Retired power battery digital lossless recombination method of reconfigurable battery network

Abstract

The invention relates to the technical field of battery reconfiguration of an energy storage battery system, in particular to a digital lossless reconfiguration method for retired power batteries of a reconfigurable battery network. The invention discloses a decommissioned power battery digital lossless recombination method of a reconfigurable battery network. The method comprises the following steps: S100, designing a reconfigurable battery network topological structure; S110, establishing the reconfigurable battery network; s120, establishing an optimization control framework; s130, estimating a battery equivalent circuit model and a parameter state; s140, establishing a dynamic optimization model; s200, controlling the reconfigurable battery network; S210, establishing a control framework; s220, arranging an energy storage strategy based on AOE control configuration; and S230, performing AOE-based energy storage system power optimization distribution double-layer control. A decommissioned power battery digital lossless recombination method of a reconfigurable battery network has the advantages that an energy storage battery pack intermittently undertakes charging and discharging tasks, the charging and discharging times are effectively reduced, SOC balance between energy storage battery modules is achieved, overcharging and overdischarging of energy storage batteries are avoided, and battery aging is slowed down.

Description

Technical Field

[0001] The present invention relates to the technical field of battery reconstruction of energy storage battery systems, and in particular to a digital lossless reconstruction method for retired power batteries of a reconfigurable battery network. Background Art

[0002] Nowadays, new energy has been widely used in all walks of life, especially energy storage battery systems, which are superior to other new energy sources in terms of technology, cost and performance. An energy storage battery system is a device that can store electrical energy, usually composed of batteries and parallel voltage converters, and is used to store and release electrical energy for a long time. The main function is to convert electrical energy into chemical energy and store the charge in the battery so that it can be released when needed. This system is usually designed for long-term energy storage and charging and discharging, and plays an important role in aspects such as grid dispatching, peak load reduction and power management. The key features of energy storage batteries include high capacity, long cycle life and stable performance. Although battery cells can achieve a cycle life of nearly 10,000 "full charge and discharge" times, with high safety and stability, there are differences among battery cells, which leads to the short board effect of the energy storage battery system. The short board effect of a battery cell will lead to the retirement of the entire battery module, resulting in great waste.

[0003] In view of the above defects, the life of the battery module is usually extended by battery reconstruction. For example, the Chinese invention patent with publication number CN112510792A discloses a reconfigurable converter for a retired battery energy storage system and a control method thereof, wherein the converter includes a battery pack selection circuit, a combined bidirectional converter, and a control unit, and the control unit is respectively connected to the battery pack selection circuit and the combined bidirectional converter. The retired battery energy storage system sequentially selects the battery pack selection circuit and the combined bidirectional converter, and then leads out the positive and negative terminals for connecting the load or charging equipment; the battery pack selection circuit is used to control the connection between each battery pack in the retired battery energy storage system and the combined bidirectional converter, and the bidirectional combined converter is used to control the opening of the balance. The converter and control method of the patent are used to effectively reconstruct and control the retired batteries in the energy storage battery system. However, the battery energy storage system of the patent must undertake the charging and discharging tasks in real time, which is easy to cause overcharging and over-discharging of the battery, accelerate battery aging, and shorten the battery life. Summary of the invention

[0004] In view of this, the present invention aims to provide a method for digital lossless reconstruction of retired power batteries in a reconfigurable battery network. The method adopts a lossless reconstruction method of designing a reconfigurable battery network topology and controlling the reconfigurable battery network to solve the problem that the battery energy storage system must bear the charging and discharging tasks in real time, which easily leads to overcharging and over-discharging of the battery, accelerates battery aging, and shortens the battery life.

[0005] In order to solve the above problems, the present invention provides a digital lossless reorganization method for retired power batteries of a reconfigurable battery network, comprising:

[0006] S100, design reconfigurable battery network topology:

[0007] S110, establishing a reconfigurable battery network:

[0008] The reconfigurable battery network includes a battery energy storage system composed of K battery clusters, each battery cluster is composed of N parallel units connected in series, each parallel unit includes a battery module and a controllable switch, the controllable switch includes a parallel switch and a series switch, each battery module is composed of a battery cell connected in series with a series switch, and M battery modules are connected in parallel with a parallel switch to form a parallel unit;

[0009] S120, establish an optimized control architecture:

[0010] By measuring the terminal voltage U and terminal current I of each parallel unit in the dynamic reconfigurable battery network, a measurement state set M = {U, I} is formed;

[0011] The parameters and states of each battery in the dynamic reconfigurable battery network are jointly estimated, including the battery state of charge, resistance and capacitance parameters, to form a state and parameter set S = {R 0 , R 1 , C 1 , SOC}, where R 0 Represents the equivalent internal resistance of the battery, R 1 Represents polarization internal resistance, C 1 represents polarized capacitance;

[0012] The topological connection control variable Z of the reconfigurable battery network is optimized and controlled to generate a battery network topology generation scheme that meets the optimization objectives and constraints;

[0013] S130, estimating the battery equivalent circuit model and parameter status:

[0014] The Thevenin equivalent circuit model of the battery module is established, and the mathematical model is expressed by the following formula (1):

[0015]

[0016] Among them, P k is the estimated error covariance matrix, and E is the unit matrix. By continuously observing the variable assignment, the battery SOC can be obtained;

[0017] S140. Establish a dynamic optimization model for a reconfigurable battery network:

[0018] In a reconfigurable battery network, the dynamic optimization model of the battery cluster can be established as:

[0019]

[0020] Among them, z ij is the switch state, taking the value of {0,1}; L(i,j), U(i,j) is the SOC of the battery module in the i-th row and j-th column in the m-parallel and n-series battery cluster at the end of charging / discharging;

[0021] S150, reconfigurable battery network model solution:

[0022]

[0023] S200, control of reconfigurable battery network:

[0024] S210. Establish a reconfigurable battery network control architecture:

[0025] The bottom layer is the retired power battery system. Each local controller obtains the status information of each power battery through status measurement and uploads it to the communication network. The communication network nodes communicate with neighboring nodes, and through mutual cooperation and joint decision-making, the distributed energy management of the retired power battery system is realized, and then the control instructions are issued to the retired power batteries through the local controllers.

[0026] S220, Energy storage strategy arrangement based on AOE control configuration:

[0027] S221, constructing an AOE network, the AOE network is a directed acyclic graph, the AOE includes multiple nodes, each two nodes are connected by an edge, each node represents an event, each edge represents an action, the direction of the edge represents the progressive and migration relationship between the events represented by the nodes, and this relationship is determined by the action represented by the edge;

[0028] S222, configure AOE, use the AOE network configuration file to define the basic information, variables, events and actions of the AOE network, so as to describe the constructed AOE network and the designed control strategy;

[0029] S223, configure measuring points;

[0030] S224, configuring a communication channel;

[0031] S225, parsing and executing the AOE network;

[0032] S230, dual-layer control of power optimization distribution of energy storage system based on AOE.

[0033] Furthermore, in step S130, the method for obtaining the mathematical model of formula (1) is:

[0034] Write down the equation of the continuous time of the battery module equivalent Thevenin model, and its mathematical expression is:

[0035]

[0036] Among them, the open circuit voltage of the battery is U ocv (t) is a variable related to the battery dynamic state of charge SOC(t), and can be expressed in ampere-hour integration method. SOC 0 is the initial state value. The battery current i(t) is affected by the battery polarization capacitance C 1 and the voltage u of the polarized capacitor C The battery terminal voltage u(t), u C (t), U ocv (t) and equivalent internal resistance R 0 The voltage forms a closed voltage loop;

[0037] The state and parameter estimation method of lithium-ion battery based on extended Kalman filter is expressed by equation (5) as the discrete state and observation equation of lithium-ion battery, and the discrete time is Δt=1s:

[0038]

[0039] Among them, S N is the rated capacity of the battery, I 1 is the polarization resistance current, h is the battery charge and discharge efficiency, β k =exp(-1 / t(k)), t=R 1 C 1 is the battery time constant;

[0040] The OCV of the battery is the terminal voltage of the battery under long-term static conditions. It has a relatively fixed functional relationship with the battery SOC, which is recorded as OCV = f(SOC), q = N(0,Q), r = N(0,R) is Gaussian white noise;

[0041] The I of formula (5) 1 (k), S oc (k), R 0 (k), C 1 (k) and R 1 (k) The expression is expressed as:

[0042] x k =g k-1 (x k-1 )+q k-1 ;

[0043] The expression of U(k) is written as:

[0044] y k =h k(x k )+r k ;

[0045] x=[I 1 S oc R 0 C 1 R 1 ];

[0046] y=U;

[0047] G k =g k '(x)|x=x k ;

[0048] H k =h k '(x)|x=x k ;

[0049] The iterative formula (1) is obtained by the battery parameter and SOC joint estimation method based on the extended Kalman filter.

[0050] Furthermore, in step S140, the method for obtaining formula (2) is:

[0051] In a reconfigurable battery network,

[0052] When the parallel switch Sj of the jth parallel unit is closed, all the series switches Sij of the M battery modules are disconnected, where 1≤i≤m. At this time, the output voltage of the jth parallel unit is 0;

[0053] When the parallel switch Sj of the jth parallel unit is disconnected, only one series switch among the series switches Sij of the M battery modules is closed, where 1≤i≤m;

[0054] The fluctuation of the port output voltage of the reconfigurable battery network is controlled within the range of the following formula (6):

[0055]

[0056] Among them, U re is the cluster target output voltage or load voltage, e rr is the maximum allowable deviation.

[0057] Furthermore, the value of λ includes:

[0058] When λ∈(0,1), the balancing control part of the battery module considers the impact of the battery balancing control process on the matching load demand of the energy storage system, which can be used for the active balancing control method.

[0059] When λ=1, the balancing process of the battery module has the same effect on the energy storage system meeting the load voltage requirements, and can be used for energy storage systems where the balancing current of the battery module is equal to the working current of the energy storage system. Since the battery topology network transformation is used to meet the balance between battery modules and the voltage matching of the energy storage system to the load, the reconfigurable battery network system is suitable for the case of λ=1.

[0060] When λ∈(1,+∞), the extreme case of λ∝+∞ is considered, and the impact of the balancing control between battery modules on the load matching of the energy storage system is ignored.

[0061] Furthermore, in step S150, the method for obtaining formula (3) is:

[0062] The battery is calibrated for available capacity, SOC and resistance and capacitance parameters, which are recorded as SN ij , soc ij (0), R ij (0) and C ij (0);

[0063] The available capacity of the battery SN ij The resistance and capacitance parameters R and C are constant within a fixed time interval and are functions of soc and temperature T. Then, the battery charge can be modeled as:

[0064] O cvij (τ m+1 )≈O cvij (τ m )-k ij (τ m )ΔS ocij (τ m )…………………………(7);

[0065] Among them, Q represents the charge of the battery, and τ is the discrete time interval;

[0066] When the same amount of charge flows, the battery SOC changes differently, and the battery capacity difference matrix (8) is constructed:

[0067]

[0068] in:

[0069] Snorm is the standard capacity of the battery cell, which is taken as the average value of the sum of all battery capacities in the selected cluster;

[0070] The discharge rate ρ is defined as:

[0071]

[0072] When the same amount of electricity Q flows through each unit in the network, the SOC change of each unit can be expressed as:

[0073] S ocij (τ m+1 )=S ocij (τ m )-ρD(i, j)……………………(10);

[0074] Assuming that the jth series cell is in the on state, the output voltage of the jth series cell is u j (t), and within the time u of formula (11) j (t) value is basically unchanged, that is, u j (t)≈u j (τm). Through system hardware design, the transient process of the capacitor and the voltage change caused by the transient process of the capacitor can be ignored. Therefore, within the τ time slot, the output voltage of the battery is expressed by formula (12):

[0075]

[0076] The system optimization objective function is simplified to minimizing the sum of the product of the absolute value of the remaining SOC and the voltage deviation during charging / discharging and the proportional factor λ.

[0077] Furthermore, in step S210, the method for establishing a reconfigurable battery network control architecture further includes:

[0078] The retired power battery nodes communicate with each other bidirectionally through the communication network. All such communication relationships are described by an undirected graph containing the connection relationships between agents, which is recorded as: G = {N, E};

[0079] Where N = {v 1 , v 2 , ..., v n} represents the set of all agents in the communication network, E = {ε 1 , ε 2 , ..., ε m} represents the set of all communication connections in the communication network, where a group communication connection is an unordered set of two agents;

[0080] If {v i , v j}∈E, then there is a communication relationship between agents i and j, and the set of all neighbors of agent i is Ω i = {v j |{v i , v j}∈E, j=1,...,n}, the total number of neighbors of agent i is n i , ni =|Ω i |;

[0081] For an undirected graph G, let A represent the association matrix of graph G. ij ) n×n , if {v i , v j}∈E, then α ij =1, otherwise, α ij =0.

[0082] Further, in step S222, the method for configuring AOE includes: AOE network declaration, AOE node declaration and AOE edge declaration.

[0083] Further, in step S225, the method of parsing and executing the AOE network includes:

[0084] S2251. Save the policy obtained after parsing the AOE network configuration file as an AOE vector, where the AOE vector contains one or more AOE networks;

[0085] S2252, parsing the measuring point configuration file and the channel configuration file to allocate memory space for storing the measuring point data, and establishing a communication connection according to the communication mode declared in the channel configuration file;

[0086] S2253, according to the trigger condition of the AOE network obtained by the analysis, the start of the control strategy is determined, and the AOE network using simple repetition drive and time drive is executed according to the trigger condition parameters of the AOE network defined in the AOE network configuration file; and for the AOE network using event drive, the event condition of the first node is used as the condition for whether the entire AOE network is triggered, and the first node represents a node with an in-degree of 0 in the AOE network;

[0087] S2254, monitor the change of the measurement point value according to the attributes and the starting address in the measurement point configuration file; when the measurement point value changes and causes the first node event to be triggered, the AOE network starts and executes control;

[0088] S2255, the AOE network pops up a head node from the current topological sorting queue, and waits until an event of the head node occurs or times out; the head node refers to the node at the front of the current topological sorting queue;

[0089] S2256. When the event of the pop-up node times out, if the timeout is not due to receiving an external AOE cancellation instruction or failure, return to step S2255;

[0090] S2257, when a node event occurs, execute all actions triggered by the event, that is, the actions on all edges with the node as the head node in the AOE network are executed, and wait for the return result of the action execution. For actions with unsuccessful results, take measures such as ignoring, retrying, stopping the affected nodes, and stopping AOE execution according to the failure mode of each action;

[0091] S2258. When all actions are executed, if the topological sorting queue is not empty, return to step S2255, otherwise it means that all nodes have been popped out, the AOE network is executed, and the control process ends.

[0092] Furthermore, in step S230, the method for optimizing and allocating the dual-layer control includes:

[0093] S231, select the best discharge battery pack. The principle of selecting the battery pack can be described as:

[0094] max(SOC 1 ,SOC 2 ,L,SOC n );

[0095] S232, judging the discharge power, described as:

[0096]

[0097] Among them, P D_limit Represents the maximum allowable discharge power of the energy storage battery pack at time t; P D_ max represents the maximum discharge characteristic determined by the materials constituting the energy storage battery pack; E(t-1), E min They represent the remaining power of the energy storage battery pack at the last moment and the minimum value of the remaining power allowed; η D represents the discharge efficiency; Δt represents the calculation window time;

[0098] Determine the maximum allowable discharge power P of the selected optimal discharge battery pack D_limit (t) and command power P ref Size relationship:

[0099] When P D_limit (t)>P ref , then the optimal battery pack is selected for discharge;

[0100] When P D_limit (t)<P ref , repeat step S231, select the next target battery pack from the remaining battery packs, and compare the sum of the maximum allowable discharge power with the command power until the command power is met;

[0101] S233, determine the number m of selected discharge battery groups, and evenly divide the power among the energy storage battery groups. The calculation formula is:

[0102]

[0103] To achieve optimal allocation of lower-layer power, it can be expressed as:

[0104]

[0105] in, represents the SOC of the i-th energy storage battery module at the current moment, P Di represents the discharge power of the i-th energy storage battery module, E i Indicates the remaining power of the i-th energy storage battery module; SOC ave Indicates the average SOC of all energy storage battery modules.

[0106] Furthermore, the constraints of the two-layer control are:

[0107] Discharge power constraint: 0≤P Di ≤P D_limit (t);

[0108] SOC Constraint: SOC i，min ≤SOC i ≤SOC i，max ;

[0109] Total output constraint:

[0110] Compared with the prior art, the digital lossless reorganization method of retired power batteries in a reconfigurable battery network described in the present invention has the following advantages:

[0111] The advantage of this technical solution is that it adopts a method of lossless reorganization of the design of a reconfigurable battery network topology and control of the reconfigurable network, and uses the optimal charging and discharging battery pack selection as the upper-level control strategy to enable the energy storage battery pack to intermittently undertake the charging and discharging tasks, without the need for all battery packs to undertake the charging and discharging tasks in real time, effectively reducing the number of charging and discharging times. Using power optimization allocation as the lower-level control strategy, SOC balance between energy storage battery modules is achieved, overcharging and over-discharging of energy storage batteries is avoided, and battery aging is slowed down. The control strategy is implemented based on AOE, and the implementation process is simple and flexible, with high versatility, high computational efficiency, and low learning difficulty. BRIEF DESCRIPTION OF THE DRAWINGS

[0112] Figure 1 is a schematic diagram of a reconfigurable battery network cluster topology of the present invention;

[0113] Figure 2 A schematic diagram of a reconfigurable battery network topology optimization control framework of the present invention;

[0114] Figure 3 A schematic diagram of the Thevenin model of a lithium-ion battery cell of the present invention;

[0115] Figure 4 A schematic diagram of any parallel unit of the reconfigurable battery network cluster topology of the present invention;

[0116] Figure 5 It is a curve diagram of the relationship between the battery SOC and the number of reconstruction times (discharge) when Ure=85V of the present invention;

[0117] Figure 6 It is a curve diagram of the relationship between the battery SOC and the number of reconstruction times (discharge) when Ure=90V of the present invention;

[0118] Figure 7 is a curve diagram of the relationship between the output voltage and the number of reconstruction times (discharge) when σ1=0.05 of the present invention;

[0119] Figure 8 is a curve diagram of the relationship between the output voltage and the number of reconstruction times (discharge) when σ1=0.2 of the present invention;

[0120] Fig. 9 It is a schematic diagram of the architecture of the distributed energy management and control system for retired power batteries of the present invention;

[0121] Fig.10 It is a flow chart of power distribution control between battery modules of a reconfigurable battery network of the present invention;

[0122] Fig.11 It is the AOE topology diagram of the reconfigurable battery network control strategy of the present invention;

[0123] Fig.12 A graph showing the command power of the reconfigurable battery network and the output power of the energy storage power station of the present invention;

[0124] Fig.13 A curve diagram showing the change in the working state of the energy storage battery module of the first embodiment of the reconfigurable battery network of the present invention;

[0125] Fig.14 is a curve diagram of the output power of the first embodiment of the reconfigurable battery network of the present invention;

[0126] Fig.15 A curve diagram showing the change in the working state of the energy storage battery module of the second embodiment of the reconfigurable battery network of the present invention;

[0127] Fig.16 is a curve diagram of the output power of the second embodiment of the reconfigurable battery network of the present invention;

[0128] Fig.17A curve diagram of the SOC change of the double-layer control strategy of the energy storage battery module of the first embodiment of the reconfigurable battery network of the present invention;

[0129] Fig.18 A curve diagram of the SOC change of the double-layer control strategy of the energy storage battery module of the second embodiment of the reconfigurable battery network of the present invention;

[0130] Fig.19 It is a curve diagram of the SOC change of the reconfigurable battery network energy storage battery module equalization control strategy of the present invention. DETAILED DESCRIPTION

[0131] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0132] The technical solutions between the various embodiments can be combined with each other, but it must be based on the fact that ordinary technicians in the field can implement it. When the technical solutions between the embodiments can be combined, they are all within the protection scope required by the present invention.

[0133] The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0134] A digital lossless reorganization method for retired power batteries of a reconfigurable battery network, comprising:

[0135] S100, design reconfigurable battery network topology:

[0136] S110, establishing a reconfigurable battery network:

[0137] like Figure 1 As shown, the reconfigurable battery network includes a battery energy storage system composed of K battery clusters, each battery cluster is composed of N parallel units connected in series, each parallel unit includes a battery module and a controllable switch, the controllable switch includes a parallel switch and a series switch, each battery module is composed of a battery cell connected in series with a series switch, and M battery modules are connected in parallel with a parallel switch to form a parallel unit;

[0138] The design and operation of a dynamically reconfigurable battery network (DRBN) system involves dynamic optimization and control of complex networks. The dynamic generation of the optimal topology of the battery network is an important factor affecting the characteristics of large-scale battery networks. The energy storage system dynamically reconstructs the topology between battery cells by controlling the state of the switch, forming a topological connection strategy that meets the actual application requirements under different control objectives. How to efficiently generate an energy storage system topology that helps balance the battery modules through the coupled collaborative control of the battery module and the power electronic topology, realize differentiated battery management and avoid the "short board effect" of the system, is an urgent problem to be solved for the safe and reliable operation of the battery energy storage system.

[0139] S120, establish an optimized control architecture:

[0140] like Figure 2 As shown in FIG. 1 , by measuring the terminal voltage U and terminal current I of each parallel unit in the dynamic reconfigurable battery network, a measurement state set M = {U, I} is formed. The parameters and states of each battery in the dynamic reconfigurable battery network are jointly estimated, including the battery state of charge (SOC), resistance and capacitance parameters, to form a state and parameter set S = {R 0 , R 1 , C 1 , SOC}, where R 0 Represents the equivalent internal resistance of the battery, R 1 Represents polarization internal resistance, C 1 Represents polarized capacitance. The topological connection control variable Z of the reconfigurable battery network is optimized and controlled to generate a battery network topology generation scheme that meets the optimization objectives and constraints;

[0141] Based on this optimization control framework, we first study the battery state estimation and key parameter extraction methods based on reconfigurable battery networks and equivalent circuit models. Then, we take the balance between battery modules in the energy storage system as the optimization goal, so that the state set S of the energy storage system satisfies the physical constraints. The topology of the DRBN is found from the optimization feasible domain, and finally, the balancing effect of the algorithm is verified through a semi-physical simulation platform.

[0142] S130, estimating the battery equivalent circuit model and parameter status:

[0143] In a reconfigurable battery network, the battery module can be disconnected from the energy storage system, and the terminal voltage, current and other electrical parameters of the battery module can be actually measured. Therefore, in the battery network topology optimization, the reconstruction process of the battery network needs to consider the electrical properties of the battery module. There are many ways to model lithium-ion batteries. Since the Thevenin model is consistent with the actual electrical properties of the battery and the calculation modeling is simple, the Thevenin equivalent circuit model of the battery module is established. The equivalent Thevenin model of the battery and the reference direction of voltage and current are as follows: Figure 3 As shown, the equation of the continuous time of the battery module equivalent Thevenin model is written, and its mathematical expression is:

[0144]

[0145] Among them, the open circuit voltage of the battery is U ocv (t) is a variable related to the battery dynamic state of charge SOC(t), and can be expressed in ampere-hour integration method. SOC 0 is the initial state value. The battery current i(t) is affected by the battery polarization capacitance C 1 and the voltage u of the polarized capacitor C The battery terminal voltage u(t), u C (t), U ocv (t) and equivalent internal resistance R 0 The voltage forms a closed voltage loop.

[0146] The state and parameter estimation method of lithium-ion battery based on extended Kalman filter is expressed by equation (5) as the discrete state and observation equation of lithium-ion battery, and the discrete time is Δt=1s:

[0147]

[0148] Among them, S N is the rated capacity of the battery, I 1 is the polarization resistance current, h is the battery charge and discharge efficiency, β k =exp(-1 / t(k)), t=R 1 C 1 is the battery time constant;

[0149] The OCV of the battery is the terminal voltage of the battery under long-term static conditions. It has a relatively fixed functional relationship with the battery SOC, which is recorded as OCV = f(SOC), q = N(0,Q), r = N(0,R) is Gaussian white noise;

[0150] The I of formula (5) 1 (k), S oc (k), R 0 (k), C 1(k) and R 1 (k) The expression is expressed as:

[0151] x k =g k-1 (x k-1 )+q k-1 ;

[0152] The expression of U(k) is written as:

[0153] y k =h k (x k )+r k ;

[0154] x=[I 1 S oc R 0 C 1 R 1 ];

[0155] y=U;

[0156] G k =g k '(x)|x=x k ;

[0157] H k =h k '(x)|x=x k ;

[0158] The mathematical model of the battery parameter and SOC joint estimation method based on the extended Kalman filter is expressed by the following iterative formula (1):

[0159]

[0160] Among them, P k is the estimated error covariance matrix, and E is the unit matrix. By continuously observing the variable assignment, the battery SOC can be obtained;

[0161] S140. Establish a dynamic optimization model for a reconfigurable battery network:

[0162] The battery cluster needs to follow continuous-time control constraints during operation. When establishing the continuous-time model of the cluster under this operation strategy, without loss of generality, we assume that the switches are ideal switches.

[0163] In reconfigurable battery networks, such as Figure 4 As shown,

[0164] When the parallel switch S of the jth parallel unit j When closed, all series switches S of the M battery modules ijAll are disconnected, where 1≤i≤m. At this time, the output voltage of the jth parallel unit is 0;

[0165] When the parallel switch S of the jth parallel unit j When disconnected, the series switch S of the M battery modules ij Only one series switch is closed, where 1≤i≤m;

[0166] The fluctuation of the port output voltage of the reconfigurable battery network is controlled within the range of the following formula (6):

[0167]

[0168] Among them, Ure is the target output voltage or load voltage of the cluster, and err is the maximum allowable deviation. The selection of Ure is related to the composition of the battery cluster, such as the number of reconstructed batteries in the cluster, the external characteristics of the battery, etc. Ure can be set to the optimal estimated value that satisfies a certain utilization rate of the battery in the cluster, indicating the output voltage value that can be achieved in the rated state.

[0169] In order to make each battery module in the battery cluster achieve SOC balance while maintaining the output voltage basically unchanged, so that they can be charged and discharged at the same time, the objective function can be the sum of the battery SOC variance and the voltage deviation multiplied by the proportional factor λ, which is shown in the first line of formula (2). Among them, the value of λ is used to balance the effect of battery module SOC consistency management and energy storage system matching external voltage requirements. According to the proportional relationship, it can be divided into the following situations:

[0170] When λ=0, the objective function only performs the balancing control of the battery module, and does not consider the matching of the energy storage system to the external voltage during the balancing process of the battery module. This situation can be used when the balancing current of the battery module is small and the balancing of the battery module has little effect on the matching of the energy storage system to the external load. This method is common in passive balancing control.

[0171] When λ∈(0,1), the balancing control part of the battery module considers the impact of the battery balancing control process on the matching load demand of the energy storage system, which can be used for the active balancing control method.

[0172] When λ=1, the balancing process of the battery module has the same effect on the energy storage system to meet the load voltage requirements, and can be used for energy storage systems where the balancing current of the battery module is equal to the working current of the energy storage system. Since the battery topology network transformation is used to meet the balance between battery modules and the voltage matching of the energy storage system to the load, the reconfigurable battery network system is suitable for the case of λ=1.

[0173] When λ∈(1,+∞), the extreme case of λ∝+∞ is considered, and the impact of the balancing control between battery modules on the load matching of the energy storage system is ignored. This case is equivalent to only controlling the output voltage when the energy storage system is running, which is suitable for situations where only external load requirements are met.

[0174] The dynamic optimization model of the battery cluster can be established as:

[0175]

[0176] Among them, z ij is the switch state, taking the value of {0,1}; L(i,j), U(i,j) is the SOC of the battery module in the i-th row and j-th column of the m-parallel and n-series battery cluster at the end of charge / discharge. The above problem is a large-scale nonlinear dynamic optimization problem, fij is a nonlinear function, and is mixed with a large number of 0 and 1 variables, so it is difficult to obtain an analytical solution for the above problem.

[0177] S150, reconfigurable battery network model solution:

[0178] The battery is calibrated for available capacity, SOC and resistance and capacitance parameters, which are recorded as SN ij , soc ij (0), R ij (0) and C ij (0). The above problem is a large-scale nonlinear dynamic optimization problem. fij is a nonlinear function and contains a large number of 0 and 1 variables. Therefore, it is difficult to obtain an analytical solution for the above problem. The available capacity of the battery SN ij The resistance and capacitance parameters R and C are constant within a fixed time interval and are functions of soc and temperature T. Then, the battery charge can be modeled as:

[0179] O cvij (τ m+1 )≈O cvij (τ m )-k ij (τ m )ΔS ocij (τ m )…………………………(7);

[0180] Among them, Q represents the charge of the battery, and τ is the discrete time interval;

[0181] Since the battery capacities in the battery cluster are different, when the same amount of charge flows through, the battery SOC changes differently. The battery capacity difference matrix (8) is constructed:

[0182]

[0183] in:

[0184] Snorm is the standard capacity of the battery cell, which is taken as the average value of the sum of the capacities of all batteries in the selected cluster. The discharge rate ρ is defined as:

[0185]

[0186] When the same amount of electricity Q flows through each unit in the network, the SOC change of each unit can be expressed as:

[0187] S ocij (τ m+1 )=S ocij (τ m )-ρD(i, j)……………………(10);

[0188] Usually, in actual operation, the current i(t) flowing through the battery network changes continuously, and the change of i(t) does not appear suddenly. Therefore, when Q is selected reasonably, the reconstruction time interval of the battery network is in the order of seconds or minutes. Assuming that the jth series cell is in the on state, the output voltage of the jth series cell is u j (t), and within the time u of formula (11) j (t) value is basically unchanged, that is, u j (t)≈u j (τm). Through system hardware design, the transient process of the capacitor and the voltage change caused by the transient process of the capacitor can be ignored. Therefore, within the τ time slot, the output voltage of the battery is expressed by formula (12):

[0189]

[0190] The system optimization objective function is simplified to minimizing the sum of the product of the absolute value of the remaining SOC and the voltage deviation during charging / discharging and the proportional factor λ.

[0191] In summary, by transforming the complex nonlinear optimization model in continuous time into a linear {0,1} programming model in discrete time (the absolute value objective function can be transformed into a linear one), when the battery network is running, the objective function, constraints and recursive relationship of the optimization problem are summarized as formula (3):

[0192]

[0193] Validation of Reconfigurable Battery Network Optimization Model

[0194] The reconfigurable battery network optimization model is verified through the following example.

[0195] The experimental verification environment was built using retired power battery modules from Yangtze River pure electric vehicles. The battery module uses a single cell of 3.2V / 13.5Ah, and 40 series and 1 parallel form the minimum battery module. The nominal voltage of each battery module is 16V. The system consists of 69 battery modules connected in 23 series and 3 parallels, with a total of 2070 single cells. The circuit topology of the system is as follows: Figure 1 As shown, where n=40 and m=1.

[0196] During the operation, the SOC-OCV fitting function was linearized piecewise and divided into 80 segments according to the SOC from 0.1 to 0.9. During the experiment, the battery module was tested and the ambient temperature was controlled at 25°C. During the experiment, the Arbin device was used to load the battery with current and record parameters such as voltage, current, and temperature. The average data recording interval was 1s.

[0197] Without loss of generality, in this experiment, the SOC upper limit U = 0.9 and the SOC lower limit L = 0.2 of all battery modules are set; the battery network output voltage and load voltage U re Maximum deviation e rr Take it as 0.05, and the proportional factor λ is flexibly selected as 1. The balancing effect of all battery modules in the discharge process of the battery energy storage system is taken as the main research object, and different load demands U re The impact of the difference σ between the initial battery module and the energy storage system balancing effect.

[0198] Set the discharge rate ρ = 0.01, and measure the statistical average μ of the battery module SOC to be 0.65, and the statistical standard deviation σ to be 0.05. Set two different load requirements, namely, set U re The values ​​are 90V and 85V to verify the balancing effect of the reconfigurable battery network topology reconstruction measurement on the battery module in the energy storage system under different load voltage requirements. The curve experimental results of the battery module SOC and the number of reconstructions are shown in Figure 1. Figure 5 , Figure 6 The experimental results show that under the premise of keeping the system output voltage unchanged, the battery cluster balancing time undergoes about 50 reconstructions, and the SOC between battery modules is well balanced; at the same time, it is verified that the system has good adaptability to the 85V load voltage and can achieve battery module balance during 30 reconstructions.

[0199] In order to verify the compatibility of the energy storage system with the differences of battery modules, the discharge rate ρ is set to 0.01 and the load demand voltage U re The value is 90V. The statistical average μ of the SOC of the same battery module is 0.65, and the statistical standard deviation σ is 1 The impact of σ2 on the balancing effect of the energy storage system is 0.05 and σ2 is 0.2. As the number of reconstructions increases, the SOC statistics of the battery module are as follows: Figure 7 , Figure 8 As shown in the figure. The experimental results show that as the differences of battery modules increase, the system can achieve the balance of all battery modules due to the balance control effect of the battery modules. And through reasoning, if the statistical standard deviation σ of the battery module increases, and at the same time, if you want to get a better balance effect, you need to reduce the output voltage of the cluster.

[0200] S200, control of reconfigurable battery network:

[0201] S210. Establish a reconfigurable battery network control architecture:

[0202] In view of the high safety requirements of the retired battery system for cascade utilization, the distributed battery energy management and control system architecture is studied based on the reconfigurable battery network technology.

[0203] At present, the energy management and control systems for massive differentiated retired power batteries mostly adopt centralized control, which has high processing pressure on the core unit, low fault tolerance, and is limited by computing performance, monitoring data collection, communication and other conditions. Compared with centralized control, the distributed energy management and control system has the characteristics of providing local real-time computing capabilities, avoiding the bandwidth cost and delay problems caused by frequent interaction with cloud computing centers. It is based on a series of local controllers, with low reliability requirements for a single controller, reducing the requirements for cloud computing centers and communication systems, and strengthening the control capabilities of retired power batteries.

[0204] The present invention designs a distributed energy management and control system architecture for retired power batteries, as follows: Fig. 9 As shown in the figure, the bottom layer is the retired power battery system. Each local controller obtains the status information of each power battery through status measurement and uploads it to the communication network. The communication network nodes communicate with neighboring nodes, and through mutual cooperation and joint decision-making, the distributed energy management of the retired power battery system is realized, and then the control instructions are issued to the retired power battery through the local controller.

[0205] Among them, multiple functions such as data collection and processing, on-site analysis and decision-making, and operation status detection are integrated into the local controller to maximize the natural advantages of information capture accuracy and decision-making response speed.

[0206] The architecture has the following features:

[0207] (1) Decentralization: A fully distributed architecture is adopted to realize data exchange between the distributed energy management and control systems of retired power batteries based on the communication network.

[0208] (2) Distributed collaborative optimization algorithm: Design a distributed optimization model and a high-convergence fast iterative solution algorithm to achieve collaborative optimization.

[0209] (3) Communication method and function design: Establish a device communication framework that takes into account communication delay, communication error, and redundant control.

[0210] The retired power battery nodes communicate with each other bidirectionally through the communication network. All such communication relationships are described by an undirected graph containing the connection relationships between agents, which is recorded as: G = {N, E};

[0211] Where N = {v 1 , v 2 , ..., v n} represents the set of all agents in the communication network, E = {ε 1 , ε 2 , ..., ε m} represents the set of all communication connections in the communication network, where a group communication connection is an unordered set of two agents;

[0212] If {v i , v j}∈E, then there is a communication relationship between agents i and j, and the set of all neighbors of agent i is Ω i = {v j |{v i , v j}∈E, j=1,...,n}, the total number of neighbors of agent i is n i , n i =|Ω i |;

[0213] For an undirected graph G, let A represent the association matrix of graph G. ij ) n×n , if {v i , v j}∈E, then α ij =1, otherwise, α ij = 0. represents the degree matrix of the graph, where is the number of edges connected to node i. Then the Laplace matrix of G is.

[0214] The Laplace matrix L is a semi-positive definite matrix that can reflect the connectivity of the graph

[433] ,

[434] . The number of 0s in the L eigenvector represents the number of connected components in the graph G. The next smallest eigenvalue represents the tightness of the communication network connection. This study assumes that the communication network formed by the energy storage in the microgrid contains only one connected component, that is, the L eigenvector contains only one 0.

[0215] S220, Energy storage strategy arrangement based on AOE control configuration:

[0216] S221. Construct an AOE network. The AOE network is a directed acyclic graph. The AOE includes multiple nodes. Every two nodes are connected by an edge. Each node represents an event, and each edge represents an action. The direction of the edge represents the progressive and migration relationship between the events represented by the nodes. This relationship is determined by the action represented by the edge. The variables that actions and nodes rely on in calculation can only come from measurement points, variable definitions, or variables calculated in actions that have been executed.

[0217] After determining the order of each action and the events in which the action occurs and is completed, use nodes to represent the events. The actions executed due to the event are directed edges starting from this node. The node pointed to by each directed edge represents the event after the action is completed. Organize all nodes and edges according to the above logic to form an AOE network.

[0218] S222. Configure AOE. AOE configuration is to establish an AOE network to implement the required control logic and functions. The AOE network configuration file is used to define the basic information, variables, events and actions of the AOE network, thereby describing the constructed AOE network and the designed control strategy. File configuration is completed in a standardized format by entering table information, without the need to write code, so it is a low-code expression method.

[0219] 1) AOE Network Statement

[0220] The AOE network declaration part defines the number of AOE networks in the table, overall information, and some variable initial values. The trigger conditions are divided into three categories: SimpleRepeat (timing trigger, you need to set the timing yourself), TimeDriven (time-driven), and EventDriven (event-driven); the definition of variable initial values ​​is: <variable name 1>: <initial value 1>; <variable name 2>: <initial value 2>; ...

[0221] 2) AOE node declaration

[0222] Node types are divided into two categories: Switch and Condition. The Switch node is a two-branch logic judgment, and the trigger logic is: when the expression is true, the edge with sequence number 1 of the node performs an action; when the expression is false, the edge with sequence number 2 performs an action. When the Switch node is a node with an AOE network in-degree of 0 (the first node), the Switch node is directly triggered. If the expression is true within the timeout period, action 1 is executed. If it is false or a timeout event occurs, action 2 is executed. The expression of the Condition node is the event triggering condition. If the expression is true within the timeout period, the node event occurs, and all actions initiated by the node are executed in parallel.

[0223] 3) AOE side declaration

[0224] AOE edge declaration is to divide the control strategy into several actions according to the order of execution. The edge needs to declare the first and last nodes in the format of <last node ID>; <first node ID>. The direction of the edge is: last node → first node. The action types involved in this method are the following two: SetPoints, which means setting points. This action can set the measurement point value to a digital quantity or define the variable as an analog quantity; Milp, which means solving sparse mixed integer linear programming problems.

[0225] S223. Configure the measuring point. The measuring point configuration is used to configure the parameters of the controller input and output signal points. The writing of the measuring point configuration file is to list the attributes required to describe a measuring point. The content of the measuring point configuration table includes the point number, name, alias, whether it is discrete, whether it is a calculation point, and the default value of the measuring point. The point number is the number given to this measuring point in the program. You can define it yourself. When using it, the measuring point and the point number of the communication channel correspond one by one to complete the docking. The alias is a combination of English letters and underscores. It can be used as a variable name for writing expressions in the AOE table, so that it can be expressed without a period, making the meaning of the expression clearer. The format of the measuring point configuration file is shown in Table 1.

[0226] Table 1

[0227]

[0228] S224, configure the communication channel. The configuration of the communication channel is to enable the controller to establish modbus communication with the energy storage battery and realize information interaction between the energy storage battery and the controller. The channel configuration file is used to describe the communication method between the controller and the controlled object, and according to the measurement point attributes in the measurement point configuration file, the register information and address for storing the measurement point value are given. The channel configuration file includes two parts: communication mode configuration and measurement point register configuration. Modbus, IEC104, MQTT, and serial port are supported as communication methods. The communication mode configuration includes server address, server port, subscription topic, and push topic; serial port communication is adopted, and the communication mode configuration includes port number, baud rate, check bit, data bit, and stop bit; the measurement point register configuration is used to give the address information for storing the measurement point data, including serial number, register type, starting address, data type, new request flag, polling cycle, and point number.

[0229] Taking the channel using Modbus communication as an example, the channel configuration file includes channel name, number of connections, service port, connection name, number of measurement points, client IP, client port, slave ID, communication protocol, upper limit of the number of registers read at one time, upper limit of the number of switches read at one time, upper limit of the number of registers written at one time, upper limit of the number of switches written at one time, (default) polling cycle (ms), timeout (ms), channel status measurement point number, sequence number, register type, start address, data type, new request flag, polling cycle, and point number. The channel name is represented by a string, which can contain Chinese characters; the number of connections is represented by a positive integer; the service port is represented by a positive integer; the connection name is represented by a string; the number of measurement points is represented by a positive integer; the client IP is represented by a string in the IPv4 address format; the client port is a positive integer; the communication protocol is represented by one of XA and ENCAP, XA and ENCAP represent two communication protocols in Modbus communication (XA represents the Modbus TCP protocol, corresponding to the Modbus TCP in the slave; ENCAP is the most primitive Modbus RTU, which means that only TCP is used as the carrier byte organization, corresponding to the Modbus RTU over in the slave TCP); the upper limit of the number of registers read at one time is expressed by a positive integer; the upper limit of the number of switches read at one time is expressed by a positive integer; the upper limit of the number of registers written at one time is expressed by a positive integer; the upper limit of the number of switches written at one time is expressed by a positive integer; the default polling period (ms) is expressed by a positive integer; the timeout (ms) is expressed by a positive integer; the channel status measurement point number is expressed by a positive integer; the sequence number is expressed by a positive integer, and the measurement points are generally numbered continuously starting from 1; the register type is expressed by one of the four COILS, DISCRETE, HOLDING and INPUT, among which COILS represents the coil register, which can actually be compared to the switch quantity, DISCRETE represents the discrete input register, which can only read the switch quantity, which is equivalent to the read-only mode of the coil register, HOLDING represents the holding register, the unit of this register is no longer bit but two bytes, which can store a specific amount of data and is readable and writable, INPUT represents the input register, which can be regarded as a holding register in read-only mode; the starting address is expressed by a positive integer, and the address offset of the measurement point is determined according to the data type.

[0230] S225, parse and execute the AOE network, parse the AOE network configuration file, channel configuration file and measurement point configuration file, and apply the control strategies described in these files to the actual industrial control scenario to achieve industrial control. Specifically:

[0231] S2251. Save the policy obtained after parsing the AOE network configuration file as an AOE vector, where the AOE vector contains one or more AOE networks;

[0232] S2252, parsing the measuring point configuration file and the channel configuration file to allocate memory space for storing the measuring point data, and establishing a communication connection according to the communication mode declared in the channel configuration file;

[0233] S2253, according to the trigger condition of the AOE network obtained by the analysis, the start of the control strategy is determined, and the AOE network using simple repetition drive and time drive is executed according to the trigger condition parameters of the AOE network defined in the AOE network configuration file; and for the AOE network using event drive, the event condition of the first node is used as the condition for whether the entire AOE network is triggered, and the first node represents a node with an in-degree of 0 in the AOE network;

[0234] S2254, monitor the change of the measurement point value according to the attributes and the starting address in the measurement point configuration file; when the measurement point value changes and causes the first node event to be triggered, the AOE network starts and executes control;

[0235] S2255, the AOE network pops up a head node from the current topological sorting queue, and waits until an event of the head node occurs or times out; the head node refers to the node at the front of the current topological sorting queue;

[0236] S2256. When the event of the pop-up node times out, if the timeout is not due to receiving an external AOE cancellation instruction or failure, return to step S2255;

[0237] S2257, when a node event occurs, execute all actions triggered by the event, that is, the actions on all edges with the node as the head node in the AOE network are executed, and wait for the return result of the action execution. For actions with unsuccessful results, take measures such as ignoring, retrying, stopping the affected nodes, and stopping AOE execution according to the failure mode of each action;

[0238] S2258. When all actions are executed, if the topological sorting queue is not empty, return to step S2255, otherwise it means that all nodes have been popped out, the AOE network is executed, and the control process ends.

[0239] S230, dual-layer control of power optimization allocation of energy storage system based on AOE.

[0240] Considering the retired battery system, each energy storage battery pack has a large difference in charge state due to different charging and discharging operations. In order to reduce the number of charge and discharge cycles, prevent overcharging and over-discharging, and maintain safe and stable operation of the battery pack, the present invention proposes a two-layer control strategy for energy storage system power optimization allocation based on AOE. The upper layer takes the selection of the optimal working battery pack as the objective function, and selects the optimal charge and discharge battery pack according to the relationship between the command power and the actual charge and discharge power of the energy storage battery pack. The lower layer takes the SOC balance as the objective function to achieve power distribution between battery modules. The control flow chart is as follows: Fig.10 shown.

[0241] The following takes the command power, that is, the energy storage battery pack is in the discharge working state as an example to describe the upper control strategy in detail.

[0242] S231. Select the optimal discharge battery pack. Considering that the energy storage system has a large number of battery packs, it is particularly important to select the optimal discharge battery pack based on the command power and the battery pack discharge capacity in order to reduce the number of discharges. Since the battery discharge capacity is inversely proportional to the SOC, the principle of selecting the battery pack can be described as:

[0243] max(SOC 1 ,SOC 2 ,L,SOC n );

[0244] S232, determine the discharge power. The maximum allowable discharge power of the energy storage battery pack is determined by its own discharge characteristics and the remaining power, which can be described as:

[0245]

[0246] Among them, P D _ limit Represents the maximum allowable discharge power of the energy storage battery pack at time t; P D _max represents the maximum discharge characteristic determined by the materials constituting the energy storage battery pack; E(t-1), E min They represent the remaining power of the energy storage battery pack at the last moment and the minimum value of the remaining power allowed; η D represents the discharge efficiency; Δt represents the calculation window time;

[0247] Determine the maximum allowable discharge power P of the selected optimal discharge battery pack D _ limit (t) and command power P ref Size relationship:

[0248] When P D_limit (t)>P ref , then the optimal battery pack is selected for discharge;

[0249] When PD_limit (t)<P ref , repeat step S231, select the next target battery pack from the remaining battery packs, and compare the sum of the maximum allowable discharge power with the command power until the command power is met;

[0250] S233, determine the number m of selected discharge battery groups, and evenly divide the power among the energy storage battery groups. The calculation formula is:

[0251]

[0252] The lower power optimization allocation layer is the optimal discharge battery group and the total power borne by each energy storage battery group obtained according to the upper control strategy. The objective function is to minimize the SOC balance within the energy storage battery group, that is, between the energy storage battery modules, to achieve the lower power optimization allocation, which can be expressed as:

[0253]

[0254] in, represents the SOC of the i-th energy storage battery module at the current moment, P Di represents the discharge power of the i-th energy storage battery module, E i Indicates the remaining power of the i-th energy storage battery module; SOC ave Indicates the average SOC of all energy storage battery modules.

[0255] Discharge power constraint: 0≤P Di ≤P D_limit (t);

[0256] SOC Constraint: SOC i，min ≤SOC i ≤SOC i，max ;

[0257] Total output constraint:

[0258] When the command power changes, the power optimization allocation double-layer control strategy is executed cyclically in the set window time to meet the requirements of the dispatching instructions and coordinate operation with new energy.

[0259] The following is an example of verifying the reconfigurable battery network control method.

[0260] In order to verify the proposed two-layer control strategy for power optimization allocation of energy storage system, this paper takes a certain energy storage power station as an example, uses Simulink to establish a simulation model of the energy storage power station, and implements the control strategy based on AOE to realize the simulation analysis of the energy storage power station.

[0261] The energy storage power optimization distribution control strategy based on AOE can be realized by simply writing an Excel configuration file. It has many advantages such as low difficulty in use, simple writing, intuitive control process, high calculation and operation efficiency, etc. The above control strategy is represented by AOE, such as Fig.11 The following is a detailed introduction to the measurement point configuration, AOE configuration and communication channel configuration.

[0262] The measurement point configuration is mainly used to configure the parameters of the controller input and output signals and define the properties of each measurement point. This example includes the remote control issued by the main control system, that is, the system active command power measurement point, each energy storage battery module startup signal measurement point, SOC measurement point and actual power measurement point, and defines their names, whether they are discrete, whether they are calculation points, and default values. The measurement point configuration file is shown in Table 2.

[0263] AOE configuration is to describe the control strategy by defining the basic information, variables, events and actions of AOE. This example includes 6 nodes and 5 actions.

[0264] Table 2

[0265]

[0266] Node 1 is a Simulink communication event node, which is an event trigger type. When the energy storage system starts working, the node is triggered to start executing the control strategy. Action 1 is to calculate the intermediate quantity, which is executed after the communication event node is triggered to calculate the SOC of the energy storage battery pack. Action 2 is the optimal battery pack selection. According to the SOC of the energy storage battery pack and the objective function, the charging and discharging battery pack is optimized. Action 3 is to solve the power optimization allocation problem of the energy storage battery module. When there is an optimal solution, execute action 4 to issue the optimization strategy. When there is no optimal solution, execute action 5 to issue the default equal distribution strategy. When there is no optimal solution, it means that the command power is unreasonable or the dispatchable capacity of the energy storage is too small.

[0267] The communication channel configuration is used to describe the communication mode between the controller and the Simulink energy storage power station simulation model, and to provide the measurement point register information and address. In this example, the controller is used as a slave station, and Modbus is used to realize the communication between the controller and the simulation model. The measurement point data are all of EightByteFloat data type, occupying 4 bits of registers, and then the address is allocated according to the data type.

[0268] The specific parameters of the simulation model are set as follows: the rated capacity of the energy storage module is 1000kWh, the maximum allowable charging and discharging power is 1500kW, and the maximum and minimum SOC values ​​are 0.9 and 0.1 respectively.

[0269] Run the controller and simulation model to obtain the command power and total power output curve of the energy storage power station as shown below: Figures 12 to 16 As shown in the figure, it can be seen that the actual output power of the energy storage power station can accurately follow the change of the command power. Although the output power fluctuates slightly due to the characteristics of the energy storage battery, its fluctuation value is within 200W.

[0270] like Fig.17 and Fig.18 As shown in the figure, among the energy storage battery groups 1, 2, and 3, the SOC of energy storage battery group 1 is the largest. It can be seen from the figure that according to the upper-level control strategy, when the command power is met, the energy storage battery group 1 is in working state, and the battery groups 2 and 3 are in shutdown state. In the time range of 0.2-0.4h, 0.6-0.8h, 1.4-2.8h, etc., because the command power is greater than the maximum allowable discharge power, the energy storage battery group 2 is put into operation; in the event range of 0.4-0.6h, 0.8-1.4h, 2.8-3.4h, etc., the command power is less than the maximum allowable discharge power, and the energy storage battery group 2 exits operation, realizing the optimal charge and discharge battery group selection. Under the strategy of this subject, the battery group intermittently undertakes the charge and discharge tasks, reducing the number of cycle charge and discharge.

[0271] like Fig.19 As shown, and with Fig.17 and Fig.18 Comparison of power sharing control strategies. From the figure, we can see that the lower-level control strategy achieves reasonable power distribution between energy storage battery modules. After 2 hours, SOC reaches basic balance. Therefore, the strategy of this project achieves power balance between energy storage battery modules, reduces the difference in charge state, avoids overcharging and over-discharging, and slows down the aging of energy storage batteries.

[0272] Through simulation analysis, the following conclusions can be drawn:

[0273] (1) The optimal charging and discharging battery group selection is used as the upper-level control strategy to enable the energy storage battery group to intermittently undertake the charging and discharging tasks. It is not necessary for all battery groups to undertake the charging and discharging tasks in real time, which effectively reduces the number of charging and discharging times.

[0274] (2) With power optimization allocation as the lower-level control strategy, SOC balance between energy storage battery modules is achieved, overcharging and over-discharging of energy storage batteries are avoided, and battery aging is slowed down.

[0275] (3) The control strategy is implemented based on AOE. The control strategy can be written through Excel. The implementation process is simple and flexible. It has many advantages such as high versatility, high calculation efficiency, and low learning difficulty. It is conducive to allowing more industrial users to participate in the dual carbon strategy.

[0276] In the future, we will consider the impact of energy storage costs, balance of energy storage battery utilization times, etc. on the optimal allocation of energy storage power, and further optimize the control strategy.

[0277] Although the present invention is disclosed as above, the present invention is not limited thereto. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the scope defined by the claims.

Claims

1. A digital lossless reorganization method for retired power batteries in a reconfigurable battery network, characterized in that: include: S100, design reconfigurable battery network topology: S110, establishing a reconfigurable battery network: The reconfigurable battery network includes a battery energy storage system composed of K battery clusters, each battery cluster is composed of N parallel units connected in series, each parallel unit includes a battery module and a controllable switch, the controllable switch includes a parallel switch and a series switch, each battery module is composed of a battery cell connected in series with a series switch, and M battery modules are connected in parallel with a parallel switch to form a parallel unit; S120, establish an optimized control architecture: By measuring the terminal voltage U and terminal current I of each parallel unit in the dynamically reconfigurable battery network, a measurement state set M = {U, I} is formed; The parameters and states of each battery in the dynamically reconfigurable battery network are jointly estimated, including the battery state of charge, resistance and capacitance parameters, to form a state and parameter set S = {R0, R1, C1, SOC}, where R0 represents the battery equivalent internal resistance, R1 represents the polarization internal resistance, and C1 represents the polarization capacitance; The topological connection control variable Z of the reconfigurable battery network is optimized and controlled to generate a battery network topology generation scheme that meets the optimization objectives and constraints; S130, estimating the battery equivalent circuit model and parameter status: The Thevenin equivalent circuit model of the battery module is established, and the mathematical model is expressed by the following formula (1): Among them, P k is the estimated error covariance matrix, and E is the unit matrix. By continuously observing the variable assignment, the battery SOC can be obtained; S140. Establish a dynamic optimization model for a reconfigurable battery network: In a reconfigurable battery network, the dynamic optimization model of the battery cluster can be established as: Among them, z ij is the switch state, taking the value of {0,1}; L(i,j), U(i,j) is the SOC of the battery module in the i-th row and j-th column in the m-parallel and n-series battery cluster at the end of charging / discharging; S150, reconfigurable battery network model solution: S200, control of reconfigurable battery network: S210. Establish a reconfigurable battery network control architecture: The bottom layer is the retired power battery system. Each local controller obtains the status information of each power battery through status measurement and uploads it to the communication network. The communication network nodes communicate with neighboring nodes, and through mutual cooperation and joint decision-making, the distributed energy management of the retired power battery system is realized, and then the control instructions are issued to the retired power batteries through the local controllers. S220, Energy storage strategy arrangement based on AOE control configuration: S221, constructing an AOE network, the AOE network is a directed acyclic graph, the AOE includes multiple nodes, each two nodes are connected by an edge, each node represents an event, each edge represents an action, the direction of the edge represents the progressive and migration relationship between the events represented by the nodes, and this relationship is determined by the action represented by the edge; S222, configure AOE, use the AOE network configuration file to define the basic information, variables, events and actions of the AOE network, so as to describe the constructed AOE network and the designed control strategy; S223, configure measuring points; S224, configuring a communication channel; S225, parsing and executing the AOE network; S230, dual-layer control of power optimization distribution of energy storage system based on AOE.

2. The method for digital lossless reorganization of retired power batteries in a reconfigurable battery network according to claim 1 is characterized in that: In step S130, the method for obtaining the mathematical model of formula (1) is: Write down the equation of the continuous time of the equivalent Thevenin model of the battery module, and its mathematical expression is: Among them, the open circuit voltage of the battery is U ocv (t) is a variable related to the battery's dynamic state of charge SOC(t), and can be expressed in ampere-hour integration, with SOC0 being the initial state value. The battery's current i(t) is affected by the battery's polarization capacitor C1 and the voltage u of the polarization capacitor. C The battery terminal voltage u(t), u C (t), U ocv (t) and the equivalent internal resistance R0 voltage form a closed voltage loop; The state and parameter estimation method of lithium-ion battery based on extended Kalman filter is expressed by equation (5) as the discrete state and observation equation of lithium-ion battery, and the discrete time is Δt=1s: Among them, S N is the rated capacity of the battery, I1 is the polarization resistance current, h is the battery charge and discharge efficiency, β k =exp(-1 / t(k)), t=R1C1 is the battery time constant; The OCV of the battery is the terminal voltage of the battery under long-term static conditions. It has a relatively fixed functional relationship with the battery SOC, which is recorded as OCV = f(SOC), q = N(0,Q), r = N(0,R) is Gaussian white noise; The I1(k) and S oc (k), R0(k), C1(k) and R1(k) are expressed as: x k =g k-1 (x k-1 )+q k-1 ; The expression of U(k) is written as: y k =h k (x k )+r k ; x=[I1 S oc R0 C1 R1]; y=U; G k =g k '(x)|x=x k ; H k =h k '(x)|x=x k ; The iterative formula (1) is obtained by the battery parameter and SOC joint estimation method based on the extended Kalman filter.

3. The method for digital lossless reorganization of retired power batteries in a reconfigurable battery network according to claim 1 is characterized in that: In step S140, the method for obtaining formula (2) is: In a reconfigurable battery network, When the parallel switch Sj of the jth parallel unit is closed, all the series switches Sij of the M battery modules are disconnected, where 1≤i≤m. At this time, the output voltage of the jth parallel unit is 0; When the parallel switch Sj of the jth parallel unit is disconnected, only one series switch among the series switches Sij of the M battery modules is closed, where 1≤i≤m; The fluctuation of the port output voltage of the reconfigurable battery network is controlled within the range of the following formula (6): Among them, U re is the cluster target output voltage or load voltage, e rr is the maximum allowable deviation.

4. The method for digital lossless reorganization of retired power batteries in a reconfigurable battery network according to claim 3 is characterized in that: The values ​​of λ include: When λ∈(0,1), the balancing control part of the battery module considers the impact of the battery balancing control process on the matching load demand of the energy storage system, which can be used for the active balancing control method. When λ=1, the balancing process of the battery module has the same effect on the energy storage system meeting the load voltage requirements, and can be used for energy storage systems where the balancing current of the battery module is equal to the working current of the energy storage system. Since the battery topology network transformation is used to meet the balance between battery modules and the voltage matching of the energy storage system to the load, the reconfigurable battery network system is suitable for the case of λ=1. When λ∈(1,+∞), the extreme case of λ∝+∞ is considered, and the impact of the balancing control between battery modules on the load matching of the energy storage system is ignored.

5. The digital lossless reorganization method for retired power batteries of a reconfigurable battery network according to claim 1 is characterized in that: In step S150, the method for obtaining formula (3) is: The battery is calibrated for available capacity, SOC and resistance and capacitance parameters, which are recorded as SN ij , soc ij (0), R ij (0) and C ij (0); The available capacity of the battery SN ij The resistance and capacitance parameters R and C are constant within a fixed time interval and are functions of soc and temperature T. Then, the battery charge can be modeled as: The cvij (t m+1 )≈O cvij (t m )-k ij (t n )△S ocij (t m )......................(7); Among them, Q represents the charge of the battery, and τ is the discrete time interval; When the same amount of charge flows, the battery SOC changes differently, and the battery capacity difference matrix (8) is constructed: in: Snorm is the standard capacity of the battery cell, which is taken as the average value of the sum of all battery capacities in the selected cluster; The discharge rate ρ is defined as: When the same amount of electricity Q flows through each unit in the network, the SOC change of each unit can be expressed as: S ocij (t m+1 )=S ocij (t m )-ρD(i,j)......................(10); Assuming that the jth series cell is in the on state, the output voltage of the jth series cell is u j (t), and within the time u of formula (11) j (t) value is basically unchanged, that is, u j (t)≈u j (τm). Through system hardware design, the transient process of the capacitor and the voltage change caused by the transient process of the capacitor can be ignored. Therefore, within the τ time slot, the output voltage of the battery is expressed by formula (12): The system optimization objective function is simplified to minimizing the sum of the product of the absolute value of the remaining SOC and the voltage deviation during charging / discharging and the proportional factor λ.

6. The method for digital lossless reorganization of retired power batteries in a reconfigurable battery network according to claim 1 is characterized in that: In step S210, the method for establishing a reconfigurable battery network control architecture further includes: The retired power battery nodes communicate with each other in a two-way manner through the communication network, and all such communication relationships are described by an undirected graph containing the connection relationships between agents, denoted as: G = {N, E}. Where N = {v1, v2, ..., v n } represents the set of all agents in the communication network, E = {ε1, ε2, ..., ε m } represents the set of all communication connections in the communication network, where a group communication connection is an unordered set of two agents; If {v i , v j }∈E, then there is a communication relationship between agents i and j, and the set of all neighbors of agent i is Ω i = {v j |{v i , v j }∈E, j=1,...,n}, the total number of neighbors of agent i is n i , n i =|Ω i |; For an undirected graph G, let A represent the association matrix of graph G. ij ) n×n , if {v i , v j }∈E, then α ij =1, otherwise, α ij =0.

7. The method for digital lossless reorganization of retired power batteries in a reconfigurable battery network according to claim 1, characterized in that: In step S222, the method for configuring AOE includes: AOE network declaration, AOE node declaration and AOE edge declaration.

8. The method for digital lossless reorganization of retired power batteries in a reconfigurable battery network according to claim 1 is characterized in that: In step S225, the method of parsing and executing the AOE network includes: S2251. Save the policy obtained after parsing the AOE network configuration file as an AOE vector, where the AOE vector contains one or more AOE networks; S2252, parsing the measuring point configuration file and the channel configuration file to allocate memory space for storing the measuring point data, and establishing a communication connection according to the communication mode declared in the channel configuration file; S2253, according to the trigger condition of the AOE network obtained by the analysis, the start of the control strategy is determined, and the AOE network using simple repetition drive and time drive is executed according to the trigger condition parameters of the AOE network defined in the AOE network configuration file; and for the AOE network using event drive, the event condition of the first node is used as the condition for whether the entire AOE network is triggered, and the first node represents a node with an in-degree of 0 in the AOE network; S2254, monitor the change of the measurement point value according to the attributes and the starting address in the measurement point configuration file; when the measurement point value changes and causes the first node event to be triggered, the AOE network starts and executes control; S2255, the AOE network pops up a head node from the current topological sorting queue, and waits until an event of the head node occurs or times out; the head node refers to the node at the front of the current topological sorting queue; S2256. When the event of the pop-up node times out, if the timeout is not due to receiving an external AOE cancellation instruction or failure, return to step S2255; S2257, when a node event occurs, execute all actions triggered by the event, that is, the actions on all edges with the node as the head node in the AOE network are executed, and wait for the return result of the action execution. For actions with unsuccessful results, take measures such as ignoring, retrying, stopping the affected nodes, and stopping AOE execution according to the failure mode of each action; S2258. When all actions are executed, if the topological sorting queue is not empty, return to step S2255, otherwise it means that all nodes have been popped out, the AOE network is executed, and the control process ends.

9. The method for digital lossless reorganization of retired power batteries in a reconfigurable battery network according to claim 1, characterized in that: In step S230, the method for optimizing and allocating dual-layer control includes: S231, select the best discharge battery pack. The principle of selecting the battery pack can be described as: max(SOC1,SOC2,L,SOC n ); S232, judging the discharge power, described as: Among them, P D_limit Represents the maximum allowable discharge power of the energy storage battery pack at time t; P D_ max represents the maximum discharge characteristic determined by the materials constituting the energy storage battery pack; E(t-1), E min They represent the remaining power of the energy storage battery pack at the last moment and the minimum value of the remaining power allowed; η D represents the discharge efficiency; Δt represents the calculation window time; Determine the maximum allowable discharge power P of the selected optimal discharge battery pack D_limit (t) and command power P ref Size relationship: When P D_limit (t)>P ref , then the optimal battery pack is selected for discharge; When P D_limit (t)<P ref , repeat step S231, select the next target battery pack from the remaining battery packs, and compare the sum of the maximum allowable discharge power with the command power until the command power is met; S233, determine the number m of selected discharge battery groups, and evenly divide the power among the energy storage battery groups. The calculation formula is: To achieve optimal allocation of lower-layer power, it can be expressed as: in, represents the SOC of the i-th energy storage battery module at the current moment, P Di represents the discharge power of the i-th energy storage battery module, E i Indicates the remaining power of the i-th energy storage battery module; SOC ave Indicates the average SOC of all energy storage battery modules.

10. The method for digital lossless reorganization of retired power batteries in a reconfigurable battery network according to claim 9, characterized in that: Constraints of two-layer control: Discharge power constraint: 0≤P Di ≤P D_limit (t); SOC Constraint: SOC i，min ≤SOC i ≤SoC i，max ; Total output constraint:

Images