of charge of the battery.,

[0013] The method of the invention makes it possible both to achieve a cost reduction objective, for example energy or environmental, linked to the operation of the radio transceiver connected to the electricity network and equipped with a battery and an electrical energy production device (for example a photovoltaic or wind device), and to maximize the lifespan of the battery.

[0014] The constraint(s) for protecting the battery against aging can be: to require that the state of charge of the battery belongs to an authorized range limited by a lower limit between 10% and 30%, advantageously between 15% and 25%, and an upper limit between 75% and 95%, advantageously between 80% and 90%; to set a maximum rate of charge or discharge of the battery.

[0015] The maximum set charge or discharge rate can be determined based on the battery temperature and / or the current state of charge of the battery.

[0016] In a first embodiment, the objective function is a function which associates a price of electricity from the electricity network (purchase price by an electricity consumer and / or sale price by an electricity producer) with a possible operating strategy of the battery during the period of fixed duration T by integrating over time, over the fixed duration T, the product of a price of electricity from the electricity network applicable during an elementary duration Δ t and of the quantity of electrical energy supplied by the electricity network to the radio access system during said elementary duration Δ t , and in that the control device determines, by optimizing the objective function, a battery operating strategy which minimizes said electricity price during the determined duration T.

[0017] Advantageously, the quantity of electrical energy supplied by the electrical network during an elementary duration Δ t is determined by a mathematical operation of adding a quantity of electrical energy consumed by the radio transceiver during the elementary duration Δt and a quantity of electrical energy charged into the battery during the elementary duration Δ t , in the case where said battery is charging during the elementary duration Δ t , and of subtracting a quantity of electrical energy produced by the energy production device during the elementary duration Δ t and a quantity of electrical energy discharged from the battery in the case where said battery is discharging during the elementary duration Δ t .

[0018] Advantageously, the objective function O is defined by the equation: O = ∑ t = 0 T P e t ⋅ P BS t + P Bat SOC t , SOC t + 1 − P PV t × Δ t , Or t and t+1 represent two successive instants separated by the elementary duration Δ t ; Either ( t) represents the purchase price by a consumer of the electricity supplied by the electricity network (6) to the radio access system (1) at time t or the sale price of the electricity to the electricity network by the radio access system (1); P PV ( t ) represents the electrical power produced by the production device (4) at time t; P BS ( t ) represents the electrical power consumed by the transceiver (2) at time t; P Bat (( SOC ( t ) , SOC ( t + 1)) represents the electrical power consumed, if it is positive, or produced, if it is negative, by the battery (5) at time t; SOC(t) And SOC(t + 1) represent the states of charge of the battery (5) at times t and t+1 respectively.

[0019] The optimization may be based on data on the evolution during the period of determined duration T of state variables relating to the electrical energy consumed by the radio transceiver, to the state of charge of the battery and to the price (of sale and / or purchase) of electricity from the electrical network, which is predicted from previous data on the evolution of said state variables during a past period. It may use an interior points method.

[0020] In a second embodiment, the optimization uses a stochastic method implementing a learning mechanism.

[0021] In an alternative embodiment, the objective function associates a carbon footprint cost with a possible operating strategy of the battery for a given duration T of the radio access system and in that the optimization aims to minimize said carbon footprint over the duration T.

[0022] The invention also relates to a control device for a radio access system to a communication network comprising a radio transceiver, a device for connection to an electrical network, a device for producing electrical energy and a battery for storing electrical energy produced by said device for producing and supplying electricity to the radio transceiver, characterized in that it comprises an optimization module intended to determine an operating strategy for the battery by optimizing an objective function which associates a cost with a possible operating strategy for the battery for a determined duration T while respecting at least one constraint for protecting the battery against aging and a battery control module intended to control the flows of electrical energy entering and leaving the battery as a function of the determined operating strategy,and characterized in that a battery protection constraint is to limit the duration of non-use of the battery to a predefined maximum duration, the control device being further characterized in that it determines the maximum duration of non-use of the battery as a function of the temperature of the battery and / or the current state of charge of the battery.,

[0023] The invention also relates to a radio access system to a communication network comprising a radio transceiver, a device for connection to an electrical network, a device for producing electrical energy, a battery for storing electrical energy produced by said device for producing and supplying electricity to the radio transceiver and a control device according to the preceding claim. Summary description of the drawings

[0024] The invention will be better understood with the aid of the following description of a particular embodiment of a method and a system for managing the flow of electrical energy within a radio access system to a telecommunications network according to the invention, with reference to the appended drawings in which: There figure 1 represents an architecture comprising a GSC (Green Small Cell) system connected to an electrical network and an energy management entity, of the prior art; The figure 2 represents standardized profiles of temporal evolution, over a period of 24 hours, of the selling price of electricity from the electricity network, solar irradiation and the load (or electrical consumption) of the GSC; The figure 3represents an architecture comprising a GSC (Green Small Cell) system connected to an electrical network and an energy management entity, according to a particular embodiment of the invention; The figure 4 represents a flowchart of the steps of the management method of the invention according to a particular embodiment; The figure 5 represents a functional diagram illustrating the interaction between the radio access system, or GSC system, connected to the electrical network (belonging to the “Environment” set) and the control device (“Agent”). Detailed description of particular embodiments of the invention

[0025] From the outset, it should be noted that, in the various figures, identical, analogous or corresponding elements bear the same references, unless specifically indicated otherwise.

[0026] On the figure 3, a radio access system 1 to a communication network is shown, connected to a smart electricity network 6. For example, the radio access system 1 belongs to the radio access part or "RAN" (from the English "Radio Access Network") of a mobile telephone network. However, the invention applies to the radio access part of any other communication network.

[0027] Radio Access System 1, also known as the Green Small Cell (GSC) system, includes: a radio transceiver 2, or relay antenna, here a base station, covering a cell 3, an electrical energy production device 4, in this case a photovoltaic device, an electrical energy storage battery 5, a control device 100.

[0028] Device 4 is intended to locally produce electrical energy from green energy, in this case solar energy. Device 4 could use another renewable energy, for example wind energy, to produce electricity.

[0029] The battery 5 is intended on the one hand to store electrical energy produced by the device 4 and / or from the electrical network 6, and, on the other hand, to supply electrical energy to the radio transceiver 2 and / or to the electrical network 6.

[0030] The control device 100 integrates a device for connection to the electrical network 6 (not shown in the figure 3 ) which allows the radio access system 1 to be connected to the electrical network 6 via: of a link 8A for transporting incoming and outgoing electrical energy flows, allowing the bidirectional exchange of electrical energy flows between the electrical network 6 and the radio access system 1; of a data transport link 8B allowing the bidirectional exchange of data and / or information between the electrical network 6 and the radio access system 1.

[0031] The control device 100 is also connected to the radio transceiver 2 and to the local electrical energy production device 4, here via a power bus 9A and a data bus 9B respectively allowing the transmission of electrical energy flows and the transmission of data between these different elements 100, 4 and 2. On the figure 3, the arrows connecting the elements - control device 100, base station 2 and electrical energy production device 4 - to the power bus 9A and to the data bus 9B indicate the directions of the corresponding data flows and the corresponding electrical energy flows.

[0032] Furthermore, the control device 100 is directly connected to the battery 5 and intended to control the operation of the battery 5, in other words the incoming (or charging) electrical energy flows and the outgoing (or discharging) electrical energy flows from the battery 5. Its role is to directly control the battery 5.

[0033] In reference to the figure 3 , we will now describe the different flows of electrical energy likely to circulate within the radio access system 1 in operation.

[0034] The local production device 4 produces electrical energy intermittently, here depending on the sunshine. The flow of electrical energy thus produced locally by the device 4 can be supplied to the battery 5 and stored therein at least temporarily.

[0035] In operation, the radio transceiver 2 consumes electrical energy. For this purpose, it receives a flow of electrical energy consisting of one or both of the following flows: a first flow of electrical energy coming from the electrical network 6; a second flow of electrical energy coming from the battery 5.

[0036] The photovoltaic device 4 could directly provide a third flow of electrical energy to the radio transceiver 2 and / or to the electrical network 6.

[0037] In operation, the battery 5 can, on the one hand, receive an incoming flow of electrical energy from the photovoltaic device 4 and / or the electrical network 6, and, on the other hand, provide an outgoing flow of electrical energy to power the radio transceiver 2 and / or the electrical network 6.

[0038] The control device 100 is intended to control or pilot the different flows of electrical energy which have just been described within the radio access system 1 by implementing the method for controlling or managing the flows of electrical energy within a radio access system 1 which will now be described with reference to the figure 4 .

[0039] The method for controlling the flow of electrical energy within a radio access system 1, implemented by the control device 100, aims to control the flow of energy between the following entities: the base station 2, the electrical energy production device 4, the battery 5 and the electrical network 6. It is based on direct control of the battery 5 adapted to optimize both an objective function and the life expectancy of the battery 5, as will be explained later.

[0040] The operating strategy of the battery 5 (i.e. the strategy or procedure or method for charging and discharging the battery 5) determines the strategy for controlling the electrical energy flows between the electrical network 6 and the radio access system 1 which comprises the electrical energy production device 4. Indeed, controlling only the battery 5 is sufficient to define the electrical energy flows between the components within the radio access system 1 and between the electrical network 6 and the radio access system 1.This results from the energy balance at the level of the radio access system 1 between what is “produced” - by the electricity network 6 and / or by the local production device 4 and / or, where applicable, by the battery 5 when the latter is discharging - and what is consumed - by the base station 2 and / or by the electricity network 6 (electrical energy is sold to the electricity network 6) and, where applicable, by the battery 5 when the latter is charging.

[0041] The control of battery 5 depends on an optimization operation E1 consisting of optimizing an objective function while respecting one or more constraints relating to battery 5 intended to protect them against excessively rapid aging (i.e. premature or accelerated).

[0042] In mathematical optimization, an objective function is a function that serves as a criterion for determining the best solution to an optimization problem. It associates a value with an instance of an optimization problem. In the context of the invention, the objective function associates a cost with an operating strategy (or charging and discharging strategy) of the battery 5 for a given duration T.

[0043] The term "cost" is understood to mean a charge that is linked to the operation of the radio access system 1, or GSC system, during the determined duration T. This cost can be energetic (economic) or ecological (environmental). For example, the charge is either the price of electricity purchased from the electricity grid 6 by the radio access system 1, or the carbon footprint (i.e. the quantity of CO 2 emitted) of the radio access system 1, induced by an operating strategy of the battery 5.

[0044] The operating strategy of the battery 5 during a given period T can be defined, at each instant t of this period T, by an electrical power P batt of the battery 5. At an instant t, this electrical power P batt is either positive, when it corresponds to an electrical power consumed by the battery 5 (i.e. when the battery is charging and receives an incoming flow of electrical energy), or negative, when it corresponds to an electrical power produced by the battery 5 (i.e. when the battery is discharging and provides an outgoing flow of electrical energy), or zero if the battery is at rest. The goal of the optimization is to find an optimal operating strategy (i.e. charging and discharging) of the battery 5 which minimizes the cost.

[0045] Generally, the control device 100 interacts with an environment comprising the radio access system 1 and the smart grid 6, the state of which at the moment t can be represented by the state vector st. This state vector st groups together state variables corresponding to the energy consumed by the base station 2, the energy produced by the photovoltaic device 4, the state of charge SOC(t) of the battery 5 and the price of electricity P e (t). At each instant t, the control device 100 selects an action a t corresponding to a charge or discharge instruction of the battery 5. The aim is to optimize an objective function O able to determine in the long term a cost linked to the operation of the radio access system 1 while maximizing the duration of the battery 5.

[0046] All the states of the environment perceived by the control device 100 can be described as follows: S = B × E × R × P Or SPIT And P are sets composed of elements constituted by scalars, relating respectively to the electrical power consumed by the base station 2, to the state of charge of the battery 5, to the electrical power produced by the photovoltaic device 4 and to the price of electricity on the electrical network 6; S is the set constituted by the Cartesian product, or the combination, of the sets SPIT And P , composed of elements consisting of vectors of size 4.

[0047] Given an initial state s ∈ S, the optimization problem to be considered for a given duration of time is as follows: Trouver π * = argmin π ∈ S × A O π s Or A is the set of charge rates (positive), discharge rates (negative) and rest rate (zero), of battery 5, π is a possible strategy for battery 5 which associates with each state of set S a charge or discharge or rest rate of set A, i.e. a charge or discharge or rest action of battery 5, and O π< ( s ) represents the value of the overall optimization objective for the strategy π , starting from a state s.

[0048] In a first example of realization, the objective function O associates a price of electricity on the electricity network 6 with a strategy π for operating the battery 5 during a given period or duration T. This objective function O can be defined by the following equation (1): O = ∑ t = 0 T P e t ⋅ E t t , t + 1 Or Et(t, t + 1) represents the quantity of electrical energy supplied by the electrical network 6 to the radio access system 1 during an elementary duration Δ t , between the times “t” and “t+1” (with “ t +1" = t + Δ t ), this amount of energy Et(t, t + 1) being positive when the radio access system 1 consumes (or buys) electrical energy on the network 6, negative when the radio access system 1 supplies (or sells) electrical energy on the network 6, or zero when the radio access system 1 and the electrical network 6 do not exchange electrical energy; and Either ( t ) represents the price of electricity on the electricity network, valid at time t and applicable between times t and t+1 - more specifically Either ( t ) represents the purchase price by a consumer from the electricity network 6 if Et(t, t + 1) is positive (electrical energy purchased on network 6), and the selling price of electricity to the electricity network S when Et(t, t + 1) is negative (electrical energy sold to network 6).

[0049] The elementary duration Δ t is advantageously adapted to the speed of evolution or variation of the environment, which includes in particular the price of electricity, the production of electrical energy and the consumption of electrical energy. For example, if the price of electricity is likely to vary every 30 minutes, the elementary duration Δ t can be set to 30 minutes.

[0050] Thus, the objective function O as defined by equation (1) integrates over time, over the fixed duration T, the product of a price Either ( t ) of the electricity from the applicable electrical network during the elementary duration Δ t , and of the quantity of electrical energy Et(t, t + 1) supplied by the electrical network to the radio access system 1 during this elementary duration Δ t .

[0051] The energy balance at a time t at the level of the radio access system 1 between: the electrical power received or produced, which includes o the positive electrical power supplied by the electrical network 6 to the radio access system 1, i.e. the electrical power that the radio access system 1 purchases from the electrical network 6; o the electrical power P PV ( t) produced by the production device 4; and o where applicable, the electrical power supplied by the battery 5 when the latter is discharging (i.e. produces an outgoing flow of electrical energy) and the consumed electrical power, which includes o the negative electrical power supplied by the electrical network 6 to the radio access system 1, i.e. the electrical power that the radio access system sells to the electrical network 6; o the electrical power consumed by the base station 2; and o where applicable, the electrical power consumed by the battery 5 when the latter is charging (i.e. receives an incoming flow of electrical energy) can be translated by the following equation (2): E t t , t + 1 Δ t + P PV t = P BS t + P Bat SOC t , SOC t + 1 Or E t t , t + 1 Δ t represents the electrical power (positive, negative or zero as previously explained) supplied by the electrical network 6 to the radio access system 1 at time t; P PV ( t) represents the electrical power produced by the photovoltaic device 4 at time t; P BS ( t ) represents the electrical power consumed by base station 2 at time t; P Batt (( SOC ( t ) , SOC ( t + 1)) represents the electrical power consumed (if it is positive, battery 5 is then charging) or produced (if it is negative, battery 5 is then discharging) by battery 5 at time t, this power being zero when the battery is at rest; SOC ( t ) And SOC(t + 1) represent the states of charge of battery 5 at times t and t+1 respectively.

[0052] The following equation (3) results from equation (2): E t t , t + 1 = P BS t + P Bat SOC t , SOC t + 1 − P PV t × Δ t

[0053] The quantity of electrical energy (positive, negative or zero) supplied by the electrical network 6 to the GSC system 1 between two times t and t+1 therefore depends on the evolution of the state of charge of the battery between these two times t and t+1, in other words: E t t , t + 1 = E t SOC t , SOC t + 1

[0054] It results from the energy balance at the level of the radio access system 1, that the quantity of electrical energy (positive, negative or zero) supplied by the electrical network 6 during the elementary duration Δ t , noted Et(t, t + 1) or And t ( SOC ( t ) , SOC ( t + 1)), is determined by a mathematical operation of addition of a quantity of electrical energy consumed by the radio transmitter-receiver 2 during the elementary duration Δ t , noted P BS ( t ) × Δ t , and, in the case where the battery 5 is charging (i.e. receives an incoming flow of electrical energy from the production device 4) during the elementary duration Δ t , of a quantity of electrical energy charged into the battery 5 during the elementary duration Δ t and changing the state of charge of the battery 5 from SOC(t) to SOC(t+1), noted P Bat ( SOC ( t ) ,SOC ( t + 1)), and subtraction of a quantity of electrical energy produced by the energy production device 4 during the elementary duration Δ t , noted P PV ( t ) × Δ t , and, in the case where the battery 5 is discharging during the elementary duration Δ t (i.e. delivers an outgoing flow of electrical energy to the base station 2), of a quantity of electrical energy discharged from the battery 5 and causing the state of charge of the battery 5 to change from SOC(t) to SOC(t+1), noted | P Bat ( SOC ( t ) , SOC (t + 1)) | .

[0055] Taking into account equality (3), the objective function O , defined by equation (1), can be formulated equivalently by the following equation (4): O = ∑ t = 0 T P e t ⋅ P BS t + P Bat SOC t , SOC t + 1 − P PV t × Δ t

[0056] During the optimization operation E1, the control device 100 seeks to solve the optimization problem of minimizing the objective function O by applying a strategy π for charging and discharging battery 5 (or operating strategy of battery 5) for a given duration T. In other words, it seeks to determine which strategy π among the set of possible strategies π for charging and discharging battery 5 makes it possible to minimize the value of the objective function O in order to determine thus argminO { π} . Ultimately, optimization allows us to determine the charge states SOC(t)of battery 5, at successive times t, t+1, t+2,...., during the time period of duration T going from t=0 to T, which make it possible to minimize the objective function O corresponding here to the price of electricity over period T.

[0057] According to the invention, the control device 100 seeks to solve the previously defined optimization problem by satisfying at least one constraint of protection of the battery against accelerated aging.

[0058] In a first embodiment of the invention, a battery protection constraint is to impose that the state of charge SOC(t) of the battery 5 at each instant t belongs to a range of authorized values ​​limited by a lower limit between 10% and 30%, advantageously between 15% and 25%, and an upper limit between 75% and 95%, advantageously between 80% and 90%. For example, the constraint to be respected is that the state of charge of the battery is between 20% and 90%, in other words: 20 % ≤ SOC t ≤ 90 % .

[0059] In a second embodiment of the invention, a battery protection constraint is to set a maximum charge rate and / or discharge rate of the battery 5. In other words, a maximum charge current and / or a maximum discharge current of the battery 5 is set. The maximum charge and / or discharge rate set can be determined as a function of the current state of charge of the battery and / or the temperature of the battery.

[0060] In a third embodiment of the invention, a protection constraint of the battery 5 is to limit the duration of non-use of the battery 5 to a predefined maximum duration. The maximum duration of non-use of the battery 5 may be fixed. Alternatively, the control device 100 determines the maximum authorized duration of non-use of the battery 5 as a function of the temperature of the battery and / or the current state of charge of the battery, from stored reference data. This constraint aims to prevent the battery 5 from remaining inactive for too long, which would have the effect of degrading it, in particular if the outside temperature is high.

[0061] Consideration could be given to imposing other constraints intended to protect the battery 5 against accelerated aging, in other words to increase the lifetime of the battery 5.

[0062] The control device 100 can perform the optimization operation E1 while respecting one or more constraints to protect the battery against aging.

[0063] Note that if we set constraints on battery quantities, such as its temperature, its state of charge, its charge and / or discharge rate, these quantities must appear in the set of environmental states as battery state variables, for the definition of the optimization problem.

[0064] To solve the previously defined optimization problem, the control device 100 can use any suitable mathematical optimization method.

[0065] In a first embodiment, the control device 100 determines, or predicts, the evolution over time - during the determined period T - of the state variables Either ( t ) , P BS ( t ) , P PV ( t ) And SOC(t) contained in equation (4), based on previous data on the evolution of these state variables during a past period. If the period T has a duration of one day, i.e. 24 hours, the evolution of the state variables during a day J can be predicted from the evolution of these state variables during the previous day J-1. For example, we consider that the evolution of the state variables Either ( t ) , P BS ( t ) , P PV ( t ) And SOC(t) during a day is identical to their evolution during the previous day. Thus, we consider that the evolution of the variables Either ( t ) , P BS ( t ) , P PV ( t ) during a day is known on the basis of previous data on the evolution of these variables. In this case, the control device 100 can solve the optimization problem using, for example, interior point optimization methods.

[0066] The optimal charging and discharging strategy determined by optimization E1 is then used by the control device 100 to control the flows of electrical energy entering the battery 5 and the flows of electrical energy leaving the battery 5, during a step E2 of controlling the battery 5.

[0067] In a second embodiment, the control device 100 uses a stochastic optimization method intended to solve the optimization problem on the basis of a learning mechanism or process which makes it possible to estimate the evolution of the state variables. Stochastic optimization has the advantage of not requiring any model.

[0068] On the figure 5, a functional diagram is shown illustrating an example of a learning mechanism that can be implemented to optimize the cost linked to the operation of the radio access system 1 while respecting one or more constraints on the battery aimed at protecting it from aging.

[0069] In reference to the figure 5 , an agent, constituted by the control device 100, observes a current state s t at a current time t of the environment comprising the battery 5, the base station 2, the photovoltaic panel 4 and the price of electricity and consequently chooses an action a t . In response to the chosen action a t , the environment evolves towards a new state s t+1 and gives rise to a reward r t+1, or a sanction, which the agent seeks to maximize, respectively minimize, over time. The agent's goal is therefore to optimize an objective function Oby exploiting the rewards and / or sanctions obtained to evaluate the quality of actions in a given state. The objective function O must represent, over the long term (typically over a period of time corresponding to the theoretical lifetime of the battery), the cost associated with operating the radio access system 1 while maximizing the lifetime of the battery 5.

[0070] We can match the function O π ( s ) to the sum of rewards / penalties accumulated during the application of the strategy π from the state s in the long run. Therefore, it is possible to use the Bellman equation as follows: π ∗ = argmin π ∈ S × A O π s = argmin π ∈ S × A ∑ a ∑ s ′ P s , s ′ a R s , s ′ a + O π s ′ Or : R s , s ′ a is the cost (reward / penalty) associated with selecting the action a to state s then access to state s' P s , s ′ a is the transition probability from state s to state s' by applying the action a, which depends on environmental hazards (production, consumption and price of energy) and is not known a priori; O π< ( s' ) is the sum of rewards / penalties accumulated during the application of strategy π from state s'.

[0071] A learning algorithm, or Q-learning, allows the optimal value to be found O* ( s ) and the associated strategy by interacting with the environment, without prior knowledge of P s , s ′ a The only conditions are to correctly define R s , s ′ a and explore enough possibilities (state - action) to converge towards the optimal solution.

[0072] For example, the immediate cost (reward / penalty), i.e. resulting directly from an action a to move from state s to state s', can be formulated as follows: R s , s ′ a = E BS s + E Bat a s − E PV s P e s + Γ bat s , a , s ′ Or: R s , s ′ a is the immediate cost of selecting an actiona to the state s accessing the state s', The BS is the energy consumed by base station 2, The Bat is the energy consumed (if it is positive) or produced (if it is negative) by battery 5, The PV is the energy produced by the photovoltaic device 4, Either is the price of electricity on the electricity network 6; Γ bat is the protection constraint of battery 5 against accelerated aging, allowing the life of battery 5 to be extended, this constraint increasing as one moves away from a recommended operating interval.

[0073] During learning, the control device 100 explores different strategies according to the Q-learning algorithm until it finds the optimal strategy which minimizes the cost in the long term, i.e. over a period of duration T. The Q-learning algorithm associates a Q-value with each state-action pair. Q ( s, a) evaluating the impact of choosing an action a at state s on the long-term cost. By exploring different actions for different states, the agent collects rewards / penalties which it uses to update the Q-values ​​until reaching equilibrium (no more Q-values ​​are changed).

[0074] The control device 100 thus makes it possible to optimize, without a priori knowledge of the fluctuation models of the production, consumption and price of electrical energy, the long-term cost of a radio access system 1 connected to the electrical network 6 and the life expectancy of the battery 5. Initially, this control device uses a learning process to find the optimal operating strategy (charging and discharging).

[0075] Once learning is complete, the Q-value matrix is ​​used to determine the optimal action to choose for each state, during a step E2 of controlling battery 5.

[0076] The control device 100 comprises an optimization module 101 intended to determine an operating strategy of the battery 5 by optimizing an objective function which associates a cost with a possible operating strategy of the battery for a determined duration T while respecting at least one constraint of protection of the battery against accelerated aging and a control module 102 of the battery 5 intended to control the incoming and outgoing electrical energy flows of the battery 5 according to the determined operating strategy. The optimization module 101 and the control module 102 are respectively intended to implement the optimization step E1 and the control step E2.

[0077] When a new radio access system is deployed, it can exploit the operating strategies of neighboring radio access systems that have already completed the learning process. A new learning process should only be performed in the case of significant changes such as a new base station or battery technology or a new local power generation device, or a significant change in the energy market.

[0078] In an alternative embodiment, the objective function associates a cost in carbon footprint (or more generally in ecological footprint) with an operating strategy of the battery 5 for a determined duration T. The optimization aims to determine the optimal operating strategy of the battery 5 which minimizes the carbon footprint (or the ecological footprint).

Claims

1. Method for managing flows of electrical energy within a radio access system (1) for a communication network including a radio transceiver (2), a device for connecting to an electrical grid (6), an electrical energy production device (4) and a battery (5) for storing electrical energy produced by said production device (4) and for electrically powering the radio transceiver (2), characterized in that a control device (100) determines an operating strategy for the battery (5) by optimizing (E1) an objective function (O) that associates a cost with a possible operating strategy for the battery (5) during a period of determined duration T and while respecting at least one constraint for protecting the battery (5) against accelerated ageing, and manages (E2) the incoming and outgoing flows of electrical energy of the battery (5) depending on the determined operating strategy, wherein one constraint for protecting the battery (5) is to restrict the duration of disuse of the battery (5) to a predefined maximum duration and wherein the maximum duration of disuse of the battery (5) is determined depending on the temperature of the battery and / or on the current state of charge of the battery.

2. Method according to Claim 1, characterized in that one constraint for protecting the battery (5) is to impose that the state of charge of the battery belongs to a permitted range restricted by a lower bound of between 10% and 30%, advantageously between 15% and 25%, and an upper bound of between 75% and 95%, advantageously between 80% and 90%.

3. Method according to either of the preceding claims, characterized in that one constraint for protecting the battery (5) is to set a maximum rate of charge or discharge of the battery.

4. Method according to the preceding claim, characterized in that the set maximum rate of charge or discharge is determined depending on the temperature of the battery.

5. Method according to Claim 3 or 4, characterized in that the set maximum rate of charge or discharge is determined depending on the current state of charge of the battery (5).

6. Method according to one of the preceding claims, characterized in that the objective function (O) is a function that associates a price of electricity of the electrical grid (6) with a possible operating strategy for the battery (5) during the period of determined duration T by integrating over time, over the determined duration T, the product - of a price of electricity of the electrical grid that is applicable during an elementary duration Δt and - the amount of electrical energy supplied by the electrical grid to the radio access system during said elementary duration Δt, and in that the control device (100) determines, by optimizing the objective function, an operating strategy for the battery (5) that minimizes said price of electricity during the determined duration T.

7. Method according to the preceding claim, characterized in that the amount of electrical energy supplied by the electrical grid during an elementary duration Δt is determined by a mathematical operation • of adding an amount of electrical energy consumed by the radio transceiver during the elementary duration Δt and an amount of electrical energy charged to the battery during the elementary duration Δt, if said battery is charging over the elementary duration Δt, and • of subtracting an amount of electrical energy produced by the energy production device during the elementary duration Δt and an amount of electrical energy discharged from the battery if said battery is discharging over the elementary duration Δt.

8. Method according to one of the preceding claims, characterized in that the objective function O is defined by the equation O = ∑ t = 0 T P e t ⋅ P BS t + P BAT SOC t , SOC t + 1 − P PV t × Δ t , where • t and t+1 represent two successive instants separated by the elementary duration Δt; • Pe(t) represents the purchase price for a consumer of electricity supplied by the electrical grid (6) to the radio access system (1) at the instant t, or the sale price of electricity to the electrical grid by the radio access system (1); • PPV(t) represents the electric power produced by the production device (4) at the instant t; • PBS(t) represents the electric power consumed by the transceiver (2) at the instant t; • PBat((SOC(t), SOC(t + 1)) represents the electric power consumed, if it is positive, or produced, if it is negative, by the battery (5) at the instant t; • SOC(t) and SOC(t + 1) represent the states of charge of the battery (5) at the instants t and t+1, respectively.

9. Method according to one of Claims 6 to 8, characterized in that the optimization is based on data on the evolution, over the period of determined duration T, of state variables relating to the electrical energy consumed by the radio transceiver, to the state of charge of the battery and to the price of electricity on the electrical grid, which is predicted on the basis of previous data on the evolution of said state variables over a past period.

10. Method according to the preceding claim, characterized in that the optimization uses an interior point method.

11. Method according to one of Claims 1 to 7, characterized in that the optimization uses a stochastic method implementing a learning mechanism.

12. Method according to one of Claims 1 to 10, characterized in that the objective function associates a carbon footprint cost with a possible operating strategy for the battery, during a determined duration T, of the radio access system, and in that the optimization aims to minimize said carbon footprint over the duration T.

13. Control device for a radio access system for a communication network including a radio transceiver (2), a device for connecting to an electrical grid (6), an electrical energy production device (4) and a battery (5) for storing electrical energy produced by said production device (4) and for electrically powering the radio transceiver (2), characterized in that it comprises an optimization module (101) intended to determine an operating strategy for the battery (5) by optimizing (E1) an objective function that associates a cost with a possible operating strategy for the battery during a determined duration T while respecting at least one constraint for protecting the battery against ageing, and a module (102) for managing the battery (5) intended to control the incoming and outgoing flows of electrical energy of the battery (5) depending on the determined operating strategy, wherein one constraint for protecting the battery (5) is to restrict the duration of disuse of the battery (5) to a predefined maximum duration and wherein the control device is determining the maximum duration of disuse of the battery (5) depending on the temperature of the battery and / or on the current state of charge of the battery.

14. Radio access system for a communication network including a radio transceiver (2), a device for connecting to an electrical grid (6), an electrical energy production device (4), a battery (5) for storing electrical energy produced by said production device (4) and for electrically powering the radio transceiver (2), and a control device (100) according to the preceding claim.