A consistent-based adaptive distributed hierarchical control method for energy storage system

Abstract

The application discloses a consistent-based adaptive distributed hierarchical control method for an energy storage system, and comprises the following steps: a photovoltaic water electrolysis hydrogen production system model composed of photovoltaic cells, lithium batteries, electrolytic cells, constant power loads and DC / DC converters is established; in the communication layer, the energy storage units in the system are regarded as agents, an average state estimator is designed based on multi-agent consistency, the average values of various state variables are calculated, and the average values are transmitted to adjacent energy storage units through an inter-neighbor communication network, so that the dependence on a centralized controller is reduced, and the communication burden is reduced; in the main control layer, an improved droop control method is adopted, the droop coefficient is improved based on a sigmoid function, adaptive adjustment of load current distribution is realized to achieve state of charge balance, and an acceleration factor improved based on a normal distribution function is added; in the secondary control layer, a unit virtual pressure drop balancing term is designed to eliminate the influence of line impedance, and an average bus voltage compensation term is designed to compensate the average bus voltage to a reference value.

Description

Technical Field

[0001] This invention belongs to the field of energy storage system control in new energy microgrids, and relates to an adaptive distributed hierarchical control method for energy storage systems based on consistency, and particularly to an adaptive state-of-charge balancing method for energy storage systems. Background Technology

[0002] Currently, establishing off-grid photovoltaic (PV) electrolysis hydrogen production systems is considered an effective strategy for absorbing PV power, eliminating the impact of PV grid connection on the stability and power quality of the main grid, and reducing hydrogen production costs. In the absence of main grid support, PV hydrogen production systems often incorporate energy storage systems to balance power fluctuations between PV generation and electrolysis loads. As the scale of off-grid hydrogen production expands, energy storage systems typically consist of multiple parallel energy storage units to increase storage capacity and improve the reliability of the energy storage system's power supply. Currently, the control method for energy storage units often employs a droop control strategy. However, under long-term operation, the fixed droop coefficient and differences in line impedance create an inherent contradiction between voltage dips and precise current distribution in droop control. This leads to an unbalanced state of charge (SOC) of the energy storage units, causing some units to prematurely shut down due to overcharging and over-discharging, affecting the stability of the electrolysis hydrogen production system. Therefore, achieving SOC balance among energy storage units is currently a necessary measure to maintain the stability of PV hydrogen production systems. This goal can be achieved by developing and implementing a state-of-charge balancing strategy for energy storage systems, which involves eliminating the effects of line impedance to achieve precise current shunting, while setting up compensation terms to recover from bus voltage drops. Summary of the Invention

[0003] To address the issues of unbalanced state of charge (SOC) in off-grid photovoltaic hydrogen production systems, low current distribution accuracy, and bus voltage dips in energy storage units, this invention employs the following technical solution: an adaptive distributed hierarchical control method for energy storage systems based on consistency, comprising the following steps:

[0004] A model of a photovoltaic water electrolysis hydrogen production system consisting of photovoltaic cells, lithium batteries, electrolyzers, constant power loads, and DC / DC converters was established.

[0005] Based on the off-grid photovoltaic water electrolysis hydrogen production system model, an improved droop-level control method for energy storage systems based on multi-agent consensus is proposed to achieve balanced state of charge of the energy storage system, eliminate the influence of line impedance to achieve precise current shunting, and simultaneously restore bus voltage dip.

[0006] Furthermore: The improved droop-level control method for energy storage systems based on the off-grid photovoltaic water electrolysis hydrogen production system model and multi-agent consensus achieves balanced state of charge of the energy storage system, eliminates the influence of line impedance to achieve precise current shunting, and simultaneously restores the bus voltage drop process as follows:

[0007] At the communication layer, the energy storage units in the photovoltaic water electrolysis hydrogen production system are regarded as intelligent agents. Based on the multi-agent consensus design average state estimation method, the average values ​​of the lithium battery state of charge, unit virtual voltage and output voltage are calculated, and the average values ​​are transmitted to the neighboring energy storage units through the neighbor communication network.

[0008] In the main control layer, an improved droop control method is adopted. Based on the sigmoid function, the droop coefficient is improved to achieve adaptive adjustment of the lithium battery output current distribution to achieve state of charge balance. An acceleration factor based on the normal distribution function is added to accelerate the later stage of the lithium battery state of charge balance.

[0009] In the secondary control layer, a virtual voltage drop balancing term is designed to eliminate the influence of line impedance and realize the output current distribution according to the rated capacity of the lithium battery. An average bus voltage compensation term is designed to compensate the average bus voltage to the reference value.

[0010] Furthermore: the improved droop control method, which improves the droop coefficient based on the sigmoid function to achieve adaptive adjustment of the lithium battery output current distribution to achieve state of charge balance, and the addition of an acceleration factor based on the normal distribution function to accelerate the later stage of lithium battery state of charge balance, also includes:

[0011] Based on dual closed-loop control of voltage and current, improved droop control, unit virtual voltage equalization term, and average bus voltage compensation term are added after the reference voltage in the outer voltage loop. The output voltage of the energy storage unit is fed back by the voltage loop. The reference current of the inner current loop is obtained through the first PI controller, and then the difference is calculated with the inductor current of the DC / DC converter. The difference is then used to obtain the PWM modulation signal through the second PI controller. The PWM modulation signal controls the switching on and off of the IGBT in the DC / DC converter to achieve rapid equalization of the state of charge of the lithium battery, accurate distribution of the output current according to the rated capacity, and average bus voltage compensation.

[0012] Furthermore: The improved droop control method, which improves the droop coefficient based on the sigmoid function, and achieves adaptive adjustment of the lithium battery output current distribution to achieve state-of-charge balance, is as follows:

[0013] During lithium battery charging, the difference in charging depth of each cell is established based on the sigmoid function, that is, the relationship between the difference between the charging depth and its average value and the droop coefficient. At the same time, the ratio of the balance coefficient to the rated capacity is introduced into the droop coefficient.

[0014] During lithium battery discharge, the relationship between the difference in discharge depth of each cell, i.e. the difference between the discharge depth and its average value, and the droop coefficient is established based on the sigmoid function. At the same time, the ratio of the balance coefficient to the rated capacity is introduced into the droop coefficient.

[0015] Furthermore: The process of adding an acceleration factor based on a normal distribution function to accelerate the later-stage balancing speed of the lithium battery's state of charge is as follows:

[0016] During lithium battery charging, an acceleration factor based on a normal distribution function is introduced into the improved droop coefficient. The acceleration factor is a variable with the square of the difference in charging depth. At the same time, an adjustment factor is added to change the speed in the later stage of state-of-charge equilibrium.

[0017] During lithium battery discharge, an acceleration factor based on a normal distribution function is introduced into the improved droop coefficient. The acceleration factor is a variable with the square of the difference in discharge depth. At the same time, an adjustment factor is added to change the speed in the later stage of state-of-charge equilibrium.

[0018] Furthermore: The virtual voltage drop balancing term of the design unit eliminates the influence of line impedance, and the process of distributing the output current according to the rated capacity of the lithium battery is as follows:

[0019] The virtual voltage drop is obtained by multiplying the droop factor by the output current of the energy storage unit.

[0020] The ratio of virtual voltage drop to the maximum deviation of bus voltage is the unit virtual voltage drop. A PI controller is used to equalize the unit virtual voltage drop to the average unit virtual voltage drop. The unit virtual voltage drop of each energy storage unit is consistent, so that the output current distribution is only related to the droop coefficient. This eliminates the influence of line impedance on the accurate current splitting and state of charge balance of the output current. When the state of charge is balanced, the output current can be distributed proportionally according to the rated capacity of each energy storage unit.

[0021] Furthermore: the process of compensating the average bus voltage to the reference value using the aforementioned design average bus voltage compensation term is as follows:

[0022] The average bus voltage is the average value of the output voltage of all energy storage units. An integral controller is used to make the average bus voltage approach the bus voltage reference value and maintain the stability of the bus voltage.

[0023] Furthermore, the improved droop coefficient is characterized as follows:

[0024] (6)

[0025] (7)

[0026] in: R vi ( S i ), ρ ( S i )and k D It is an energy storage unit iImproved droop coefficient, variable acceleration factor and balance factor, μ It is the regulating factor of the acceleration factor; the charge / discharge deviation value within it. ζ and σ Represented as:

[0027] (8)

[0028] (9)

[0029] in, S avgi and D avgi It is an energy storage unit i of S i and D i The dynamic average value.

[0030] This invention provides an adaptive distributed hierarchical control method for energy storage systems based on consensus. This method is an adaptive hierarchical control strategy for parallel energy storage units in an electrolytic hydrogen production system to improve power supply reliability. In this strategy, distributed communication and a multi-agent consensus design average state estimation method are employed to obtain the average value of state variables. In the main control layer, a sigmoid function is proposed to improve the droop coefficient to achieve state-of-charge (POC) balance, and a novel acceleration factor based on a normal distribution function is designed to accelerate the speed of subsequent POC balance. In the secondary control layer, a unit virtual voltage drop balancing term and an average bus voltage compensation term are designed to eliminate the influence of line impedance and restore the average bus voltage deviation. Stability analysis confirms that the proposed method has strong stability. Finally, a photovoltaic hydrogen production simulation model is established on the Matlab / Simulink platform to verify the proposed control strategy. The results show that the proposed control strategy can achieve rapid POC balance and accurate load current distribution under various complex operating conditions, and has excellent average bus voltage compensation capability. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0032] Figure 1 The research object of this invention is an off-grid photovoltaic water electrolysis hydrogen production system;

[0033] Figure 2This invention presents an adaptive distributed hierarchical control structure for energy storage systems based on consistency.

[0034] like Figure 3 (a) shows the improved droop coefficient proposed in this invention. R vi Differences in state of charge ζ The changing image, (b) is the novel acceleration factor proposed in this invention. ρ ( S i Differences in state of charge ζ Changing images;

[0035] Figure 4 This is the equivalent linearized model of the control strategy of the present invention, taking two energy storage units as an example;

[0036] Figure 5 The root locus of important parameters of the system of this invention is shown in Figure (a), which shows the variation with the state of charge Δ. S = S 1 -S 2. The root locus varies with the cutoff frequency. Figure (b) shows the root locus as a function of the cutoff frequency ω c The root locus changes, and Figure (c) shows the change with the equilibrium factor. k D The root locus of the change, Figure (d) shows the change with the adjustment factor. μ The changing root locus;

[0037] Figure 6 The following are simulation results for Case 1 of the present invention: (a) shows the state of charge change curves of the three energy storage units in Case 1 of the present invention; (b) shows the output current change curves of the three energy storage units in Case 1 of the present invention; and (c) shows the average bus voltage change curves of the present invention in Case 1 of the present invention.

[0038] Figure 7 The following are simulation results for Case 2 of the present invention: (a) shows the state of charge change curves of the three energy storage units in Case 2 of the present invention; (b) shows the output current change curves of the three energy storage units in Case 2 of the present invention; and (c) shows the average bus voltage change curves of the present invention in Case 2 of the present invention.

[0039] Figure 8 The following are simulation results for Case 3 of the present invention: (a) shows the state of charge change curves of the three energy storage units in Case 3 of the present invention; (b) shows the output current change curves of the three energy storage units in Case 3 of the present invention; and (c) shows the average bus voltage change curves of the present invention in Case 3 of the present invention. Detailed Implementation

[0040] It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and embodiments.

[0041] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the present invention or its application or use. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0042] An adaptive distributed hierarchical control method for a consensus-based energy storage system includes the following steps:

[0043] S1: Establish an off-grid photovoltaic water electrolysis hydrogen production system model using Matlab / Simulink, including photovoltaic cells, lithium batteries, electrolyzers, DC loads, and DC / DC converters; perform accurate simulation of the system and configure appropriate capacity and initial values ​​for each unit;

[0044] S2: Based on the off-grid photovoltaic water electrolysis hydrogen production system model, an improved droop-level control method for energy storage system based on multi-agent consistency is proposed to achieve balanced state of charge of energy storage system, eliminate the influence of line impedance to achieve precise current shunting, and restore bus voltage drop.

[0045] Steps S1 and S2 are executed sequentially;

[0046] like Figure 1 The diagram shows the composition of an off-grid photovoltaic water electrolysis hydrogen production system. This system represents the object of the distributed hierarchical control method implemented within the energy storage system. It consists of photovoltaic cells, lithium batteries, an electrolyzer, DC loads, AC loads, and a DC / DC converter. The photovoltaic system acts as a distributed power source, connected to the DC bus via an electronic power converter to achieve clean power generation. The energy storage unit is connected to the bus via a bidirectional DC / DC converter to maintain voltage stability and balance system power fluctuations. Simultaneously, the DC microgrid can also be connected to the AC grid via a DC / AC converter to achieve AC-DC energy exchange. This invention only considers the islanded operation mode of the DC microgrid and does not consider grid connection.

[0047] like Figure 2The diagram illustrates an adaptive distributed hierarchical control structure for a consensus-based energy storage system. It primarily comprises a communication layer, a main control layer, and a secondary control layer. In the communication layer, each energy storage unit is treated as an agent, establishing a neighbor-to-neighbor communication network. An average state estimation method based on multi-agent consensus is used. The droop coefficient is improved using the sigmoid function, allowing it to adaptively adjust with changes in the state of charge (SOC), ultimately leading to the SOC of each energy storage unit converging to its average value. In the secondary control layer, a virtual voltage term is used to distribute the output current of the energy storage units proportionally to their capacity, while an average bus voltage compensation term recovers from bus voltage dips.

[0048] Through secondary control correction, the output voltage value of the lithium battery can be expressed as:

[0049] (1)

[0050] in, δu 1i It is an energy storage unit i The unit virtual voltage balance term, δu 2i It is an energy storage unit i Average bus voltage compensation term.

[0051] Communication Layer: In a multi-agent system, consensus refers to the process by which all agents eventually converge to a unified state. This consensus algorithm does not rely on a central controller or strong communication links, but is achieved solely through local information processing and neighbor-to-neighbor communication between agents. This invention studies an undirected strongly connected network with N agents. Its nodes are... , For the node sequence pairs in this energy storage system, i.e. ,like This means i and j These are adjacent nodes that can transmit data to each other. (Network) Laplace matrix It can be represented as:

[0052] (2)

[0053] in, a ij It is the communication weight between nodes. If i and j If they are adjacent nodes, then a ij =1, otherwise a ij =0. The adaptive distributed hierarchical control method for energy storage systems based on consistency can be expressed as:

[0054] (3)

[0055] in, and They are intelligent agents i and intelligent agents j The state variable, and It is a local input signal. In the event of communication failure, if the undirected strongly connected network of the energy storage system contains at least one spanning tree, then (3) can satisfy... .

[0056] This invention selects a ring topology as the communication topology between energy storage units, where each energy storage unit communicates only with its adjacent units. An average state estimator is designed based on multi-agent consensus to estimate the average value of each state variable. As can be seen from the above principle, even if communication between two energy storage units fails, the average value of the state variables can still be ensured to converge to the same value.

[0057] At the main control layer: This invention integrates the state of charge (SOC) parameters into droop control to achieve SOC balance, proposing a distributed SOC balance control method. The control objective of the SOC balance strategy can be stated as follows:

[0058] (4)

[0059] For ease of explanation, the state of charge will be referred to in the following text. Written ; (energy storage unit) i Depth of discharge ) written as Assuming the charging and discharging efficiencies of the two energy storage units are equal, taking the differential of equation (3) yields the ratio of the rates of change of state of charge of the two energy storage units:

[0060] (5)

[0061] As can be seen from formula (5), to achieve the control objective of state-of-charge balance, it is necessary to eliminate the interference of line impedance on state-of-charge balance, and at the same time consider the influence of the rated capacity of the energy storage unit in the droop coefficient. Therefore, this invention, under the premise of considering the rated capacity of the energy storage unit, uses the sigmoid function to establish the relationship between the state-of-charge difference and the droop coefficient. The improved droop coefficient can be expressed as:

[0062] (6)

[0063] (7)

[0064] in: R vi (S i ), ρ ( S i )and k D It is an energy storage unit i Improved droop coefficient, variable acceleration factor and balance factor, μ It is the regulating factor of the acceleration factor; the charge / discharge deviation value within it. ζ and σ Represented as:

[0065] (8)

[0066] (9)

[0067] in: S avgi and D avgi It is an energy storage unit i of S i and D i The dynamic average value.

[0068] The dynamic average state of charge of each energy storage unit is estimated using a consensus-based average state estimation method, through the unit's... S i and the dynamic average value of adjacent energy storage units S avgj To update the local estimated dynamic average S avgi Combining equation (3), the average state of charge can be expressed as follows. Furthermore, the methods for average DoD and other average state parameters are similar.

[0069] (10)

[0070] In the secondary control layer: the control loop designed in this invention mainly serves to eliminate the influence of line impedance on charge state balance and output current shunting, while also solving the bus voltage offset problem caused by droop control. Its control objective can be defined as:

[0071] (11)

[0072] (12)

[0073] After the output current of each energy storage unit reaches state-of-charge equilibrium, it can be distributed according to its rated capacity, and the average bus voltage can also be restored to its nominal value. To achieve the above objectives, the secondary control layer in this section includes a unit virtual voltage drop balancing term and an average bus voltage compensation term. The unit virtual voltage drop balancing term calculates the virtual voltage drop of the unit locally:

[0074] (13)

[0075] Wherein, ΔU max This is the maximum bus voltage deviation value, which is taken as ΔU in this application. max =5%U nom Meanwhile, the average unit virtual voltage drop Δ is calculated based on consistency. u * avg i Δ is adjusted using a PI controller. u * i and Δ u * avg i The difference between them, hence the additional item δu 1i It can be written as the following formula:

[0076] (14)

[0077] in, K PL and K IL It is the proportional and integral gain of the unit virtual voltage drop balance term, Δ u * avg i It can be represented as:

[0078] (15)

[0079] The balance terms designed by equations (14) and (15) can ultimately make the unit virtual voltage drop of each energy storage unit tend to be consistent, which can be simplified to the following equation:

[0080] (16)

[0081] When the state of charge of each energy storage unit reaches equilibrium R vi= 1 / C i Equation (15) can be simplified to:

[0082] (17)

[0083] The average voltage compensation term restores the average bus voltage to the nominal bus voltage. Through integrator adjustment, the average bus voltage is compensated under the control of the correction term.

[0084] (18)

[0085] in K IU It is the integral gain of the average voltage compensation term, and the average bus voltage. u avgoi It can be represented as:

[0086] (19)

[0087] like Figure 3 (a) shows the improved droop coefficient proposed in this invention. R vi Differences in state of charge ζ The changing image, (b) is the novel acceleration factor proposed in this invention. ρ ( S i Differences in state of charge ζ The changing graph. For energy storage cells with different initial states of charge, their droop coefficients vary with... k = k D / C i The increase in droop coefficients leads to a greater difference in power distribution among energy storage units, thereby accelerating the balancing of the state of charge (SOC) among the energy storage units. However, in the later stages of SOC balance, the difference in droop coefficients among energy storage units gradually decreases, requiring only minor modifications. k The value is insufficient to achieve the necessary rapid state-of-charge equilibrium. For example... Figure 3 As shown in (a), while maintaining k While keeping the value constant, gradually increase ρ Value, with ζ As the droop coefficient approaches zero, its slope gradually increases. This change exacerbates the difference in droop coefficients between energy storage units, thereby increasing power distribution discrepancies. However, excessively high droop coefficients... ρ This value would hinder the equilibrium velocity during the initial equilibrium phase of state-of-charge regulation. Therefore, this study introduces an acceleration factor based on a normal distribution function, expressed as: ρ ( S i It replaces the traditional droop coefficient. ρ Value. By adjusting the regulating factor in the acceleration factor. μ , ρ The value remains low during the initial equilibrium phase of state of charge regulation and gradually increases during subsequent equilibrium phases, thereby accelerating the overall state of charge equilibrium rate.

[0088] like Figure 4This is an equivalent linearized model of the control strategy of this invention, taking two energy storage units as an example. Assuming there is no delay in the communication system, the closed-loop transfer function of the DC voltage can be expressed as follows:

[0089] (20)

[0090] in, τ This is the time constant under switching action. The converter output current is filtered by a low-pass filter and input into the control system loop. Its transfer function can be expressed as:

[0091] (twenty one)

[0092] in, ω c It is the cutoff frequency of the low-pass filter.

[0093] The transfer functions of the PI regulator for the unit virtual voltage drop balancing term and the integrator for the average voltage compensation term are:

[0094] (twenty two)

[0095] (twenty three)

[0096] from Figure 4 The following relationship can be derived from this:

[0097] (twenty four)

[0098] from Figure 4 The following expression can be obtained:

[0099] (25)

[0100] (26)

[0101] Because the bandwidth of the droop control loop is designed to be much smaller than the switching frequency, therefore τ The value is very small, and its delay effect can be ignored. Other system parameters are shown in Table 1. Combining (20)-(26), the closed-loop transfer function between the output voltage of energy storage unit 1 and the reference voltage can be expressed as:

[0102] (27)

[0103] in, , .

[0104] The process of adopting an improved droop control method, which improves the droop coefficient based on the sigmoid function to achieve adaptive adjustment of the lithium battery output current distribution to achieve state of charge balance, and adding an acceleration factor based on the normal distribution function to accelerate the later stage of lithium battery state of charge balance, also includes:

[0105] Based on dual closed-loop control of voltage and current, improved droop control, unit virtual voltage equalization term, and average bus voltage compensation term are added after the reference voltage in the outer voltage loop. The output voltage of the energy storage unit is fed back by the voltage loop. The reference current of the inner current loop is obtained through the first PI controller, and then the difference is calculated with the inductor current of the DC / DC converter. The difference is then used to obtain the PWM modulation signal through the second PI controller. The PWM modulation signal controls the switching on and off of the IGBT in the DC / DC converter to achieve rapid equalization of the state of charge of the lithium battery, accurate distribution of the output current according to the rated capacity, and average bus voltage compensation.

[0106] The process of adopting an improved droop control method, which improves the droop coefficient based on the sigmoid function, to adaptively adjust the lithium battery output current distribution to achieve state-of-charge balance is as follows:

[0107] During lithium battery charging, the difference in charging depth of each cell is established based on the sigmoid function, that is, the relationship between the difference between the charging depth and its average value and the droop coefficient. At the same time, the ratio of the balance coefficient to the rated capacity is introduced into the droop coefficient.

[0108] During lithium battery discharge, the relationship between the difference in discharge depth of each cell, i.e. the difference between the discharge depth and its average value, and the droop coefficient is established based on the sigmoid function. At the same time, the ratio of the balance coefficient to the rated capacity is introduced into the droop coefficient.

[0109] The process of adding an acceleration factor based on a normal distribution function to accelerate the later-stage state-of-charge balancing speed of lithium batteries is as follows:

[0110] During lithium battery charging, an acceleration factor based on a normal distribution function is introduced into the improved droop coefficient. The acceleration factor is a variable with the square of the difference in charging depth. At the same time, an adjustment factor is added to change the speed in the later stage of state-of-charge equilibrium.

[0111] During lithium battery discharge, an acceleration factor based on a normal distribution function is introduced into the improved droop coefficient. The acceleration factor is a variable with the square of the difference in discharge depth. At the same time, an adjustment factor is added to change the speed in the later stage of state-of-charge equilibrium.

[0112] The process by which the design unit uses a virtual voltage drop balancing term to eliminate the influence of line impedance and achieves output current distribution according to the rated capacity of the lithium battery is as follows:

[0113] The virtual voltage drop is obtained by multiplying the droop factor by the output current of the energy storage unit.

[0114] The ratio of virtual voltage drop to the maximum deviation of bus voltage is the unit virtual voltage drop. A PI controller is used to equalize the unit virtual voltage drop to the average unit virtual voltage drop. The unit virtual voltage drop of each energy storage unit is consistent, so that the output current distribution is only related to the droop coefficient. This eliminates the influence of line impedance on the accurate current splitting and state of charge balance of the output current. When the state of charge is balanced, the output current can be distributed proportionally according to the rated capacity of each energy storage unit.

[0115] The process of compensating the average bus voltage to the reference value using the design average bus voltage compensation item is as follows:

[0116] The average bus voltage is the average value of the output voltage of all energy storage units. An integral controller is used to make the average bus voltage approach the bus voltage reference value and maintain the stability of the bus voltage.

[0117] like Figure 5 The figure shows the root locus of important system parameters. Figure (a) shows the variation with the state of charge Δ. S = S 1 -S 2. The root locus varies with the cutoff frequency. Figure (b) shows the root locus as a function of the cutoff frequency ω c The root locus changes, and Figure (c) shows the change with the equilibrium factor. k D The root locus of the change, Figure (d) shows the change with the adjustment factor. μ The changing root locus;

[0118] The parameters of each power supply unit selected for the adaptive distributed hierarchical control method of consistency-based energy storage system are shown in Table 1.

[0119] Table 1. Parameters of the Adaptive Distributed Hierarchical Control Method for Consistency-Based Energy Storage Systems

[0120]

[0121] from Figure 5 (a) It can be seen that the system can remain stable when the difference in state of charge between energy storage units changes. From Figure 5 (b) It can be seen that the cutoff frequency ω c Increasing the value of will cause all poles to move away from the imaginary axis, thereby improving the stability margin of the system. However, conjugate poles... p 1 and p 2. Motion away from the real axis indicates a decrease in the system's damping ratio, which may lead to current oscillations and system instability. Therefore, the cutoff frequency must be adjusted. ω c Keep it within a reasonable range. Figure 5 (c) and (d) show that, with the balance factork D and acceleration factor μ The increase of the adjustment coefficient, the pole p As the system gradually approaches the imaginary axis, its stability margin decreases, thus necessitating control. k D and μ This prevents them from becoming too large, thereby ensuring the stability of the system.

[0122] like Figure 6 As shown, Case 1 presents simulation results under load switching and changes in light intensity. The initial state of charge of the energy storage units is set to 80%, 70%, and 60%, respectively. Furthermore, in this case, the simulated capacity of all energy storage units is 8 Ah. From... Figure 6 (a) It can be seen that the convergence speed of the state of charge (SOC) does not slow down due to the decrease in SOC difference, but rather accelerates continuously under the regulation of the acceleration factor. This proves the rapid SOC balancing capability of the proposed method. The SOC of all energy storage units converges at approximately t=20s, with the SOC difference approaching 0%. From Figure 6 (b) It can be seen that state-of-charge balance is achieved through the distribution of load current among energy storage units. From Figure 6 As shown in (c), the average bus voltage can recover to the nominal value of 700V in steady state, and only a small voltage deviation occurs during changes in RES, EL, and RL power. The results of simulation case 1 demonstrate that the proposed control strategy can achieve dynamic state-of-charge balance, accurate load current distribution, and stable bus voltage recovery under varying photovoltaic and load power conditions.

[0123] like Figure 7 As shown, Case 2 presents simulation results for energy storage units with different rated capacities under varying photovoltaic and load power conditions. Figure 7 (a) It can be seen that the proposed method still has a relatively fast convergence speed in the later stages of state-of-charge equilibrium. Compared with Case 1, due to the reduction in the total capacity of the energy storage units, each energy storage unit achieves the state-of-charge equilibrium target at approximately t=14s. From Figure 7 (b) It can be seen that the output current of each energy storage unit is according to i o1 : i o2 : i o3 The ratio of 4:3:2 indicates that the proposed control strategy has the ability to proportionally distribute the load current. Figure 7(c) It can be seen that the average bus voltage can maintain the nominal value of 700V under steady state. Although voltage deviations may occur during transient processes of wind, solar, and load power fluctuations, these deviations are quickly corrected back to the nominal value. The results of simulation case 2 show that, despite the different rated capacities of the energy storage units, the proposed control strategy can achieve the control objectives of rapid state-of-charge balance, proportional distribution of load current, and stable bus voltage recovery.

[0124] like Figure 8 As shown, Case 3 presents simulation results under conditions of communication and equipment failure. From Figure 8 (a) It can be seen that after achieving state-of-charge balance at t=14s, the proposed method can maintain this balance even during communication failures. After energy storage unit 3 prematurely terminated operation, its state of charge stabilized at 64.07%, while the states of charge of energy storage units 1 and 2 quickly rebalanced. From Figure 8 (b) It can be seen that although communication between energy storage unit 1 and energy storage unit 2 fails, the load current distribution between the energy storage units is unaffected after allocation according to the rated capacity. In the case of early termination of energy storage unit 3, the output current of energy storage unit 1 and energy storage unit 2 increases and continues according to... i o1 : i o2 A 4:3 ratio distribution. From Figure 8 (c) It can be seen that in the event of a communication failure, the average bus voltage can still remain near the nominal bus voltage. However, after energy storage unit 3 prematurely exits operation, the remaining two energy storage units bear all the load power, resulting in only a slight decrease in the average bus voltage of about 2.3V. The results of simulation case three show that the proposed control strategy can withstand the risks of communication failures and equipment disconnection. Under these conditions, it can still maintain rapid state-of-charge balance, accurate load current distribution, and stable bus voltage recovery.

[0125] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A consensus-based adaptive distributed hierarchical control method for energy storage systems, characterized in that: The method comprises the following steps: S1: establishing an off-grid photovoltaic water electrolysis hydrogen production system model comprising a photovoltaic cell, a lithium battery, an electrolytic cell, a constant power load and a DC / DC converter; S2: based on the off-grid photovoltaic water electrolysis hydrogen production system model, an improved droop hierarchical control method for the energy storage system based on multi-agent consistency is used to realize the state of charge balance of the energy storage system, eliminate the influence of line impedance to realize accurate shunt, and recover the bus voltage drop at the same time; The process of the improved droop hierarchical control method for the energy storage system based on multi-agent consistency, realizing the state of charge balance of the energy storage system, eliminating the influence of line impedance to realize accurate shunt, and recovering the bus voltage drop at the same time is as follows: In the communication layer, the energy storage units in the photovoltaic water electrolysis hydrogen production system are regarded as agents, an average state estimation method is designed based on multi-agent consistency, the average values of the lithium battery state of charge, unit virtual voltage and output voltage are calculated, and the average values are transmitted to adjacent energy storage units through an inter-neighbor communication network; In the main control layer, an improved droop control method is used, the droop coefficient is improved based on the sigmoid function, the lithium battery output current distribution is adaptively adjusted to achieve state of charge balance, and an acceleration factor improved based on the normal distribution function is added to accelerate the state of charge balance speed of the lithium battery in the later stage; In the secondary control layer, a unit virtual voltage drop balance term is designed to eliminate the influence of line impedance, the output current is distributed according to the rated capacity of the lithium battery, and an average bus voltage compensation term is designed to compensate the average bus voltage to the reference value; The process of the improved droop control method, the droop coefficient improved based on the sigmoid function, the adaptive adjustment of the lithium battery output current distribution to achieve state of charge balance, and the acceleration factor improved based on the normal distribution function to accelerate the state of charge balance speed of the lithium battery in the later stage further comprises: Voltage-current double-loop control is used as the basis, in the voltage outer loop, the improved droop control, the unit virtual voltage balance term and the average bus voltage compensation term are added after the reference voltage, the energy storage unit output voltage is fed back to the voltage loop, the reference current of the current inner loop is obtained through the first PI controller, and then the reference current is subtracted from the DC / DC converter inductance current to obtain the PWM modulation signal through the second PI controller, the PWM modulation signal controls the on-off of the IGBT in the DC / DC converter to realize the fast state of charge balance of the lithium battery, the accurate distribution of the output current according to the rated capacity and the compensation of the average bus voltage.

2. The adaptive distributed hierarchical control method for a consistent-based energy storage system of claim 1, wherein: The process of the improved droop control method, the droop coefficient improved based on the sigmoid function, and the adaptive adjustment of the lithium battery output current distribution to achieve state of charge balance is as follows: When the lithium battery is charged, the relationship between the charging depth difference of each unit, i.e., the difference between the charging depth and the average value, and the droop coefficient is established based on the sigmoid function, and the ratio of the balance coefficient to the rated capacity is introduced into the droop coefficient; When the lithium battery is discharged, the relationship between the discharge depth difference of each unit, i.e., the difference between the discharge depth and the average value, and the droop coefficient is established based on the sigmoid function, and the ratio of the balance coefficient to the rated capacity is introduced into the droop coefficient.

3. The adaptive distributed hierarchical control method for a consistent-based energy storage system of claim 1, wherein: The process of adding the acceleration factor based on the normal distribution function to improve the equalization speed of the lithium battery state of charge in the later stage is as follows: During charging of the lithium battery, an acceleration factor based on the normal distribution function is introduced into the improved droop coefficient, the acceleration factor takes the square value of the charging depth difference as a variable, and a regulating factor is added to change the speed in the later stage of the state of charge equalization; During discharging of the lithium battery, an acceleration factor based on the normal distribution function is introduced into the improved droop coefficient, the acceleration factor takes the square value of the discharging depth difference as a variable, and a regulating factor is added to change the speed in the later stage of the state of charge equalization.

4. The adaptive distributed hierarchical control method for a consistent-based energy storage system of claim 1, wherein: The process of designing a unit virtual pressure drop balancing term to eliminate the influence of the line impedance and realize the output current distribution according to the rated capacity of the lithium battery is as follows: The product of the droop coefficient and the output current of the energy storage unit is a virtual pressure drop, The ratio of the virtual pressure drop to the maximum bus voltage offset is a unit virtual pressure drop, a PI controller is used to make the unit virtual pressure drop equal to the average unit virtual pressure drop, and the unit virtual pressure drops of all energy storage units are consistent, so that the output current distribution is only related to the droop coefficient, thereby eliminating the influence of the line impedance on the accurate shunt of the output current and the state of charge equalization, and the output current can be distributed in proportion to the rated capacity of each energy storage unit when the state of charge is equalized.

5. The adaptive distributed hierarchical control method for a consistent-based energy storage system of claim 1, wherein: The process of designing an average bus voltage compensation term to compensate for the average bus voltage to the reference value is as follows: The average bus voltage is the average value of the output voltages of all energy storage units, and an integral controller is used to make the average bus voltage tend to the bus voltage reference value, thereby maintaining the stability of the bus voltage.

6. The adaptive distributed hierarchical control method for a consistent-based energy storage system of claim 1, wherein: The improved droop coefficient is represented as: (6) (7) Wherein: R vi ( S i ), ρ ( S i ) and k D is the improved droop coefficient of energy storage unit i , variable acceleration factor and balance factor, μ is the adjustment factor of acceleration factor; wherein the charge and discharge deviation value ζ and σ is expressed as: (8) (9) wherein S avgi is an energy storage unit i of a dynamic average value, D avgi is an energy storage unit i of a dynamic value; SoCi is the state of charge of the energy storage unit i; SoD is the depth of discharge of the energy storage unit i.

Images