A method and system for peak-shaving dispatching of energy storage power stations in a power system

Abstract

This invention relates to the field of power energy storage control technology, and discloses a peak-shaving dispatching method and system for energy storage power stations in power systems. The method includes: firstly, collecting operating data of parallel battery racks, calculating the equivalent number of parallel branches characterizing the load-sharing state, and deriving the current redistribution step coefficient after a single rack disconnects; then, predicting the cascade disconnection protection margin of the remaining battery racks by combining impedance and thermal parameters; finally, generating power regulation coefficients using a continuous smooth mapping model, correcting peak-shaving commands, and converting them into the upper limit of DC bus current to control the energy storage converter. This invention can effectively suppress current step impacts at the discharge end, prevent cascading disconnection of parallel battery racks, and ensure the stability of the power grid peak-shaving process.

Description

Technical Field

[0001] This invention relates to the field of power energy storage control technology, and more specifically, to a peak-shaving dispatching method and system for energy storage power stations in a power system. Background Technology

[0002] With the construction of new power systems, electrochemical energy storage power stations, represented by lithium-ion batteries, have become a key resource for peak shaving and valley filling on the power grid side. In practical engineering applications of utility-grade energy storage power stations, a multi-battery rack parallel bus topology is typically adopted, where multiple battery racks are connected to the same DC bus via a DC combiner, and then connected to the AC grid via a power storage converter (PCS).

[0003] In grid peak shaving scenarios, dispatching instructions typically require energy storage power stations to maintain a constant active power output for extended periods. To ensure operational safety, each battery rack is equipped with an independent battery management system (BMS) and protection devices (such as contactors and fuses). When any single cell is detected to have excessively low voltage, excessively high temperature, or excessive current, the protection mechanism will trigger a physical disconnection of the battery rack, causing it to be taken out of operation.

[0004] However, existing peak-shaving scheduling strategies typically treat the entire energy storage stack as a whole, or perform simple power allocation based solely on the state of charge (SOC), ignoring the discrete structural properties and electrical mechanisms within the parallel system. This traditional control method presents significant safety hazards and control blind spots in the final stages of long-duration constant-power discharge:

[0005] First, due to differences in cell aging, line impedance, and thermal environment, the current distribution among parallel battery racks is often uneven. Especially during the discharge cutoff phase, the rapid drop in voltage causes the current to passively increase in order to maintain constant power, which in turn exacerbates the heterogeneity of current distribution. Some aged or high-temperature battery racks are very likely to reach the protection boundary first.

[0006] Secondly, existing scheduling methods lack the ability to predict discrete disconnection effects. In constant power discharge mode, once a critical battery rack disconnects due to undervoltage or overheating, its original current load will instantly transfer to the remaining online battery racks. This step-like redistribution of current causes the terminal voltage of the remaining battery racks to drop instantaneously (due to a sudden increase in internal resistance voltage) and the temperature rise rate to accelerate. Given the already minimal voltage margin at the end of discharge, this shock can easily induce the remaining battery racks to successively trigger their protection thresholds, resulting in continuous disconnections from the grid.

[0007] This cascading disconnection phenomenon can cause unpredictable phase-change collapse in the output power of energy storage power stations, and in severe cases, it can even cause the entire station to shut down. Not only will it fail to complete the intended peak-shaving task, but it will also impact the stability of the power grid. Summary of the Invention

[0008] This invention provides a peak-shaving dispatching method and system for energy storage power stations in a power system, which solves the technical problems mentioned in the background art.

[0009] Firstly, a peak-shaving dispatch method for an energy storage power station in a power system, applied to an energy storage power station comprising a DC combiner unit and a multi-battery rack parallel structure, including:

[0010] Real-time acquisition of voltage, current, and temperature data of each parallel battery rack, as well as DC bus voltage;

[0011] The equivalent number of parallel branches characterizing the load sharing state is calculated based on the current data of each parallel battery rack, and the current redistribution step coefficient after a single battery rack is disconnected is determined based on the equivalent number of parallel branches.

[0012] By combining the current redistribution step coefficient with the impedance and thermal parameters of each parallel battery rack, the cascade disconnection protection margin for the remaining battery rack to trigger cascading disconnection is predicted.

[0013] The cascade disconnection protection margin is converted into a power regulation coefficient using a continuous smooth mapping model. The preset peak-shaving power command is then corrected, and the corrected command is converted into the upper limit of DC bus current to control the energy storage converter.

[0014] Secondly, a peak-shaving dispatch system for an energy storage power station in a power system is applied to an energy storage power station comprising a DC combiner unit and a multi-battery rack parallel structure, to realize the peak-shaving dispatch method for an energy storage power station in any of the claims, including:

[0015] The data acquisition module is used to collect the voltage, current, and temperature data of each parallel battery rack, as well as the DC bus voltage, in real time.

[0016] The step coefficient determination module is used to calculate the equivalent number of parallel branches characterizing the load sharing state based on the current data of each parallel battery rack, and to determine the current redistribution step coefficient after a single battery rack is disconnected based on the equivalent number of parallel branches.

[0017] The protection margin prediction module is used to combine the current redistribution step coefficient with the impedance and thermal parameters of each parallel battery rack to predict the cascade disconnection protection margin that triggers the interlocking disconnection of the remaining battery rack.

[0018] The peak shaving command control module is used to convert the cascade disconnection protection margin into a power regulation coefficient using a continuous smooth mapping model, correct the preset peak shaving power command, and convert the corrected command into the upper limit value of DC bus current to control the energy storage converter.

[0019] The beneficial effects of this invention are as follows: by constructing a model that includes the equivalent number of parallel branches and the current redistribution step coefficient, the electrical impact risk of a multi-branch parallel architecture after a single-point disconnection is quantified, and the cascade disconnection protection margin is predicted by combining electrothermal coupling parameters; by using a continuous smooth mapping model to transform this margin into a definite upper limit of DC bus current, active defense against the risk of cascading disconnection at the end of peak shaving is achieved, effectively avoiding the phase-change collapse of power output of energy storage power stations, and significantly improving the continuity and security of power system peak shaving services. Attached Figure Description

[0020] Figure 1 This is a flowchart of a peak-shaving dispatch method for an energy storage power station in a power system according to the present invention;

[0021] Figure 2 This is a block diagram of a peak-shaving dispatching system for an energy storage power station in a power system according to the present invention. Detailed Implementation

[0022] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0023] Example 1: As Figure 1 As shown, a peak-shaving dispatch method for an energy storage power station in a power system is applied to an energy storage power station including a DC combiner unit and a multi-battery rack parallel structure, comprising:

[0024] Real-time acquisition of voltage, current, and temperature data of each parallel battery rack, as well as DC bus voltage;

[0025] The equivalent number of parallel branches characterizing the load sharing state is calculated based on the current data of each parallel battery rack, and the current redistribution step coefficient after a single battery rack is disconnected is determined based on the equivalent number of parallel branches.

[0026] By combining the current redistribution step coefficient with the impedance and thermal parameters of each parallel battery rack, the cascade disconnection protection margin for the remaining battery rack to trigger cascading disconnection is predicted.

[0027] The cascade disconnection protection margin is converted into a power regulation coefficient using a continuous smooth mapping model. The preset peak-shaving power command is then corrected, and the corrected command is converted into the upper limit of DC bus current to control the energy storage converter.

[0028] In a preferred embodiment, the equivalent number of parallel branches characterizing the load-sharing state is calculated based on the current data of each parallel battery rack, including:

[0029] The equivalent number of parallel branches is calculated using the following formula:

[0030] ;

[0031] In the formula, for The equivalent number of parallel branches at any given time. For the first A parallel battery rack in Real-time current value at any given moment. This represents the total number of parallel battery racks currently online. This indicates a summation operation.

[0032] Preferably, the effective load branch number is calculated based on the impact of the uniformity of current distribution in parallel battery racks on the load sharing state. In a parallel circuit structure, the current distribution of each battery rack directly relates to its actual load sharing effect. The more uniform the current distribution in each battery rack, the closer the actual number of effective branches is to its actual number; if there are differences in current distribution, the number of effective branches will decrease as the differences increase. Specifically, firstly, for... Real-time current of each parallel battery rack Summing the results reveals the total current level of the current parallel battery rack; squaring this total current strengthens the correlation between the total current and the current in each branch; and for each First, square the currents and then sum them. The result reflects the dispersion of the currents in each branch; the greater the dispersion, the higher the value of the result. Then, divide the square of the total current by the sum of the squares of the currents in each branch. This allows for the quantification of the number of effective load branches at the current moment.

[0033] In specific implementation, At any given time, first obtain the real-time current of each parallel battery rack. ,in The value range is 1 to Next, calculate the numerator: [contain all...] Perform a summation operation to obtain Then, square the summation result to obtain... Then calculate the denominator: for each Perform squaring operations on each side to obtain... Then put all Perform a summation operation to obtain Finally, divide the calculated result of the numerator by the calculated result of the denominator to obtain the numerical value. Equivalent number of parallel branches at time .in, The value is determined by the real-time detected status of the online battery rack, and there is no need to preset a fixed value; Data is acquired through a real-time acquisition device, with a preferred acquisition frequency of 1Hz. This value is determined based on the scheduling system's requirement for real-time current data to ensure that the data can reflect the operating status of the battery rack in a timely manner.

[0034] In a preferred embodiment, determining the current redistribution step coefficient after a single battery rack is disconnected based on the equivalent number of parallel branches includes:

[0035] First, calculate the number of smooth parallel branches, using the following formula:

[0036] ;

[0037] Next, the current redistribution step coefficient is calculated using the following formula:

[0038] ;

[0039] In the formula, for The current redistribution step coefficient at time 1. for The number of smooth parallel branches at any given time. for The equivalent number of parallel branches at any given time. For natural logarithm operations, It is a natural constant.

[0040] Preferably, after a single battery rack is disconnected, the remaining battery rack needs to carry the current of the original rack. The discrete changes in the number of equivalent parallel branches can cause abrupt changes in current distribution. Therefore, it is necessary to smooth the number of equivalent parallel branches to eliminate the impact of these abrupt changes. Specifically, a formula for smoothing the number of parallel branches is constructed: by summing 1 with the natural logarithm (the logarithm is 1 plus the natural exponent, and the exponent is the number of equivalent parallel branches minus 1), the discrete changes in the number of equivalent parallel branches are transformed into a continuous, smoothed number of parallel branches, avoiding abrupt changes in subsequent coefficients. Then, to quantify the current amplification of the remaining rack after the single battery rack is disconnected, a formula for the current redistribution step coefficient is constructed using the smoothed number of parallel branches, i.e., the ratio of the smoothed number of parallel branches to (smoothed number of parallel branches minus 1). The coefficient obtained in this way can continuously reflect the changing trend of current distribution. In specific implementation, at time t, the previously calculated number of equivalent parallel branches at time t is first obtained. Substitute it into the formula Among them It is a natural constant (with a value of approximately 2.71828). The number of smooth parallel branches at time t is calculated using the natural logarithm. ; then, Substitute into the formula In the calculation, the current redistribution step coefficient at time t is obtained. .

[0041] In a preferred embodiment, by combining the current redistribution step coefficient with the impedance and thermal parameters of each parallel battery rack, the cascade disconnection protection margin for triggering interlocking disconnection of the remaining battery rack is predicted, including:

[0042] First, calculate the predicted voltage margin using the following formula:

[0043] ;

[0044] ;

[0045] ;

[0046] Next, the cascade disconnection protection margin is calculated using the following formula:

[0047] ;

[0048] In the formula, For the first Predicted current values ​​for each parallel battery rack. The step coefficient for the current redistribution. This is the real-time current value. To account for the predicted equivalent internal resistance after thermal feedback, This is the equivalent open-circuit voltage. To predict the terminal voltage, This is the undervoltage protection threshold. For the first Predicted voltage margin of a parallel battery rack, For the cascade disconnection protection margin, To smooth the polymerization coefficient, For natural logarithm operations, It is a natural constant. This indicates a summation operation.

[0049] Preferably, after a single battery rack is disconnected, the current of the remaining battery racks will be amplified by the current redistribution step factor. At the same time, the internal resistance of the battery racks will change due to temperature variations, which will further affect the terminal voltage of the battery racks. If the terminal voltage is lower than the undervoltage protection threshold, a new disconnection will be triggered. Therefore, it is necessary to first predict the terminal voltage after the current amplification and its margin relative to the protection threshold, and then integrate the margins of all remaining racks to quantify the risk of cascading disconnection. Therefore, the predicted current is first obtained by multiplying the current redistribution step factor and the real-time current, and then the predicted terminal voltage is calculated by combining the internal resistance after thermal feedback and the open-circuit voltage to obtain the predicted voltage margin. Finally, the cascading disconnection protection margin is obtained through smoothing aggregation calculation.

[0050] Specifically, at time t, first obtain the previously obtained current redistribution step coefficient at time t. Real-time current value of the i-th parallel battery rack Predicted equivalent internal resistance considering thermal feedback Equivalent open-circuit voltage Undervoltage protection threshold and smoothing polymerization coefficient (The preferred value is 20-50, which is a range that can achieve a balance between smoothing the fluctuations in the margin of each battery rack and the differences in retention risk in multiple simulations); First, and Substitute into the formula The predicted current value of the i-th parallel battery rack is calculated. Next, , and Substitute into the formula The predicted terminal voltage of the i-th parallel battery rack is calculated. ; then and Substitute into the formula The predicted voltage margin of the i-th parallel battery rack is calculated. Finally, With each Substitute into the formula (in is a natural constant, with a value of approximately 2.71828. (This is calculated using natural logarithm operations) to obtain the cascade disconnection protection margin at time t. .

[0051] In a preferred embodiment, the cascade disconnection protection margin is converted into a power regulation coefficient using a continuous smooth mapping model, including:

[0052] First, calculate the distance error value using the following formula:

[0053] ;

[0054] Next, the power regulation coefficient is calculated using the following formula:

[0055] ;

[0056] In the formula, This represents the distance error value. For the cascade disconnection protection margin, The preset safe distance threshold, The power regulation coefficient is... This is the soft saturation kurtosis coefficient. For natural logarithm operations, It is a natural constant.

[0057] Preferably, the cascade disconnection protection margin needs to be converted into a power regulation coefficient to correct the peak-shaving command. However, direct numerical conversion can easily lead to abrupt changes in the regulation command. Therefore, the deviation (distance error) between the protection margin and the preset safety benchmark is first quantified. Then, through soft saturation calculations involving natural exponents and logarithms, the deviation is converted into a continuously changing power regulation coefficient to achieve smooth command correction. The corresponding calculation process is constructed based on this approach. In specific implementation, at time t, the previously obtained cascade disconnection protection margin at time t is first obtained. Preset safe distance threshold (Its preferred value is determined based on the allowable margin fluctuation range of the system, usually ranging from 0.2 to 0.5), and the soft saturation kurtosis coefficient. (The preferred value is 20 to 50, which ensures a fast adjustment response while avoiding abrupt changes in the coefficient); First, and Substitute into the formula The distance error at time t is calculated. ; then , Substitute into the formula (in is a natural constant, with a value of approximately 2.71828. (This is a natural logarithm operation), which calculates the power regulation coefficient at time t. .in, The calculation results for the cascade disconnection protection margin are as follows: These are pre-set safety baseline values. These are pre-configured coefficients used to control the steepness of the soft saturation function.

[0058] In a preferred embodiment, modifying the preset peak-shaving power command includes:

[0059] The corrected peak-shaving power command is calculated using the following formula:

[0060] ;

[0061] In the formula, for The peak power command after time correction. for Preset peak-shaving power command at any time. The power regulation coefficient is denoted as .

[0062] Preferably, the preset peak-shaving power command is generated based on normal operating conditions. However, in actual operation, there is a risk of cascading disconnection of the battery racks. Therefore, it needs to be corrected by incorporating a power regulation coefficient that reflects the risk state, so that the peak-shaving power command adapts to the current safe operating boundary. Specifically, the preset command is combined with the regulation coefficient through multiplication to obtain an execution command that fits the actual state. In specific implementation, at time t, the preset peak-shaving power command at time t is first obtained. This value is the peak-shaving power target value corresponding to time t, which is pre-issued by the scheduling system; then, the power regulation coefficient at time t, which was calculated earlier, is obtained. ; then and Substitute into the formula The symbol “·” represents a multiplication operation, which calculates the corrected peak power command at time t. This instruction is the actual peak-shaving power executed according to the current operating status of the battery rack. The values ​​are pre-configured and issued by the scheduling system; The power regulation coefficient is denoted as .

[0063] In a preferred embodiment, converting the modified command into a DC bus current upper limit value to control the energy storage converter includes:

[0064] The upper limit value of the DC bus current is calculated using the following formula:

[0065] ;

[0066] In the formula, for The upper limit of DC bus current at any given time. This refers to the revised peak-shaving power command. for DC bus voltage at any given time;

[0067] The calculated The current limiting parameters are sent to the energy storage converter.

[0068] Preferably, the corrected peak-shaving power command is a power parameter, while the operation of the energy storage converter needs to be controlled by current parameters. Therefore, it is necessary to combine the real-time DC bus voltage to convert the power command into a corresponding upper limit current value, which serves as the current-limiting constraint for the converter, ensuring that the current executed by the converter matches the corrected power command. Specifically, a corresponding conversion formula is constructed. In practical implementation, at time t, the previously calculated corrected peak-shaving power command at time t is first obtained. Then obtain the DC bus voltage at time t. The voltage value is acquired by a real-time acquisition device, and the acquisition frequency is consistent with the overall data acquisition frequency of the system; subsequently, and Substitute into the formula The upper limit of the DC bus current at time t is calculated. ; then, the calculated As a current-limiting parameter, it is sent to the energy storage converter to constrain the converter's operating current. Among these parameters... The result of the correction calculation is to adjust the preset peak power command. These are the operating parameters collected in real time.

[0069] It should be noted that the sum of the squared currents is an intermediate variable used to calculate the equivalent number of parallel branches. It is obtained by squaring the sum of the real-time currents of all parallel battery racks. Its function is to enhance the correlation between the total current and the current in each branch.

[0070] It should be noted that the sum of squared currents is an intermediate variable used to characterize the dispersion of current distribution in each parallel battery rack. It is obtained by squaring the real-time current of each parallel battery rack individually and then summing the results. The more uneven the current distribution, the larger this value. When used in conjunction with the square of the total current, it can accurately reflect the degree of heterogeneity in load sharing.

[0071] It should be noted that dividing the square of the total current by the sum of the squares of the currents yields the equivalent number of parallel branches. By using the ratio of the square of the total current to the sum of the squares of the currents in each branch, the abstract current distribution uniformity is transformed into a scalar, avoiding the limitations of traditional methods that judge the load status solely by the current of a single battery rack. This method can comprehensively reflect the coordinated load status of all parallel battery racks.

[0072] It should be noted that the equivalent number of parallel branches It is used for quantification The effective load branch number of the parallel battery rack at any given time is calculated using the following formula: Its value corresponds to the uniformity of current distribution; that is, when the current is completely evenly distributed across all battery racks. Equal to the total number of battery racks currently online When the current is concentrated in a few battery racks, It will be significantly smaller than For example, if 3 out of 8 parallel battery racks carry 70% of the current, It may drop to around 4, reflecting the effectiveness of load sharing.

[0073] It should be noted that the real-time current value of the i-th parallel battery rack at time t... This refers to the individual battery racks that make up a parallel system at a specific moment. Real-time current monitoring data, including The value range is from 1 to the current total number of online parallel battery racks. .

[0074] It should be noted that the total number of parallel battery racks currently online... It means This refers to the number of battery racks that are always in normal operation and participating in parallel operation, rather than the total number of battery racks installed in the energy storage power station. This value changes dynamically as the protection of a battery rack is disconnected or restored. For example, if an energy storage power station has a total of 12 battery racks... At any given time, one circuit will trip due to undervoltage protection. This parameter affects the calculation results of the equivalent parallel branch number and the current redistribution step coefficient.

[0075] It should be noted that the smoothing function, which includes operations on the natural logarithm and natural exponent, has the following form: ,in For natural logarithm operations, It is a natural constant, and its function is to... Discrete jumps (such as from a single frame disconnection) Become This transforms the data into a continuous and smooth numerical change, avoiding abrupt changes in the step coefficient of subsequent current redistribution, thus ensuring the continuity and stability of risk prediction. Its smoothing characteristic stems from the combined effect of the natural exponential and logarithmic functions, preserving... While dispersing information, it also eliminates discrete disturbances.

[0076] It should be noted that the logic of smoothing the equivalent parallel branch count by subtracting 1 and then adding 1 is a step in constructing a smoothed parallel branch count, which first calculates... By focusing on the change in the degree of parallel connection, and then processing it as a continuous function through a smoothing function, and finally adding 1 to restore it to the order of the number of branches, the problem of control mutation caused by discrete changes in the number of branches in parallel systems can be solved in a targeted manner.

[0077] It should be noted that the number of smooth parallel branches The smoothing result used to handle the equivalent number of parallel branches is calculated using the following formula: By eliminating The discrete characteristics, for example when When it suddenly drops from 7 to 6 The value will transition continuously from close to 7 to close to 6, rather than jumping abruptly, ensuring the step factor for subsequent current redistribution. The continuous changes in parameters are used to avoid oscillations in peak-shaving commands caused by sudden changes in parameters.

[0078] It should be noted that the current redistribution step coefficient The formula used to characterize the current amplification factor of the remaining battery rack after a single battery rack is disconnected is as follows: This reflects the degree of current impact on the remaining system caused by the disconnection of a single frame, i.e. The smaller, The larger the current, the more intense the current surge, for example when hour, This means that if one frame is disconnected, the current in the remaining frame will be amplified by 1.33 times; when hour, The current surge is significantly aggravated.

[0079] It should be noted that the current redistribution step coefficient after a single battery rack is disconnected refers to the coefficient that quantifies the process of a step redistribution of current borne by the remaining battery racks after a single parallel battery rack is disconnected due to protection trigger.

[0080] It should be noted that the method of updating the equivalent internal resistance based on the predicted temperature rise is a design that quantifies the thermo-electrical correlation effect. The internal resistance change is inferred by predicting the temperature rise caused by the predicted current, and then this change is incorporated into the internal resistance parameter to form the predicted equivalent internal resistance. This addresses the shortcomings of traditional dispatching methods that ignore the impact of temperature on electrical parameters; specifically, when a single frame disconnects, the amplified current not only directly impacts the voltage but also... Heat generation raises the battery rack temperature, which in turn increases internal resistance, creating a positive feedback risk. The specific process involves first predicting the current... The thermal model calculates the predicted temperature rise within a preset time window, and then uses a temperature-internal resistance correlation model (such as...) , The initial equivalent internal resistance is updated using the temperature resistance coefficient, and the predicted equivalent internal resistance considering thermal feedback is finally obtained.

[0081] It should be noted that the predicted equivalent internal resistance... It refers to the first The equivalent internal resistance of a parallel battery rack after a single rack is disconnected, and after being affected by temperature rise within a predicted time window, is a dynamically optimized version of the traditional equivalent internal resistance. Its core value lies in capturing the chain reaction of current amplification → temperature rise → increased internal resistance, avoiding voltage prediction errors caused by using a fixed internal resistance. For example, the initial equivalent internal resistance of a battery rack is... The predicted current amplification after a single frame disconnects will cause a temperature rise. ,like Then the equivalent internal resistance is predicted. By substituting the predicted equivalent internal resistance into the predicted terminal voltage formula, the voltage calculation can reflect the shrinkage of the safety boundary caused by thermal feedback.

[0082] It should be noted that the predicted terminal voltage This is a parameter set to determine whether the battery rack triggers undervoltage protection after a single rack is disconnected. The calculation formula is as follows: ,in This is the equivalent open-circuit voltage. To predict the equivalent internal resistance, To predict the current. Its significance is that after a single frame disconnects, the first... The predicted terminal voltage of a battery rack under the combined effects of current surge and thermal feedback is the basis for assessing the safety status of a single battery rack. If the undervoltage protection threshold is met, the battery rack will not trigger protection; otherwise, there is a risk of it being deactivated. Compared with traditional terminal voltage calculation, by incorporating the instantaneous impact of current redistribution and the hysteresis effect of thermal feedback, the prediction accuracy is more in line with actual operating conditions.

[0083] It should be noted that the predicted voltage margin It is a parameter that quantifies the voltage safety boundary of a single battery rack, and the calculation formula is as follows: Reflecting the first The voltage safety redundancy of each battery rack after a single rack is disconnected. A larger predicted voltage margin indicates a more ample safety margin and a lower likelihood of triggering undervoltage protection; a value ≤ 0 indicates that the battery rack will be tripped up when a single rack is disconnected. For example, if the predicted terminal voltage of a battery rack... Undervoltage protection threshold Then predict voltage margin This indicates that the battery rack has a 0.3V redundancy to withstand sudden fluctuations; if the predicted terminal voltage drops to 2.4V, the margin is -0.1V, indicating that protection will be triggered.

[0084] It should be noted that the smoothing function used for aggregating the predicted voltage margin has the following form: ,in To smooth the polymerization coefficient, This represents the total number of currently online parallel battery racks. Its function is to aggregate the individual predicted voltage margins of all parallel battery racks into a system-level cascade disconnection protection margin, while avoiding control oscillations caused by traditional minimum-value aggregation. Specifically, through a combined operation of natural exponential and logarithmic calculations, the margins of all battery racks are weighted and aggregated; that is, battery racks with higher risk (smaller margins) receive higher weights, making the aggregation result sensitive to the risk of the weakest link in the system. The value of balances risk sensitivity and aggregation smoothness.

[0085] It should be noted that the cascade disconnection protection margin This is used to quantify the cascading disconnect risk of the entire parallel battery rack system after the failure of a single rack. Its calculation formula is a smoothed aggregation function, which is a smoothed representation of the minimum voltage safety margin at the system level, i.e. This indicates that even if a single battery rack fails, the system as a whole still has safety redundancy, and the remaining battery racks will not be deactivated in a chain reaction. This indicates that the system has entered a danger zone; the disconnection of a single aircraft will trigger a cascading shutdown of multiple aircraft. For example, when... When the system is in a safe state; when When a single frame disconnects, it will trigger a cascading disconnection.

[0086] It should be noted that the smoothing polymerization coefficient This parameter is set to optimize the smoothing effect of the polymer. The preferred value is 20~50. Its function is to control the sensitivity of the smoothing function to the voltage margin of a single battery cell. The larger the value, the closer the smoothing function is to taking all values. The effect of minimizing the risk is more sensitive to the battery rack where the risk is weakest, and it can quickly detect local risks. The smaller the value, the smoother the aggregation result, avoiding over-adjustment caused by instantaneous fluctuations in individual battery racks. This value range was determined through multiple simulations, ensuring both the system's sensitivity to real-world risks and filtering out invalid fluctuations, thus guaranteeing cascade disconnection protection margin. Stability and reliability.

[0087] It should be noted that the preset prediction time window This parameter is set for calculating and predicting temperature rise. It refers to the length of time used to predict the temperature rise of the battery rack after a single rack is disconnected. The preferred value is 10 seconds, which balances the accuracy of temperature rise prediction with the real-time performance of the calculation. This time window is chosen because the temperature rise caused by the current surge after a single rack is disconnected can be fully reflected within 10 seconds, and this duration is much shorter than the control cycle of peak shaving scheduling (such as 1 second or 2 seconds), so it will not affect the real-time performance of scheduling commands. If the time window is too short (such as 1 second), the complete temperature rise amplitude cannot be captured, resulting in an underestimation of the equivalent internal resistance; if the time window is too long (such as 30 seconds), it will introduce uncertainty in future operating conditions and reduce the accuracy of the prediction.

[0088] It should be noted that the equivalent open-circuit voltage of the parallel battery rack It refers to the first The equivalent open-circuit voltage of a parallel battery rack differs from the open-circuit voltage of a traditional battery cell or the overall open-circuit voltage of an energy storage stack. It is obtained through online identification using the Thevenin equivalent model. The equivalent open-circuit voltage of a parallel battery rack reflects the chemical state (e.g., SOC) and aging state of an individual rack, rather than the overall average state. Specifically, differences in SOC and aging levels among parallel battery racks will result in different equivalent open-circuit voltages. Using the overall open-circuit voltage would lead to inaccuracies in the voltage prediction of individual battery racks.

[0089] It should be noted that the undervoltage protection threshold of the parallel battery rack... This refers to the lower limit voltage of undervoltage protection for a single parallel battery rack, not the protection threshold for the entire energy storage stack. Specifically, traditional undervoltage protection thresholds are mostly for individual battery cells or the entire energy storage stack. However, each parallel battery rack has an independent BMS and protection devices, requiring a dedicated protection threshold to be set for each rack to adapt to the discrete protection logic of the parallel system. For example, the undervoltage protection threshold for a typical lithium iron phosphate battery cell is 2.5V. If a parallel battery rack consists of 100 cells connected in series, then the undervoltage protection threshold for that rack is... This ensures that the protection threshold is applied to parallel battery racks, guaranteeing that undervoltage protection can accurately trigger the disconnection of a single rack, thus avoiding affecting the operation of other normal racks.

[0090] It should be noted that the distance error value is obtained by subtracting the preset safe distance threshold from the cascade disconnection protection margin, which is achieved by introducing the safe distance threshold. This provides additional safety redundancy for the system, avoiding misjudgments caused by prediction errors. Specifically, it includes cascade disconnection protection margin. These are theoretical values ​​predicted by the model, and therefore contain a certain degree of prediction error. If we directly use... As a risk threshold, a risk may occur but not be identified due to prediction errors. Therefore, setting a safe distance threshold is necessary. ,pass Calculate the distance error value, that is This indicates that the system is not only theoretically safe, but also has sufficient redundancy to withstand prediction errors; This indicates that the system has approached or exceeded safety standards and power needs to be reduced to mitigate risks.

[0091] It should be noted that the distance error value Used to quantify cascade disconnection protection margin Compared with the preset safety benchmark (safe distance threshold) The deviation is calculated using the following formula: Its significance lies in the difference between the system's actual safety margin and the preset safety benchmark, serving as a crucial bridge connecting the system's risk state and power regulation commands. For example, if a preset safety distance threshold... Cascade disconnection protection margin Then the distance error value The system has sufficient safety redundancy to maintain the current power; if ,but The system is approaching the safety boundary and power needs to be reduced.

[0092] It should be noted that the preset safe distance threshold This refers to the extra voltage margin reserved to mitigate prediction errors and sudden fluctuations, preferably ranging from 0.2 to 0.5V. Its function is to create a safety buffer for the system. The value is determined by comprehensively considering factors such as model prediction errors (e.g., voltage prediction error ±0.1V) and sudden current fluctuations (e.g., ±5%), ensuring that even with certain deviations, the system can still avoid accidental cascading disconnection. For example, if the maximum prediction error is 0.1V, then (…). =0.3V) can ensure that even if the actual margin is 0.1V smaller than the predicted value, there is still a redundancy of 0.2V; if the value is too small (such as 0.1V), it cannot resist the prediction error; if the value is too large (such as 0.8V), it will excessively limit the power output and affect the peak shaving efficiency. The range of 0.2 to 0.5V can achieve a balance between safety and efficiency.

[0093] It should be noted that the soft saturation function, which includes the difference between two logarithmic and exponential terms, is a mapping model designed to achieve continuous and smooth power regulation, and its form is: ,in This is the soft saturation kurtosis coefficient. This represents the distance error value. The difference between the two logarithmic-exponential terms is used to calculate the distance error for any range. It continuously and smoothly maps to the [0,1] interval, avoiding power spikes caused by traditional hard threshold control. When hour, The system has sufficient safety redundancy to maintain the preset peak-shaving power; when hour, The system risk is extremely high, requiring a significant reduction in power; when hour, Follow Linear variation enables smooth power regulation, avoiding impact on the power grid and energy storage equipment.

[0094] It should be noted that the soft saturation kurtosis coefficient This parameter is set to optimize the adjustment characteristics of the soft saturation function. The preferred value is 20~50. Its function is to control the steepness of the soft saturation function and balance the adjustment response speed and stability. The larger the value, the closer the function is to the value of the function. The effect of truncating to [0,1] is that the more sensitive the response, the faster it can deal with changes in risk. The smaller the value, the smoother the function and the more stable the adjustment process, but the slower the response to risks. This value range was determined through multiple control simulations, ensuring that... When the safety boundary is exceeded, the power regulation coefficient can decrease rapidly to avoid risks, while also preventing [the situation from worsening]. To prevent power oscillations caused by minor fluctuations, and to ensure the stability of peak-shaving power.

[0095] It should be noted that the power regulation coefficient This function is used to convert system-level risk quantification results into executable power correction instructions. Its value range is [0,1], and the calculation formula is a soft saturation function. Its meaning is the preset correction ratio for the peak-shaving power instruction, i.e. The closer it is to 1, the lower the system risk and the higher the peak-shaving power that can be maintained; The closer the value is to 0, the higher the system risk, and the greater the need to reduce peak-shaving power. For example, when... , hour, This indicates that the preset peak-shaving power command needs to be corrected to 50% of its original value; when hour, It can maintain the preset power.

[0096] It should be noted that using a soft saturation function to map the distance error value to obtain the power regulation coefficient is a design for achieving continuous and smooth peak-shaving control. This is achieved by leveraging the nonlinear mapping characteristics of the soft saturation function to map the distance error value... Power regulation coefficient converted to the [0,1] interval This avoids the power surges and grid impacts caused by traditional hard threshold control. Specifically, it is divided into three stages: when At that time, the mapping result The system has sufficient safety redundancy and requires no power adjustment; when At that time, the mapping result follows Linear variation enables gradual power adjustment; when At that time, the mapping result The system poses an extremely high risk and requires a rapid reduction in power.

[0097] It should be noted that the revised peak-shaving power command This is used to balance the preset peak-shaving demand with system risk control. It involves quantifying the abstract risk result (power regulation coefficient). This is transformed into an executable power command, unlike traditional fixed commands based solely on grid demand. This command dynamically adapts to the real-time risk status of the parallel battery racks, i.e., when the system risk is low (…). When the system risk is close to 1), the command approaches the preset peak-shaving power to ensure peak-shaving efficiency; when the system risk is high ( When the value approaches 0, the command is significantly reduced to avoid the risk of cascading failure. For example, if the preset peak-shaving power command requires the power plant to output full power, but the system detects an increased risk of cascading failure, If the value drops to 0.6, the corrected command will adjust the power to 60% of the preset value, which will both meet some of the peak-shaving requirements and avoid equipment safety accidents.

[0098] It should be noted that the preset peak-shaving power command This refers to the initial peak-shaving power target value issued by the power grid dispatching system based on overall load forecasting. Specifically, this instruction does not consider the discrete structural attributes of parallel battery racks and the risk of cascading disconnection; it is formulated solely based on the power grid supply and demand balance. Direct execution of this instruction could easily lead to safety hazards. For example, the power grid issues a peak load target value based on peak load demand. It may require the energy storage power station to output full power, but the instruction does not take into account the actual state that some battery racks are close to the protection threshold. By combining it with the power regulation coefficient, the instruction can be adapted to the scenario, filling the matching gap between the initial instruction and the operating boundary of the parallel system.

[0099] It should be noted that the upper limit of DC bus current... It serves as a crucial bridge connecting power commands and current limiting constraints. By transforming the abstract power correction result into a current threshold that can be directly applied to the energy storage converter, it fundamentally limits the total current flowing through the DC combiner unit, preventing a step amplification of the current in the remaining racks after a single rack disconnects. This parameter is used to achieve closed-loop control from risk quantification to power correction to physical current limiting, unlike traditional indirect control based solely on voltage or SOC. For example, after the corrected peak-shaving power command is determined, the upper limit of the current is calculated using the real-time DC bus voltage. The DC loop controller of the energy storage converter will strictly limit the total current within this threshold. Even if a battery rack disconnects, the current shared by the remaining racks will not exceed the safety boundary, thus blocking the trigger chain of cascading disconnections.

[0100] It should be noted that the DC bus voltage This refers to the real-time bus voltage after the DC bus unit aggregates the current of all parallel battery racks, which is different from the terminal voltage of a single battery rack or the general DC bus voltage. Specifically, the DC bus voltage is not a fixed value, but is affected by the terminal voltage, current distribution, and line impedance of all parallel battery racks, directly reflecting the overall electrical state of the parallel system.

[0101] It should be noted that energy storage power stations with multiple parallel battery racks and DC combiner unit structures face the risk of single rack disconnection leading to current redistribution and cascade shutdown. General peak-shaving commands are not designed to address this risk. For example, for energy storage devices with a single battery rack structure, there is no parallel current redistribution issue, and the initial command can be executed directly. However, for parallel-structured power stations, scenario-specific modifications are necessary to ensure operational safety.

[0102] It should be noted that the voltage fluctuations after multiple battery racks are connected in parallel reflect the uniformity of current distribution and the consistency of battery rack status in the parallel system. That is, when the current distribution is uneven or some battery racks are in abnormal condition, the DC bus voltage will exhibit irregular fluctuations, which directly affect the accuracy of the current upper limit calculation. This can be addressed by real-time data acquisition in this scenario. This ensures that the upper limit of the current can dynamically adapt to changes in system status, avoiding control deviations caused by the use of general voltage acquisition logic.

[0103] It should be noted that, for multi-battery rack parallel structures, controlling the DC loop of the energy storage converter through the upper limit of the DC bus current is a risk-averse measure for multi-battery rack parallel structures. The upper limit of the DC bus current is used as a control signal to act on the DC loop controller of the energy storage converter. Specifically, to address the risk of current step amplification in parallel systems, the current shared by each parallel battery rack is indirectly controlled by limiting the total current, thereby blocking the triggering conditions for cascading disconnection. For example, general-purpose converter control may only focus on whether the AC side output power meets the standard, while simultaneously constraining the DC side state through the current upper limit to ensure the safe operation of the parallel system, achieving dual protection for both AC side peak-shaving needs and DC side system safety.

[0104] Example 2: Figure 2 As shown, a peak-shaving dispatch system for an energy storage power station in a power system is applied to an energy storage power station including a DC combiner unit and a multi-battery rack parallel structure, to realize the peak-shaving dispatch method for an energy storage power station in any of the claims, including:

[0105] The data acquisition module is used to collect the voltage, current, and temperature data of each parallel battery rack, as well as the DC bus voltage, in real time.

[0106] The step coefficient determination module is used to calculate the equivalent number of parallel branches characterizing the load sharing state based on the current data of each parallel battery rack, and to determine the current redistribution step coefficient after a single battery rack is disconnected based on the equivalent number of parallel branches.

[0107] The protection margin prediction module is used to combine the current redistribution step coefficient with the impedance and thermal parameters of each parallel battery rack to predict the cascade disconnection protection margin that triggers the interlocking disconnection of the remaining battery rack.

[0108] The peak shaving command control module is used to convert the cascade disconnection protection margin into a power regulation coefficient using a continuous smooth mapping model, correct the preset peak shaving power command, and convert the corrected command into the upper limit value of DC bus current to control the energy storage converter.

[0109] The peak-shaving dispatch system process of a multi-battery rack parallel structure energy storage power station includes two links: energy flow and data flow. At the energy level, the power grid is connected to the energy storage converter (PCS) via a transformer. The PCS is connected to multiple sets of parallel battery racks through a DC bus and a DC combiner unit to realize the transfer of charging and discharging energy between the power grid and the energy storage power station. At the data and control level, the data acquisition module collects the voltage, current, and temperature data of each parallel battery rack in real time, and the DC bus voltage acquisition module acquires the DC bus voltage data synchronously. Both types of data are transmitted to the peak-shaving dispatch controller. The controller internally performs the following processes sequentially: equivalent parallel branch number calculation (based on battery rack current data) → current redistribution step coefficient determination → cascade disconnection protection margin prediction (combined with impedance and thermal parameters) → continuous smooth mapping and peak power command correction. Finally, the upper limit value of DC bus current is calculated and sent to the DC loop controller of the PCS. The controller then constrains the operation of the PCS according to the upper limit of current, thereby limiting the total DC bus current. This not only meets the peak demand of the power grid but also avoids the risk of cascade disconnection caused by current step amplification of parallel battery racks.

[0110] It should be specifically noted that all input data described in the above specific implementation methods are acquired in real time through legal and compliant hardware interfaces with the user's full knowledge, explicit consent, and active cooperation. The preset parameters, prior constants, and statistical means are all derived from publicly available scientific literature data, de-identified general scientific research datasets, or calibration data in a laboratory environment, and do not contain any unauthorized sensitive third-party information. The system's data processing is limited to local or volatile memory computation transmitted via an encrypted channel. There is no situation of illegally collecting, stealing, or retaining user biometric data or infringing on user privacy without the user's knowledge. All parameter calls and generation comply with the principles of data minimization and legality, legitimacy, and necessity.

[0111] The embodiments of this example have been described above. However, this example is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of this example, and all of them are within the protection scope of this example.

Claims

1. A peak-shaving dispatch method for an energy storage power station in a power system, applied to an energy storage power station comprising a DC combiner unit and a multi-battery rack parallel structure, characterized in that, include: Real-time acquisition of voltage, current, and temperature data of each parallel battery rack, as well as DC bus voltage; The equivalent number of parallel branches characterizing the load sharing state is calculated based on the current data of each parallel battery rack, and the current redistribution step coefficient after a single battery rack is disconnected is determined based on the equivalent number of parallel branches. By combining the current redistribution step coefficient with the impedance and thermal parameters of each parallel battery rack, the cascade disconnection protection margin for the remaining battery rack to trigger cascading disconnection is predicted. The cascade disconnection protection margin is converted into a power regulation coefficient using a continuous smooth mapping model. The preset peak-shaving power command is then corrected, and the corrected command is converted into the upper limit of DC bus current to control the energy storage converter.

2. The peak-shaving dispatch method for an energy storage power station in a power system according to claim 1, characterized in that, The equivalent number of parallel branches characterizing the load sharing state is calculated based on the current data of each parallel battery rack, including: Obtain the real-time current value of each parallel battery rack at the current moment; The real-time current values ​​of all parallel battery racks are summed, and the summation result is squared to obtain the square value of the total current. The real-time current value of each parallel battery rack is squared, and all the squared values ​​are summed to obtain the total sum of squared current values. Divide the square of the sum of currents by the sum of the squares of the currents to obtain the equivalent number of parallel branches.

3. The peak-shaving dispatch method for an energy storage power station in a power system according to claim 2, characterized in that, Determining the current redistribution step coefficient after a single battery rack is disconnected based on the equivalent number of parallel branches includes: Construct a smoothing function that includes natural logarithm and natural exponent operations. Use the smoothing function to smooth the difference between the equivalent number of parallel branches and the value of one, and add the value of one to obtain the smoothed number of parallel branches. Calculate the difference between the number of smooth parallel branches and the value one to obtain the branch number difference; The current redistribution step coefficient is obtained by dividing the number of smooth parallel branches by the difference in the number of branches.

4. The peak-shaving dispatch method for an energy storage power station in a power system according to claim 3, characterized in that, Combining the current redistribution step coefficient with the impedance and thermal parameters of each parallel battery rack, the cascade disconnection protection margin for triggering interlocking disconnection of the remaining battery rack is predicted, including: Multiply the current redistribution step coefficient by the real-time current value of each parallel battery rack to obtain the predicted current value of each parallel battery rack after a single rack is disconnected. Using the predicted current value and the preset prediction time window, the predicted temperature rise is calculated in combination with the thermal model of each parallel battery rack, and the equivalent internal resistance of each parallel battery rack is updated based on the predicted temperature rise to obtain the predicted equivalent internal resistance. The predicted terminal voltage is obtained by subtracting the product of the predicted equivalent internal resistance and the predicted current value from the equivalent open-circuit voltage of each parallel battery rack. Subtract the preset undervoltage protection threshold from the predicted terminal voltage to obtain the predicted voltage margin for each parallel battery rack. A smoothing function is constructed to aggregate the predicted voltage margins of all parallel battery racks to obtain the cascade disconnection protection margin.

5. The peak-shaving dispatch method for an energy storage power station in a power system according to claim 4, characterized in that, The cascade disconnection protection margin is converted into a power regulation coefficient using a continuous smooth mapping model, including: Subtract the preset safe distance threshold from the cascade disconnection protection margin to obtain the distance error value; A soft-saturated function containing the difference between two log-exponential terms is constructed as the continuous smooth mapping model; The distance error value is calculated using the soft saturation function, and the distance error value is mapped to a value between zero and one to obtain the power adjustment coefficient.

6. The peak-shaving dispatch method for an energy storage power station in a power system according to claim 5, characterized in that, The preset peak-shaving power command is modified, including: Obtain the current preset peak-shaving power command; The preset peak-shaving power command is multiplied by the power adjustment coefficient to obtain the corrected peak-shaving power command.

7. The peak-shaving dispatch method for an energy storage power station in a power system according to claim 6, characterized in that, The revised command is converted into an upper limit value for the DC bus current to control the energy storage converter, including: Obtain the current DC bus voltage; Divide the corrected peak power command by the DC bus voltage to obtain the upper limit value of the DC bus current; The upper limit value of the DC bus current is sent to the DC loop controller of the energy storage converter to limit the total current flowing through the DC combiner unit.

8. A peak-shaving dispatch system for an energy storage power station in a power system, applied to an energy storage power station including a DC combiner unit and a multi-battery rack parallel structure, to realize the peak-shaving dispatch method for an energy storage power station in a power system as described in any one of claims 1-7, characterized in that, include: The data acquisition module is used to collect the voltage, current, and temperature data of each parallel battery rack, as well as the DC bus voltage, in real time. The step coefficient determination module is used to calculate the equivalent number of parallel branches characterizing the load sharing state based on the current data of each parallel battery rack, and to determine the current redistribution step coefficient after a single battery rack is disconnected based on the equivalent number of parallel branches. The protection margin prediction module is used to combine the current redistribution step coefficient with the impedance and thermal parameters of each parallel battery rack to predict the cascade disconnection protection margin that triggers the interlocking disconnection of the remaining battery rack. The peak shaving command control module is used to convert the cascade disconnection protection margin into a power regulation coefficient using a continuous smooth mapping model, correct the preset peak shaving power command, and convert the corrected command into the upper limit value of DC bus current to control the energy storage converter.

Images