Micro-grid cluster-oriented adaptive optimization scheduling method for energy storage system full life cycle

Abstract

The application relates to the field of power supply circuits and electric energy storage systems, and discloses a full-life-cycle adaptive optimization scheduling method for an energy storage system of a micro-grid cluster, which comprises the following steps: acquiring polarization impedance data of each energy storage branch representing electrode interface charge transfer resistance and an operating state, and decomposing bus net load power into a transient high-frequency sequence and a steady-state low-frequency sequence; distributing high-frequency fluctuation according to the reciprocal of the polarization impedance, adjusting the virtual impedance gain of a power conversion unit control loop to make the output current change rate lower than the physical boundary of ion diffusion rate; and introducing carbon emission intensity weight and loss cost function to correct the power limit value. The application establishes the physical correlation between the battery kinetics and the power fluctuation frequency, smoothes the current waveform by using the virtual impedance adjustment technology, suppresses the concentration polarization and lithium precipitation phenomenon of the electrode interface, and essentially prolongs the service life of the energy storage system.

Description

Technical Field

[0001] This invention belongs to the technical field of power supply circuits and energy storage systems, and particularly relates to an adaptive optimization scheduling method for the entire life cycle of energy storage systems for microgrid clusters. Background Technology

[0002] With the increasing penetration of distributed energy, microgrid systems, including stationary energy storage stations and electric vehicle clusters, typically employ scheduling methods to maintain power balance. Existing methods, based on the power balance principle, utilize prediction tools such as long short-term memory networks to determine power commands and combine rainflow counting to count the equivalent cycle number to calculate battery health status and carbon emission indicators. For example, Chinese invention patent CN108092306B discloses a droop control method for low-voltage microgrid energy storage systems that considers mismatched line resistance. Addressing the poor power distribution accuracy caused by impedance mismatch in microgrid lines, it proposes an improved strategy based on dynamic virtual impedance, adjusting the virtual impedance in real time to eliminate the influence of line parameter differences, and combining it with SOC droop control to achieve balanced convergence of the state of charge of each energy storage unit.

[0003] This scheduling logic based on overall energy management ignores the response characteristics of the internal electrochemical processes of the battery. Because the second-level power fluctuation frequency generated on the microgrid side is higher than the solid-phase diffusion rate of ions inside the battery, concentration polarization occurs at the electrode interface. Under the continuous action of high-frequency commands, overpotential accumulation occurs inside the battery, inducing the precipitation of active lithium on the electrode surface. Such physical damage caused by kinetic mismatch cannot be identified by overall statistical methods, resulting in irreversible degradation of the battery structure. To address this problem, simply increasing the scale of energy storage configuration or increasing the sampling frequency of the controller cannot change the charge migration resistance inside the battery. When the rate of change of external current exceeds the ion diffusion boundary, the battery exhibits polarization characteristics. Analysis shows that the existing technology has the following shortcomings: 1. Mismatch between the ion diffusion rate inside the energy storage unit and the frequency of external power fluctuations; 2. The scheduling logic lacks the ability to sense transient physical losses at the electrode interface; 3. There is no real-time coupling adjustment mechanism between the converter control parameters and the battery kinetic characteristics.

[0004] Therefore, how to adjust the electrical response characteristics of the energy storage converter to match the boundary of the internal electrochemical reaction of the battery is the technical problem to be solved by this invention. Summary of the Invention

[0005] This invention provides an adaptive optimization scheduling method for the entire lifecycle of energy storage systems for microgrid clusters, comprising the following steps:

[0006] Step S1: Obtain the voltage and current signals of each energy storage branch in the microgrid cluster, and calculate the polarization impedance data, state of charge, and health status of each energy storage branch.

[0007] Step S2: Extract the net load power fluctuation signal at the microgrid bus end, and decompose the net load power fluctuation signal into a short-time high-frequency power sequence and a long-time low-frequency power sequence.

[0008] Step S3: Quantify the electrochemical kinetic carrying capacity of each energy storage branch based on the polarization impedance data. Based on the magnitude of the electrochemical kinetic carrying capacity, allocate the short-term high-frequency power sequence to the energy storage branch with a polarization impedance lower than the first preset threshold, and allocate the long-term low-frequency power sequence to the energy storage branch with a polarization impedance higher than the second preset threshold.

[0009] Step S4: By adjusting the virtual impedance gain parameter in the control loop of the bidirectional power conversion unit corresponding to each energy storage branch, a damping correction component is introduced into the power loop to limit the rate of change of the output current.

[0010] Step S5: Obtain the carbon emission intensity weight and calculate the real-time loss cost function using the battery cycle life loss model. Use the carbon emission intensity weight and loss cost function as the correction criteria for the charge and discharge power operation limit. Based on the corrected charge and discharge power operation limit, execute the closed-loop issuance of scheduling instructions to each energy storage branch.

[0011] Preferably, the method for obtaining polarization impedance data in step S1 includes: acquiring voltage and current signals through acquisition sensors installed at the ports of each energy storage branch in the microgrid cluster, and acquiring disturbance response signals of the energy storage branch in real time within the frequency range of 0.1Hz to 1000Hz; separating the complex impedance component in the disturbance response signal through a parameter identification process; and determining the real part value of the complex impedance component as polarization impedance data.

[0012] Preferably, in step S2, the process of extracting the net load power fluctuation signal from the microgrid bus terminal through the energy management system and decomposing the net load power fluctuation signal includes: decomposing the net load power fluctuation signal using a bandpass filter bank and setting the cutoff frequency of the bandpass filter bank; extracting signal components with a change period of less than 10s from the net load power fluctuation signal and defining them as short-time high-frequency power sequences; and extracting signal components with a change period of more than 60s and defining them as long-time low-frequency power sequences.

[0013] Preferably, in step S3, a positive correlation is established between the reciprocal of the polarization impedance data and the allocated power, so that the energy storage branch with smaller polarization impedance can bear a higher amplitude of power fluctuation during the scheduling cycle.

[0014] Preferably, the logic for adjusting the virtual impedance gain parameter in step S4 includes: calculating the real-time change rate of the power command; when the real-time change rate exceeds the physical boundary value of the ion diffusion rate, linearly increasing the virtual impedance gain value of the bidirectional power conversion unit controller so that the overpotential amplitude of the electrode interface is maintained above the lithium plating critical voltage.

[0015] Preferably, the high-frequency power allocation coefficient of each energy storage branch is determined according to the following formula: ,in, is the high-frequency power allocation coefficient of the i-th energy storage branch; n is the total number of energy storage branches participating in the dispatch within the microgrid cluster; i and j are both energy storage branch numbers, and j is then summed from 1 to n. and These are the polarization impedance data for the i-th and j-th energy storage branches, respectively. and These are the reciprocals of the corresponding polarization impedance data; and These represent the health status of the i-th and j-th energy storage branches, respectively.

[0016] Preferably, the method of receiving the carbon emission intensity weight in step S5 includes: obtaining the real-time carbon intensity value of the external input electrical energy; when the real-time carbon intensity value exceeds 500 g / kWh, reducing the charging upper limit in the charging and discharging power operation limit by reducing the DC side current setpoint of the bidirectional power conversion unit.

[0017] Preferably, the battery cycle life loss model used in step S5 is a battery cycle life loss model based on the rainflow counting method. The method for calculating the real-time loss cost function includes: counting the number of battery charge-discharge cycles; substituting the number of battery charge-discharge cycles into the capacity decay equation to calculate the lifetime occupancy; converting the lifetime occupancy into an equivalent impedance correction coefficient, and superimposing the equivalent impedance correction coefficient as a correction factor into the control command of the bidirectional power conversion unit.

[0018] Preferably, the energy storage branch includes a station-type energy storage system and a mobile electric vehicle battery connected through a bidirectional charging and discharging interface. In step S3, the transient power output amplitude of the mobile electric vehicle battery is dynamically limited by polarization impedance data.

[0019] Preferably, the scheduling method adopts a multi-timescale control architecture. Within a minute-level scheduling cycle, the power command is corrected based on the offset of the state of charge relative to the preset reference value, so that the state of charge of each energy storage branch is maintained within the range of 20% to 80%.

[0020] Compared with existing technologies, the adaptive optimization scheduling method for the entire lifecycle of energy storage systems in microgrid clusters, as proposed in this invention, has the following advantages:

[0021] 1. In the adaptive optimization scheduling of the entire life cycle of the energy storage system, by identifying the polarization impedance of the energy storage unit online, a physical relationship is established between the internal ion diffusion rate of the battery and the power fluctuation frequency of the external power grid. When the power request change rate generated by the microgrid side exceeds the kinetic boundary of the electrochemical reaction, the virtual impedance parameter of the converter is adjusted to form a flexible damping effect at the circuit port, so as to smooth the current waveform flowing through the battery body, avoid concentration polarization and lithium metal deposition at the electrode interface, and block the abnormal degradation of the life of the energy storage battery from the electrochemical reaction mechanism level.

[0022] 2. Based on multi-dimensional electrochemical impedance state perception, heterogeneous energy storage resources in the microgrid cluster are hierarchically scheduled according to their dynamic response potential. By allocating low polarization impedance battery cells to high-frequency fluctuation smoothing tasks and high polarization impedance aging battery cells to steady-state base load tasks, the working characteristics of each energy storage resource in the cluster are matched with its physical state. This ensures the thermal stability of the batteries while improving the overall operational efficiency of the microgrid cluster throughout its entire life cycle.

[0023] 3. By utilizing virtual inertial reconfiguration technology in the converter control loop, the physical buffering capacity of the energy storage system against large disturbances in the source load is improved. When dealing with instantaneous fluctuations in wind power or photovoltaic power, the system exhibits non-rigid electrical response characteristics by adaptively adjusting the converter control parameters, suppressing voltage flicker and frequency oscillations at the point of common coupling, and enhancing the physical stability of the microgrid cluster under extreme conditions without altering the grid topology. Attached Figure Description

[0024] Figure 1 This is an execution flowchart of the adaptive optimization scheduling of the energy storage system throughout its entire lifecycle, as described in this invention.

[0025] Figure 2 This is a diagram showing the multi-level control architecture and signal interaction of the energy storage system of this invention. Detailed Implementation

[0026] The technical solutions in the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0027] It should be noted that all directional and positional terms used in this invention, such as: up, down, left, right, front, back, vertical, horizontal, inner, outer, top, low, lateral, longitudinal, center, etc., are only used to explain the relative positional relationship and connection between components in a specific state (as shown in the accompanying drawings). They are only for the convenience of describing this invention and do not require that this invention be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting this invention. In addition, the descriptions of "first," "second," etc., in this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly indicating the number of technical features indicated.

[0028] In the description of this invention, unless otherwise explicitly specified and limited, the terms installation, connection, and linking should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to mechanical connections; they can refer to direct connections or indirect connections through an intermediate medium; they can refer to the internal connection of two components. For those skilled in the art, the specific meaning of the above terms in this invention can be understood according to the specific circumstances.

[0029] In the description of this specification, references to the terms "an embodiment," "some embodiments," "illustrative embodiments," "examples," "specific examples," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example, and the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0030] An adaptive optimization scheduling method for the entire lifecycle of energy storage systems for microgrid clusters includes the following steps:

[0031] Step S1: Obtain the voltage and current signals of each energy storage branch in the microgrid cluster, and calculate the polarization impedance data, state of charge, and health status of each energy storage branch.

[0032] Step S2: Extract the net load power fluctuation signal at the microgrid bus end, and decompose the net load power fluctuation signal into a short-time high-frequency power sequence and a long-time low-frequency power sequence.

[0033] Step S3: Quantify the electrochemical kinetic carrying capacity of each energy storage branch based on the polarization impedance data. Based on the magnitude of the electrochemical kinetic carrying capacity, allocate the short-term high-frequency power sequence to the energy storage branch with a polarization impedance lower than the first preset threshold, and allocate the long-term low-frequency power sequence to the energy storage branch with a polarization impedance higher than the second preset threshold.

[0034] Step S4: By adjusting the virtual impedance gain parameter in the control loop of the bidirectional power conversion unit corresponding to each energy storage branch, a damping correction component is introduced into the power loop to limit the rate of change of the output current.

[0035] Step S5: Obtain the carbon emission intensity weight and calculate the real-time loss cost function using the battery cycle life loss model. Use the carbon emission intensity weight and loss cost function as the correction criteria for the charge and discharge power operation limit. Based on the corrected charge and discharge power operation limit, execute the closed-loop issuance of scheduling instructions to each energy storage branch.

[0036] Preferably, the method for obtaining polarization impedance data in step S1 includes: acquiring voltage and current signals through acquisition sensors installed at the ports of each energy storage branch in the microgrid cluster, and acquiring disturbance response signals of the energy storage branch in real time within the frequency range of 0.1Hz to 1000Hz; separating the complex impedance component in the disturbance response signal through a parameter identification process; and determining the real part value of the complex impedance component as polarization impedance data.

[0037] Preferably, in step S2, the process of extracting the net load power fluctuation signal from the microgrid bus terminal through the energy management system and decomposing the net load power fluctuation signal includes: decomposing the net load power fluctuation signal using a bandpass filter bank and setting the cutoff frequency of the bandpass filter bank; extracting signal components with a change period of less than 10s from the net load power fluctuation signal and defining them as short-time high-frequency power sequences; and extracting signal components with a change period of more than 60s and defining them as long-time low-frequency power sequences.

[0038] Preferably, in step S3, a positive correlation is established between the reciprocal of the polarization impedance data and the allocated power, so that the energy storage branch with smaller polarization impedance can bear a higher amplitude of power fluctuation during the scheduling cycle.

[0039] Preferably, the logic for adjusting the virtual impedance gain parameter in step S4 includes: calculating the real-time change rate of the power command; when the real-time change rate exceeds the physical boundary value of the ion diffusion rate, linearly increasing the virtual impedance gain value of the bidirectional power conversion unit controller so that the overpotential amplitude of the electrode interface is maintained above the lithium plating critical voltage.

[0040] Preferably, the high-frequency power allocation coefficient of each energy storage branch is determined according to the following formula: ,in, is the high-frequency power allocation coefficient of the i-th energy storage branch; n is the total number of energy storage branches participating in the dispatch within the microgrid cluster; i and j are both energy storage branch numbers, and j is then summed from 1 to n. and These are the polarization impedance data for the i-th and j-th energy storage branches, respectively. and These are the reciprocals of the corresponding polarization impedance data; and These represent the health status of the i-th and j-th energy storage branches, respectively.

[0041] Preferably, the method of receiving the carbon emission intensity weight in step S5 includes: obtaining the real-time carbon intensity value of the external input electrical energy; when the real-time carbon intensity value exceeds 500 g / kWh, reducing the charging upper limit in the charging and discharging power operation limit by reducing the DC side current setpoint of the bidirectional power conversion unit.

[0042] Preferably, the battery cycle life loss model used in step S5 is a battery cycle life loss model based on the rainflow counting method. The method for calculating the real-time loss cost function includes: counting the number of battery charge-discharge cycles; substituting the number of battery charge-discharge cycles into the capacity decay equation to calculate the lifetime occupancy; converting the lifetime occupancy into an equivalent impedance correction coefficient, and superimposing the equivalent impedance correction coefficient as a correction factor into the control command of the bidirectional power conversion unit.

[0043] Preferably, the energy storage branch includes a station-type energy storage system and a mobile electric vehicle battery connected through a bidirectional charging and discharging interface. In step S3, the transient power output amplitude of the mobile electric vehicle battery is dynamically limited by polarization impedance data.

[0044] Preferably, the scheduling method adopts a multi-timescale control architecture. Within a minute-level scheduling cycle, the power command is corrected based on the offset of the state of charge relative to the preset reference value, so that the state of charge of each energy storage branch is maintained within the range of 20% to 80%.

[0045] Example 1: In the operation of a microgrid cluster containing distributed power sources and energy storage units in different health states, voltage and current signals are acquired in real time by sensors installed at the ports of each energy storage branch. Based on the recursive least squares method with a forgetting factor, online parameter identification is performed on the voltage and current signals to calculate polarization impedance data characterizing the charge transfer resistance and diffusion hysteresis at the electrode interface. The system combines the open-circuit voltage method and the ampere-hour integral method to obtain the current state of charge and health status of each energy storage branch. The energy management system extracts the net load power fluctuation signal at the microgrid bus end and uses a parameterized bandpass filter bank with a cutoff frequency to physically decompose the net load power fluctuation signal into a short-time high-frequency power sequence with a change period of less than 10s and a long-time low-frequency power sequence with a change period of more than 60s. Based on the polarization impedance data, the system quantifies the electrochemical dynamic carrying capacity of each energy storage branch. The system executes a power hierarchical allocation logic based on physical characteristics, allocating the short-time high-frequency power sequence to energy storage branches with polarization impedance data lower than the first preset threshold and allocating the long-time low-frequency power sequence to energy storage branches with polarization impedance data higher than the second preset threshold.

[0046] After receiving the power command allocated to each energy storage branch, the system calculates the corresponding current change rate in real time. When the current change rate exceeds the physical boundary value of the ion diffusion rate determined by the polarization impedance data, the system adjusts the virtual impedance gain parameter in the control loop of the bidirectional power conversion unit corresponding to each energy storage branch. This adjustment process linearly increases the virtual inertia of the bidirectional power conversion unit controller while introducing a damping correction component in the current inner loop feedback path, so that the change rate of the output current of the bidirectional power conversion unit is lower than the physical boundary value of the ion diffusion rate. This keeps the overpotential amplitude at the electrode interface above the lithium plating critical voltage. The physical boundary value of the ion diffusion rate in step S4... Based on the offline calibration database, by applying current pulses of different slopes to individual battery cells and recording the rate of change of the critical current when the voltage drop exceeds 50mV, a system based on polarization impedance was constructed. A three-dimensional lookup table mapping state of charge (SOC) to health state (H) is used. Parameters are read during the controller's operating cycle and the lookup table is retrieved to determine the state of charge. When the rate of change of the power command current di / dt exceeds this value, the proportional coefficient is... The proportional regulator calculates the virtual inductance correction, which is then superimposed on the bidirectional power conversion unit control loop to limit the output current rise rate. Simultaneously, the system receives carbon emission intensity weights from the upstream distribution network and uses a battery cycle life loss model based on rainflow counting to calculate the real-time loss cost function by counting the number of battery charge-discharge cycles. The carbon emission intensity weights and loss cost function are used as correction criteria for the charge-discharge power operating limits. When the real-time carbon intensity value of the acquired external input electrical energy exceeds 500 g / kWh, or when the equivalent impedance correction coefficient of the calculated lifetime occupancy indicates a high-loss state, the system reduces the charging upper limit in the charge-discharge power operating limits by lowering the DC-side current setpoint of the bidirectional power conversion unit. In step S5, the cycle depth is extracted using the rainflow counting method. The equivalent impedance correction factor is calculated using the fatigue damage equation with a power exponent of 1.8. When the real-time carbon intensity value input from the external power grid exceeds 500 g / kWh, according to the formula... Calculate the upper limit of the charging current, where This is the revised upper limit of the DC-side charging current for the bidirectional power converter. This is the reference value for the rated current of the corresponding energy storage branch; The carbon sensitivity coefficient is calibrated to 0.002 kWh / g. This is a real-time carbon intensity value; the calculation results are consistent with... The determined lifetime loss correction terms work together on the bidirectional power conversion unit control command limiter to reduce the DC side charging current setpoint, and execute closed-loop issuance of scheduling commands to each energy storage branch based on the corrected charging and discharging power operating limits.

[0047] Example 2: To objectively verify the proposed adaptive optimization scheduling method for the entire lifecycle of energy storage systems for microgrid clusters, and to assess its dynamic response characteristics and safety under extreme grid fluctuations, as well as the effectiveness of the carbon constraint mechanism, a microgrid test environment was built, including a real-time digital simulator and a power hardware-in-the-loop platform. In simulating real-world physical scenarios, when the net load power at the microgrid bus terminal undergoes a drastic step change, the test platform was configured with high-precision wideband current probes and voltage sensors to collect electrical signals from each branch. The data acquisition frequency was set to 10kHz, and Gaussian white noise with a signal-to-noise ratio of 30dB was artificially superimposed on the acquisition channels to realistically simulate the complex electromagnetic interference in industrial environments. The test subjects were three groups of lithium iron phosphate battery clusters with different aging levels. Their initial states were precisely calibrated using an electrochemical workstation: the first battery cluster was in a brand-new state with a reference polarization impedance of 12.5 mΩ; the second battery cluster was in a semi-aged state with a reference polarization impedance of 24.8 mΩ; and the third battery cluster was in an aged state with a reference polarization impedance of 45.2 mΩ. At the start of the test, the system applied a pulsed load surge command with an amplitude of 80% of the rated power and a rise time of 200 ms to the microgrid bus to simulate extreme short-term high-frequency power fluctuations. The test samples executed a complete set of adaptive optimization scheduling logic. The recursive least squares algorithm with a forgetting factor identified the polarization impedance data of the three battery clusters in real time under noise interference. The voltages converged to 12.8 mΩ, 25.1 mΩ, and 45.9 mΩ respectively, with relative errors controlled within 5%. Based on the physical frequency domain decoupling mechanism of the bandpass filter bank, high-frequency power components with periods less than 10 s were separated from the total load command. According to the quantification results of electrochemical kinetic carrying capacity, the first battery cluster was allocated to carry 70% of the amplitude of the high-frequency component, the second battery cluster carried 25%, and the third battery cluster with the highest polarization impedance carried only 5%. Real-time monitoring data showed that at the moment of load change, the terminal voltage drop of the third battery cluster was only 45 mV, and its electrode interface overpotential was always maintained within the safe threshold, without triggering the lithium plating risk alarm.

[0048] To verify the necessity of the virtual impedance adjustment mechanism, a control group was set up by removing the virtual impedance gain adjustment function in step S4. Under the same 200ms pulse excitation, the current command in the control group, which was not subject to the rate of change limit, was directly sent to the converter, causing the output current change rate of the first battery cluster to instantly reach 3.5C / s, exceeding the physical boundary value of 2.0C / s for the ion diffusion rate derived from its polarization impedance data. Oscilloscope recordings showed that this instantaneous impact caused a transient overshoot of up to 320mV in the terminal voltage of the first battery cluster, accompanied by obvious voltage ringing, indicating that a large amount of charge that could not diffuse in time accumulated at the electrode interface, triggering concentration polarization. In contrast, the sample of this invention automatically intervened with a damping correction component when it detected that the current change rate was trying to break through the 2.0C / s boundary, physically clamping the rising slope of the actual output current to 1.95C / s. Although the response time was delayed from 200ms to 350ms, the voltage overshoot amplitude was suppressed to within 80mV, avoiding kinetic mismatch. Further carbon emission constraint verification was introduced. The test was conducted by simulating a sudden change in the real-time carbon intensity value from 300 g / kWh to 600 g / kWh from the upper-level power grid. Upon receiving this high-carbon signal, the sample group of this invention, combined with the cycle life loss function calculated by the rainflow counting method, triggered a reduction in the charging power limit. The DC-side charging current setpoint of the bidirectional power conversion unit was automatically reduced from 100A to 20A, physically blocking the large-scale storage of high-carbon energy. The control group, which did not enable this carbon constraint logic, continued to charge at full power with a current of 100A. After 24 hours of accelerated aging cycle testing, the equivalent cycle life loss of the sample group of this invention was reduced by 18.5% compared with the control group, and the weighted average carbon density of the stored energy was reduced by 42%. The test data objectively confirmed that by physically coupling the electrochemical impedance characteristics with the power frequency domain allocation and virtual inertial control, the micro-polarization effect that causes battery damage can be effectively mitigated on a millisecond time scale, ensuring that the energy storage system operates within the dual safety boundaries of electrochemical and low-carbon throughout its entire life cycle.

[0049] Example 3: In the low-level algorithm deployment and parameter calibration stage of the embedded controller of the energy storage system, in order to eliminate uncertainties in the online parameter identification and lifetime loss calculation process, an initial state of the computing environment based on a digital signal processor (DSP) was defined. The system set the analog-to-digital conversion resolution of the analog sampling channel to 16 bits, and the sampling period was locked at 100μs to meet the Nyquist sampling theorem's requirement for capturing high-frequency impedance characteristics. This was specifically applied to polarization impedance data. For online acquisition, the system does not use the general open-loop calculation, but instead executes a closed recursive least squares (RLS) operation procedure with a forgetting factor. This procedure initializes the covariance matrix as an identity matrix with diagonal elements of 1000 and sets the forgetting factor λ to 0.98. This value is selected based on the experimental optimization results of balancing the convergence speed and smoothness of the parameters under the condition of a signal-to-noise ratio of 20dB. In each sampling period, the processor reads the current terminal voltage and loop current data, constructs the observation vector, calculates the prediction error using the parameter estimate value of the previous moment, and updates the gain vector and covariance matrix accordingly, finally outputting the real-time polarization impedance estimate value.

[0050] After establishing the source of the impedance parameters, the system is configured with a proportional controller having dead-zone characteristics to achieve a deterministic mapping between the current change rate and virtual inertia, based on the virtual impedance gain adjustment logic mentioned in this invention. The controller internally sets the current change rate threshold corresponding to the physical boundary value of the ion diffusion rate as... When the real-time monitored rate of change of current di / dt is less than or equal to this threshold, the virtual inductance value... The value remains zero, without interfering with system dynamics; once di / dt exceeds this threshold, the controller uses the formula... Calculate the required virtual inductance, where, To dynamically introduce adaptive virtual inductance into the control loop of the bidirectional power conversion unit; damping coefficient It is calibrated to 50 mH / (A / s). The transient rate of change of the power command current as monitored in real time; The current change rate threshold is the physical boundary value of the preset ion diffusion rate. This linear control law ensures that the magnitude of the damping correction component is proportional to the degree of exceeding the limit, thereby forcibly smoothing the transient changes in current in the physical circuit and eliminating ambiguity in the control logic. For the construction of the battery cycle life loss model, the system performs a quantitative conversion process from rainflow counting results to economic costs. The system allocates dedicated memory space to record the discharge depth of each charge-discharge half-cycle identified by the rainflow counting method. The loss cost function is no longer an abstract concept, but is instantiated into a calculation process based on the power-law damage accumulation theory, that is, the loss cost of a single cycle. It exhibits a non-linear positive correlation with the depth of discharge, specifically following the... The functional relationship is as follows: the baseline coefficient α is set to 0.5 yuan / cycle based on the full life cycle amortized cost provided by the battery manufacturer, while the power exponent β is calibrated to 1.8 based on the fatigue characteristics of lithium iron phosphate material. Through this explicit calculation logic, the system can transform discrete physical cycle data into continuous economic loss indicators in real time, providing a mathematically deterministic decision basis for subsequent optimization scheduling.

[0051] Example 4: To ensure polarization impedance data To assess the quantification accuracy under different temperatures and aging conditions, the system performs standardized offline calibration and data filling procedures before being put into closed-loop operation. Typical energy storage battery cells are tested under full operating conditions in a constant temperature environment chamber. The temperature gradient is set from -20℃ to 50℃ in 10℃ intervals, and the state of charge gradient is set from 10% to 90% in 10% intervals. At each operating point, a perturbation excitation of 0.1Hz to 1000Hz is applied using an electrochemical workstation. The reference polarization impedance value is extracted by analyzing the perturbation response signal, and a multi-dimensional impedance lookup table is constructed and burned into the non-volatile memory of the embedded controller as the initial convergence value of the online parameter identification algorithm. This eliminates the risk of transient oscillation and non-convergence caused by random selection of initial values ​​during the cold start phase of the algorithm.

[0052] To address the environmental differences in system deployment across various microgrid sites, a pre-deployment calibration procedure was established. During the initial power-on phase, the self-test program automatically activates the background noise assessment module, continuously collecting voltage and current channel noise floor data under no-load conditions. It calculates the power spectral density and dynamically adjusts the passband gain and cutoff frequency of the bandpass filter bank accordingly, ensuring that the signal-to-noise ratio of the effective signal is always higher than the preset threshold. Simultaneously, the system automatically performs a small-amplitude charge-discharge pulse test, measuring the total response time of the circuit and fine-tuning the update frequency of the virtual impedance control loop to compensate for phase lag caused by differences in cable length and connection impedance, ensuring that the virtual inertial control action remains synchronized with physical grid fluctuations.

[0053] Example 5: To address the potential parameter drift and adaptive degradation issues that may arise in adaptive scheduling algorithms during engineering deployment, an offline generation and verification process for adaptive parameter matrices based on actual microgrid operating data was established. This process collected full-time operating data of the target microgrid over a historical year, including power output sequences of photovoltaic and wind power, load demand curves, and charging and discharging behavior records of the energy storage system. The sampling frequency was set to 1 minute. Based on the collected data, a test scenario set incorporating source-load uncertainty characteristics was constructed. The K-means clustering algorithm was used to categorize the massive historical scenarios into several typical operating condition classes. For each typical operating condition class, a particle swarm optimization algorithm was used to optimize the cutoff frequency of the bandpass filter bank and the damping coefficient of the virtual impedance controller. The optimal parameter combination under each operating condition is obtained by optimizing the power exponent β in the battery life loss model, with the objective function being the minimum total cost of the system throughout its entire life cycle, thus forming an adaptive parameter matrix.

[0054] After constructing the parameter matrix, a hardware-in-the-loop verification of the matrix was performed using a hardware-in-the-loop simulation platform. The hardware interface of the actual controller was connected to a real-time digital simulator, and the generated typical operating condition test sequences were injected. The parameter calling logic and dynamic response performance of the controller during the switching process of different operating conditions were monitored. The focus was on whether the system could accurately identify the current scenario and switch to the corresponding optimal parameter group in milliseconds during sudden changes in operating conditions, and whether unexpected power oscillations or voltage flicker were introduced during the parameter switching process. Only when the control error in all test scenarios was lower than the preset threshold and the system stability index met the grid connection standard was the adaptive parameter matrix allowed to be burned into the non-volatile storage area of ​​the field controller as a benchmark database for online operation. This ensured that the algorithm always had optimal control performance and stability in complex and ever-changing engineering environments.

[0055] The embodiments of this application have been described above with reference to the accompanying drawings. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. This application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. A method for adaptive optimization scheduling of energy storage systems throughout their entire lifecycle for microgrid clusters, characterized in that, Includes the following steps: Step S1: Obtain the voltage and current signals of each energy storage branch in the microgrid cluster, and calculate the polarization impedance data, state of charge, and health status of each energy storage branch. Step S2: Extract the net load power fluctuation signal at the microgrid bus end, and decompose the net load power fluctuation signal into a short-time high-frequency power sequence and a long-time low-frequency power sequence. Step S3: Quantify the electrochemical kinetic carrying capacity of each energy storage branch based on the polarization impedance data. Based on the magnitude of the electrochemical kinetic carrying capacity, allocate the short-term high-frequency power sequence to the energy storage branch with a polarization impedance lower than the first preset threshold, and allocate the long-term low-frequency power sequence to the energy storage branch with a polarization impedance higher than the second preset threshold. Step S4: By adjusting the virtual impedance gain parameter in the control loop of the bidirectional power conversion unit corresponding to each energy storage branch, a damping correction component is introduced into the power loop to limit the rate of change of the output current. Step S5: Obtain the carbon emission intensity weight and calculate the real-time loss cost function using the battery cycle life loss model. Use the carbon emission intensity weight and loss cost function as the correction criteria for the charge and discharge power operation limit. Based on the corrected charge and discharge power operation limit, execute the closed-loop issuance of scheduling instructions to each energy storage branch.

2. The adaptive optimization scheduling method for the entire lifecycle of an energy storage system for microgrid clusters according to claim 1, characterized in that, The method for obtaining polarization impedance data in step S1 includes: acquiring voltage and current signals through acquisition sensors installed at the ports of each energy storage branch in the microgrid cluster, and acquiring disturbance response signals of the energy storage branch in real time within the frequency range of 0.1Hz to 1000Hz; separating the complex impedance component in the disturbance response signal through the parameter identification process; and determining the real part value of the complex impedance component as polarization impedance data.

3. The adaptive optimization scheduling method for the entire lifecycle of an energy storage system for microgrid clusters according to claim 1, characterized in that, In step S2, the process of extracting the net load power fluctuation signal at the microgrid bus end through the energy management system and decomposing the net load power fluctuation signal includes: decomposing the net load power fluctuation signal using a bandpass filter bank and setting the cutoff frequency of the bandpass filter bank; extracting the signal components with a change period of less than 10s from the net load power fluctuation signal and defining them as short-time high-frequency power sequences; and extracting the signal components with a change period of more than 60s and defining them as long-time low-frequency power sequences.

4. The adaptive optimization scheduling method for the entire lifecycle of an energy storage system for microgrid clusters according to claim 1, characterized in that, In step S3, a positive correlation is established between the reciprocal of the polarization impedance data and the allocated power, so that the energy storage branch with smaller polarization impedance can bear a higher amplitude of power fluctuation during the scheduling cycle.

5. The adaptive optimization scheduling method for the entire lifecycle of an energy storage system for microgrid clusters according to claim 1, characterized in that, The logic for adjusting the virtual impedance gain parameter in step S4 includes: calculating the real-time change rate of the power command; when the real-time change rate exceeds the physical boundary value of the ion diffusion rate, linearly increasing the virtual impedance gain value of the bidirectional power conversion unit controller so that the overpotential amplitude of the electrode interface is maintained above the critical lithium plating voltage.

6. The adaptive optimization scheduling method for the entire lifecycle of an energy storage system for microgrid clusters according to claim 4, characterized in that, The high-frequency power allocation coefficient for each energy storage branch is determined according to the following formula: ,in, is the high-frequency power allocation coefficient of the i-th energy storage branch; n is the total number of energy storage branches participating in the dispatch within the microgrid cluster; i and j are both energy storage branch numbers, and j is then summed from 1 to n. and These are the polarization impedance data for the i-th and j-th energy storage branches, respectively. and These are the reciprocals of the corresponding polarization impedance data; and These represent the health status of the i-th and j-th energy storage branches, respectively.

7. The adaptive optimization scheduling method for the entire lifecycle of an energy storage system for microgrid clusters according to claim 1, characterized in that, The method of receiving the carbon emission intensity weight in step S5 includes: obtaining the real-time carbon intensity value of the external input electrical energy; when the real-time carbon intensity value exceeds 500g / kWh, reducing the charging upper limit in the charging and discharging power operation limit by reducing the DC side current setpoint of the bidirectional power conversion unit.

8. The adaptive optimization scheduling method for the entire lifecycle of an energy storage system for microgrid clusters according to claim 1, characterized in that, The battery cycle life loss model used in step S5 is a battery cycle life loss model based on the rainflow counting method. The method for calculating the real-time loss cost function includes: counting the number of battery charge and discharge cycles; substituting the number of battery charge and discharge cycles into the capacity decay equation to calculate the lifetime occupancy; converting the lifetime occupancy into an equivalent impedance correction coefficient, and superimposing the equivalent impedance correction coefficient as a correction factor into the control command of the bidirectional power conversion unit.

9. The adaptive optimization scheduling method for the entire lifecycle of an energy storage system for microgrid clusters according to claim 1, characterized in that, The energy storage branch includes a station-type energy storage system and a mobile electric vehicle battery connected through a bidirectional charging and discharging interface. In step S3, the transient power output amplitude of the mobile electric vehicle battery is dynamically limited by polarization impedance data.

10. The adaptive optimization scheduling method for the entire lifecycle of an energy storage system for microgrid clusters according to claim 1, characterized in that, The scheduling method adopts a multi-timescale control architecture. Within a minute-level scheduling cycle, the power command is corrected based on the offset of the state of charge relative to the preset reference value, so that the state of charge of each energy storage branch is maintained within the range of 20% to 80%.

Images