Energy storage soc adaptive equalization method based on consistency algorithm and tanh function

Abstract

The application discloses a kind of energy storage SOC adaptive equalization methods based on consistency algorithm and Tanh function, belong to distributed photovoltaic energy storage grid-connected system control technical field.The method of the present application calculates the power difference between the total output power of photovoltaic and load power, obtains the parameters of each energy storage unit based on distributed communication network, calculates SOC deviation and DC side port voltage deviation through consistency algorithm;Active power instruction of each energy storage grid-connected converter and DC side port voltage instruction of each energy storage grid-connected converter are calculated;Through PI control and model predictive control, the switching control signal of bidirectional DC / DC converter and energy storage grid-connected converter is generated.The application greatly accelerates the SOC equalization convergence speed of battery energy storage unit, maintains equalization stability under the condition such as frequent load fluctuation, and has the function of predictive limiting battery extreme charging and discharging, suitable for multiple distributed photovoltaic energy storage grid-connected system collaborative operation control scene.

Description

Technical Field

[0001] This invention relates to the technical field of multi-distributed photovoltaic-storage grid-connected systems, and specifically to an adaptive balancing method for energy storage SOC based on a consensus algorithm and the Tanh function. Background Technology

[0002] With the rapid popularization of distributed generation technology in new energy sources, distributed photovoltaic-storage grid-connected systems have become the mainstream application solution for industrial and commercial distributed power supply and industrial park microgrids due to their advantages such as flexible dispatch, absorption of excess photovoltaic power, and smoothing of load fluctuations. Because solar irradiance varies significantly with time of day and weather, photovoltaic power generation inherently possesses intermittency and random fluctuations. Relying solely on photovoltaic grid-connected systems cannot achieve stable power supply to loads and the grid. Therefore, distributed energy storage grid-connected systems are necessary to achieve bidirectional energy buffering and maintain AC bus power balance and voltage stability. During long-term parallel operation, the state of charge (SOC) of the battery storage units in distributed energy storage grid-connected systems will continuously deviate due to factors such as differences in factory capacity, number of charge-discharge cycles, inconsistent operating temperature, and inconsistent initial SOC. Without an effective SOC balancing control strategy, battery energy storage units with excessively high SOC will prematurely enter overcharge protection, while those with excessively low SOC will prematurely trigger discharge lower limit lockout, leading to a significant decrease in the overall usable capacity of the system. Furthermore, long-term unbalanced charging and discharging will accelerate battery aging, shorten the lifespan of battery energy storage units, and in severe cases, even cause safety hazards such as thermal runaway. Therefore, designing a fast SOC collaborative balancing control scheme adaptable to multiple energy storage capacities is a core technical requirement for the stable and reliable operation of distributed photovoltaic-energy storage grid-connected systems.

[0003] In existing technologies, several solutions exist for the SOC balancing problem among battery energy storage units in multi-distributed photovoltaic-energy storage grid-connected systems: 1. Titled “Distributed optimal control of DC microgrid considering balance of chargestate”, Huang BN, Zheng S, Wang R, et al., IEEE Transactions on Energy Conversion, 2022, 37(3): 2162-2174. This paper proposes an improved droop control strategy based on dual-quadrant SOC, introducing a fixed acceleration coefficient. As the SOC difference between battery energy storage units decreases, the system SOC balancing speed will become slower and slower.

[0004] 2. The article titled "State of charge balancing strategy for energy storage system in islanded DC microgrid based on micro-tuning virtual resistance" by Mi Y, Deng J, Chang JF, et al., *Electric Power Systems Research*, 2022, 209: 107921, introduces a virtual fine-tuning resistor to eliminate the influence of line impedance in the microgrid, while achieving SOC balancing. However, the acceleration effect is only significant when the SOC deviation between battery energy storage units is large, and the SOC balancing speed is slow.

[0005] 3. An article titled "A Multi-Energy Storage Coordinated Operation Control Strategy Without a Voltage Observer," by Dai Guanggui, He Jin, and Wu Fulei, published in *Journal of Power Supply*, Vol. 24, No. 1, 2026, pp. 146-156, proposes a hierarchical control method. However, this method does not fully consider the impact of practical factors such as inconsistent initial SOC values ​​and capacities among the various battery energy storage units during actual operation, and therefore cannot fully guarantee the long-term reliable operation of the system.

[0006] In summary, existing technologies generally suffer from the following shortcomings: 1. Insufficient consideration of energy storage unit capacity differences: Most methods assume that all energy storage units have the same capacity, which is difficult to adapt to the application scenarios of heterogeneous battery packs that are common in actual systems.

[0007] 2. SOC equalization speed needs to be improved: The existing method has a long SOC equalization convergence time, especially at the end of the equalization period, which affects the system response performance. Summary of the Invention

[0008] The technical problem to be solved by this invention is to overcome the shortcomings of the prior art. It provides an adaptive SOC equalization method for energy storage based on consensus algorithm and Tanh function. In multi-distributed photovoltaic-energy storage grid-connected systems, when the capacity of the battery energy storage units is different and the initial SOC is different, it can effectively realize the accurate allocation of power of the distributed energy storage grid-connected system, significantly accelerate the SOC equalization process of the battery energy storage units, and improve the operational reliability and economy of multi-distributed photovoltaic-energy storage grid-connected systems.

[0009] The technical solution of the present invention is as follows.

[0010] An adaptive SOC balancing method for energy storage based on consensus algorithm and Tanh function is proposed. The system involved in this balancing method consists of multiple distributed energy storage grid-connected systems, multiple photovoltaic power generation grid-connected systems, and grid-connected units. Each grid-connected unit includes an AC bus, electrical loads, and the power grid. The multiple distributed energy storage grid-connected systems comprise n parallel distributed energy storage grid-connected systems. Each distributed energy storage grid-connected system consists of a battery energy storage unit, a bidirectional DC / DC converter, and an energy storage grid-connected converter connected in parallel. The output terminals of all n distributed energy storage grid-connected systems are connected to the AC bus, thus integrating the multiple distributed energy storage systems into the grid. Any distributed energy storage grid-connected system in the system is denoted as the i-th distributed energy storage grid-connected system, where i belongs to [1, n]. The multi-photovoltaic grid-connected system includes m parallel photovoltaic grid-connected systems. Each photovoltaic grid-connected system consists of photovoltaic units, BOOST converters, and photovoltaic grid-connected inverters connected in parallel. The output terminals of all m photovoltaic grid-connected systems are connected to the AC bus. Any photovoltaic grid-connected system in the multi-photovoltaic grid-connected system is denoted as the j-th photovoltaic grid-connected system, where j belongs to [1, m]. The energy storage adaptive SOC equalization method includes the following steps: Step 1: The photovoltaic unit is continuously operated in MPPT mode, the battery energy storage unit realizes charge and discharge control, and the grid-connected unit realizes grid-connected / voltage stabilization control; Step 2: Based on the communication network of the distributed hierarchical control architecture, the output current I of the photovoltaic unit in the j-th photovoltaic power generation grid-connected system at the current moment is collected in real time. pvj and output voltage U pvj And calculate the output power P of the j-th photovoltaic unit at the current time. pvj and the total output power P of m photovoltaic units pv ; Step 3: Real-time acquisition of the power P of the electrical load in the system at the current moment. load Calculate the power difference of the multi-distributed optical-storage grid-connected system at the current moment, and denot it as the system power difference ΔP, where ΔP = P pv -P load ; Step 4: Based on the communication network of the distributed hierarchical control architecture, collect in real time the capacity C of the battery energy storage unit in the i-th distributed energy storage grid-connected system at the current moment. i Real-time State of Charge (SOC) i and the DC side port voltage V of the grid-connected energy storage converter dci The deviation u of the battery energy storage unit relative to the average SOC in the i-th distributed energy storage grid-connected system is obtained through a consensus algorithm. soci And the deviation u of the DC-side port voltage of the grid-connected energy storage converter in the i-th distributed energy storage grid-connected system relative to the average voltage. vi ; Step 5: Introduce the Tanh function and adaptive acceleration coefficient to distribute the system power difference ΔP of the multi-distributed photovoltaic-storage grid-connected system at the current moment among the grid-connected converters in each distributed energy storage grid-connected system, and obtain the active power command value P of the i-th distributed energy storage grid-connected converter i at the current moment. refi ; Step 6: Correct the initial droop coefficient to obtain the adaptive droop coefficient R of the i-th battery energy storage unit at the current time. di Then, the actual command value V of the DC side port voltage of the i-th energy storage grid-connected converter at the current moment is calculated. refi According to the DC side port voltage V of the i-th energy storage grid-connected converter dci and the actual instruction value V refi The control signal for the bidirectional DC / DC converter is generated by the predictive controller using the MPC model; based on the active power command value P of the i-th energy storage grid-connected converter... refi By using power / current dual closed-loop control and PWM modulation, control signals are generated for the power switching devices of the energy storage grid-connected converter.

[0011] Preferably, the output power P of the j-th photovoltaic unit at the current time in step 2 is... pvj and the total output power P of m photovoltaic units pv The calculation formulas are as follows: P pvj =I pvj U pvj , .

[0012] Preferably, step 4 is implemented as follows: Step 4.1: Equate each distributed energy storage grid-connected system to an independent node in a communication network and construct an undirected connected communication topology G=(V,E,A), where V={1,2,…,n} is the set of nodes; E is the set of communication edges, used to describe whether there is a direct communication connection between the nodes. This is the communication weight matrix; Let be the communication weight between node i and its neighbor node y; Let the i-th distributed energy storage grid-connected system be node i, and define the set of nodes that can directly exchange state information with it as the neighbor set n. i Define its number of neighbors as ,symbol Represents set n i The number of elements contained in it; let any distributed energy storage grid-connected system adjacent to the i-th distributed energy storage grid-connected system be node y, y∈[1,n] i The set of nodes with which a node can directly exchange state information is defined as the neighbor set n. yDefine its number of neighbors as ,symbol Represents set n y The number of elements contained in it; the communication weight between node i and its neighbor node y is set to The node's own weight is set to For nodes that are neither node i itself nor belong to its neighbor set, their communication weight is set to 0, thereby ensuring that the sum of node i's communication weights to itself and all its neighbor nodes is equal to 1. Node i calculates its consistency correction based on the weighted deviation between its own state and the states of other communicable nodes. The calculation formula is:

[0013] In the formula, Let be the consistency correction amount for node i at time k. Let i represent the local consistency state of node i. Neighboring nodes A consistent state, Let y be the relative state deviation between node y and node i. Node i performs a weighted sum of the relative state deviations of all its neighbors to obtain the consistency correction direction and magnitude for its local state: when the weighted state of a neighboring node is higher than the local state of node i, the consistency correction is positive, driving node i's state to increase; when the weighted state of a neighboring node is lower than the local state of node i, the consistency correction is negative, driving node i's state to decrease. By repeatedly executing local state information exchange and consistency correction, each node gradually eliminates the state deviation between nodes under the condition of a connected communication topology, achieving consistent convergence of the states of each node. when When representing the battery energy storage unit in the i-th distributed energy storage grid-connected system, the consistency correction amount is used to determine the SOC deviation between each battery energy storage unit and adjust its charging and discharging power accordingly; when When representing the DC side port voltage of the grid-connected converter in the i-th distributed energy storage grid-connected system, the consistency correction amount is used to calculate the DC voltage deviation between nodes and to provide a basis for DC bus voltage compensation control. Step 4.2: Based on the aforementioned consensus algorithm, calculate the deviation u of the battery energy storage unit relative to the average SOC in the i-th distributed energy storage grid-connected system. soci And the deviation u of the DC-side port voltage of the grid-connected energy storage converter in the i-th distributed energy storage grid-connected system relative to the average voltage. vi The calculation formula is:

[0014] In the formula, SOC y State of charge (SOC) is the state of charge (SOC) of a battery energy storage unit in any distributed energy storage grid-connected system adjacent to the i-th distributed energy storage grid-connected system. avg V is the capacity-weighted average SOC; dci V is the DC-side port voltage of the grid-connected converter in the i-th distributed energy storage grid-connected system; dcy Let be the DC port voltage of the grid-connected converter in any distributed energy storage grid-connected system adjacent to the i-th distributed energy storage grid-connected system.

[0015] Preferably, step 5 is implemented as follows: Step 5.1: Introduce the adaptive acceleration factor m of the i-th battery energy storage unit. i The calculation formula is as follows:

[0016] In the formula, m min Let m be the minimum acceleration coefficient of the i-th battery energy storage unit, used to ensure basic regulation capability during small deviation phases. max u is the maximum acceleration coefficient of the i-th battery energy storage unit, used to limit the maximum adjustment intensity during large deviation phases. db This is the SOC dead zone threshold; Step 5.2: Introduce the tanh function to calculate the active power command value P of the i-th energy storage grid-connected converter. refi The calculation formula is:

[0017] In the formula, The power required by the i-th battery energy storage unit is expressed as:

[0018] In the formula, ΔP>0 indicates that the battery energy storage unit is in a charging state, and ΔP<0 indicates that the battery energy storage unit is in a discharging state.

[0019] Preferably, step 6 is implemented as follows: Step 6.1: Introduce the tanh function to calculate the adaptive droop coefficient R of the i-th distributed energy storage grid-connected system at the current time in the multi-distributed energy storage grid-connected system. di The calculation formula is:

[0020] In the formula, R di0 Let m be the initial droop coefficient of the i-th distributed energy storage grid-connected system;i is the adaptive acceleration factor for the i-th battery energy storage unit; Step 6.2, the secondary voltage compensation value ΔV of the DC side port voltage of the i-th energy storage grid-connected converter at the current moment. i The calculation formula is:

[0021] In the formula, t is the running time; Step 6.3, the actual command value V of the DC side port voltage of the i-th energy storage grid-connected converter at the current moment. refi The calculation formula is:

[0022] In the formula, V nom i is the rated value of the DC side port voltage of the grid-connected converter. btai This represents the output current of the i-th battery energy storage unit at the current moment. Step 6.4, convert the DC side port voltage V of the i-th energy storage grid-connected converter. dci The actual command value V of the DC side port voltage of the i-th energy storage grid-connected converter at the current moment. refi The port voltage deviation E is obtained by subtracting the input voltage. Then, the port voltage deviation E is passed through the output of the PI control and the output voltage V of the i-th battery energy storage unit at the current moment. bati Output current i bati As input to the MPC model prediction, the control signal for the bidirectional DC / DC converter is generated by the MPC model prediction controller; Step 6.5: Set the active power command value P of the i-th energy storage grid-connected converter. refi As the command value for the power control loop of the grid-connected energy storage converter, the control signal for the power switching device of the grid-connected energy storage converter is generated through power / current dual closed-loop control and PWM modulation.

[0023] Compared with the prior art, the beneficial effects of this invention are reflected in: 1. Adaptable to heterogeneous energy storage, precise power allocation, and significantly accelerated SOC equalization convergence speed: This invention introduces a capacity-weighted consensus algorithm to calculate the average SOC, fully distinguishing the capacity differences of each energy storage unit. Power allocation strictly combines the dual factors of battery capacity and real-time SOC deviation. An adaptive acceleration factor with dead zone threshold constraint is designed, and the droop coefficient and power allocation weight are dynamically adjusted by the Tanh nonlinear function to solve the problems of tailing at the end of equalization and insufficient adjustment strength of the traditional fixed acceleration coefficient strategy.

[0024] 2. Equipped with an active overcharge and over-discharge protection mechanism to effectively extend the overall lifespan of the energy storage system: The integrated model predictive control (MPC) can predict the SOC change trend brought about by the power command at the next moment, and correct and limit the charging and discharging power in advance, actively avoiding the battery from entering the dangerous conditions of deep overcharge and deep over-discharge, reducing irreversible damage to the battery, and slowing down the aging rate of the energy storage unit.

[0025] 3. The system is highly robust and adaptable to complex operating conditions with severe photovoltaic load fluctuations: It adopts a distributed hierarchical communication architecture for coordinated and consistent voltage compensation control, which can smooth out power disturbances caused by sudden changes in photovoltaic output and frequent load switching in real time. Under extreme conditions such as sudden changes in illumination and large load fluctuations, the SOC equilibrium convergence trend remains unaffected, and the power allocation accuracy is not affected, making it suitable for the actual complex operating scenarios of distributed photovoltaic-storage grid-connected systems. Attached Figure Description

[0026] Figure 1 This invention relates to the structure of a multi-distributed optical energy storage grid-connected system.

[0027] Figure 2 This is a block diagram of the overall control structure of the system of the present invention.

[0028] Figure 3 This invention describes the SOC balancing process for four battery energy storage units with different capacities under normal charging and discharging conditions, using the optimized control method of this invention.

[0029] Figure 4 The power distribution effect of the optimized control method of this invention is shown for four battery energy storage units with different capacities under normal charging and discharging conditions.

[0030] Figure 5 This describes the SOC balancing process for four battery energy storage units with different capacities under normal charging and discharging conditions, using existing control methods.

[0031] Figure 6 The power distribution effect of existing control methods on four battery energy storage units of the same capacity under normal charging and discharging conditions is shown.

[0032] Figure 7 This invention describes the SOC balancing process for four batteries of different capacities operating under conditions of frequent fluctuations in photovoltaic power and load, using the optimized control method of this invention.

[0033] Figure 8 The power distribution effect of the optimized control method of this invention is demonstrated for four batteries of different capacities under conditions of frequent fluctuations in photovoltaic power and load. Detailed Implementation

[0034] The following is a detailed description of this embodiment with reference to the accompanying drawings.

[0035] Figure 1 This is the structure of the multi-distributed optical-storage grid-connected system of the present invention. (The structure is derived from...) Figure 1 As can be seen, the system involved in the balancing method of this invention consists of multiple distributed energy storage grid-connected systems, multiple photovoltaic power generation grid-connected systems, and grid-connected units. The grid-connected unit includes an AC bus, electrical loads, and a power grid. The multiple distributed energy storage grid-connected systems include n parallel distributed energy storage grid-connected systems. Each distributed energy storage grid-connected system consists of a battery energy storage unit, a bidirectional DC / DC converter, and an energy storage grid-connected converter connected in parallel. The output terminals of all n distributed energy storage grid-connected systems are connected to the AC bus. Any one of the distributed energy storage grid-connected systems in the multiple distributed energy storage grid-connected systems is denoted as the i-th distributed energy storage grid-connected system, where i belongs to [1, n]. The multi-photovoltaic grid-connected system includes m parallel photovoltaic grid-connected systems. Each photovoltaic grid-connected system consists of a photovoltaic unit, a BOOST converter, and a photovoltaic grid-connected inverter connected in parallel. The output terminals of all m photovoltaic grid-connected systems are connected to the AC bus. Any photovoltaic grid-connected system in the multi-photovoltaic grid-connected system is denoted as the j-th photovoltaic grid-connected system, where j belongs to [1, m].

[0036] exist Figure 1 Above, Load is the electrical load, L is the filter inductor, and C is the filter capacitor. That is, an LC filter is set before the electrical load.

[0037] The photovoltaic units operate continuously in maximum power point tracking (MPPT) mode and transmit energy to the AC bus via a BOOST converter and a grid-connected inverter. The battery energy storage unit is connected to the AC bus via a bidirectional DC / DC converter and a grid-connected inverter, enabling bidirectional energy flow (charging and discharging).

[0038] In this embodiment, n=4, m=4, the inductance of the filter inductor L is 9mH, the capacitance of the filter capacitor C is 500uF, and the rated phase voltage of the three-phase AC output voltage is 220V / 50Hz.

[0039] Figure 2 This is a block diagram of the overall control structure of the system of the present invention. Figure 2 As can be seen, the energy storage adaptive SOC equalization method of the present invention includes the following steps: Step 1: The photovoltaic unit is continuously operated in MPPT mode, the battery energy storage unit realizes charge and discharge control, and the grid-connected unit realizes grid-connected / voltage stabilization control.

[0040] Step 2: Based on the communication network of the distributed hierarchical control architecture, the output current I of the photovoltaic unit in the j-th photovoltaic power generation grid-connected system at the current moment is collected in real time. pvj and output voltage U pvj And calculate the output power P of the j-th photovoltaic unit at the current time. pvjand the total output power P of m photovoltaic units pv .

[0041] In this embodiment, the output power P of the j-th photovoltaic unit at the current time is... pvj and the total output power P of m photovoltaic units pv The calculation formulas are as follows: P pvj =I pvj U pvj , .

[0042] Step 3: Real-time acquisition of the power P of the electrical load in the system at the current moment. load Calculate the power difference of the multi-distributed optical-storage grid-connected system at the current moment, and denot it as the system power difference ΔP, where ΔP = P pv -P load .

[0043] Step 4: Based on the communication network of the distributed hierarchical control architecture, collect in real time the capacity C of the battery energy storage unit in the i-th distributed energy storage grid-connected system at the current moment. i Real-time State of Charge (SOC) i and the DC side port voltage V of the grid-connected energy storage converter dci The deviation u of the battery energy storage unit relative to the average SOC in the i-th distributed energy storage grid-connected system is obtained through a consensus algorithm. soci And the deviation u of the DC-side port voltage of the grid-connected energy storage converter in the i-th distributed energy storage grid-connected system relative to the average voltage. vi .

[0044] In this embodiment, step 4 is implemented as follows: Step 4.1: Equate each distributed energy storage grid-connected system to an independent node in a communication network and construct an undirected connected communication topology G=(V,E,A), where V={1,2,…,n} is the set of nodes; E is the set of communication edges, used to describe whether there is a direct communication connection between the nodes. This is the communication weight matrix; Let be the communication weight between node i and its neighbor node y; Let the i-th distributed energy storage grid-connected system be node i, and define the set of nodes that can directly exchange state information with it as the neighbor set n. i Define its number of neighbors as ,symbol Represents set n i The number of elements contained in it; let any distributed energy storage grid-connected system adjacent to the i-th distributed energy storage grid-connected system be node y, y∈[1,n] iThe set of nodes with which a node can directly exchange state information is defined as the neighbor set n. y Define its number of neighbors as ,symbol Represents set n y The number of elements contained in it; the communication weight between node i and its neighbor node y is set to The node's own weight is set to For nodes that are neither node i itself nor belong to its neighbor set, their communication weight is set to 0, thereby ensuring that the sum of node i's communication weights to itself and all its neighbor nodes is equal to 1. Node i calculates its consistency correction based on the weighted deviation between its own state and the states of other communicable nodes. The calculation formula is:

[0045] In the formula, Let be the consistency correction amount for node i at time k. Let i represent the local consistency state of node i. Neighboring nodes A consistent state, Let y be the relative state deviation between node y and node i; node i performs a weighted sum of the relative state deviations of all its neighbors to obtain the consistency correction direction and magnitude for its local state: when the weighted state deviation of a neighboring node is higher than the local state of node i, the consistency correction amount is...

[0046] If the value is positive, the state of node i increases; when the weighted state of a neighboring node is lower than the local state of node i, the consistency correction is negative, and the state of node i decreases; each node gradually eliminates the state deviation between nodes by repeatedly executing local state information exchange and consistency correction under the condition of communication topology connectivity, and achieves consistent convergence of the states of each node. when When representing the battery energy storage unit in the i-th distributed energy storage grid-connected system, the consistency correction amount is used to determine the SOC deviation between each battery energy storage unit and adjust its charging and discharging power accordingly; when When representing the DC side port voltage of the grid-connected converter in the i-th distributed energy storage grid-connected system, the consistency correction amount is used to calculate the DC voltage deviation between nodes and to provide a basis for DC bus voltage compensation control.

[0047] Step 4.2: Based on the aforementioned consensus algorithm, calculate the deviation u of the battery energy storage unit relative to the average SOC in the i-th distributed energy storage grid-connected system. sociAnd the deviation u of the DC-side port voltage of the grid-connected energy storage converter in the i-th distributed energy storage grid-connected system relative to the average voltage. vi The calculation formula is:

[0048] In the formula, SOC y State of charge (SOC) is the state of charge (SOC) of a battery energy storage unit in any distributed energy storage grid-connected system adjacent to the i-th distributed energy storage grid-connected system. avg V is the capacity-weighted average SOC; dci V is the DC-side port voltage of the grid-connected converter in the i-th distributed energy storage grid-connected system; dcy Let be the DC port voltage of the grid-connected converter in any distributed energy storage grid-connected system adjacent to the i-th distributed energy storage grid-connected system.

[0049] Step 5: Introduce the Tanh function and adaptive acceleration coefficient to distribute the system power difference ΔP of the multi-distributed photovoltaic-storage grid-connected system at the current moment among the grid-connected converters in each distributed energy storage grid-connected system, and obtain the active power command value P of the i-th distributed energy storage grid-connected converter i at the current moment. refi .

[0050] In this embodiment, step 5 is implemented as follows: Step 5.1: Introduce the adaptive acceleration factor m of the i-th battery energy storage unit. i The calculation formula is as follows:

[0051] In the formula, m min Let m be the minimum acceleration coefficient of the i-th battery energy storage unit, used to ensure basic regulation capability during small deviation phases. max u is the maximum acceleration coefficient of the i-th battery energy storage unit, used to limit the maximum adjustment intensity during large deviation phases. db This is the SOC dead zone threshold; Step 5.2: Introduce the tanh function to calculate the active power command value P of the i-th energy storage grid-connected converter. refi The calculation formula is:

[0052] In the formula, The power required by the i-th battery energy storage unit is expressed as:

[0053] In the formula, ΔP>0 indicates that the battery energy storage unit is in a charging state, and ΔP<0 indicates that the battery energy storage unit is in a discharging state.

[0054] Step 6: Correct the initial droop coefficient to obtain the adaptive droop coefficient R of the i-th battery energy storage unit at the current time. di Then, the actual command value V of the DC side port voltage of the i-th energy storage grid-connected converter at the current moment is calculated. refi According to the DC side port voltage V of the i-th energy storage grid-connected converter dci and the actual instruction value V refi The control signal for the bidirectional DC / DC converter is generated by the predictive controller using the MPC model; based on the active power command value P of the i-th energy storage grid-connected converter... refi By using power / current dual closed-loop control and PWM modulation, control signals are generated for the power switching devices of the energy storage grid-connected converter.

[0055] In this embodiment, step 6 is implemented as follows: Step 6.1: Introduce the tanh function to calculate the adaptive droop coefficient R of the i-th distributed energy storage grid-connected system at the current time in the multi-distributed energy storage grid-connected system. di The calculation formula is:

[0056] In the formula, R di0 Let m be the initial droop coefficient of the i-th distributed energy storage grid-connected system; i is the adaptive acceleration factor for the i-th battery energy storage unit; Step 6.2, the secondary voltage compensation value ΔV of the DC side port voltage of the i-th energy storage grid-connected converter at the current moment. i The calculation formula is:

[0057] In the formula, t is the running time; Step 6.3, the actual command value V of the DC side port voltage of the i-th energy storage grid-connected converter at the current moment. refi The calculation formula is:

[0058] In the formula, V nom i is the rated value of the DC side port voltage of the grid-connected converter. btai This represents the output current of the i-th battery energy storage unit at the current moment.

[0059] Step 6.4, convert the DC side port voltage V of the i-th energy storage grid-connected converter. dci The actual command value V of the DC side port voltage of the i-th energy storage grid-connected converter at the current moment.refi The port voltage deviation E is obtained by subtracting the input voltage. Then, the port voltage deviation E is passed through the output of the PI control and the output voltage V of the i-th battery energy storage unit at the current moment. bati Output current i bati As input to the MPC model prediction, the control signal for the bidirectional DC / DC converter is generated by the MPC model prediction controller; Step 6.5: Set the active power command value P of the i-th energy storage grid-connected converter. refi As the command value for the power control loop of the grid-connected energy storage converter, the control signal for the power switching device of the grid-connected energy storage converter is generated through power / current dual closed-loop control and PWM modulation.

[0060] To demonstrate the technical effectiveness of this invention, simulation tests were conducted.

[0061] Figure 3 This invention describes the SOC balancing process for four battery energy storage units of different capacities under normal charging and discharging conditions, using the optimized control method of this invention. Figure 4 The power distribution effect of the optimized control method of this invention is demonstrated for four battery energy storage units of different capacities under normal charging and discharging conditions. Figure 5 This describes the State of Charge (SOC) balancing process for four battery energy storage units of different capacities under normal charge and discharge conditions, using existing control methods. Figure 6 The power distribution effect of existing control methods on four battery energy storage units with different capacities under normal charging and discharging conditions is shown. Figures 3-6 Simulation results comparing the SOC balancing process and power distribution of four battery energy storage units with different capacities under normal charge and discharge conditions, using the optimized control method of this invention and existing control methods. The simulation test conditions were as follows: the initial SOCs of the four battery energy storage units were 40%, 30%, 45%, and 25%, respectively; the capacities were 3 Ah, 2 Ah, 2.5 Ah, and 1.5 Ah, respectively; the illuminance of each photovoltaic unit was 1000 ppm; and the output power was 230 kW. load It is 100kW.

[0062] It can be seen that although existing control methods can achieve SOC equalization, this strategy mainly relies on improved droop control for power regulation, and the dynamic coupling of energy storage capacity differences and SOC deviation is not fully considered. Therefore, the equalization speed is relatively limited when the SOC difference is large. When the system enters the final stage of equalization, the SOC curve still exhibits a slow convergence process, showing a certain convergence tail phenomenon. The proposed distributed SOC equalization strategy based on consistency and Tanh adaptive adjustment shows better equalization effect under this condition. Under the strategy proposed in this paper, the SOC of the four batteries can converge to the consistent state more quickly, and no obvious oscillations occur during the equalization process, indicating that Tanh adaptive adjustment can effectively improve the problem of slow convergence at the end of the traditional consistency control. Furthermore, the strategy proposed in this paper can adjust the power ratio undertaken by each energy storage unit according to the capacity and SOC deviation. The energy storage unit with larger capacity has stronger energy regulation capability, so it undertakes a relatively larger power in power allocation; the energy storage unit with smaller capacity undertakes a relatively smaller power, thereby avoiding overload of small-capacity batteries.

[0063] The present invention also selected four battery energy storage units with different capacities to verify the SOC balancing process and power distribution effect of the distribution method of the present invention under the condition of frequent fluctuations in photovoltaic power and load. Figure 7 This invention describes the SOC balancing process for four battery energy storage units of different capacities operating under conditions of frequent fluctuations in photovoltaic power and load, using the optimized control method of this invention. Figure 8 The power distribution effect of the optimized control method of this invention is demonstrated for four battery energy storage units with different capacities under conditions of frequent fluctuations in photovoltaic power and load. Figure 7 and Figure 8 The power distribution effect and SOC balancing process simulation results of four battery energy storage units with different capacities under frequent fluctuations in photovoltaic power and load, using the optimized control method of this invention, are presented. The simulation test conditions are as follows: the initial SOCs of the four battery energy storage units are 40%, 30%, 45%, and 25%, respectively, and their capacities are 3Ah, 2Ah, 2.5Ah, and 1.5Ah, respectively. Initially, the illuminance of each photovoltaic unit is 400 ppm, the output power is 90 kW, and the AC load is P. load The output power is 230kW, with a light intensity of 1000 ppm for each photovoltaic unit at 2s. load 400kW, P at 4s load With a power output of 100kW, at 6s, one photovoltaic unit maintains an illuminance of 1000 and an output power of 230kW; with three photovoltaic units, the illuminance is 0 and the output power is 0. load The power output decreased to 200kW, and P was at 8s. load It has been increased to 400kW.

[0064] It can be seen that the SOC reaches equilibrium at t=4.5s, and despite drastic power fluctuations during this period, the equilibrium effect remains good. Although the photovoltaic output and load power of the system change at multiple moments, the overall SOC of the four batteries still maintains a trend of convergence in the same direction, without divergence or significant oscillation due to power disturbances. When the photovoltaic power is sufficient, the system allocates excess power to the energy storage units for absorption, with low SOC units receiving higher charging priority; when the photovoltaic power is insufficient or the load increases, the energy storage units enter the discharge state, and high SOC units and large-capacity units undertake more power support tasks.

Claims

1. An adaptive SOC balancing method for energy storage based on consensus algorithm and Tanh function, characterized in that, The system involved in this equalization method consists of multiple distributed energy storage grid-connected systems, multiple photovoltaic power generation grid-connected systems, and grid-connected units. Each grid-connected unit includes an AC bus, electrical loads, and the power grid. The multiple distributed energy storage grid-connected systems comprise n parallel distributed energy storage grid-connected systems. Each distributed energy storage grid-connected system consists of a battery energy storage unit, a bidirectional DC / DC converter, and an energy storage grid-connected converter connected in parallel. The output terminals of all n distributed energy storage grid-connected systems are connected to the AC bus. Any one of the distributed energy storage units in the multiple distributed energy storage grid-connected systems can be connected to the grid. The system is denoted as the i-th distributed energy storage grid-connected system, where i belongs to [1, n]. The multi-photovoltaic grid-connected system includes m parallel photovoltaic grid-connected systems. Each photovoltaic grid-connected system consists of photovoltaic units, BOOST converters, and photovoltaic grid-connected inverters connected in parallel. The output terminals of all m photovoltaic grid-connected systems are connected to the AC bus. Any photovoltaic grid-connected system in the multi-photovoltaic grid-connected system is denoted as the j-th photovoltaic grid-connected system, where j belongs to [1, m]. The energy storage adaptive SOC equalization method includes the following steps: Step 1: The photovoltaic unit is continuously operated in MPPT mode, the battery energy storage unit realizes charge and discharge control, and the grid-connected unit realizes grid-connected / voltage stabilization control; Step 2: Based on the communication network of the distributed hierarchical control architecture, the output current I of the photovoltaic unit in the j-th photovoltaic power generation grid-connected system at the current moment is collected in real time. pvj and output voltage U pvj And calculate the output power P of the j-th photovoltaic unit at the current time. pvj and the total output power P of m photovoltaic units pv ; Step 3: Real-time acquisition of the power P of the electrical load in the system at the current moment. load Calculate the power difference of the multi-distributed optical-storage grid-connected system at the current moment, and denot it as the system power difference ΔP, where ΔP = P pv -P load ; Step 4: Based on the communication network of the distributed hierarchical control architecture, collect in real time the capacity C of the battery energy storage unit in the i-th distributed energy storage grid-connected system at the current moment. i Real-time State of Charge (SOC) i and the DC side port voltage V of the grid-connected energy storage converter dci The deviation u of the battery energy storage unit relative to the average SOC in the i-th distributed energy storage grid-connected system is obtained through a consensus algorithm. soci And the deviation u of the DC-side port voltage of the grid-connected energy storage converter in the i-th distributed energy storage grid-connected system relative to the average voltage. vi ; Step 5: Introduce the Tanh function and adaptive acceleration coefficient to distribute the system power difference ΔP of the multi-distributed photovoltaic-storage grid-connected system at the current moment among the grid-connected converters in each distributed energy storage grid-connected system, and obtain the active power command value P of the i-th distributed energy storage grid-connected converter i at the current moment. refi ; Step 6: Correct the initial droop coefficient to obtain the adaptive droop coefficient R of the i-th battery energy storage unit at the current time. di Then, the actual command value V of the DC side port voltage of the i-th energy storage grid-connected converter at the current moment is calculated. refi According to the DC side port voltage V of the i-th energy storage grid-connected converter dci and the actual instruction value V refi The control signal for the bidirectional DC / DC converter is generated by the predictive controller using the MPC model; based on the active power command value P of the i-th energy storage grid-connected converter... refi By using power / current dual closed-loop control and PWM modulation, control signals are generated for the power switching devices of the energy storage grid-connected converter.

2. The adaptive SOC balancing method for energy storage based on consensus algorithm and Tanh function according to claim 1, characterized in that, Step 2 describes the output power P of the j-th photovoltaic unit at the current time. pvj and the total output power P of m photovoltaic units pv The calculation formulas are as follows: P pvj =I pvj U pvj , .

3. The adaptive SOC balancing method for energy storage based on consensus algorithm and Tanh function as described in claim 1, characterized in that, The implementation process of step 4 is as follows: Step 4.1: Equate each distributed energy storage grid-connected system to an independent node in a communication network and construct an undirected connected communication topology G=(V,E,A), where V={1,2,…,n} is the set of nodes; E is the set of communication edges, used to describe whether there is a direct communication connection between the nodes. This is the communication weight matrix; Let be the communication weight between node i and its neighbor node y; Let the i-th distributed energy storage grid-connected system be node i, and define the set of nodes that can directly exchange state information with it as the neighbor set n. i Define its number of neighbors as ,symbol Represents set n i The number of elements contained in it; let any distributed energy storage grid-connected system adjacent to the i-th distributed energy storage grid-connected system be node y, y∈[1,n] i The set of nodes with which a node can directly exchange state information is defined as the neighbor set n. y Define its number of neighbors as ,symbol Represents set n y The number of elements contained in it; the communication weight between node i and its neighbor node y is set to The node's own weight is set to For nodes that are neither node i itself nor belong to its neighbor set, their communication weight is set to 0, thereby ensuring that the sum of node i's communication weights to itself and all its neighbor nodes is equal to 1. Node i calculates its consistency correction based on the weighted deviation between its own state and the states of other communicable nodes. The calculation formula is: In the formula, Let be the consistency correction amount for node i at time k. Let i represent the local consistency state of node i. For neighboring nodes A consistent state, Let y be the relative state deviation between node y and node i. Node i performs a weighted sum of the relative state deviations of all its neighbors to obtain the consistency correction direction and magnitude for its local state: when the weighted state of a neighboring node is higher than the local state of node i, the consistency correction is positive, driving node i's state to increase; when the weighted state of a neighboring node is lower than the local state of node i, the consistency correction is negative, driving node i's state to decrease. By repeatedly executing local state information exchange and consistency correction, each node gradually eliminates the state deviation between nodes under the condition of a connected communication topology, achieving consistent convergence of the states of each node. when When representing the battery energy storage unit in the i-th distributed energy storage grid-connected system, the consistency correction amount is used to determine the SOC deviation between each battery energy storage unit and adjust its charging and discharging power accordingly; when When representing the DC side port voltage of the grid-connected converter in the i-th distributed energy storage grid-connected system, the consistency correction amount is used to calculate the DC voltage deviation between nodes and to provide a basis for DC bus voltage compensation control. Step 4.2: Based on the aforementioned consensus algorithm, calculate the deviation u of the battery energy storage unit relative to the average SOC in the i-th distributed energy storage grid-connected system. soci And the deviation u of the DC-side port voltage of the grid-connected energy storage converter in the i-th distributed energy storage grid-connected system relative to the average voltage. vi The calculation formula is: In the formula, SOC y State of charge (SOC) is the state of charge (SOC) of a battery energy storage unit in any distributed energy storage grid-connected system adjacent to the i-th distributed energy storage grid-connected system. avg V is the capacity-weighted average SOC; dci V is the DC-side port voltage of the grid-connected converter in the i-th distributed energy storage grid-connected system; dcy Let be the DC port voltage of the grid-connected converter in any distributed energy storage grid-connected system adjacent to the i-th distributed energy storage grid-connected system.

4. The adaptive SOC balancing method for energy storage based on consensus algorithm and Tanh function as described in claim 1, characterized in that, The implementation process of step 5 is as follows: Step 5.1: Introduce the adaptive acceleration factor m of the i-th battery energy storage unit. i The calculation formula is as follows: In the formula, m min Let m be the minimum acceleration coefficient of the i-th battery energy storage unit, used to ensure basic regulation capability during small deviation phases. max u is the maximum acceleration coefficient of the i-th battery energy storage unit, used to limit the maximum adjustment intensity during large deviation phases. db This is the SOC dead zone threshold; Step 5.2: Introduce the tanh function to calculate the active power command value P of the i-th energy storage grid-connected converter. refi The calculation formula is: In the formula, The power required by the i-th battery energy storage unit is expressed as: In the formula, ΔP>0 indicates that the battery energy storage unit is in a charging state, and ΔP<0 indicates that the battery energy storage unit is in a discharging state.

5. The adaptive SOC balancing method for energy storage based on consensus algorithm and Tanh function according to claim 1, characterized in that, The implementation process of step 6 is as follows: Step 6.1: Introduce the tanh function to calculate the adaptive droop coefficient R of the i-th distributed energy storage grid-connected system at the current time in the multi-distributed energy storage grid-connected system. di The calculation formula is: In the formula, R di0 Let m be the initial droop coefficient of the i-th distributed energy storage grid-connected system; i is the adaptive acceleration factor for the i-th battery energy storage unit; Step 6.2, the secondary voltage compensation value ΔV of the DC side port voltage of the i-th energy storage grid-connected converter at the current moment. i The calculation formula is: In the formula, t is the running time; Step 6.3, the actual command value V of the DC side port voltage of the i-th energy storage grid-connected converter at the current moment. refi The calculation formula is: In the formula, V nom i is the rated value of the DC side port voltage of the grid-connected converter. btai This represents the output current of the i-th battery energy storage unit at the current moment. Step 6.4, convert the DC side port voltage V of the i-th energy storage grid-connected converter. dci The actual command value V of the DC side port voltage of the i-th energy storage grid-connected converter at the current moment. refi The port voltage deviation E is obtained by subtracting the input voltage. Then, the port voltage deviation E is passed through the output of the PI control and the output voltage V of the i-th battery energy storage unit at the current moment. bati Output current i bati As input to the MPC model prediction, the control signal for the bidirectional DC / DC converter is generated by the MPC model prediction controller. Step 6.5: Set the active power command value P of the i-th energy storage grid-connected converter. refi As the command value of the power control loop of the energy storage grid-connected converter, the control signal of the power switching device of the energy storage grid-connected converter is generated through power / current dual closed-loop control and PWM modulation.

Images