Battery management system for a hybrid energy storage system based on soc estimation

Abstract

The present application relates to the technical field of electric energy storage system, in particular to a composite energy storage system battery management system based on SOC estimation. The system focuses on the distribution of the sustained frequency modulation time period, the current oscillation time period and the conventional current time period in the to-be-analyzed time period, quantifies the oscillation frequency index, the adjustment time scale and the power supply rate of the energy storage unit at the real-time moment. Based on the evaluation accuracy, calculation resources and anti-current fluctuation of the SOC estimation algorithm, the real-time characteristics of the battery are matched to select the optimal SOC estimation algorithm for execution. By analyzing the time period distribution characteristics of the to-be-analyzed time period at the real-time moment, the real-time state of the energy storage battery is quantified, so that the finally recommended SOC estimation algorithm can improve the display accuracy of the residual capacity percentage value of the energy storage battery in the composite energy storage system on the basis of meeting the hard requirements of the power grid system, thereby improving the battery management efficiency of the composite energy storage system.

Description

Technical Field

[0001] This invention relates to the field of energy storage system technology, and more specifically to a battery management system for a composite energy storage system based on SOC estimation. Background Technology

[0002] Current hybrid energy storage systems, by centrally combining different types of energy storage technologies and leveraging supporting technologies such as hardware configuration, energy management, and coordinated dispatch, can efficiently provide active regulation support across multiple time scales. Compared to the decentralized and independent configuration of different types of energy storage systems, centralized hybrid energy storage systems can improve the overall performance and utilization efficiency of the energy storage system through efficient synergy and interaction among various elements. Their comprehensive multi-time scale regulation capability is significantly higher than the simple sum of corresponding independent energy storage capacities. Furthermore, as a unified regulation support entity, hybrid energy storage has comprehensive regulation functions, can autonomously track the grid operating status, and optimize and decompose regulation plans across different time scales in real time, achieving active adaptive support.

[0003] When the aforementioned hybrid energy storage system operates, multiple energy storage batteries are simultaneously deployed to cope with the changing grid environment. Therefore, the accurate display of the remaining battery capacity (SOC estimation) of each energy storage battery is particularly important, as it directly affects the system strategy for subsequent hybrid battery deployment. Existing SOC estimation methods are diverse, but the selection process often relies solely on the algorithm's accuracy, computational resources, and resistance to current fluctuations, without effectively incorporating real-time energy storage unit status for further evaluation. This results in selected algorithms being unsuitable for the current energy storage unit's state, failing to accurately display the percentage of remaining capacity, reducing the effective management of wasted energy from the energy storage batteries, and ultimately lowering the battery management efficiency of the hybrid energy storage system. Summary of the Invention

[0004] To address the technical problem that existing technologies fail to consider the matching between real-time status and the algorithm when selecting SOC estimation algorithms for energy storage batteries in hybrid energy storage systems, resulting in the inability to select the most suitable SOC algorithm and consequently the inability to reasonably display the percentage of remaining capacity, thus reducing the effective management of wasted energy and lowering the battery management efficiency of hybrid energy storage systems, this invention aims to provide a battery management system for hybrid energy storage systems based on SOC estimation. The specific technical solution adopted is as follows: This invention proposes a battery management system for a composite energy storage system based on SOC estimation, the system comprising: The energy storage battery real-time detection module is used to acquire the continuous frequency modulation period, current oscillation period, and normal current period within a preset analysis time period prior to the real-time moment. The real-time characteristic quantification module of the energy storage battery is used to obtain the oscillation frequency index at real time based on the occurrence interval variation characteristics of the current oscillation time period; to obtain the adjustment duration scale at real time based on the length variation characteristics of the continuous frequency adjustment time period; and to obtain the power supply rate at real time based on the current value between the normal current time period and the real time time. The oscillation frequency index, adjustment duration scale, and power supply rate are used as the real-time characteristics of the battery. The SOC estimation algorithm screening module is used to obtain the matching degree of the real-time characteristics of the battery based on the evaluation accuracy, computing resources and resistance to current fluctuations of each SOC estimation algorithm pre-stored in the memory, and to screen out the optimal SOC estimation algorithm for execution based on the matching degree.

[0005] Furthermore, the continuous frequency modulation period, current oscillation period, and normal current period within the preset analysis time period prior to obtaining the real-time time include: The continuous frequency modulation period is determined by searching the frequency modulation operation log within the time period to be analyzed. Before each continuous frequency modulation period, the current variance within the time range of the sliding window is analyzed under a preset time domain sliding window. If the current variance at the sliding window position meets the preset value condition, the sliding stops, and the time range of the sliding window is the current oscillation period. Other time periods besides the current oscillation period and the continuous frequency modulation period are the regular current time periods.

[0006] Furthermore, the method for obtaining the oscillation frequency index includes: If there are only one or zero current oscillation periods within the time period to be analyzed, then the oscillation frequency index is set to a preset minimum value; If there are two current oscillation periods within the time period to be analyzed, the negative correlation mapping and normalization result between the time intervals of the two current oscillation periods is used as the oscillation frequency index. If there are three or more current oscillation time periods within the time period to be analyzed, the first time interval between the first two current oscillation time periods in the time domain is obtained, and the second time interval between the last two current oscillation time periods in the time domain is obtained. The absolute value of the difference between the first time interval and the second time interval is used as the numerator, and the second time interval is used as the denominator. The resulting ratio is normalized to obtain the oscillation frequency index.

[0007] Furthermore, the method for obtaining the adjustment duration scale includes: If the time period to be analyzed contains only one or zero continuous frequency modulation time periods, then the adjustment duration scale is set to the preset maximum value; If the time period to be analyzed contains two or more continuous frequency modulation (FM) time periods, the average duration of all continuous FM time periods is calculated, and the durations of the continuous FM time periods are arranged in chronological order to obtain a duration sequence. The backward difference sequence of the duration sequence is calculated, and the average value of the backward difference sequence is used as the change coefficient. The average duration of the continuous FM time periods is weighted and adjusted using the change coefficient, and the weighted result is normalized to obtain the adjusted duration scale.

[0008] Furthermore, the method for obtaining the power supply rate includes: The power supply rate is obtained by normalizing the average current value between the closest normal current time period and the real time. If there is no normal current time period in the time period to be analyzed, the average current value in the time period to be analyzed is normalized to obtain the power supply rate.

[0009] Further, the optimal SOC estimation algorithm is selected and executed based on the matching degree, including: The first matching degree corresponding to the oscillation frequency index, the second matching degree corresponding to the adjustment duration scale, and the third matching degree corresponding to the power supply rate are obtained respectively. After weighting and summing the first matching degree, the second matching degree, and the third matching degree according to the preset weight, the optimization degree of each SOC estimation algorithm is obtained. The optimal SOC estimation algorithm is selected according to the optimization degree.

[0010] Furthermore, the methods for obtaining the first degree of matching include: The normalized evaluation accuracy and the normalized resistance to current fluctuations are weighted and summed according to preset weights to obtain the first algorithm feature; the difference between the first algorithm feature and the oscillation frequency index is negatively correlated to obtain the first matching degree.

[0011] Furthermore, the methods for obtaining the second degree of matching include: The normalized computing resources are used as the second algorithm feature, and the second algorithm feature is negatively correlated with the difference in the adjustment time scale to obtain the second matching degree.

[0012] Furthermore, the methods for obtaining the third matching degree include: The product of normalized current fluctuation resistance and normalized computing resources is used as the numerator, and the normalized evaluation accuracy is used as the denominator to obtain the third algorithm feature. The third algorithm feature is negatively correlated with the difference in power supply rate to obtain the third matching degree.

[0013] Furthermore, the SOC estimation algorithm with the highest degree of optimization is selected as the optimal SOC estimation algorithm.

[0014] The present invention has the following beneficial effects: To consider the real-time state of energy storage batteries when selecting SOC estimation algorithms, this invention first statistically analyzes the state distribution within the analysis period. Considering that in composite energy storage systems, frequency regulation is required to address current oscillations in energy storage units to ensure grid current stability, this invention focuses on the distribution of continuous frequency regulation periods, current oscillation periods, and normal current periods within the analysis period. Based on this distribution information, the oscillation frequency index, adjustment duration scale, and power supply rate of the energy storage unit at real-time can be effectively quantified. These three real-time battery characteristics characterize the real-time state of the energy storage unit. Then, the evaluation accuracy, computational resources, and resistance to current fluctuations of the SOC estimation algorithm are matched with the real-time battery characteristics. Based on the degree of matching, the optimal SOC estimation algorithm is selected for execution. By analyzing the time-period distribution characteristics within the analysis period, this invention quantifies the real-time state of energy storage batteries, ensuring that the ultimately recommended SOC estimation algorithm not only meets the rigid requirements of the grid system but also further satisfies the composite usage of batteries. This improves the display accuracy of the remaining battery percentage in composite energy storage systems, thereby enhancing battery management efficiency. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a block diagram of a battery management system for a composite energy storage system based on SOC estimation, provided as an embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a battery management system for a composite energy storage system based on SOC estimation proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] The following description, in conjunction with the accompanying drawings, details a specific scheme for a battery management system for a composite energy storage system based on SOC estimation provided by the present invention.

[0020] Please see Figure 1 The diagram illustrates a block diagram of a composite energy storage system battery management system based on SOC estimation according to an embodiment of the present invention. The system includes a real-time energy storage battery detection module 101, a real-time energy storage battery feature quantization module 102, and an SOC estimation algorithm screening module 103.

[0021] It should be noted that the system provided in this embodiment of the invention is for each energy storage battery in a composite energy storage system. The system in this embodiment of the invention relies on a basic grid management system and can call the current signal of the energy storage battery and the operation log in the memory through data call commands. The existing grid management operation is a technical means well known to those skilled in the art, and will not be described in detail here.

[0022] In a power grid scenario, the current state within the grid is primarily affected by various factors, resulting in a dynamic change in current that periodically exhibits unstable fluctuations (i.e., oscillations). To ensure grid current stability, existing methods actively control the grid's frequency through frequency regulation equipment (i.e., frequency regulation operation) to stabilize the grid's power generation and ensure its normal operation. Therefore, directly using the real-time current magnitude of the energy storage battery cannot accurately and effectively assess the battery's state characteristics at any given moment; further evaluation based on frequency regulation operation is necessary.

[0023] The system in this embodiment of the invention first utilizes the real-time detection module 101 of the energy storage battery to acquire the continuous frequency regulation time period, the current oscillation time period, and the normal current time period within a preset analysis time period prior to the real-time moment. The division of each time period depends on the current signal of the energy storage battery during the analysis time period. The current signal can be retrieved from data in the basic power grid management system. In this embodiment, the current signal sampling frequency is once per second, and the analysis time period can be set to 10 minutes. Specific settings can be made according to the implementer, and will not be elaborated upon in this embodiment. The final segmented result should present the following sequence: normal current time period – current oscillation time period – continuous frequency regulation time period, that is, the three time periods alternate in this order.

[0024] Preferably, in this embodiment of the invention, because the system can directly access the operation logs of the energy storage battery in the power grid management system, the unstable fluctuation phenomenon of the current in the power grid after being disturbed (oscillation phenomenon) is observed. Then, to maintain stability, the power grid takes proactive adjustment measures (frequency regulation operation). Compared with the oscillation phenomenon, the frequency regulation operation is explicitly recorded by the power grid system. Therefore, the continuous frequency regulation period can be determined by searching the frequency regulation operation logs within the time period to be analyzed. The oscillation phenomenon occurs before the continuous frequency regulation period and the current has significant fluctuation characteristics. Therefore, a time-domain sliding window can be pre-set before each continuous frequency regulation period, and the current variance corresponding to the time-domain sliding window is statistically analyzed. If the variance meets the preset value condition during the sliding process, it indicates that the corresponding time is the start time of the current oscillation period, and the sliding stops; that is, the sliding time area is the current oscillation period. Other time periods besides the current oscillation period and the continuous frequency regulation period are the conventional current time periods. In this embodiment of the invention, the value condition is set to the variance being less than a variance threshold, where the variance threshold is set to 5.

[0025] In the real-time monitoring module 101 for energy storage batteries, the continuous frequency regulation time period, the current oscillation time period, and the normal current time period can reflect the battery state within the time period to be analyzed based on their distribution. For example, if the current oscillation time periods are more numerous and have smaller intervals, it indicates that the battery is in a high-frequency current oscillation state. If the duration of the continuous frequency regulation time period is getting longer, it indicates that the grid current needs to be adjusted quickly, and the battery is in a low-sensitivity and safe state. The current during the normal current time period can reflect the current level of the battery. Therefore, the real-time feature quantification module 102 for energy storage batteries can further statistically analyze the distribution of various state time periods within the time period to be analyzed. Based on the variation characteristics of the occurrence interval of the current oscillation time period, the oscillation frequency index at the real time is obtained; based on the variation characteristics of the length of the continuous frequency regulation time period, the adjustment duration scale at the real time is obtained; based on the current value between the normal current time period and the real time, the power supply rate at the real time is obtained. By using the oscillation frequency index, the adjustment duration scale, and the power supply rate as real-time battery features, the real-time state of the energy storage battery can be quantified.

[0026] Preferably, in this embodiment of the invention, because the distribution of current oscillation time periods within the time period to be analyzed can reflect the state of the energy storage battery at real time, and the number of current oscillation time periods within the time period to be analyzed is not fixed, it is necessary to analyze them separately for different situations. Specifically, the method for obtaining the oscillation frequency index includes: If there are only one or zero current oscillation periods within the analysis period, the oscillation frequency index is set to a preset minimum value. This preset minimum value should be the minimum value in the final obtained oscillation frequency index value range, indicating that the oscillation of the energy storage battery is not frequent at this time. The oscillation frequency index in this embodiment of the invention needs to be normalized, that is, the value range is between 0 and 1. Therefore, in this embodiment of the invention, the preset value here can be set to 0.

[0027] If there are two current oscillation periods within the time period to be analyzed, the negative correlation mapping and normalization result between the time intervals of the two current oscillation periods is used as the oscillation frequency index; that is, the smaller the time interval between the two current oscillation periods, the more frequent the oscillation may be, and the larger the final oscillation frequency index. In this embodiment of the invention, the time interval between the two current oscillation periods is first normalized using range standardization, and then the negative correlation mapping and normalization are achieved by subtracting the normalization result from the positive integer 1. The method of normalization using range standardization is a well-known technique to those skilled in the art and will not be described in detail here.

[0028] If there are three or more current oscillation periods within the time period to be analyzed, the first time interval between the first two current oscillation periods in the time domain is obtained, and the second time interval between the last two current oscillation periods in the time domain is obtained. The absolute value of the difference between the first and second time intervals is used as the numerator, and the second time interval is used as the denominator. The resulting ratio is then normalized to obtain the oscillation frequency index. That is, in this scenario, the smaller the second time interval, the more frequently the two current oscillation periods closer to the real time occur. At the same time, the greater the difference compared to the first time interval in the initial stage, the more significant the trend of the second time interval decreasing is, and the larger the final ratio will be, i.e., the larger the oscillation frequency index will be. It should be noted that the normalization process here can also be implemented using range standardization, which will not be elaborated here.

[0029] Preferably, similar to the method for obtaining the oscillation frequency index, the analysis of adjusting the duration scale requires analyzing the distribution of continuous frequency modulation time periods, and also requires targeted analysis for different quantities, specifically including: If the analysis period contains only one or zero continuous frequency modulation periods, it indicates that there is only one frequency modulation operation or no frequency modulation operation within the analysis period. This suggests that there is ample time available for grid frequency regulation, and the degree of need for rapid adjustment of the grid current is relatively small. In this case, the adjustment duration scale is set to the preset maximum value. Similar to the preset minimum value of the oscillation frequency index, since the adjustment duration scale also needs to be eventually normalized to between 0 and 1, the preset maximum value here is set to 1.

[0030] If the time period to be analyzed contains two or more continuous frequency regulation (FM) periods, the average duration of all continuous FM periods is calculated. The durations of the continuous FM periods are then arranged in chronological order to obtain a duration sequence. The backward difference sequence of this duration sequence is then calculated. The backward difference sequence is the difference sequence formed by subtracting the previous element from the next element in the duration sequence. The larger the element in the backward difference sequence, the greater the duration of the continuous FM period in the chronological direction. Therefore, the average value of the backward difference sequence is used as a coefficient of variation. A larger coefficient of variation indicates that the duration of the continuous FM period generally increases in the chronological direction, suggesting that the real-time adjustment period should have a longer operating adjustment period, the degree of need for rapid grid current response adjustment is weaker, and a longer frequency regulation time is reserved at the current real-time point. Therefore, the duration of the continuous FM period is weighted and adjusted using the aforementioned coefficient of variation, and the weighted result is normalized to obtain the adjustment duration scale.

[0031] In this embodiment of the invention, after normalizing the variation coefficient, the result of adding it to a positive integer 1 is used as the adjustment weight. This adjustment weight is then multiplied by the average duration to obtain a weighted result. A larger variation coefficient indicates an overall increasing trend in the continuous frequency modulation (FM) time period; in this case, the adjustment weight is a larger value between 1 and 2, ultimately increasing the average duration. Conversely, if the variation coefficient is 0 or negative, it indicates no change or a decrease in the FM time period; in this case, the adjustment weight is a smaller value between 1 and 2, resulting in a smaller adjustment magnitude. By adjusting the adjustment weight, the average duration with a clear increasing trend can be significantly enhanced, effectively characterizing the features of the adjustment duration scale.

[0032] It should be noted that the normalization methods in the embodiments of the present invention can all be implemented through range standardization. The maximum and minimum values ​​under each data dimension can be determined by analyzing historical data and other methods, thereby realizing the normalization operation of each data within the dimension. These are technical means well known to those skilled in the art and will not be elaborated here.

[0033] Preferably, in this embodiment of the invention, the power supply rate reflects the power demand at the current moment; that is, the larger the current value, the greater the power demand, and thus the larger the power supply rate. Therefore, the average current value between the most recent normal current time period and the real time is normalized to obtain the power supply rate. If there is no normal current time period within the time period to be analyzed, the average current value within the time period to be analyzed is normalized to obtain the power supply rate. Similarly, the normalization method here can be implemented by range standardization, which will not be elaborated further.

[0034] The real-time feature quantification module 102 of the energy storage battery can obtain the oscillation frequency index, adjustment duration scale, and power supply rate as real-time battery features at real time. Then, in the SOC estimation algorithm screening module 103, the matching degree of the battery's real-time features is obtained based on the evaluation accuracy, computing resources, and resistance to current fluctuations pre-stored in the memory for each SOC estimation algorithm. The optimal SOC estimation algorithm is then selected and executed based on this matching degree. The SOC estimation algorithm screening module 103 can directly call various indicators of each SOC estimation algorithm from the prior database. In this embodiment of the invention, the various indicators of the algorithm include the normalized results of evaluation accuracy, computing resources, and resistance to current fluctuations. Please refer to Table 1, which shows the prior data corresponding to each SOC estimation algorithm in this embodiment of the invention.

[0035] Table 1 As shown in Table 1, the real-time feature quantification module for energy storage batteries can directly retrieve features from the prior database for processing. Among these features, current fluctuation resistance refers to the degree to which the corresponding SOC estimation algorithm adapts to severe current fluctuations in the system; computational resources refer to the amount of computational resources occupied by the corresponding SOC estimation algorithm in the system; the greater the resource consumption, the longer the computation time; and evaluation accuracy reflects the effectiveness of the algorithm when applied to the power grid system. It should be noted that all the above features are prior data stored in memory and can be directly retrieved without further computation.

[0036] It should be noted that, for different states of energy storage batteries, the SOC algorithm with the highest evaluation accuracy is not necessarily the best, nor is the algorithm with the least computational resources necessarily the best. The most suitable algorithm should be selected for different states. Therefore, it is necessary to screen the optimal SOC estimation algorithm by utilizing the degree of matching between the algorithm and the real-time characteristics of the battery.

[0037] Preferably, in this embodiment of the invention, the optimal SOC estimation algorithm is selected and executed based on the matching degree, including: The first matching degree corresponding to the oscillation frequency index, the second matching degree corresponding to the adjustment duration scale, and the third matching degree corresponding to the power supply rate are obtained respectively. After weighting and summing the first matching degree, the second matching degree, and the third matching degree according to the preset weight, the optimization degree of each SOC estimation algorithm is obtained. The optimal SOC estimation algorithm is selected according to the optimization degree.

[0038] In this embodiment of the invention, the weights of the first matching degree are set to 0.4, the weights of the second matching degree are set to 0.4, and the weights of the third matching degree are set to 0.2. The degree of preference is expressed by the formula: Where m represents the degree of optimization, f represents the first degree of matching, g represents the second degree of matching, and h represents the third degree of matching. A higher degree of matching indicates a better match; therefore, a higher degree of optimization indicates that the algorithm better matches the current real-time battery characteristics. Thus, the SOC estimation algorithm with the highest degree of optimization is selected as the optimal SOC estimation algorithm. Finally, the remaining capacity of the energy storage battery can be estimated by executing the optimal SOC estimation algorithm, and the estimated remaining capacity result can be displayed in the corresponding battery's power display module, completing the intelligent management of the energy storage battery.

[0039] Preferably, in this embodiment of the invention, the method for obtaining the first matching degree includes: The normalized evaluation accuracy and the normalized current fluctuation resistance are weighted and summed according to preset weights to obtain the first algorithm feature. The difference between the first algorithm feature and the oscillation frequency index is negatively correlated to obtain the first matching degree. In this embodiment of the invention, the weight of the normalized evaluation accuracy is set to 0.3, and the weight of the normalized current fluctuation resistance is set to 0.7. That is, the current fluctuation resistance should have a large reference value when analyzing the matching of the oscillation frequency index. For example, under the condition of high oscillation frequency index, the battery current will oscillate frequently. At this time, the SOC estimation algorithm should have high current fluctuation resistance. Only with high current fluctuation resistance can the basic evaluation accuracy be guaranteed. Therefore, the first matching degree considers both current fluctuation resistance and evaluation accuracy, and sets the current fluctuation resistance to have a larger weight and the evaluation accuracy to have a smaller weight. The first algorithm feature obtained is finally compared with the oscillation frequency index. If the similarity between the two is large, the first matching degree is large.

[0040] In this embodiment of the invention, since all feature indicators have been normalized, the normalization algorithm can also be implemented using range standardization. Therefore, the absolute value of the difference between the first algorithm feature and the oscillation frequency index can be directly calculated. The absolute value of the difference is also a data between 0 and 1. Subtracting the absolute value of the difference from the positive integer 1 will achieve the final negative correlation mapping, and the first matching degree obtained is also between 0 and 1.

[0041] Preferably, in this embodiment of the invention, the method for obtaining the second matching degree includes: Since the size of computing resources represents the time required for algorithm execution, the normalized computing resources can be directly used as the second algorithm feature. The second algorithm feature is then negatively correlated with the difference in the adjusted time scale to obtain the second matching degree. It should be noted that the negative correlation mapping and normalization algorithms involved in the second matching degree have been described in the above embodiments and will not be repeated here.

[0042] Preferably, in this embodiment of the invention, the method for obtaining the third matching degree includes: In scenarios with high power supply rates, the high power supply rate occurs because the difference between the remaining power and the rated current is significant. Therefore, current flows into the energy storage battery more quickly, resulting in a larger current volume. This indicates a longer energy storage time required. In this scenario, the selected algorithm needs to ensure high resistance to current fluctuations. Furthermore, due to the longer storage time, the algorithm has a higher fault tolerance, thus requiring less precision and computational resources. Based on this logic, the product of normalized resistance to current fluctuations and normalized computational resources is used as the numerator, and the normalized evaluation precision is used as the denominator to obtain the third algorithm feature. This third algorithm feature is then negatively correlated with the difference in power supply rate to obtain the third matching degree. It should be noted that although the evaluation precision is normalized, none of the SOC algorithms selected in this embodiment have a precision of 0; that is, there is no case where the denominator is 0 in the above calculation process.

[0043] Preferably, in this embodiment of the invention, it is further considered that the current change state in the power grid system is mainly divided into two stages: short-term high-frequency oscillation and medium-to-long-term low-frequency oscillation; the corresponding frequency regulation operations are mainly: short-term high-frequency regulation commands and medium-to-long-term low-frequency regulation commands; energy storage batteries are mainly divided into two categories: energy-type energy storage batteries and power-type energy storage batteries. Among them, energy-type energy storage batteries mainly undertake peak-shaving tasks. If they frequently receive short-term high-frequency regulation commands, their battery life will be greatly damaged. Power-type energy storage batteries mainly undertake rapid regulation tasks such as primary and secondary frequency regulation. However, their capacity is relatively small and they cannot accept medium-to-long-term regulation commands. In an ideal scenario, energy-type energy storage batteries should handle medium-to-long-term low-frequency oscillations, and power-type energy storage batteries should handle short-term high-frequency oscillations. However, in actual use, both types of energy-type energy storage batteries are usually used in combination. In order to achieve efficient management of the energy storage system batteries, it is necessary to identify the type of battery currently in use and recommend a targeted SOC estimation algorithm.

[0044] If the current energy storage battery is an energy-type energy storage battery, it means that the oscillation it can accept should be a medium-to-long-term low-frequency oscillation. Therefore, it should exhibit the characteristic of a smaller first algorithm feature and a larger second algorithm feature. Thus, the difference between the second and first algorithm features is further normalized to obtain a fourth matching degree. The product of the optimal matching degree and the fourth matching degree is used as the final optimal matching degree to select the optimal SOC estimation algorithm.

[0045] If the current energy storage battery is a power type energy storage battery, it means that the type of oscillation it can receive is short-time high-frequency oscillation. In contrast to the energy type energy storage battery, the difference between the first algorithm feature and the second algorithm feature is normalized to obtain the fourth matching degree, and the final optimization degree is calculated in the same way.

[0046] In summary, this invention focuses on the distribution of continuous frequency regulation time periods, current oscillation time periods, and conventional current time periods within the analysis time period. It quantifies the oscillation frequency index, adjustment duration scale, and power supply rate of the energy storage unit at real-time. Based on the matching of the SOC estimation algorithm's evaluation accuracy, computational resources, and resistance to current fluctuations with the battery's real-time characteristics, the optimal SOC estimation algorithm can be selected for execution according to the degree of matching. This invention quantifies the real-time state of the energy storage battery by analyzing the time period distribution characteristics within the analysis time period. This allows the finally recommended SOC estimation algorithm to not only meet the rigid requirements of the power grid system but also further satisfy the composite usage of the battery, improving the display accuracy of the remaining battery percentage value in the composite energy storage system, thereby improving the battery management efficiency of the composite energy storage system.

[0047] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0048] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A battery management system for a composite energy storage system based on SOC estimation, characterized by, The system includes: The energy storage battery real-time detection module is used to acquire the continuous frequency modulation period, current oscillation period, and normal current period within a preset analysis time period prior to the real-time moment. The real-time characteristic quantification module of the energy storage battery is used to obtain the oscillation frequency index at real time based on the occurrence interval variation characteristics of the current oscillation time period; to obtain the adjustment duration scale at real time based on the length variation characteristics of the continuous frequency adjustment time period; and to obtain the power supply rate at real time based on the current value between the normal current time period and the real time time. The oscillation frequency index, adjustment duration scale, and power supply rate are used as the real-time characteristics of the battery. The SOC estimation algorithm screening module is used to obtain the matching degree of the real-time characteristics of the battery based on the evaluation accuracy, computing resources and resistance to current fluctuations of each SOC estimation algorithm pre-stored in the memory, and to screen out the optimal SOC estimation algorithm for execution based on the matching degree. The continuous frequency modulation period, current oscillation period, and normal current period within the preset analysis time period prior to obtaining the real-time time include: The continuous frequency modulation period is determined by searching the frequency modulation operation log within the time period to be analyzed; before each continuous frequency modulation period, the current variance within the time range of the sliding window is analyzed under a preset time domain sliding window; if the current variance at the sliding window position meets the preset value condition, the sliding stops, and the time range of the sliding window is the current oscillation period; other time periods besides the current oscillation period and the continuous frequency modulation period are the regular current period. The method for obtaining the oscillation frequency index includes: If there are only one or zero current oscillation periods within the time period to be analyzed, then the oscillation frequency index is set to a preset minimum value; If there are two current oscillation periods within the time period to be analyzed, the negative correlation mapping and normalization result between the time intervals of the two current oscillation periods is used as the oscillation frequency index. If there are three or more current oscillation time periods within the time period to be analyzed, the first time interval between the first two current oscillation time periods in the time domain is obtained, and the second time interval between the last two current oscillation time periods in the time domain is obtained; the absolute value of the difference between the first time interval and the second time interval is used as the numerator, and the second time interval is used as the denominator. The ratio obtained is normalized to obtain the oscillation frequency index. The method for obtaining the adjustment duration scale includes: If the time period to be analyzed contains only one or zero continuous frequency modulation time periods, then the adjustment duration scale is set to the preset maximum value; If the time period to be analyzed contains two or more continuous frequency modulation (FM) time periods, the average duration of all continuous FM time periods is calculated, and the durations of the continuous FM time periods are arranged in chronological order to obtain a duration sequence. The backward difference sequence of the duration sequence is calculated, and the average value of the backward difference sequence is used as the change coefficient. The average duration of the continuous FM time periods is weighted and adjusted using the change coefficient, and the weighted result is normalized to obtain the adjusted duration scale.

2. The battery management system for a composite energy storage system based on SOC estimation according to claim 1, wherein, The method for obtaining the power supply rate includes: The power supply rate is obtained by normalizing the average current value between the closest normal current time period and the real time. If there is no normal current time period in the time period to be analyzed, the average current value in the time period to be analyzed is normalized to obtain the power supply rate.

3. The battery management system for a composite energy storage system based on SOC estimation according to claim 1, wherein, The optimal SOC estimation algorithm is selected based on the matching degree and executed, including: The first matching degree corresponding to the oscillation frequency index, the second matching degree corresponding to the adjustment duration scale, and the third matching degree corresponding to the power supply rate are obtained respectively. After weighting and summing the first matching degree, the second matching degree, and the third matching degree according to the preset weight, the optimization degree of each SOC estimation algorithm is obtained. The optimal SOC estimation algorithm is selected according to the optimization degree.

4. The battery management system for a composite energy storage system based on SOC estimation according to claim 3, wherein, Methods for obtaining the first degree of matching include: The normalized evaluation accuracy and the normalized resistance to current fluctuations are weighted and summed according to preset weights to obtain the first algorithm feature; the difference between the first algorithm feature and the oscillation frequency index is negatively correlated to obtain the first matching degree.

5. The SOC estimation based composite energy storage system battery management system of claim 3, wherein, Methods for obtaining the second degree of matching include: The normalized computing resources are used as the second algorithm feature, and the second algorithm feature is negatively correlated with the difference in the adjustment time scale to obtain the second matching degree.

6. The SOC estimation based composite energy storage system battery management system of claim 3, wherein, The methods for obtaining the third matching degree include: The product of normalized current fluctuation resistance and normalized computing resources is used as the numerator, and the normalized evaluation accuracy is used as the denominator to obtain the third algorithm feature. The third algorithm feature is negatively correlated with the difference in power supply rate to obtain the third matching degree.

7. The SOC estimation based composite energy storage system battery management system of claim 3, wherein, The SOC estimation algorithm with the highest degree of preference is selected as the optimal SOC estimation algorithm.

Images