A direct-current micro-grid adaptive impedance reshaping power flattening method based on multi-dimensional health perception

Abstract

The application discloses a DC micro-grid adaptive impedance remodeling power smoothing method based on multi-dimensional health degree perception, relates to the technical field of DC micro-grid, obtains the capacity health degree and impedance health degree of a storage battery, determines a virtual resistance component based on the capacity health degree, determines a virtual inductance component based on the impedance health degree, introduces the virtual resistance component and the virtual inductance component into a droop control equation, generates a reference voltage instruction of a converter, remodels the output impedance characteristics of a storage battery branch, determines a dynamic time constant associated with the impedance health degree and a super capacitor state of charge, decomposes a total power instruction into a low-frequency component borne by the storage battery and a high-frequency component borne by the super capacitor by using the dynamic time constant, and the application can multi-dimensionally perceive the battery health degree, effectively smooths the dynamic current impact of an aging battery and delays the performance degradation of the battery.

Description

Technical Field

[0001] This invention relates to the field of DC microgrid technology, and in particular to an adaptive impedance reshaping power smoothing method for DC microgrids based on multidimensional health perception. Background Technology

[0002] The penetration rate of distributed new energy sources, represented by photovoltaics and wind power, in DC microgrids is increasing. New energy power generation has significant intermittency and volatility. If directly connected to the grid, it will cause severe fluctuations in DC bus voltage, seriously affecting power supply quality.

[0003] To address this issue, hybrid energy storage systems (HESS) consisting of high-energy-density batteries and high-power-density supercapacitors are widely used. However, most existing HESS control technologies rely on simple power distribution adjustments based on the state of charge (SOC). For example, when the battery's SOC is high, the droop factor is reduced to increase output power. In reality, with increasing operating time, batteries not only experience capacity degradation but also a significant increase in internal resistance and a decrease in dynamic response. Traditional droop control exhibits purely resistive characteristics and cannot provide dynamic buffering for aging batteries. When the load changes abruptly, even if the steady-state current of the aging battery is limited, the instantaneous current surge can still cause a sudden drop in the terminal voltage of the aging battery or even induce thermal runaway. Summary of the Invention

[0004] To overcome the shortcomings of the prior art, this invention provides a DC microgrid adaptive impedance reshaping power smoothing method based on multidimensional health perception. By decoupling the capacity decay and internal resistance increase of the battery through multidimensional health perception, the method introduces both into the dynamic adjustment of virtual resistance, virtual inductance and adaptive filter time constant. While achieving optimized power allocation between and within devices, it actively predicts and controls thermal risks, thereby significantly smoothing power fluctuations, delaying battery aging, reducing the risk of thermal runaway, and improving the power supply quality and intrinsic safety level of the DC microgrid throughout its entire life cycle.

[0005] To achieve the above objectives, the present invention adopts the following technical solution, including the following steps: a multi-dimensional health perception step: obtaining the capacity health of the battery. and resistance health Inter-device power allocation steps: based on capacity health Determine the virtual resistance components based on impedance health. The virtual inductance component is determined, and the virtual resistance and virtual inductance components are introduced into the droop control equation to generate the reference voltage command of the converter, so as to reshape the output impedance characteristics of the battery branch; the power spectrum allocation steps in the device are as follows: determine the dynamic time constant associated with the impedance health β and the state of charge of the supercapacitor, and use the dynamic time constant to decompose the total power command into the low-frequency component borne by the battery and the high-frequency component borne by the supercapacitor.

[0006] Preferably, the multi-dimensional health perception step specifically includes: during the battery charging stage, calculating the incremental capacity curve based on the terminal voltage sequence and current sequence, and extracting the voltage offset of the characteristic peak as the first key aging feature; simultaneously calculating the temperature rise rate under the same current rate as the second key aging feature; constructing the first key aging feature, the second key aging feature, the historical cumulative throughput, and the current ambient temperature into an input feature vector, inputting the input feature vector into a long short-term memory network, and decoupling and outputting the capacity health α and impedance health β by the long short-term memory network.

[0007] Preferably, in the inter-device power distribution step, the droop control equation after introducing virtual resistance and virtual inductance components is constructed as follows: ,in, Reference voltage command, This is the rated voltage of the DC bus. For output current, Based on the droop coefficient, For virtual resistance components, Let t be the virtual inductance component, and t be time.

[0008] Preferably, virtual resistance component The calculation formula is: ,in, This is the gain coefficient. To adjust the index.

[0009] Preferably, the virtual inductance component The magnitude of is determined by the impedance health β, and its functional relationship is set as follows: ,in, This is the virtual inertia coefficient.

[0010] Preferably, in the power spectrum allocation step within the device, the reference current of the battery... Calculated using the following formula: ,in, For the Laplace transform operator, This is the total current command. The dynamic time constant, This is the state of charge of the supercapacitor.

[0011] Preferably, the dynamic time constant The regulation law is defined as: ,in, The initial time constant, and These are the weighting coefficients.

[0012] Preferably, it also includes a thermoelectric co-management step: based on impedance health β, combined with the battery's reference current. and the rated internal resistance of the battery Predict the heat generation capacity at the next moment. : .

[0013] Preferably, the thermoelectric co-management steps further include: when the predicted heat generation power exceeds a preset threshold, increasing the power output of the cooling system in advance to perform feedforward heat dissipation control.

[0014] Preferably, the thermoelectric co-management step further includes: generating a forced power derating factor when the predicted heat generation power exceeds a safety threshold and the heat dissipation system has reached its maximum power. Its value range is Apply the derating factor to the current power distribution stage to calculate the actual output current of the battery. : .

[0015] The advantages of this invention are: (1) Achieved precise decoupling and multi-dimensional quantitative perception of the aging state of energy storage batteries: This invention overcomes the technical defects of traditional ampere-hour integration methods or single voltage parameter methods in distinguishing between the two independent aging dimensions of capacity decay and internal resistance increase by integrating incremental capacity analysis and long short-term memory network algorithms. It can independently and quantitatively output capacity health indicators and impedance health indicators by extracting the characteristic peak shift and temperature rise rate of the charging voltage curve, which significantly improves the confidence and accuracy of the system's estimation of the battery's full life cycle state.

[0016] (2) Unlike traditional droop control, which only possesses pure resistive characteristics and can only regulate steady-state current, this invention introduces a virtual inductance component that dynamically changes with impedance health into the control loop. This technique utilizes the suppression effect of the virtual inductance on the rate of current change to effectively smooth the current surge caused by sudden load changes in the aging battery. By reducing the rate of increase of the output current of the aging battery, it effectively prevents the instantaneous drop in terminal voltage and the accumulation of internal hot spots caused by increased internal resistance, thus solving the technical problem of easy failure of aging batteries under high-frequency dynamic loads under traditional control strategies.

[0017] (3) The adaptive frequency conversion filtering strategy effectively delays the performance degradation of the battery pack: This invention changes the traditional control mode with a fixed cutoff frequency of the low-pass filter and establishes a dynamic coupling mechanism between the time constant, impedance health, and supercapacitor state of charge. That is, as the battery ages, the control system can automatically adjust the filtering parameters, compress the power fluctuation frequency band borne by the battery, and transfer more high-frequency and mid-frequency power fluctuations to the supercapacitor side. This spectrum migration strategy based on aging status ensures that severely aged batteries can operate under more stable current conditions, thereby effectively reducing the rate of internal electrochemical side reactions, delaying further degradation of battery capacity and power characteristics, and improving the overall operating life and economy of the hybrid energy storage system.

[0018] (4) An active thermoelectric co-management mechanism was constructed to improve the intrinsic safety level of the system: By directly introducing the impedance factor characterizing electrochemical aging into the thermal management control loop, feedforward prediction and control of the heat generation power of the energy storage module were realized. This mechanism effectively solves the lag problem of traditional temperature feedback control, ensuring that in the later stage of battery performance degradation, the system can strictly maintain the operating temperature within a safe range through real-time coordination of electrical derating control and active heat dissipation regulation, thereby reducing the risk of thermal runaway caused by internal resistance aging from the physical source and significantly improving the operational reliability and safety of distributed microgrid equipment. Attached Figure Description

[0019] Figure 1 This is a flowchart of a power mitigation method for DC microgrids based on multidimensional health perception.

[0020] Figure 2 A flowchart of multidimensional SOH sensing and feature extraction provided in an embodiment of this application.

[0021] Figure 3 This is a hardware connection and system topology diagram of a device provided in an embodiment of this application. Detailed Implementation

[0022] 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. 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.

[0023] like Figure 1 As shown in the figure, the DC microgrid power smoothing method based on multi-dimensional health perception provided in this embodiment includes the following steps.

[0024] S1: Multidimensional Health Perception Steps

[0025] like Figure 2 As shown, this step is used to obtain two independent aging indicators of the battery: capacity health. and resistance health .

[0026] S1.1: Triggering Conditions and Data Acquisition

[0027] When the power smoothing device is in charging mode, the multi-dimensional SOH sensing module initiates high-frequency sampling to acquire the battery's terminal voltage sequence V(t) and current sequence I(t) during the constant current charging phase. The sampling interval is denoted as... .

[0028] S1.2: Incremental Capacity Curve Calculation and Feature Extraction

[0029] The module calculates the incremental capacity curve based on the acquired terminal voltage sequence V(t) and current sequence I(t). The calculation formula is: , in, Let be the charging current at time t. The sampling interval is... and These are the terminal voltages at the current time and the previous time, respectively.

[0030] The calculated incremental capacity curve is smoothed using a Gaussian filter to eliminate noise interference. Characteristic peaks corresponding to redox reactions on the curve are identified, and the voltage offset of these peaks is extracted. As the first key aging characteristic, voltage offset It directly characterizes the degree of loss of active lithium ions inside the battery and has a strong linear correlation with capacity decay.

[0031] S1.3: Calculation of temperature rise rate

[0032] Simultaneous calculation of temperature rise rate at the same current ratio As the second key aging characteristic, due to the increase in battery internal resistance with increasing aging, following Joule's law (P=I²R), aged batteries will exhibit a faster rate of temperature rise under the same current excitation. Therefore, the rate of temperature rise... It can effectively characterize the power decay (i.e. impedance increase) characteristics of a battery.

[0033] S1.4: LSTM fusion output

[0034] The extracted first key aging feature, second key aging feature, historical cumulative throughput, and current ambient temperature are used to construct the input feature vector X.

[0035] The input feature vector X is fed into a pre-trained Long Short-Term Memory (LSTM) network. The LSTM network utilizes forget gates, input gates, and output gates to capture the non-linear dependencies of the battery aging process across long and short time series, thereby decoupling and outputting two independent health metrics: capacity health. and resistance health : Capacity health Values ​​range from 0 to 1, representing the ratio of the current maximum usable capacity to the factory rated capacity, used to measure the battery's charge storage capacity; impedance health. Values ​​≥ 1 represent the ratio of the current DC internal resistance to the factory rated internal resistance, used to measure the dynamic response loss of the battery. A larger value indicates more severe internal resistance aging.

[0036] Capacity health and resistance health The data is transmitted in real time to the adaptive impedance reshaping controller, serving as the core basis for subsequent control strategies. Traditional droop control strategies typically exhibit only pure resistive characteristics. While this strategy can adjust the steady-state current based on the SOC (State of Charge), it cannot limit the impact of sudden current changes in aging batteries with significantly increased internal impedance. Under high-frequency pulsating loads, aging batteries are highly susceptible to rapid voltage drops or internal hotspot accumulation due to excessively large instantaneous current changes. Therefore, this invention introduces virtual resistors and virtual inductors into the control loop to dynamically reshape the power supply's output impedance characteristics based on the SOH (State of Harm) state.

[0037] S2: Inter-device power allocation steps

[0038] The adaptive impedance reshaping controller receives the capacity health status. and resistance health Then, the first level of control is executed: power distribution optimization between devices. Specifically, the controller reads the rated voltage of the DC bus in real time and combines it with... and The correction factor is calculated, the reference voltage command for the DC / DC converter is generated, and the improved droop control equation is constructed as follows: , in, The reference voltage command output by the DC / DC converter. This is the rated voltage of the DC bus. For output current, Based on the droop coefficient, It is a virtual resistance component.

[0039] S2.1: Setting the virtual resistor

[0040] Virtual resistance Based on the degree of capacity decay, its calculation formula is as follows: , in, This is the gain coefficient. To adjust the index, when the battery capacity decreases (i.e. When the current is reduced, the virtual resistance increases non-linearly. This mechanism forces the device to automatically take on a smaller current share during steady-state operation, preventing overcharging and over-discharging, and achieving capacity balance of "smaller capacity, smaller responsibility".

[0041] S2.2: Setting the virtual inductor

[0042] Virtual inductance component The magnitude is directly determined by impedance health. The decision is made, and its functional relationship is set as follows: , in, This is a virtual inertia coefficient, which is relevant for cases with severe internal resistance aging (i.e., For batteries with high inductance values, the controller automatically increases the virtual inductance value. This means increased electrical inertia in this branch. When a sudden load change occurs, the larger virtual inductance generates a reverse electromotive force to counteract the drastic change in current, thereby significantly reducing the rate of increase of the output current of the aging battery. This mechanism smooths out the originally sharp power surges, making the response speed of the aging battery slower than that of a new battery, avoiding instantaneous overheating and voltage collapse caused by excessive internal resistance, and effectively extending the remaining life of the aging battery.

[0043] S3: Power Spectrum Allocation Steps within the Device

[0044] After completing the macroscopic power distribution between devices, this embodiment further executes the second level of control: the power distribution within the devices, namely the power spectrum distribution between the battery and the supercapacitor. The hybrid energy storage system uses a low-pass filter to decompose the total power command into a low-frequency component (borne by the battery) and a high-frequency component (borne by the supercapacitor). In this embodiment, the time constant of the low-pass filter is no longer a fixed value, but is designed to be related to impedance health. and the state of charge of supercapacitors dynamic time constant .

[0045] S3.1: Reference Current Calculation

[0046] Reference current of the battery Calculated using the following formula: , Where s is the Laplace transform operator, representing a complex frequency domain variable used to describe the dynamic transfer function of the low-pass filter. This is the total current command.

[0047] S3.2: Dynamic Time Constant Adjustment

[0048] Dynamic time constant The regulation law is defined as: , in, The initial time constant, and These are the weighting coefficients. It can be seen that when the battery's internal resistance increases (i.e., ...), ... When the time constant increases, Automatically increases. Due to the filter's cutoff frequency. : , The increase in frequency means a lower cutoff frequency, a narrower passband for the filter, and further compression of the power fluctuation spectrum borne by the battery. More high-frequency and mid-frequency jitter is transferred to the supercapacitor side. This allows aging batteries to operate primarily under more stable DC conditions, entering a "rest and recuperation" mode. Simultaneously, the introduction of… This can prevent the supercapacitor from overloading. When the state of charge of the supercapacitor deviates too much from the median value (i.e., (Too high or too low), this item will be appropriately limited. The excessive increase in [something] requires finding a dynamic balance between protecting aging batteries and maintaining system stability.

[0049] S4: Steps for thermoelectric co-management

[0050] This embodiment builds upon traditional temperature control logic to construct an active thermal-electric collaborative management mechanism. This mechanism utilizes the impedance health status output by the aforementioned SOH sensing module. Establish a heat generation prediction model.

[0051] S4.1: Estimated Heat Production

[0052] Based on the reference current calculated in S3.1, the current reference current of the current battery is obtained. Combined with real-time impedance Estimated heat generation capacity at the next moment: , S4.2: Feedforward heat dissipation control On the heat dissipation side, based on this predicted value, the thermoelectric co-management unit performs feedforward control on the heat dissipation side: when the predicted heat generation power... When the preset threshold is exceeded, the system increases the power output of the cooling system in advance (e.g., by increasing the fan speed or the liquid cooling pump flow rate) to achieve the control effect of "cooling before heating".

[0053] S4.3: Forced Electrical Derating Control

[0054] On the electrical side, when the predicted rate of temperature rise exceeds the safety threshold and the cooling system has reached its maximum power, the system generates a forced power derating factor. ( This coefficient directly affects the current power distribution circuit, and the actual output current of the battery. for: , At this time, the reference current of the supercapacitor Then it is: .

[0055] It should be understood that the system reduces This mechanism directly reduces the current amplitude by modifying the battery's output current, forcing the aging module's output power to be lower than its theoretically allocated value. This addresses the risk of thermal runaway caused by battery aging from a physical source, ensuring the device's thermal stability throughout its entire lifespan and guaranteeing the system's inherent safety.

[0056] like Figure 3 As shown in the figure, this embodiment provides a DC microgrid power smoothing device based on multi-dimensional health perception, including a bidirectional DC / DC converter, a hybrid energy storage unit, a multi-dimensional SOH perception module, an adaptive impedance reshaping controller, and a thermoelectric co-management unit.

[0057] The hybrid energy storage unit consists of energy-type energy storage elements (batteries in this embodiment) and power-type energy storage elements (supercapacitors in this embodiment) connected in parallel. Each energy storage element is connected to the DC bus through an independent bidirectional DC / DC converter.

[0058] The multi-dimensional SOH sensing and adaptive impedance reshaping controller, along with the thermal management execution unit, are integrated into the same main control chip. This main control chip acquires the voltage, current, and temperature signals of each energy storage element in real time through a sampling circuit, and outputs PWM drive signals to control the on / off state of the bidirectional DC / DC converter, thereby regulating the power flow and enabling plug-and-play distributed installation.

[0059] The above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for adaptive impedance reshaping power smoothing in DC microgrids based on multi-dimensional health perception, characterized in that, Includes the following steps: Multi-dimensional health perception steps: Obtain the battery's capacity health. and resistance health ; Inter-device power allocation steps: based on the aforementioned capacity health status Determine the virtual resistance component based on the impedance health. The virtual inductance component is determined, and the virtual resistance component and the virtual inductance component are introduced into the droop control equation to generate the reference voltage command of the converter, so as to reshape the output impedance characteristics of the battery branch. The power spectrum allocation steps within the device are as follows: Determine the dynamic time constant associated with the impedance health β and the state of charge of the supercapacitor, and use the dynamic time constant to decompose the total power command into a low-frequency component carried by the battery and a high-frequency component carried by the supercapacitor.

2. The adaptive impedance reshaping power smoothing method for DC microgrids based on multi-dimensional health perception as described in claim 1, characterized in that, The multidimensional health perception steps specifically include: During the battery charging phase, the incremental capacity curve is calculated based on the terminal voltage sequence and current sequence, and the voltage offset of the characteristic peak is extracted as the first key aging feature. Simultaneously calculate the temperature rise rate at the same current rate as the second key aging characteristic; The first key aging feature, the second key aging feature, the historical cumulative throughput and the current ambient temperature are constructed as an input feature vector. The input feature vector is input into a long short-term memory network, which decouples and outputs the capacity health α and the impedance health β.

3. The adaptive impedance reshaping power smoothing method for DC microgrids based on multi-dimensional health perception as described in claim 1, characterized in that, In the inter-device power distribution step, the droop control equation after introducing the virtual resistance component and the virtual inductance component is constructed as follows: ,in, Reference voltage command, This is the rated voltage of the DC bus. For output current, Based on the droop coefficient, For the virtual resistance component, Let t be the virtual inductance component, and t be time.

4. The adaptive impedance reshaping power smoothing method for DC microgrids based on multi-dimensional health perception according to claim 1 or 3, characterized in that, The virtual resistance component The calculation formula is: ,in, This is the gain coefficient. To adjust the index.

5. The adaptive impedance reshaping power smoothing method for DC microgrids based on multi-dimensional health perception according to claim 1 or 3, characterized in that, The virtual inductance component The magnitude of is determined by the impedance health β, and its functional relationship is set as follows: ,in, This is the virtual inertia coefficient.

6. The adaptive impedance reshaping power smoothing method for DC microgrids based on multi-dimensional health perception according to claim 1, characterized in that, In the power spectrum allocation step within the device, the reference current of the battery... Calculated using the following formula: ,in, For the Laplace transform operator, This is the total current command. The dynamic time constant, This is the state of charge of the supercapacitor.

7. The adaptive impedance reshaping power smoothing method for DC microgrids based on multi-dimensional health perception according to claim 1 or 6, characterized in that, The dynamic time constant The regulation law is defined as: ,in, The initial time constant, and These are the weighting coefficients. This is the state of charge of the supercapacitor.

8. The adaptive impedance reshaping power smoothing method for DC microgrids based on multi-dimensional health perception according to claim 1, characterized in that, It also includes a thermoelectric co-management step: based on the impedance health β, combined with the battery's reference current. and the rated internal resistance of the battery Predict the heat generation capacity at the next moment. : .

9. The adaptive impedance reshaping power smoothing method for DC microgrids based on multi-dimensional health perception as described in claim 8, characterized in that, The thermoelectric co-management steps also include: when the predicted heat generation power exceeds a preset threshold, increasing the power output of the cooling system in advance to execute feedforward heat dissipation control.

10. The adaptive impedance reshaping power smoothing method for DC microgrids based on multi-dimensional health perception according to claim 8, characterized in that, The thermoelectric co-management steps also include: generating a forced power derating factor when the predicted heat generation power exceeds a safety threshold and the heat dissipation system has reached its maximum power. Calculate the actual output current of the battery. : .

Images