A control method and device of a matrix direct current energy storage system under off-grid conditions

Abstract

This invention belongs to the field of control for electrochemical energy storage battery charge-discharge power conversion systems, and provides a control method and device for a matrix-type DC energy storage system under off-grid conditions. The method generates a droop voltage compensation amount based on the difference between the DC bus voltage command and the feedback voltage, correcting the matrix-level droop characteristic curve; allocates a cluster-level droop coefficient reference value based on the state of charge of each PEP cluster, generates an adjustment droop coefficient in conjunction with dynamic adjustment, and executes voltage droop control to obtain a cluster voltage reference value; the cluster voltage reference value is evenly distributed to obtain a PEP voltage reference; a balancing fine-tuning amount is generated based on the difference between the SOC of each PEP within the cluster and the average SOC of the cluster, and these are superimposed to obtain the port voltage reference value, which is then executed for control. This invention achieves stable support of the bus voltage, reasonable power distribution between clusters, and balanced adjustment of the battery state within the cluster under off-grid conditions.

Description

Technical Field

[0001] This invention belongs to the technical field of electrochemical energy storage battery charge and discharge power conversion system, and relates to a control method and device for a matrix DC energy storage system under off-grid conditions. Background Technology

[0002] Battery energy storage systems typically employ multiple individual battery cells to form battery modules, which are then connected in series to create a high-voltage system. When the State of Charge (SOC) of these series-connected battery modules is unbalanced, if one module fully charges or fully discharges, the entire cluster of modules shuts down, reducing battery efficiency. Continuing to charge or discharge will lead to overcharging or over-discharging, posing safety hazards. To address this issue, existing technologies incorporate a DC-DC converter into each battery pack. By adjusting the power of the DC-DC converter, the charging and discharging power of each battery module is adjusted, achieving balanced state control. In the DC bus of a microgrid, a matrix DC energy storage system consists of multiple power electronic pack (PEP) clusters connected in parallel, with multiple PEPs connected in series within each cluster. When grid-connected equipment on the DC bus shuts down or the external grid loses power, the DC bus voltage is controlled and supported by the matrix DC energy storage system, requiring coordinated control of the bus voltage among the multiple clusters. However, existing multi-cluster control bus voltage schemes typically employ droop control to achieve average power distribution. But in PEP series-parallel systems, to balance battery states, there are issues of inter-cluster and intra-cluster power distribution, rather than simple equal distribution. Furthermore, during power adjustments or fluctuations, the dynamic process of droop control may overshoot, lacking an effective dynamic correction mechanism. Summary of the Invention

[0003] The purpose of this invention is to solve the technical problem in the prior art that it is difficult to balance reasonable power allocation and dynamic stability in the multi-cluster coordinated control of matrix DC energy storage systems under off-grid conditions, and to provide a control method and device for matrix DC energy storage systems under off-grid conditions.

[0004] To achieve the above objectives, the present invention employs the following technical solution: The first aspect of this invention provides a control method for a matrix-type DC energy storage system under off-grid conditions. The matrix-type DC energy storage system includes a PEP cluster composed of multiple PEPs connected in series, and a PEP matrix composed of multiple PEP clusters connected in parallel. The method includes the following steps: The droop voltage compensation amount is generated based on the difference between the preset DC bus voltage command and the bus feedback voltage, and the droop voltage compensation amount is used to correct the matrix-level droop characteristic curve. Based on the charge state of each PEP cluster, the cluster-level droop coefficient baseline value of each PEP cluster is assigned according to the modified matrix-level droop characteristic curve. Each PEP cluster receives its own cluster-level droop coefficient reference value and adds the dynamic adjustment amount of its own cluster to generate an adjustment droop coefficient. Based on the adjustment droop coefficient, voltage droop control is performed to generate the cluster voltage reference value of its own cluster. The dynamic adjustment amount is generated through transient optimization of droop characteristics. The cluster voltage reference value is divided equally according to the number of PEPs within the cluster to obtain the PEP voltage reference. The equalization fine-tuning amount is generated based on the difference between the SOC of each PEP in the cluster and the average SOC of the cluster. The PEP voltage reference is superimposed with the equalization fine-tuning amount to obtain the port voltage reference value of each PEP, and port voltage control is performed.

[0005] As an embodiment of the present invention, the droop voltage compensation amount is used to shift or rotate the droop characteristic curve so that the bus voltage is compensated to the target voltage.

[0006] As an embodiment of the present invention, the transient optimization of droop characteristics specifically includes: The difference between the cluster voltage reference value and the cluster voltage feedback value is taken as the cluster voltage error; The cluster voltage error is converted into a cluster current reference value, and the difference between the cluster current reference value and the cluster current feedback value is taken as the cluster current error. A dynamic adjustment amount is generated based on the cluster current error.

[0007] As an embodiment of the present invention, the cluster current error is dynamically adjusted using a PI regulator.

[0008] As an embodiment of the present invention, the method for generating the equalization fine-tuning amount includes: The difference between the cluster voltage reference value and the cluster voltage feedback value is calculated, and the cluster voltage error is output. The cluster voltage error is normalized by performing absolute value calculation and slope limitation on the cluster voltage error. The SOC deviation is obtained based on the difference between the SOC of each PEP and the cluster average SOC. Based on the normalized cluster voltage error and the SOC deviation, a voltage allocation ratio coefficient is generated; The voltage distribution ratio coefficient is multiplied by the cluster voltage error and combined with the charging / discharging direction sign to generate a directional voltage adjustment amount, which is then used as the equalization fine-tuning amount.

[0009] As an embodiment of the present invention, the generated voltage distribution ratio coefficient is specifically as follows: Based on a preset voltage distribution ratio coefficient benchmark, the voltage distribution ratio coefficient is generated through multiplication logic or lookup table logic according to the normalized cluster voltage error and the SOC deviation; at the same time, the rate of change of the coefficient is limited by the slew rate parameter to output a stable voltage distribution ratio coefficient.

[0010] As an embodiment of the present invention, the rate of change of the voltage distribution ratio coefficient is adjusted by a preset slew rate parameter; the slew rate parameter specifies the maximum allowable change of the voltage distribution ratio coefficient per unit time.

[0011] A second aspect of the present invention provides a control device for a matrix-type DC energy storage system under off-grid conditions, comprising: A matrix-level controller is used to generate a droop voltage compensation amount based on the difference between the DC bus voltage command and the bus feedback voltage, correct the matrix-level droop characteristic curve using the droop voltage compensation amount, and allocate a cluster-level droop coefficient reference value for each PEP cluster based on the SOC of each PEP cluster and the corrected matrix-level droop characteristic curve. At least one cluster-level controller, each cluster-level controller is connected to the corresponding PEP cluster, for receiving the cluster-level droop coefficient reference value of the cluster, and generating an adjustment droop coefficient in combination with the dynamic adjustment amount of the cluster, performing voltage droop control based on the adjustment droop coefficient, generating a cluster voltage reference value of the cluster, dividing the cluster voltage reference value equally according to the number of PEPs in the cluster to obtain the PEP voltage reference, and generating an equalization fine adjustment amount based on the difference between the SOC of each PEP in the cluster and the average SOC of the cluster. Multiple unit-level controllers are provided, each connected to a corresponding PEP, for receiving the PEP voltage reference and the equalization fine-tuning amount, superimposing the two as the port voltage reference value of the PEP, and performing port voltage control.

[0012] As an embodiment of the present invention, the cluster-level controller further includes: The droop transient optimization module is used to generate the dynamic adjustment amount based on the voltage error and current error of this cluster; The voltage distribution limit dynamic adjustment module is used to generate the equalization fine-tuning amount based on the cluster voltage error and the SOC deviation of each PEP in the cluster.

[0013] A third aspect of the present invention provides a matrix-type DC energy storage system, comprising: DC bus; Multiple PEP clusters, each PEP cluster consisting of multiple PEPs connected in series, and the multiple PEP clusters are connected in parallel to the DC bus; The control device for the above-mentioned matrix DC energy storage system under off-grid conditions includes a matrix-level controller connected to the DC bus, a cluster-level controller connected to the matrix-level controller and the corresponding PEP cluster, and a single-unit controller connected to the corresponding cluster-level controller and the corresponding PEP.

[0014] Compared with the prior art, the present invention has the following beneficial effects: This invention discloses a control method for a matrix-type DC energy storage system under off-grid conditions. It adopts a three-layer collaborative control architecture of matrix level, cluster level, and individual cell level. Under off-grid conditions, it achieves individual cell control by correcting the bus voltage through droop voltage compensation, allocating cluster-level power according to SOC, and evenly distributing cluster voltage and superimposing equalization fine-tuning. It can simultaneously achieve three core objectives: stable support of DC bus voltage, reasonable power allocation between clusters, and balanced adjustment of PEP state within clusters. It solves the technical defects of traditional off-grid control that cannot take into account both voltage stability and battery balance, and significantly improves the system's operational reliability and energy utilization rate.

[0015] Furthermore, by shifting or rotating the matrix-level droop characteristic curve through droop voltage compensation, the DC bus voltage deviation can be quickly offset, and the bus voltage can be accurately compensated to the target voltage of the matrix-level droop characteristic. This achieves zero steady-state error stable control of the DC bus voltage under off-grid conditions, avoiding the voltage offset problem inherent in traditional droop control.

[0016] Furthermore, by introducing dynamic adjustment based on voltage and current errors into cluster-level control, the droop control response can be dynamically corrected in real time, effectively suppressing voltage overshoot and oscillation during load surges and power fluctuations, improving the system's dynamic response speed and anti-interference capability, and ensuring the system's stable operation in off-grid conditions.

[0017] Furthermore, by employing a dual-closed-loop transient optimization logic that converts cluster voltage error into cluster current reference and then generates dynamic adjustment amount from cluster current error, it is possible to quickly track voltage and current deviations, accurately correct droop characteristics, further reduce voltage fluctuations in the dynamic process, and make cluster-level control response faster and adjustment smoother.

[0018] Furthermore, the voltage distribution limit is dynamically adjusted based on the cluster voltage error and the deviation of PEPSOC, thereby achieving intra-cluster equalization control while ensuring cluster voltage stability. This avoids excessive equalization action from interfering with the bus voltage, achieving synergistic optimization of voltage stability and individual cell equalization. The equalization process does not affect the overall system operation.

[0019] Furthermore, through the complete equalization fine-tuning generation logic of cluster voltage error normalization processing, SOC deviation calculation, voltage distribution coefficient generation, and charging / discharging direction sign superposition, the output of each PEP port can be precisely and directionally adjusted, so that cells with high SOC can charge and discharge faster, and cells with low SOC can charge and discharge slower. The equalization direction is clear, the adjustment range is controllable, and the difference in SOC between cells within the cluster can be quickly reduced.

[0020] Furthermore, by using preset slew rate parameters to limit the rate of change of the voltage distribution ratio coefficient, system oscillations caused by sudden changes in the equalization adjustment can be avoided, allowing the equalization process to transition smoothly. An adjustable balance is established between equalization speed and system stability, improving the control adaptability under different operating conditions. Attached Figure Description

[0021] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic diagram of the topology of a matrix DC energy storage system according to an embodiment of the present invention; Figure 2 This is a general control block diagram of the matrix DC energy storage system according to an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the principle of the transient optimization step for cluster-level droop characteristics in an embodiment of the present invention. Figure 4 This is a schematic diagram illustrating the principle of dynamic adjustment of PEP port voltage distribution limit in an embodiment of the present invention. Detailed Implementation

[0023] 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 components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0024] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0025] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0026] The present invention will now be described in further detail with reference to the accompanying drawings: Example 1: This embodiment provides a control method for a matrix DC energy storage system under off-grid conditions. This method is applicable to matrix DC energy storage systems (hereinafter referred to as the system) comprising PEP clusters composed of multiple PEPs connected in series, and PEP matrices composed of multiple PEP clusters connected in parallel. Figure 1 and Figure 2 As shown, this method constructs a three-layer control architecture of matrix level, cluster level, and single-unit level, specifically including the following steps: Step S100: Generate a droop voltage compensation amount based on the difference between the DC bus voltage command and the bus feedback voltage, and use the droop voltage compensation amount to correct the matrix-level droop characteristic curve.

[0027] Specifically, under off-grid operation conditions, the stability of the DC bus voltage is the primary objective of system control. The matrix-level controller acquires the bus feedback voltage in real time, calculates the deviation between it and the preset DC bus voltage command, and obtains the droop voltage compensation amount. This droop voltage compensation amount reflects the degree of imbalance between power supply and demand in the system. By generating the droop voltage compensation amount, the system can dynamically correct the matrix-level droop characteristic curve. For example, when the bus feedback voltage is lower than the command value, the generated droop voltage compensation amount will shift or rotate the droop characteristic curve, thereby adjusting the output power of each PEP cluster to support the bus voltage to recover to the voltage compensation target value, i.e., the target voltage.

[0028] Step S200: Based on the charge state of each PEP cluster, assign a cluster-level droop coefficient baseline value to each PEP cluster according to the modified matrix-level droop characteristic curve.

[0029] Specifically, the state of battery (SOC) often differs among different PEP clusters. To prevent some PEP clusters from reaching their charge / discharge limits prematurely and affecting the overall system capacity, this embodiment introduces a power allocation mechanism based on voltage droop. The matrix-level controller determines the corresponding operating point on the corrected droop characteristic curve based on the SOC data reported by each PEP cluster, then calculates the power share each PEP cluster should bear, and converts this into a cluster-level droop coefficient reference value, which is then sent to each cluster-level controller. For example, PEP clusters with higher SOC are assigned a smaller droop coefficient reference value during discharge, thus bearing a larger output current; PEP clusters with lower SOC are assigned a larger droop coefficient reference value, bearing a smaller output current. This allocation method achieves active SOC balancing between clusters while maintaining stable bus voltage.

[0030] In step S300, each PEP cluster receives its own cluster droop coefficient reference value and generates an adjustment droop coefficient by combining it with the dynamic adjustment amount of the cluster. Based on the adjustment droop coefficient, voltage droop control is performed to generate a cluster voltage reference value for the cluster.

[0031] Specifically, the cluster-level controller, acting as an intermediate layer, not only receives the cluster-level droop coefficient reference value from the upper layer but also handles the dynamic characteristics within its own cluster. The cluster droop coefficient reference value provides the static power allocation basis, while the dynamic adjustment is used to cope with dynamic disturbances such as sudden load changes and line impedance differences. The cluster-level controller superimposes the cluster droop reference value and the dynamic adjustment value to obtain the adjustment droop coefficient, and then applies the droop control formula, such as... ,in, Indicates the cluster voltage reference value. Indicates the rated voltage. This indicates the adjustment of the droop coefficient. Indicates the output current, and calculates the reference value of the cluster voltage that this cluster should output. This step ensures that each PEP cluster, when operating in parallel off-grid mode, can both follow the upper-layer power allocation strategy and maintain its own operational stability.

[0032] Step S400: Divide the cluster voltage reference value equally according to the number of PEPs in the cluster to obtain the PEP voltage reference.

[0033] Specifically, since a PEP cluster consists of multiple PEPs connected in series, the cluster voltage reference value needs to be distributed across each PEP. In the initial stage, to simplify the control logic and ensure balanced voltage stress across individual units, the cluster-level controller divides the cluster voltage reference value by the number of PEPs connected in series within the cluster to obtain an evenly distributed PEP voltage reference. The PEP voltage reference is the target value for voltage tracking by the individual unit-level controller.

[0034] Step S500: Generate a balancing fine-tuning amount based on the difference between the SOC of each PEP in the cluster and the average SOC of the cluster. Superimpose the PEP voltage reference with the balancing fine-tuning amount to obtain the port voltage reference value of each PEP, and perform port voltage control.

[0035] Specifically, the inconsistency in the State of Charge (SOC) of individual PEPs within a cluster limits the overall capacity of the cluster. This embodiment introduces a balancing fine-tuning mechanism at the individual cell level controller. The cluster-level controller monitors the SOC of each PEP within the cluster in real time and calculates the cluster average SOC. For PEPs with an SOC higher than the average, a positive balancing fine-tuning amount is generated during discharge and superimposed on the PEP voltage reference, appropriately increasing its port voltage reference value, thereby increasing its output power and accelerating its SOC decrease; conversely, its port voltage reference value is decreased. The final generated port voltage reference value for each PEP is sent to the corresponding individual cell level controller, which precisely controls the PEP port voltage to track this reference value by adjusting the duty cycle of the DC-DC converter. Through the synergistic effect of the above three-layer control architecture, this embodiment achieves stable support of bus voltage under off-grid conditions, reasonable power distribution between clusters, and balanced adjustment of battery state within the cluster, effectively improving the system's operating efficiency and safety.

[0036] Example 2: This embodiment, based on Embodiment 1, further optimizes the correction strategy for matrix-level droop characteristics and the dynamic adjustment mechanism for cluster-level droop coefficients. Specifically, considering the diversity of off-grid operation conditions, this embodiment provides more refined control logic to solve the problems of response lag and overshoot in traditional droop control during dynamic processes.

[0037] At the matrix-level control level, the droop voltage compensation amount is used to compensate the bus voltage to the target voltage or rated voltage of the matrix-level droop characteristics.

[0038] Specifically, the voltage control target of off-grid systems is not always fixed. In a typical application scenario, to ensure that the downstream load devices always operate at the optimal voltage point, the matrix-level controller sets the target for droop voltage compensation to the rated voltage. This means that compensation allows the bus voltage to quickly recover to its rated value after load fluctuations, resulting in zero steady-state error regulation. In another application scenario, if the Energy Management System (EMS) issues a specific voltage dispatch command, the target for droop voltage compensation is set to the target voltage. In this case, the system prioritizes responding to the upper-level dispatch command, allowing the bus voltage to deviate from its rated value within a certain range to achieve a specific power allocation objective. This flexible target selection mechanism enables matrix-type DC energy storage systems to adapt to the control requirements of different off-grid scenarios.

[0039] At the cluster-level control level, to improve the dynamic response performance of droop control, this embodiment introduces a transient optimization stage for droop characteristics. The dynamic adjustment amount is generated by using the transient optimization stage based on the voltage and current errors of the cluster to dynamically correct the droop response.

[0040] Combination Figure 3 As shown, the specific implementation logic of the transient optimization step for droop characteristics is as follows: Step S301: The difference between the cluster voltage reference value and the cluster voltage feedback value is taken as the cluster voltage error. The cluster voltage error is converted into a cluster current reference value. At the same time, the cluster-level droop coefficient reference value sent by the matrix level is received as the basic parameter.

[0041] Specifically, the cluster voltage reference value is the theoretical output voltage calculated by droop control, while the cluster voltage feedback value is the actual measured voltage at the current PEP cluster port. The difference between the two is the cluster voltage error, reflecting the degree to which the current cluster output voltage deviates from the expected target. Through a voltage-to-current conversion stage—in this embodiment, the voltage-to-current conversion stage uses a proportional-integral (PI) regulator—this voltage error is converted into a cluster current reference value, which serves as the instantaneous target for the cluster output current.

[0042] Step S302: The difference between the cluster current reference value and the cluster current feedback value is taken as the cluster current error, and the dynamic adjustment amount (i.e., cluster droop coefficient adjustment amount) is generated from the cluster current error.

[0043] Specifically, the cluster current feedback value is the cluster's output current acquired in real time by sensors. Comparing the cluster current reference value with the cluster current feedback value yields a cluster current error, reflecting the actual current's deviation from the commanded current's tracking. This current error is processed by a droop coefficient fine-tuning PI controller to generate a dynamic adjustment quantity. This dynamic adjustment quantity is not directly superimposed on the voltage reference value but is used to correct the cluster-level droop coefficient or directly as a dynamic compensation component of the cluster-level droop coefficient. The cluster-level droop coefficient reference value issued at the matrix level is superimposed with the generated dynamic adjustment quantity to obtain the final cluster-level adjusted droop coefficient, i.e., the adjusted droop coefficient of the PEP cluster, which serves as the actual execution coefficient for cluster-level voltage droop control.

[0044] The design mechanism of this closed-loop logic lies in the fact that traditional static droop control relies solely on voltage deviation for adjustment. During sudden load changes, due to the inertia of voltage changes, power distribution often experiences a delay, easily leading to voltage overshoot or oscillation. This embodiment, however, introduces a dual-loop correction mechanism: converting voltage error to current reference to adjust current error, and then using current error to adjust the dynamic adjustment amount. In the initial stage of voltage deviation, the current loop responds quickly, actively adjusting the slope or intercept of the droop characteristic. This dynamic correction effectively suppresses voltage overshoot during transient processes, significantly improving the system's dynamic response speed and stability. It should be understood that... Figure 3The logic shown is only a preferred implementation. In practical applications, the generation of dynamic adjustment can also be based on the system damping requirements by introducing a differential element or a feedforward element. As long as the closed-loop dynamic correction is based on voltage error and current error, it should be considered within the scope of protection of this invention.

[0045] Example 3: This embodiment, based on Embodiment 1, provides a detailed explanation of the generation logic for intra-cluster single-unit level equalization control. Under off-grid operation conditions, equalization control not only aims to achieve uniform State of Charge (SOC) across all units but also ensures that excessive equalization actions do not affect the stability of the cluster-level voltage output. To this end, this embodiment proposes a method for generating equalization fine-tuning amounts, specifically including: dynamically adjusting the voltage distribution limits of each PEP port based on the cluster voltage error and the SOC deviation of each PEP within the cluster, thereby obtaining the equalization fine-tuning amount.

[0046] Combination Figure 4 As shown, the core of this generation logic lies in establishing a collaborative mechanism that prioritizes voltage stability and follows with equalization adjustment. Specifically, the steps for dynamically adjusting the voltage distribution limits of each PEP port are as follows: Step S501: The inputs in this step include the cluster voltage reference value, cluster voltage feedback value, preset positive limit value of cluster voltage error, SOC of each PEP individual related to intra-cluster equalization, average SOC of PEP cluster, and PEP voltage reference obtained by equalizing the cluster voltage reference value; at the same time, the slew rate parameter and the cluster voltage distribution ratio coefficient reference are configured as constraints and basic parameters for equalization adjustment.

[0047] First, the cluster voltage error is generated by using the cluster voltage reference value and the cluster voltage feedback value. Then, the absolute value of the cluster voltage error is calculated and the rising and falling slopes are limited. Finally, the error is divided by the positive limit value of the cluster voltage error to obtain the normalized cluster voltage error.

[0048] Specifically, the cluster voltage error is the difference between the cluster voltage reference value and the cluster voltage feedback value, reflecting the stability of the current cluster output voltage. First, the absolute value of the cluster voltage error is taken to eliminate directional influence, focusing only on the deviation amplitude. Then, the cluster voltage error signal is processed by limiting the rising and falling slopes. Here, slope limiting refers to restricting the rate of change of the error signal to prevent instantaneous jumps in error due to sudden load changes. After processing, the error is divided by a preset positive limit value for cluster voltage error, mapping it to the interval between 0 and 1, thus obtaining the normalized cluster voltage error. The closer this normalized value is to 1, the larger the current cluster voltage deviation, indicating that the system is in a state of large dynamic fluctuation; the closer the value is to 0, the more stable the system operation. This processing provides a weighting basis for subsequent dynamic adjustment of the balancing force.

[0049] Step S502: Based on the difference between the SOC of each PEP individual and the average SOC of the PEP cluster calculated in real time, the SOC deviation of each PEP is obtained.

[0050] Specifically, the cluster-level controller collects the SOC data of each PEP cell within the cluster in real time and calculates the arithmetic mean to obtain the average SOC of the PEP cluster. Subtracting the SOC of each PEP cell from this average value yields the SOC deviation of each PEP. This SOC deviation includes not only magnitude but also direction: a positive value indicates that the SOC of that cell is higher than the average level, requiring an increase in discharge power or a decrease in charging power; a negative value indicates the opposite.

[0051] Step S503: Based on the normalized cluster voltage error and the SOC deviation, and combined with the preset cluster voltage allocation ratio coefficient benchmark, the voltage allocation ratio coefficient of each PEP is generated through multiplication logic or table lookup logic.

[0052] Specifically, the voltage distribution ratio coefficient is a key parameter determining the magnitude of the equalization fine-tuning. This embodiment employs multiplication logic or lookup table logic, using the normalized cluster voltage error as an adjustment factor for the SOC deviation. For example, when the normalized cluster voltage error is large (e.g., close to 1), it indicates that the system is under conditions of significant voltage fluctuation. In this case, the amplitude of the equalization action should be appropriately suppressed, and the generated voltage distribution ratio coefficient will decrease accordingly, thereby reducing the equalization fine-tuning and prioritizing voltage stability. Conversely, when the system is running smoothly, the voltage distribution ratio coefficient increases, allowing for more forceful equalization adjustments. This mechanism effectively solves the problem that traditional equalization strategies may exacerbate voltage fluctuations during system dynamic processes.

[0053] Furthermore, the rate of change of the voltage distribution ratio coefficient is adjusted by a preset slew rate parameter.

[0054] Specifically, to avoid abrupt changes in the equalization fine-tuning amount due to sudden changes in the voltage distribution ratio coefficient, which could lead to system oscillations, this embodiment introduces a slew rate parameter. The slew rate parameter specifies the maximum allowable change in the voltage distribution ratio coefficient per unit time. By adjusting this slew rate parameter, the equalization adjustment process can be smoothed. For example, a larger slew rate parameter can be set in scenarios requiring rapid equalization; a smaller slew rate parameter can be set in scenarios with extremely high stability requirements.

[0055] Step S504: Multiply the voltage distribution ratio coefficient by the SOC deviation and combine it with the charging / discharging direction sign to generate a directional voltage adjustment amount, and use the voltage adjustment amount as the equalization fine-tuning amount.

[0056] Specifically, the voltage distribution ratio coefficient generated above is multiplied by the SOC deviation to obtain a controlled PEP individual port voltage adjustment. Then, the direction sign is determined based on the current charging / discharging state. For example, under discharging conditions, for PEPs with a high SOC, the direction sign is set to positive, making the final equalization fine-tuning a positive value. This value is superimposed on the PEP voltage reference, increasing the port voltage reference value of that individual unit, thereby increasing its output power and accelerating the SOC decrease. The final directional voltage adjustment is the equalization fine-tuning, which is superimposed on the PEP voltage reference to form the final PEP port voltage reference value. Through the above steps, this embodiment achieves decoupled coordination between equalization control and voltage stability, ensuring the safe and stable operation of the matrix DC energy storage system under off-grid conditions.

[0057] Example 4: This embodiment provides a control device for a matrix-type DC energy storage system under off-grid conditions. The device is used to implement the control method described in any one of embodiments 1 to 3. Its hardware architecture is designed based on a hierarchical control concept, specifically including a matrix-level controller, at least one cluster-level controller, and multiple individual-level controllers.

[0058] The matrix-level controller, as the top-level scheduling core of the system, is used to generate a droop voltage compensation amount based on the difference between the DC bus voltage command and the bus feedback voltage, correct the matrix-level droop characteristic curve using the droop voltage compensation amount, and allocate the cluster-level droop coefficient reference value of each PEP cluster based on the SOC of each PEP cluster and the corrected matrix-level droop characteristic curve.

[0059] Specifically, matrix-level controllers typically employ high-performance industrial control computers or digital signal processors (DSPs). Their inputs are connected to the DC bus via voltage sampling circuits to acquire the bus feedback voltage in real time. Their communication ports connect to each cluster-level controller via a Controller Area Network (CAN) bus, Ethernet, or fiber optic communication network. Internally, the matrix-level controller runs a top-level voltage control algorithm responsible for the overall system voltage stability and macroscopic power distribution. In off-grid mode, the matrix-level controller ensures the DC bus voltage remains within the rated range by correcting the droop characteristic curve. Simultaneously, it issues different droop coefficient reference values ​​based on the SOC differences among clusters, achieving a top-level strategy for inter-cluster SOC balancing.

[0060] The at least one cluster-level controller, each cluster-level controller being connected to a corresponding PEP cluster, is used to receive the cluster-level droop coefficient reference value of its own cluster, and generate an adjustment droop coefficient in combination with the dynamic adjustment amount of its own cluster. Based on the adjustment droop coefficient, voltage droop control is performed to generate a cluster voltage reference value of its own cluster. The cluster voltage reference value is then evenly distributed according to the number of PEPs in the cluster to obtain a PEP voltage reference, and an equalization fine-tuning amount is generated based on the difference between the SOC of each PEP in the cluster and the average SOC of the cluster.

[0061] Specifically, the cluster-level controller is a key node connecting top-level control and bottom-level execution. Each PEP cluster is equipped with one cluster-level controller, which is typically integrated into the cluster-level control cabinet. The cluster-level controller receives the cluster-level droop coefficient reference value from the matrix-level controller via a communication bus, and simultaneously collects the cluster voltage and current feedback values ​​of its own cluster through sensors. Internally, the cluster-level controller calculates the cluster voltage reference value using a droop control algorithm, and distributes it evenly as a PEP voltage reference to each individual cell-level controller. In addition, the cluster-level controller is also responsible for collecting the SOC information of each PEP within the cluster, calculating the cluster average SOC, and generating equalization fine-tuning parameters accordingly to achieve balanced management of the battery state within the cluster.

[0062] The plurality of individual level controllers are connected to the corresponding PEP and are used to receive the PEP voltage reference and the equalization fine-tuning amount, and to superimpose the two as the port voltage reference value of the PEP and perform port voltage control.

[0063] Specifically, the individual unit-level controller directly operates on the DC-DC converter within the power electronic battery pack (PEP). It receives the PEP voltage reference and equalization fine-tuning parameters from the cluster-level controller, which are then combined to form the final voltage control target. The individual unit-level controller adjusts the duty cycle of the DC-DC converter's switching transistors to control the PEP port voltage to accurately track this reference value, thereby achieving the low-level execution of power output and battery equalization. It should be understood that the individual unit-level controller and the PEP are typically integrated, with each PEP having its own independent control and communication interface.

[0064] In a preferred embodiment, the cluster-level controller further includes: a droop transient optimization module, used to generate the dynamic adjustment amount based on the voltage error and current error of the cluster; and a voltage distribution limit dynamic adjustment module, used to generate the equalization fine-tuning amount based on the cluster voltage error and the SOC deviation of each PEP within the cluster.

[0065] Specifically, combined Figure 3As shown, the droop transient optimization module is a software functional module or hardware logic circuit within the cluster-level controller. Its input receives the difference between the cluster voltage reference value and the cluster voltage feedback value (i.e., cluster voltage error) and the cluster current feedback value, and its output outputs a dynamic adjustment amount. This module achieves dynamic correction of the droop response and suppresses overshoot through the closed-loop logic described in Example 2, which converts voltage error to current reference, adjusts current error, and then uses current error to adjust the dynamic adjustment amount. The voltage distribution limit dynamic adjustment module is responsible for the equalization control logic. Its input receives the cluster voltage error and the SOC data of each PEP, and its output outputs an equalization fine-tuning amount. This module generates a directional voltage adjustment amount through the normalization processing, SOC deviation calculation, and slew rate limiting logic described in Example 3. These two modules work together to ensure voltage stability while also considering dynamic response performance and battery equalization efficiency. Through the cooperation of the above hardware architecture and functional modules, the control device of this embodiment can stably support the DC bus voltage under off-grid conditions and achieve multi-level efficient collaborative control.

[0066] Example 5: This embodiment provides a matrix-type DC energy storage system. This system aims to provide a physical platform and hardware support for the off-grid control method described in the preceding embodiments. For example... Figure 1 As shown, the system includes a DC bus, multiple PEP clusters, and a control device.

[0067] The DC bus serves as the main channel for energy transmission and distribution in the system, and its voltage level determines the operating status of the entire off-grid microgrid. The DC bus is typically connected to filter capacitors to smooth voltage fluctuations and provide a stable voltage support point for each PEP cluster.

[0068] The plurality of PEP clusters, each consisting of multiple PEPs connected in series, are connected in parallel to the DC bus.

[0069] Specifically, this matrix topology, with series connection within clusters and parallel connection between clusters, is suitable for high-voltage, high-capacity energy storage requirements. Physically, the positive and negative terminals of each PEP cluster are connected to the positive and negative busbars of the DC bus via switching devices (such as DC circuit breakers). The power ports of each PEP within a cluster are connected in series, accumulating to form a cluster-level high-voltage output. This structure allows the system to increase the voltage level by increasing the number of series-connected clusters and increase the power capacity by increasing the number of parallel clusters, exhibiting strong scalability. It should be understood that although... Figure 1 The example shows that each PEP cluster is directly connected in parallel. However, in actual engineering applications, the output terminals of each cluster may be connected in series with current-limiting reactors or fuses to suppress inter-cluster circulating currents and provide short-circuit protection. These are all equivalent transformations in this embodiment.

[0070] The control device adopts the architecture described in the foregoing embodiments, specifically including a matrix-level controller, a cluster-level controller, and a single-unit controller. The matrix-level controller is connected to the DC bus, the cluster-level controller is connected to the matrix-level controller and the corresponding PEP cluster, and the single-unit controller is connected to the corresponding cluster-level controller and the corresponding PEP.

[0071] Specifically, the physical connection topology of the control device is highly coupled with the hierarchical control logic of the system. The matrix-level controller, acting as the top-level master station, connects its voltage sampling port to the DC bus via a high-precision voltage sensor to acquire the bus feedback voltage in real time. Its communication port connects to each cluster-level controller via an industrial Ethernet or CAN bus, issuing commands such as droop coefficient reference values. The cluster-level controllers, acting as intermediate-level slave stations, are typically distributed in the local control cabinets of each PEP cluster. They acquire cluster voltage and current feedback values ​​through analog signal acquisition circuits and receive matrix-level commands via the communication bus. The individual cell-level controllers are integrated within the PEP, directly connecting to the battery module and DC-DC converter. They report individual cell SOC, temperature, and other status information to the cluster-level controllers via internal communication buses, such as CAN or a Serial Peripheral Interface (SPI), and receive port voltage reference values.

[0072] Through the aforementioned physical connections, this embodiment constructs a complete hardware closed loop: the DC bus provides voltage support, the PEP cluster performs power conversion, and the control device implements hierarchical collaborative control. This architecture ensures that the control strategies of matrix-level correction of droop curves, cluster-level dynamic optimization and equalization, and individual-level voltage tracking in the aforementioned embodiments can be accurately executed in the physical world, thereby achieving stable operation and optimized management of the system at the off-grid level.

[0073] Example 6: To verify the effectiveness of the control method for the matrix DC energy storage system provided by this invention under off-grid conditions, this embodiment takes the actual operating conditions of a certain island microgrid as an example for detailed explanation. Under normal circumstances, this island microgrid is powered by a diesel generator and a photovoltaic system. When the external power grid is interrupted due to a fault, the system enters the off-grid operation mode. At this time, the matrix DC energy storage system acts as the main voltage source to support the DC bus voltage.

[0074] In this application scenario, the matrix DC energy storage system is configured as follows: The system comprises three parallel PEP clusters, each consisting of 20 PEPs connected in series. The rated voltage of the DC bus is 750V. Initially, due to differences in battery aging and illumination, the state of charge (SOC) of each PEP cluster varies significantly: PEP cluster 1 has a SOC of 85%, PEP cluster 2 has a SOC of 75%, and PEP cluster 3 has a SOC of 65%. The initial load power is 100kW, and a load surge occurs at the 10th second, with the power increasing stepwise to 200kW.

[0075] During off-grid operation, the specific execution process of the control method described in this invention is as follows: First, at the matrix-level control layer, the system acquires the DC bus voltage in real time. During a sudden load change, the bus voltage drops instantaneously to 735V. The matrix-level controller generates a droop voltage compensation amount based on the difference between the DC bus voltage command (750V) and the bus feedback voltage (735V), and uses this compensation amount to correct the matrix-level droop characteristic curve. Specifically, in this embodiment, the target of the droop voltage compensation amount is set to compensate the bus voltage to the rated voltage of 750V to ensure the stable operation of the downstream load equipment. Through this correction, the droop characteristic curve shifts, allowing the bus voltage to be pulled back to near the rated value even with increased output power, avoiding the voltage deviation problem inherent in traditional droop control.

[0076] Secondly, at the cluster-level power allocation level, the matrix-level controller allocates cluster-level droop coefficient reference values ​​based on the SOC differences of each PEP cluster. Since PEP cluster 1 has the highest SOC (85%), it is allocated a smaller droop coefficient reference value to handle greater discharge power; PEP cluster 3 has the lowest SOC (65%), and it is allocated a larger droop coefficient reference value to handle less discharge power. This allocation strategy ensures that the SOC of each cluster gradually becomes more consistent during discharge.

[0077] Specifically, during the transient process of load abrupt change, each PEP cluster receives its own cluster-level droop coefficient reference value and generates an adjustment droop coefficient by combining it with the cluster's dynamic adjustment amount. At this time, the transient optimization stage of droop characteristics plays a crucial role. Taking PEP cluster 1 as an example, its controller uses the difference between the cluster voltage reference value and the cluster voltage feedback value as the cluster voltage error, converts this error into a cluster current reference value, and then uses the difference between the cluster current reference value and the cluster current feedback value as the cluster current error. The dynamic adjustment amount is generated from the cluster current error. This dynamic adjustment amount rapidly corrects the adjustment droop coefficient, providing additional "virtual damping" for the system. Experimental data shows that after introducing this transient optimization stage, the overshoot of the bus voltage after load abrupt change is reduced from 8% in traditional control to less than 2%, the adjustment time is shortened by about 40%, and voltage oscillations during the transient process are effectively suppressed.

[0078] Finally, at the individual-level balancing level, the cluster-level controller generates a fine-tuning amount based on the difference between the SOC of each PEP within the cluster and the cluster average SOC. Taking PEP cluster 2 as an example, the SOC of its fifth PEP is 80%, higher than the cluster average SOC (75%). The cluster-level controller performs absolute value calculation and slope limiting on the cluster voltage error to obtain a normalized cluster voltage error, and generates a voltage distribution ratio coefficient based on the SOC deviation of this PEP (+5%). During the dynamic adjustment phase after a sudden load change, the normalized cluster voltage error is large, the voltage distribution ratio coefficient is suppressed, and the balancing force automatically decreases to prioritize voltage stability. When the system enters steady state, the normalized cluster voltage error decreases, the voltage distribution ratio coefficient increases, and the balancing function is automatically enhanced. At the same time, the rate of change of the voltage distribution ratio coefficient is limited by a preset slew rate parameter to avoid sudden changes in the balancing force. Finally, the generated directional voltage adjustment amount is superimposed on the PEP voltage reference, causing the port voltage reference value of the fifth PEP to increase appropriately, thereby increasing its output power and accelerating its SOC decrease.

[0079] After a period of operation, system monitoring data showed that the DC bus voltage stabilized within the range of 750V±2V; the SOC difference between PEP clusters 1, 2, and 3 decreased from the initial 20% to less than 5%, effectively preventing system shutdown due to the depletion of power in a single cluster. This application example fully demonstrates that the method described in this invention can simultaneously achieve precise and stable control of the bus voltage, reasonable power distribution between clusters, and rapid balancing of battery states within clusters under off-grid conditions, significantly improving the power supply reliability and battery utilization of the microgrid system.

[0080] It should be understood that the above embodiments are only illustrated using island microgrids as an example. The control method described in this invention is also applicable to other off-grid application scenarios such as data center backup power supplies and independent power supply systems in remote areas. Its core control logic and beneficial effects are universal in different application scenarios.

[0081] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention, such as adjusting the specific calculation formula of the droop coefficient, changing the generation process of the dynamic adjustment amount, or replacing the specific logical operation method of the balance fine-tuning amount, should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A control method for a matrix-type DC energy storage system under off-grid conditions, wherein the matrix-type DC energy storage system comprises a PEP cluster consisting of multiple PEPs connected in series, and a PEP matrix consisting of multiple PEP clusters connected in parallel, characterized in that, Includes the following steps: The droop voltage compensation amount is generated based on the difference between the preset DC bus voltage command and the bus feedback voltage, and the droop voltage compensation amount is used to correct the matrix-level droop characteristic curve. Based on the charge state of each PEP cluster, the cluster-level droop coefficient baseline value of each PEP cluster is assigned according to the modified matrix-level droop characteristic curve. Each PEP cluster receives its own cluster-level droop coefficient reference value and adds the dynamic adjustment amount of its own cluster to generate an adjustment droop coefficient. Based on the adjustment droop coefficient, voltage droop control is performed to generate the cluster voltage reference value of its own cluster. The dynamic adjustment amount is generated through transient optimization of the droop characteristics; The cluster voltage reference value is divided equally according to the number of PEPs within the cluster to obtain the PEP voltage reference. The equalization fine-tuning amount is generated based on the difference between the SOC of each PEP in the cluster and the average SOC of the cluster. The PEP voltage reference is superimposed with the equalization fine-tuning amount to obtain the port voltage reference value of each PEP, and port voltage control is performed.

2. The control method for the matrix DC energy storage system under off-grid conditions according to claim 1, characterized in that, The droop voltage compensation amount is used to shift or rotate the droop characteristic curve so that the bus voltage is compensated to the target voltage.

3. The control method for the matrix DC energy storage system under off-grid conditions according to claim 1, characterized in that, The transient optimization of the droop characteristic specifically includes: The difference between the cluster voltage reference value and the cluster voltage feedback value is taken as the cluster voltage error; The cluster voltage error is converted into a cluster current reference value, and the difference between the cluster current reference value and the cluster current feedback value is taken as the cluster current error. A dynamic adjustment amount is generated based on the cluster current error.

4. The control method for the matrix DC energy storage system under off-grid conditions according to claim 3, characterized in that, The cluster current error is dynamically adjusted using a PI controller.

5. The control method for the matrix DC energy storage system under off-grid conditions according to claim 1, characterized in that, The methods for generating the balance fine-tuning amount include: The difference between the cluster voltage reference value and the cluster voltage feedback value is calculated, and the cluster voltage error is output. The cluster voltage error is normalized by performing absolute value calculation and slope limitation on the cluster voltage error. The SOC deviation is obtained based on the difference between the SOC of each PEP and the cluster average SOC. Based on the normalized cluster voltage error and the SOC deviation, a voltage allocation ratio coefficient is generated; The voltage distribution ratio coefficient is multiplied by the cluster voltage error and combined with the charging / discharging direction sign to generate a directional voltage adjustment amount, which is then used as the equalization fine-tuning amount.

6. The control method for the matrix DC energy storage system under off-grid conditions according to claim 5, characterized in that, The specific generated voltage allocation ratio coefficient is as follows: Based on a preset voltage distribution ratio coefficient benchmark, the voltage distribution ratio coefficient is generated through multiplication logic or lookup table logic according to the normalized cluster voltage error and the SOC deviation; at the same time, the rate of change of the coefficient is limited by the slew rate parameter to output a stable voltage distribution ratio coefficient.

7. The control method for the matrix DC energy storage system under off-grid conditions according to claim 6, characterized in that, The rate of change of the voltage distribution ratio coefficient is adjusted by a preset slew rate parameter; the slew rate parameter specifies the maximum allowable change of the voltage distribution ratio coefficient per unit time.

8. A control device for a matrix-type DC energy storage system under off-grid conditions, characterized in that, include: A matrix-level controller is used to generate a droop voltage compensation amount based on the difference between the DC bus voltage command and the bus feedback voltage, correct the matrix-level droop characteristic curve using the droop voltage compensation amount, and allocate a cluster-level droop coefficient reference value for each PEP cluster based on the SOC of each PEP cluster and the corrected matrix-level droop characteristic curve. At least one cluster-level controller, each cluster-level controller is connected to the corresponding PEP cluster, for receiving the cluster-level droop coefficient reference value of the cluster, and generating an adjustment droop coefficient in combination with the dynamic adjustment amount of the cluster, performing voltage droop control based on the adjustment droop coefficient, generating a cluster voltage reference value of the cluster, dividing the cluster voltage reference value equally according to the number of PEPs in the cluster to obtain the PEP voltage reference, and generating an equalization fine adjustment amount based on the difference between the SOC of each PEP in the cluster and the average SOC of the cluster. Multiple unit-level controllers are provided, each connected to a corresponding PEP, for receiving the PEP voltage reference and the equalization fine-tuning amount, superimposing the two as the port voltage reference value of the PEP, and performing port voltage control.

9. The control device for the matrix DC energy storage system under off-grid conditions according to claim 8, characterized in that, The cluster-level controller also includes: The droop transient optimization module is used to generate the dynamic adjustment amount based on the voltage error and current error of this cluster; The voltage distribution limit dynamic adjustment module is used to generate the equalization fine-tuning amount based on the cluster voltage error and the SOC deviation of each PEP in the cluster.

10. A matrix-type DC energy storage system, characterized in that, include: DC bus; Multiple PEP clusters, each PEP cluster consisting of multiple PEPs connected in series, and the multiple PEP clusters are connected in parallel to the DC bus; The control device for a matrix-type DC energy storage system under off-grid conditions as described in claim 8 or 9, wherein the matrix-level controller is connected to the DC bus, the cluster-level controller is connected to the matrix-level controller and the corresponding PEP cluster, and the individual-level controller is connected to the corresponding cluster-level controller and the corresponding PEP.

Images