Anti-countercurrent control method and system based on load removal feature recognition

Abstract

The invention provides an anti-countercurrent control method and system based on load removal feature recognition, and the method comprises the steps: setting an anti-countercurrent power judgment threshold value and a feature signal capturing judgment threshold value, recording an event triggering moment, a power reduction starting moment and initial power when the power of a user home-entry gateway is lower than the feature signal capturing judgment threshold value, and recording the initial power of the user home-entry gateway; collecting voltage and current waveforms within a preset time span before and after a power reduction starting moment to form a load removal characteristic waveform, calculating a time difference value and a power difference value, and storing related data in a characteristic library; scanning the voltage and current waveforms of the home-entry gateway in real time, calculating the correlation between the real-time waveform and the waveforms in the feature library through a Pearson's correlation coefficient method, updating the credibility value and the application timestamp of the matched feature waveform when the correlation coefficient exceeds a threshold value, and updating the application timestamp when the credibility value reaches a set threshold value. And generating an energy storage system power suppression instruction based on the power difference value and the time difference value. According to the invention, economic loss and power supply enterprise fines caused by reverse power transmission can be reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of energy storage system control, and more specifically, to a backflow prevention control method and system based on load shedding feature recognition. Background Art

[0002] In commercial and industrial energy storage systems, they play a crucial role in peak load shaving and valley filling. Large electricity users, in particular, can reap significant economic benefits through the appropriate deployment of energy storage systems. However, as energy storage capacity continues to expand, local power grids have explicitly mandated that user-installed energy storage equipment not feed back power to the grid to ensure safe and stable operation within distribution areas. Currently, a common solution involves installing a reverse power protection device at the customer's entrance. This device senses the power flowing through the entrance and automatically trips the customer's entrance switch if the power level falls below a set value for a sustained period. While this approach can prevent reverse power feeds caused by reduced loads and thus prevent power companies from issuing fines for reverse power feeds, the resulting sudden power outages can cause significant economic losses to the company's operations and the energy storage system's discharge revenue. Another solution involves using an energy storage EMS (Energy Management System) to detect reverse power feeds at the entrance. Once detected, it can rapidly adjust the energy storage system to reduce its generated power, mitigating the potential harm. However, this strategy involves post-detection. It takes time for the EMS to issue an anti-backflow command and for the energy storage system to adjust its output, making it difficult to completely prevent reverse flow. Power supply companies may still issue fines for detecting power reverse flow.

[0003] In the process of implementing the embodiments of the present invention, the inventors discovered that there are at least the following problems or defects in the existing technology: the existing technology either directly cuts off the user's home switch through an anti-reverse power device, causing sudden power outages and causing economic losses to the user; or it uses the energy storage EMS to detect power reverse transmission afterwards, which is difficult to completely avoid reverse transmission and cannot effectively solve the problem of the energy storage system reversely transmitting electricity to the power grid. Summary of the Invention

[0004] The present invention provides a backflow prevention control method and system based on load shedding feature recognition.

[0005] In a first aspect of the present invention, a backflow prevention control method based on load shedding feature recognition is provided, comprising:

[0006] S1: During the operation of the energy storage system, a reverse power prevention threshold and a characteristic signal capture threshold are set, wherein the characteristic signal capture threshold is greater than or equal to the reverse power prevention threshold;

[0007] S2: When it is detected that the power at the user's home gateway is less than the characteristic signal capture judgment threshold, the current time is recorded as the event trigger time, and the power reduction start time and the corresponding initial power are obtained;

[0008] S3: Collect the voltage and current waveforms at the household gateway within a preset time span before and after the power starts to decrease, and form a load shedding characteristic waveform;

[0009] S4: Calculating the time difference between the event triggering moment and the power starting to decrease moment, and the power difference between the initial power and the characteristic signal capture determination threshold;

[0010] S5: storing the load shedding characteristic waveform, initial credibility value, event triggering time, time difference and power difference as characteristic library records;

[0011] S6: Scanning the voltage waveform and current waveform at the entrance to the house in real time at intervals of the preset time span;

[0012] S7: Calculate the correlation between the real-time waveform and each characteristic waveform in the characteristic library by using the Pearson correlation coefficient method;

[0013] S8: When the correlation coefficient exceeds the correlation threshold, the credibility value of the matching feature waveform and the latest application timestamp are updated;

[0014] S9: When the credibility value reaches a set credibility threshold, a power suppression instruction for the energy storage system is generated based on the power difference and the time difference.

[0015] Furthermore, each record of the feature library includes a feature waveform field, a credibility field, an application timestamp field, a time difference field, and a power difference field, wherein the credibility field is initially assigned a value of 1.

[0016] Furthermore, the updating of the credibility value of the matching characteristic waveform includes: each time a characteristic waveform match is completed, the credibility value of the corresponding record is increased by 1, and the application timestamp is synchronously updated to the current system time.

[0017] Furthermore, generating a power suppression instruction for the energy storage system based on the power difference and the time difference includes:

[0018] Calculating a power adjustment amount in a linear proportion according to the power difference;

[0019] A response time window for power adjustment is set based on the time difference.

[0020] Furthermore, it also includes performing aging cleaning on the feature library: when the time difference between the application timestamp of the feature library record and the current system time exceeds the aging cycle threshold and the credibility value is lower than the cleaning threshold, the record is automatically deleted.

[0021] Furthermore, the correlation threshold is set to 0.8, the credibility threshold is set to 3, the aging cycle threshold is 72 hours, and the cleaning threshold is 2.

[0022] Furthermore, the preset time span is a configurable parameter, and the configuration range is 50 milliseconds to 200 milliseconds.

[0023] Furthermore, the calculation of the power difference includes dynamic compensation for the power change rate during load shedding, and the compensation factor calculation formula is:

[0024]

[0025] Where K is the compensation factor, α is the compensation coefficient, and dP / dt is the power change rate.

[0026] Furthermore, the generation delay of the power suppression instruction is less than 50 milliseconds, and the power adjustment amount is directly proportional to the power difference.

[0027] In a second aspect of the present invention, a backflow prevention control system based on load shedding feature recognition is provided, comprising:

[0028] Grid access end, distribution transformer connected through household gateway, user load end, and energy storage system connected through energy storage converter;

[0029] The energy storage converter is configured as follows:

[0030] Real-time collection of voltage and current signals at the household access point;

[0031] Perform the following steps:

[0032] During the operation of the energy storage system, a reverse power prevention threshold and a characteristic signal capture threshold are set, wherein the characteristic signal capture threshold is greater than or equal to the reverse power prevention threshold;

[0033] When it is detected that the power at the user's home gateway is less than the characteristic signal capture judgment threshold, the current time is recorded as the event trigger time, and the power reduction start time and the corresponding initial power are obtained;

[0034] Collect the voltage and current waveforms at the household gateway within a preset time span before and after the power starts to decrease, and form a load shedding characteristic waveform;

[0035] Calculating the time difference between the event triggering moment and the moment when the power starts to decrease, and the power difference between the initial power and the characteristic signal capture determination threshold;

[0036] The load shedding characteristic waveform, initial credibility value, event triggering time, time difference and power difference are stored as characteristic library records;

[0037] Scanning the voltage waveform and current waveform at the entrance to the house in real time at intervals of the preset time span;

[0038] The correlation between the real-time waveform and each characteristic waveform in the characteristic library is calculated using the Pearson correlation coefficient method;

[0039] When there is a correlation coefficient exceeding the correlation threshold, the credibility value of the matching characteristic waveform and the latest application timestamp are updated;

[0040] When the credibility value reaches a set credibility threshold, a power suppression instruction for the energy storage system is generated based on the power difference and time difference. At a voltage level of 0.4kV, the energy storage system's charge and discharge power is adjusted to prevent the reverse flow of user-side power to the grid.

[0041] The above-described embodiments of the present invention have at least the following beneficial effects: The present invention can effectively identify load shedding signature signals in advance and match them in real time, preemptively adjusting the energy storage system's power, preventing the reverse flow of user-side power to the grid, thereby minimizing the economic losses and fines imposed on power supply companies due to power reverse flow. Furthermore, by continuously updating the signature library, this method can adapt to varying load characteristics, improving control accuracy and reliability, ensuring stable operation of the energy storage system under various operating conditions, and further enhancing the application value and economic benefits of the energy storage system. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily apparent by reading the following detailed description with reference to the accompanying drawings, in which several embodiments of the present invention are shown by way of example and not limitation, in which:

[0043] Figure 1 A schematic flow chart of a backflow prevention control method based on load shedding feature recognition provided by one embodiment of the present invention;

[0044] Figure 2 A schematic structural diagram of an anti-backflow control system based on load shedding feature recognition provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0045] The principles and spirit of the present invention will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided solely to enable those skilled in the art to better understand and implement the present invention, and are not intended to limit the scope of the present invention in any way. Rather, these embodiments are provided to make the present invention more thorough and complete, and to fully convey the scope of the present invention to those skilled in the art.

[0046] Those skilled in the art will appreciate that the embodiments of the present invention may be implemented as a system, apparatus, device, method, or computer program product. Therefore, the present invention may be implemented in the following forms: entirely in hardware, entirely in software (including firmware, resident software, microcode, etc.), or in a combination of hardware and software.

[0047] It should be noted that any number of elements in the drawings is for illustration only and not for limitation, and any naming is only for distinction and does not have any limiting meaning.

[0048] Reference below Figure 1 , Figure 1 Schematic diagram of a flow chart of a backflow prevention control method based on load shedding feature recognition provided by an embodiment of the present invention. Figure 1 As shown, a backflow prevention control method based on load shedding feature recognition includes:

[0049] S1: During the operation of the energy storage system, a reverse power prevention threshold and a characteristic signal capture threshold are set, wherein the characteristic signal capture threshold is greater than or equal to the reverse power prevention threshold;

[0050] S2: When it is detected that the power at the user's home gateway is less than the characteristic signal capture judgment threshold, the current time is recorded as the event trigger time, and the power reduction start time and the corresponding initial power are obtained;

[0051] S3: Collect the voltage and current waveforms at the household gateway within a preset time span before and after the power starts to decrease, and form a load shedding characteristic waveform;

[0052] S4: Calculating the time difference between the event triggering moment and the power starting to decrease moment, and the power difference between the initial power and the characteristic signal capture determination threshold;

[0053] S5: storing the load shedding characteristic waveform, initial credibility value, event triggering time, time difference and power difference as characteristic library records;

[0054] S6: Scanning the voltage waveform and current waveform at the entrance to the house in real time at intervals of the preset time span;

[0055] S7: Calculate the correlation between the real-time waveform and each characteristic waveform in the characteristic library by using the Pearson correlation coefficient method;

[0056] S8: When the correlation coefficient exceeds the correlation threshold, the credibility value of the matching feature waveform and the latest application timestamp are updated;

[0057] S9: When the credibility value reaches a set credibility threshold, a power suppression instruction for the energy storage system is generated based on the power difference and the time difference.

[0058] It should be noted that during the operation of the energy storage system, setting the anti-reverse power judgment threshold and the characteristic signal capture judgment threshold is a basic step to achieve anti-reverse flow control. The anti-reverse power judgment threshold refers to the power value used to determine whether power reverse flow occurs. When the power at the user's home gateway is lower than this value, power reverse flow protection may be triggered. The characteristic signal capture judgment threshold is a higher threshold set to more accurately capture load shedding events and is used to trigger the collection of characteristic signals. When it is detected that the power at the user's home gateway is less than the characteristic signal capture judgment threshold, the current moment is recorded as the event triggering moment, and the power start reduction moment and the corresponding initial power are obtained. The event triggering moment refers to the moment when the power is detected to be lower than the characteristic signal capture judgment threshold, and the power start reduction moment refers to the power value when the load starts to be cut, and the initial power is the power value at this time. The recording of these parameters and moments provides basic data for subsequent characteristic waveform collection and analysis.

[0059] Specifically, the reverse power prevention threshold can be set based on the grid requirements and the operating characteristics of the energy storage system in the actual application scenario. For example, at the common 0.4kV voltage level, the threshold can be set to 10kW to effectively prevent reverse power flow when the power level drops below this value. The characteristic signal capture threshold can be adjusted based on the power variation characteristics during load shedding, for example, setting it to 15kW to ensure timely capture of characteristic signals during the initial load shedding phase. The power reduction start time and initial power are determined by real-time monitoring of power changes at the household entrance. For example, high-precision power sensors and data acquisition systems can be used to record power changes at a millisecond sampling rate to accurately determine the power reduction start time and the corresponding initial power value. Furthermore, the voltage and current waveforms at the household entrance within a preset time span before and after the power reduction start time are collected to form the load shedding characteristic waveform. This process can be performed using specialized data acquisition equipment to ensure the integrity and accuracy of the waveform data. The preset time span can be configured according to actual needs, for example, setting it to 100 milliseconds to more comprehensively capture the characteristic changes during the load shedding process.

[0060] Preferably, to improve the recognition accuracy of the characteristic signal, the collected voltage and current waveforms can be filtered to remove noise interference. For example, a low-pass filter can be used to process the waveform data to filter out high-frequency noise and ensure the accuracy of the characteristic waveform. In addition, the reliability of feature recognition can be improved by increasing the sampling time span of the characteristic signal. For example, the sampling time span can be increased from 100 milliseconds to 200 milliseconds to more comprehensively capture the characteristic changes during the load shedding process. When calculating the time difference between the event trigger moment and the moment when the power begins to decrease, high-precision clock synchronization technology can be used to ensure that the calculation accuracy of the time difference reaches the microsecond level, thereby more accurately reflecting the time characteristics of the load shedding process. At the same time, when generating the power suppression instruction of the energy storage system based on the power difference and the time difference, the power adjustment amount can be calculated according to the power difference in linear proportion, and the response time window of the power adjustment can be set based on the time difference to achieve fast and accurate power regulation, further improving the operational stability and safety of the energy storage system under load shedding conditions.

[0061] In some embodiments, each record of the feature library includes a feature waveform field, a credibility field, an application timestamp field, a time difference field, and a power difference field, wherein the credibility field is initially assigned a value of 1.

[0062] It should be noted that the feature library described in the present invention is a key data structure for storing load shedding characteristic waveforms and related parameters. Each record contains a characteristic waveform field, a credibility field, an application timestamp field, a time difference field, and a power difference field. The credibility field is initially assigned a value of 1. The feature library is established to quickly identify load shedding events that may cause power backflow during subsequent real-time monitoring by matching them with real-time waveforms, allowing for proactive action. The characteristic waveform field stores the voltage and current waveform data collected during the load shedding process, which reflects the electrical characteristics of the load shedding process. The credibility field records the number of successful characteristic waveform matches, with an initial value of 1 indicating the first time the characteristic waveform was recorded. The application timestamp field records the time the characteristic waveform was recorded, which is used for subsequent aging and cleaning operations. The time difference and power difference fields, respectively, store the time difference between the event triggering moment and the moment the power reduction begins, and the power difference between the initial power and the characteristic signal capture threshold. These parameters are crucial for the subsequent generation of power suppression instructions.

[0063] Specifically, the characteristic waveform field in the feature library stores the voltage and current waveform data collected during the load shedding process. These waveform data can reflect the electrical characteristics of the load shedding. For example, when the load is shedding, the voltage may fluctuate briefly, and the current will drop rapidly. These changes are recorded by high-precision data acquisition equipment to form a characteristic waveform. The credibility field is used to record the number of successful matches of the characteristic waveform. The initial value is 1, indicating that the characteristic waveform is recorded for the first time. As the number of subsequent matches increases, the credibility value will also increase accordingly, which helps to improve the reliability of the characteristic waveform. The application timestamp field records the time point when the characteristic waveform is recorded for subsequent aging and cleaning operations. For example, an aging cycle threshold can be set. When the time difference between the application timestamp of the characteristic waveform and the current system time exceeds the threshold, and the credibility value is lower than the cleaning threshold, the record will be automatically deleted to maintain the timeliness and accuracy of the feature library. The time difference and power difference fields respectively store the time difference between the event triggering moment and the moment when the power starts to decrease, as well as the power difference between the initial power and the characteristic signal capture judgment threshold. These parameters are crucial for the subsequent generation of power suppression instructions. For example, the power adjustment amount can be calculated in a linear proportion based on the power difference, and the response time window of the power adjustment can be set based on the time difference to achieve fast and accurate power regulation.

[0064] Preferably, in order to further improve the reliability and practicality of the feature library, the acquisition and storage process of the feature waveform can be optimized. For example, when acquiring the feature waveform, a multi-channel synchronous sampling technology can be used to ensure the synchronization and accuracy of the voltage and current waveforms. At the same time, the collected waveform data can be preprocessed, such as filtering, normalization and other operations, to remove noise and improve the comparability of the waveform. When storing the feature waveform, a compression algorithm can be used to compress the waveform data to save storage space, and the waveform data can be quickly decompressed and restored when reading. In addition, the records in the feature library can be classified and stored according to different application scenarios and load characteristics. For example, different types of load shedding events (such as single device shedding, multiple devices shedding at the same time, etc.) can be stored in different sub-libraries, so that matching feature waveforms can be found more quickly during real-time monitoring, thereby improving the efficiency and accuracy of control.

[0065] In some embodiments, updating the credibility value of the matching characteristic waveform includes: each time a characteristic waveform match is completed, increasing the credibility value of the corresponding record by 1, and synchronously updating the application timestamp to the current system time.

[0066] It should be noted that updating the credibility value of the matching characteristic waveform is an important step in the present invention for improving the accuracy of characteristic waveform recognition. When the waveform monitored in real time successfully matches a characteristic waveform in the characteristic library, that is, when the correlation coefficient calculated by the Pearson correlation coefficient method exceeds the set correlation threshold, the credibility value of the corresponding record will increase by 1, and the application timestamp will be synchronously updated to the current system time. The increase in the credibility value reflects the effectiveness of the characteristic waveform in actual operation, and the update of the application timestamp ensures the timeliness of the records in the characteristic library. This process helps the system to identify similar load shedding events more quickly in subsequent monitoring, so that measures can be taken in advance to avoid power backflow.

[0067] Specifically, the credibility value is an important parameter for each record in the feature library, which records the number of times the feature waveform has been successfully matched. The initial credibility value is 1, indicating that the feature waveform is recorded for the first time. After each successful match, the credibility value increases by 1, which indicates that the feature waveform has been verified as valid more times in actual operation, thereby improving its credibility. The application timestamp field records the time when the feature waveform was last successfully matched, which is crucial for the aging and cleanup mechanism of the feature library. For example, when the time difference between the application timestamp of a record in the feature library and the current system time exceeds the set aging cycle threshold, and the credibility value is lower than the cleanup threshold, the record will be automatically deleted to maintain the timeliness and accuracy of the feature library. The correlation threshold is an important parameter used to determine whether the real-time waveform matches the waveform in the feature library. It is usually set to 0.8 or higher, which means that the two are considered to be matched successfully only when the correlation coefficient exceeds this value.

[0068] Preferably, to further improve the recognition efficiency and accuracy of the characteristic waveform, the credibility value update mechanism can be optimized. For example, in addition to increasing by 1 after each successful match, a weighted update can also be performed based on the size of the matching correlation coefficient. If the correlation coefficient is much higher than the set threshold, for example, reaching 0.9 or higher, more credibility values ​​can be added, such as 2 or more, to more quickly improve the credibility of the characteristic waveform. In addition, for updating application timestamps, more precise time synchronization technology can be considered to ensure the accuracy of timestamps. For example, using the Network Time Protocol (NTP) or a more precise time synchronization mechanism to ensure that all recorded timestamps remain highly consistent with the system time. In actual applications, the correlation threshold can also be dynamically adjusted according to different load types and operating environments. For example, in scenarios where the load changes frequently, the correlation threshold can be appropriately lowered to increase the sensitivity of the match; while in scenarios where the load changes are relatively stable, the correlation threshold can be appropriately increased to improve the accuracy of the match.

[0069] In some embodiments, generating the energy storage system power suppression instruction based on the power difference and the time difference includes:

[0070] Calculating a power adjustment amount in a linear proportion according to the power difference;

[0071] A response time window for power adjustment is set based on the time difference.

[0072] It should be noted that generating a power suppression instruction for the energy storage system based on the power difference and the time difference is a key step in the present invention to achieve anti-backflow control. The power difference refers to the difference between the initial power and the characteristic signal capture judgment threshold, which reflects the change in power during the load shedding process; the time difference is the time interval between the event triggering moment and the moment when the power begins to decrease, which represents the duration of the load shedding process. Through these two parameters, accurate power suppression instructions can be generated to adjust the output power of the energy storage system, thereby avoiding power backflow. This process ensures that the energy storage system can respond quickly when the load is shedding, reducing the impact on the power grid.

[0073] Specifically, the calculation formula for the power difference is:

[0074] P dlt =P start -P cap

[0075] Among them, P start is the initial power when the power starts to decrease, and P cap is the characteristic signal capture judgment threshold. The calculation formula of time difference is

[0076] T dlt =T event -T start

[0077] Among them, T event is the event triggering time, and T start This is the moment when power begins to decrease. In practical applications, the power adjustment amount can be calculated linearly based on the power difference. For example, if the power difference is large, the power adjustment amount can be appropriately increased to more quickly reduce the output power of the energy storage system. At the same time, the response time window for power adjustment is set based on the time difference. For example, if the time difference is short, a shorter response time window can be set to achieve rapid adjustment. The correlation threshold is usually set to 0.8, which means that only when the correlation coefficient between the real-time waveform and the waveform in the feature library exceeds this value, the two are considered to be matched successfully, triggering the generation of the power suppression instruction.

[0078] Preferably, in order to further improve the generation accuracy and response speed of the power suppression command, the calculation and application of the power difference and time difference can be optimized. For example, when calculating the power difference, dynamic compensation for the power change rate during the load shedding process can be considered.

[0079] The calculation formula of the compensation factor is:

[0080]

[0081] Here, α is the compensation coefficient, and dP / dt is the rate of power change. By introducing a compensation factor, the dynamic characteristics of the load shedding process can be more accurately reflected, thereby generating more precise power reduction instructions. Furthermore, the response time window setting can be adjusted according to different application scenarios. For example, in scenarios with rapid load changes, a shorter response time window can be set to achieve rapid adjustment; in scenarios with slower load changes, the response time window can be appropriately extended to ensure smooth adjustment. In practical applications, the optimal compensation coefficient and response time window parameters can be determined through experimental or simulation analysis to improve overall system performance.

[0082] In some embodiments, it also includes performing aging cleaning on the feature library: when the time difference between the application timestamp of the feature library record and the current system time exceeds the aging cycle threshold and the credibility value is lower than the cleaning threshold, the record is automatically deleted.

[0083] It should be noted that performing aging cleanup on the feature library is an important mechanism in the present invention for maintaining the timeliness and accuracy of the feature library. The purpose of aging cleanup is to automatically delete feature records that no longer have reference value to ensure that the data in the feature library is always up-to-date and valid. Specifically, when the time difference between the application timestamp of the feature library record and the current system time exceeds the aging cycle threshold, and the credibility value is lower than the cleanup threshold, the record will be automatically deleted. The aging cycle threshold refers to the maximum length of time that a feature record can be retained, and the cleanup threshold is a lower limit of the credibility value. The cleanup operation will only be triggered when the credibility value is lower than this lower limit.

[0084] Specifically, the aging cycle threshold can be set according to the load change frequency and system operation characteristics in the actual application scenario. For example, for industrial scenarios with frequent load changes, the aging cycle threshold can be set to a shorter time, such as 72 hours, to ensure that the records in the feature library can promptly reflect the latest load shedding characteristics. The cleaning threshold can be adjusted according to the credibility distribution of the feature record. For example, it is set to 2, which means that only when the credibility value is lower than 2, the feature record will be considered no longer reliable, thereby triggering the cleaning operation. In actual applications, the aging cleaning mechanism can be run regularly, such as automatically executing once every 24 hours, to keep the feature library clean and efficient. In addition, the aging cleaning mechanism can also be combined with other parameters for comprehensive judgment. For example, if the credibility value of a record is higher than the cleaning threshold, but the time difference between its application timestamp and the current system time is close to the aging cycle threshold, it can also be considered for priority cleaning to further optimize the storage space of the feature library.

[0085] Preferably, in order to further improve the flexibility and adaptability of the aging cleanup mechanism, a dynamic adjustment mechanism can be introduced. For example, the aging cycle threshold and the cleanup threshold can be dynamically adjusted according to the number of records in the feature library and the usage of storage space. If the number of records in the feature library is too large, resulting in tight storage space, the aging cycle threshold can be appropriately shortened to speed up the cleanup; conversely, if there is sufficient storage space, the aging cycle threshold can be appropriately extended to retain more historical data. In addition, differentiated cleanup can be performed according to the type and importance of feature records. For example, for feature records of certain critical loads, even if their credibility value is lower than the cleanup threshold, a longer aging cycle threshold can be set to retain more useful information. In practical applications, the aging cleanup mechanism can also be optimized in combination with machine learning algorithms. For example, by analyzing historical data, the aging cycle threshold and the cleanup threshold can be automatically adjusted to achieve intelligent aging cleanup.

[0086] In some embodiments, the correlation threshold is set to 0.8, the credibility threshold is set to 3, the aging cycle threshold is 72 hours, and the cleaning threshold is 2.

[0087] It should be noted that the present invention has specifically set the correlation threshold, credibility threshold, aging cycle threshold and cleaning threshold. The reasonable configuration of these parameters is crucial to ensuring the effectiveness and reliability of the anti-backflow control method. The correlation threshold is used to judge the degree of matching between the real-time waveform and the waveform in the feature library. Only when the correlation coefficient exceeds the threshold is it considered that the two are matched successfully. The credibility threshold is used to determine the credibility of the characteristic waveform. When the credibility value of the characteristic waveform reaches the threshold, it will trigger the generation of the power suppression instruction of the energy storage system. The aging cycle threshold and the cleaning threshold are used for the aging cleaning mechanism to ensure that the records in the feature library are always timely and reliable.

[0088] Specifically, the correlation threshold is usually set to 0.8, which means that only when the correlation coefficient between the real-time waveform and the waveform in the feature library exceeds 0.8, are the two considered to have sufficient similarity, thereby triggering subsequent credibility updates and power suppression instruction generation operations. The credibility threshold is set to 3, which means that the feature waveform needs to be matched successfully at least 3 times before it is considered to have sufficient credibility and trigger the energy storage system power suppression instruction. The aging cycle threshold is set to 72 hours, which means that if the records in the feature library have not been updated for more than 72 hours, they are considered to be possibly outdated and need to be cleaned up. The cleaning threshold is set to 2, that is, the cleaning operation will only be triggered when the credibility value of the feature waveform is lower than 2. The setting of these parameters is the result of comprehensive consideration of factors such as load change frequency, system operation stability, and storage space in actual application scenarios.

[0089] To further enhance the system's adaptability and flexibility, these parameters can be dynamically adjusted based on different application scenarios. For example, in scenarios with frequent load fluctuations, the correlation threshold can be appropriately lowered, such as to 0.75, to increase matching sensitivity. In scenarios with more stable load fluctuations, the correlation threshold can be appropriately increased, such as to 0.85, to improve matching accuracy. The confidence threshold can be adjusted based on system operating experience and historical data. For example, in the early stages of system operation, the confidence threshold can be appropriately lowered to quickly accumulate valid signature waveforms. After the system stabilizes, the confidence threshold can be appropriately increased to ensure the reliability of the signature waveforms. The aging cycle threshold and cleanup threshold can also be adjusted based on the system's storage space and operating efficiency. For example, if storage space is sufficient, the aging cycle threshold can be appropriately extended to retain more historical data. If storage space is limited, the aging cycle threshold can be appropriately shortened to accelerate cleanup. Furthermore, these parameters can be optimized using machine learning algorithms. For example, by analyzing historical data and operation logs, the correlation threshold, confidence threshold, aging cycle threshold, and cleanup threshold can be automatically adjusted to achieve intelligent parameter configuration.

[0090] In some embodiments, the preset time span is a configurable parameter, and the configuration range is 50 milliseconds to 200 milliseconds.

[0091] It should be noted that the preset time span described in this invention is a key parameter for collecting the load shedding characteristic waveform. It defines the time range for collecting voltage and current waveforms around the start of power reduction. Setting the preset time span is crucial for accurately capturing the characteristic signals during the load shedding process, as it directly affects the integrity and representativeness of the characteristic waveform. By properly setting the preset time span, we can ensure that the collected waveform data accurately reflects the electrical characteristics during the load shedding process, thereby improving the accuracy of subsequent matching and control.

[0092] Specifically, the preset time span is a configurable parameter, and its configuration range is 50 milliseconds to 200 milliseconds. The setting of this range is based on a comprehensive consideration of the duration of the load shedding process and the system response speed in actual applications. For example, for some loads that are quickly shedding, a shorter time span, such as 50 milliseconds, may be required to ensure that the instantaneous characteristics of the load shedding can be quickly captured; and for some load shedding processes with a longer duration, a longer time span, such as 200 milliseconds, can be set to obtain more complete waveform data. In actual applications, adjustments can be made according to different load types and system characteristics. For example, in industrial scenarios, the load shedding process may be more complex and require a longer time span to capture the complete characteristic signal; while in commercial scenarios, the load shedding process may be faster and a shorter time span may be more appropriate.

[0093] Preferably, in order to further improve the efficiency and accuracy of characteristic waveform acquisition, the preset time span can be dynamically adjusted according to the actual application scenario. For example, an adaptive mechanism can be introduced to dynamically adjust the preset time span according to the power change rate monitored in real time. If the power change rate is fast, the preset time span can be shortened to quickly capture the instantaneous characteristics of load shedding; if the power change rate is slow, the preset time span can be appropriately extended to obtain more complete waveform data. In addition, the preset time span can be optimized in combination with a machine learning algorithm. For example, by analyzing historical data and matching results, the preset time span is automatically adjusted to achieve intelligent parameter configuration. In actual applications, multiple preset time spans can also be set, each for different load types or different monitoring stages, to improve the overall performance and adaptability of the system.

[0094] In some embodiments, the calculation of the power difference includes dynamic compensation for the power change rate during load shedding, and the compensation factor is calculated as follows:

[0095]

[0096] Where K is the compensation factor, α is the compensation coefficient, and dP / dt is the power change rate.

[0097] It should be noted that the present invention incorporates a dynamic compensation mechanism into the calculation of the power difference to more accurately reflect the power variation characteristics during load shedding. The power rate of change refers to the rate of change of power over time during load shedding. By calculating the dynamic compensation factor for the power rate of change, the power difference can be corrected, thereby more accurately generating power suppression commands for the energy storage system. This dynamic compensation mechanism effectively addresses power variations under varying load characteristics, improving control accuracy and adaptability.

[0098] Specifically, the calculation of the power difference not only considers the difference between the initial power and the characteristic signal capture judgment threshold, but also introduces dynamic compensation of the power change rate. The calculation formula of the compensation factor is

[0099]

[0100] Here, α is the compensation coefficient, which adjusts the compensation intensity; dP / dt is the power change rate, representing the rate of change of power over time. By introducing a compensation factor, the power difference can be dynamically adjusted. For example, when the power change rate is rapid, the compensation factor is increased, thereby increasing the power adjustment amount and responding more quickly to load shedding events. The specific value of the compensation coefficient α can be adjusted based on the load characteristics of the actual application scenario. For example, in scenarios with rapid load changes, the value of α can be appropriately increased to improve the compensation effect.

[0101] Preferably, to further improve the accuracy and adaptability of dynamic compensation, the compensation coefficient α can be dynamically adjusted. For example, the value of α can be dynamically adjusted based on historical data and the real-time monitored rate of power change. If the system detects frequent rapid load shedding events, the value of α can be appropriately increased to improve the compensation effect. Conversely, if the load changes are relatively stable, the value of α can be appropriately reduced to avoid overcompensation.

[0102] Furthermore, machine learning algorithms can be combined to optimize the compensation coefficient. For example, by analyzing historical data and matching results, the compensation coefficient can be automatically adjusted to achieve intelligent parameter configuration. In practical applications, multiple compensation coefficients can be introduced, each for different load types or monitoring phases, to improve the overall performance and adaptability of the system.

[0103] In some embodiments, the generation delay of the power suppression instruction is less than 50 milliseconds, and the power adjustment amount is directly proportional to the power difference.

[0104] It should be noted that the present invention specifically specifies the generation delay of the power suppression instruction and the power adjustment amount to ensure that the energy storage system can respond quickly and accurately to load shedding events. The generation delay of the power suppression instruction refers to the time interval from detecting a successful match of the load shedding characteristic waveform to generating the power suppression instruction. This delay is less than 50 milliseconds, ensuring the system's rapid response capability. The power adjustment amount is directly proportional to the power difference, which means that the size of the power adjustment amount is directly dependent on the change in power during the load shedding process, thereby achieving precise power regulation.

[0105] Specifically, the generation delay of the power suppression instruction is less than 50 milliseconds. This parameter is set to ensure that the energy storage system can respond to load shedding events in a very short time to avoid power backflow. The 50-millisecond delay is a strict performance indicator, which requires the system to have high-precision real-time monitoring and fast processing capabilities. The power adjustment amount is directly proportional to the power difference. Specifically, if the power difference is large, it indicates that the power changes significantly during the load shedding process. At this time, a larger power adjustment amount is required to suppress the output power of the energy storage system; conversely, if the power difference is small, the power adjustment amount is also reduced accordingly. This positive proportional relationship ensures that the energy storage system can be accurately adjusted according to the actual power changes, avoiding over-adjustment and ensuring the effectiveness of the adjustment.

[0106] Preferably, to further improve the efficiency of generating power suppression instructions and the accuracy of adjustment, the system's hardware and software can be optimized. For example, in hardware, high-performance processors and low-latency communication modules can be used to reduce data processing and transmission time. In software, algorithms can be optimized to reduce unnecessary computational steps and improve algorithm execution efficiency. Furthermore, a prediction mechanism can be introduced to predict the required power adjustment amount in advance after detecting a successful match with the load shedding characteristic waveform. This predicted value can then be directly applied when generating the power suppression instruction, further shortening the response time. In practical applications, the optimal delay and power adjustment proportional coefficient can be determined through experimental or simulation analysis to improve the overall performance of the system.

[0107] The above-described embodiments of the present invention have the following beneficial effects: By setting a reverse power threshold and a characteristic signal capture threshold, and when detecting that the power at a user's entry point is less than the characteristic signal capture threshold, the present invention records the relevant time and power, collects voltage and current waveforms to form a characteristic waveform, calculates the time difference and power difference, and stores this information as a characteristic database record, thereby accurately identifying and recording load shedding characteristics. Furthermore, the present invention scans the voltage and current waveforms at the entry point in real time, calculates the correlation between the real-time waveform and each characteristic waveform in the characteristic database using the Pearson correlation coefficient method, updates the confidence value and application timestamp of the matching characteristic waveform when the correlation coefficient exceeds the threshold, and generates a power suppression instruction for the energy storage system based on the power difference and time difference when the confidence value reaches the set threshold. This allows for preemptive adjustment of the energy storage system power, avoids power reverse flow, and improves the safety and economic efficiency of the energy storage system operation. Furthermore, the characteristic database is cleaned up by aging. When the time difference between the application timestamp of a characteristic database record and the current system time exceeds the aging period threshold and the confidence value falls below the cleanup threshold, the record is automatically deleted. This maintains the timeliness and accuracy of the characteristic database and further optimizes control effects. In addition, by setting parameters such as correlation thresholds, credibility thresholds, aging cycle thresholds, and cleanup thresholds, as well as optimizing preset time spans, dynamic compensation in power difference calculations, and the generation delay of power suppression instructions, the accuracy and reliability of control can be further improved, ensuring that the energy storage system can operate stably under different operating conditions and minimizing economic losses and fines imposed on power supply companies due to power backflow.

[0108] like Figure 2 As shown, in some embodiments, a backflow prevention control system based on load shedding feature recognition includes:

[0109] Grid access end, distribution transformer connected through household gateway, user load end, and energy storage system connected through energy storage converter;

[0110] The energy storage converter is configured as follows:

[0111] Real-time collection of voltage and current signals at the household access point;

[0112] Perform the following steps:

[0113] During the operation of the energy storage system, a reverse power prevention threshold and a characteristic signal capture threshold are set, wherein the characteristic signal capture threshold is greater than or equal to the reverse power prevention threshold;

[0114] When it is detected that the power at the user's home gateway is less than the characteristic signal capture judgment threshold, the current time is recorded as the event trigger time, and the power reduction start time and the corresponding initial power are obtained;

[0115] Collect the voltage and current waveforms at the household gateway within a preset time span before and after the power starts to decrease, and form a load shedding characteristic waveform;

[0116] Calculating the time difference between the event triggering moment and the moment when the power starts to decrease, and the power difference between the initial power and the characteristic signal capture determination threshold;

[0117] The load shedding characteristic waveform, initial credibility value, event triggering time, time difference and power difference are stored as characteristic library records;

[0118] Scanning the voltage waveform and current waveform at the entrance to the house in real time at intervals of the preset time span;

[0119] The correlation between the real-time waveform and each characteristic waveform in the characteristic library is calculated using the Pearson correlation coefficient method;

[0120] When there is a correlation coefficient exceeding the correlation threshold, the credibility value of the matching characteristic waveform and the latest application timestamp are updated;

[0121] When the credibility value reaches a set credibility threshold, a power suppression instruction for the energy storage system is generated based on the power difference and time difference. At a voltage level of 0.4kV, the energy storage system's charge and discharge power is adjusted to prevent the reverse flow of user-side power to the grid.

[0122] It is understandable that the modules recorded in the anti-backflow control system based on load shedding feature recognition are similar to those in the reference Figure 1 The steps in the anti-backflow control method based on load shedding feature recognition described above correspond to each other. Therefore, the operations, features and beneficial effects described above for the anti-backflow control method based on load shedding feature recognition are also applicable to the anti-backflow control system based on load shedding feature recognition and the modules contained therein, and will not be repeated here.

[0123] The above descriptions are merely some preferred embodiments of the present invention and an illustration of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present invention is not limited to the technical solutions formed by a specific combination of the above-mentioned technical features, but should also encompass other technical solutions formed by any combination of the above-mentioned technical features or their equivalents without departing from the above-mentioned inventive concept. For example, a technical solution formed by mutually replacing the above-mentioned features with (but not limited to) technical features having similar functions disclosed in the embodiments of the present invention.

Claims

1. A backflow prevention control method based on load shedding feature recognition, characterized in that: The following steps are involved: S1: During the operation of the energy storage system, a reverse power prevention threshold and a characteristic signal capture threshold are set, wherein the characteristic signal capture threshold is greater than or equal to the reverse power prevention threshold; S2: When it is detected that the power at the user's home gateway is less than the characteristic signal capture judgment threshold, the current time is recorded as the event trigger time, and the power reduction start time and the corresponding initial power are obtained; S3: Collect the voltage and current waveforms at the household gateway within a preset time span before and after the power starts to decrease, and form a load shedding characteristic waveform; S4: Calculating the time difference between the event triggering moment and the power starting to decrease moment, and the power difference between the initial power and the characteristic signal capture determination threshold; S5: storing the load shedding characteristic waveform, initial credibility value, event triggering time, time difference and power difference as characteristic library records; S6: Scanning the voltage waveform and current waveform at the entrance to the house in real time at intervals of the preset time span; S7: Calculate the correlation between the real-time waveform and each characteristic waveform in the characteristic library by using the Pearson correlation coefficient method; S8: When the correlation coefficient exceeds the correlation threshold, the credibility value of the matching feature waveform and the latest application timestamp are updated; S9: When the credibility value reaches a set credibility threshold, a power suppression instruction for the energy storage system is generated based on the power difference and the time difference.

2. The control method according to claim 1, characterized in that: Each record of the feature library includes a feature waveform field, a credibility field, an application timestamp field, a time difference field, and a power difference field, wherein the credibility field is initially assigned a value of 1.

3. The control method according to claim 2, characterized in that: The updating of the credibility value of the matching characteristic waveform includes: each time a characteristic waveform match is completed, the credibility value of the corresponding record is increased by 1, and the application timestamp is synchronously updated to the current system time.

4. The control method according to claim 1, wherein: Generating a power suppression instruction for the energy storage system based on the power difference and the time difference includes: Calculating a power adjustment amount in a linear proportion according to the power difference; A response time window for power adjustment is set based on the time difference.

5. The control method according to claim 1, characterized in that: It also includes aging cleanup of the signature database: when the time difference between the application timestamp of the signature database record and the current system time exceeds the aging cycle threshold and the credibility value is lower than the cleanup threshold, the record is automatically deleted.

6. The control method according to claim 1, characterized in that: The correlation threshold is set to 0.8, the credibility threshold is set to 3, the aging cycle threshold is 72 hours, and the cleaning threshold is 2.

7. The control method according to claim 1, characterized in that: The preset time span is a configurable parameter, and the configuration range is 50 milliseconds to 200 milliseconds.

8. The control method according to claim 1, characterized in that: The calculation of the power difference includes dynamic compensation for the power change rate during load shedding. The compensation factor calculation formula is: Where K is the compensation factor, α is the compensation coefficient, and dP / dt is the power change rate.

9. The control method according to claim 1, characterized in that: The generation delay of the power suppression instruction is less than 50 milliseconds, and the power adjustment amount is directly proportional to the power difference. 10.An anti-backflow control system based on load shedding feature recognition, characterized in that: include: Grid access end, distribution transformer connected through household gateway, user load end, and energy storage system connected through energy storage converter; The energy storage converter is configured as follows: Real-time collection of voltage and current signals at the household access point; Perform the following steps: During the operation of the energy storage system, a reverse power prevention threshold and a characteristic signal capture threshold are set, wherein the characteristic signal capture threshold is greater than or equal to the reverse power prevention threshold; When it is detected that the power at the user's home gateway is less than the characteristic signal capture judgment threshold, the current time is recorded as the event trigger time, and the power reduction start time and the corresponding initial power are obtained; Collect the voltage and current waveforms at the household gateway within a preset time span before and after the power starts to decrease, and form a load shedding characteristic waveform; Calculating the time difference between the event triggering moment and the moment when the power starts to decrease, and the power difference between the initial power and the characteristic signal capture determination threshold; The load shedding characteristic waveform, initial credibility value, event triggering time, time difference and power difference are stored as characteristic library records; Scanning the voltage waveform and current waveform at the entrance to the house in real time at intervals of the preset time span; The correlation between the real-time waveform and each characteristic waveform in the characteristic library is calculated using the Pearson correlation coefficient method; When there is a correlation coefficient exceeding the correlation threshold, the credibility value of the matching characteristic waveform and the latest application timestamp are updated; When the credibility value reaches a set credibility threshold, an energy storage system power suppression instruction is generated based on the power difference and the time difference.

Images