An uninterruptible power supply intelligent management method and system, and a storage medium

Abstract

The application relates to the technical field of power management, and discloses an intelligent management method and system of an uninterruptible power supply and a storage medium. The method comprises the following steps: updating actual power of a load end every first time length, and obtaining demand power on the basis of correcting the actual power based on a load fluctuation rate when power supply at a power supply end is started; at least one uninterruptible power supply is started based on the demand power, so that the load utilization rate of the power supply end is within a standard range; the started uninterruptible power supply is defined as a first power supply, the residual power and the power attenuation rate of the first power supply are obtained every second time length, and a high-risk power supply is determined based on the residual power and the power attenuation rate; if there is an idle uninterruptible power supply in the power supply end, the idle uninterruptible power supply is put online to replace the high-risk power supply, otherwise, a load device is closed; and after commercial power is restored, the uninterruptible power supplies in the power supply end are charged based on a charging priority, so that sustainable balance between charging safety and device service life is realized.

Description

Technical Field

[0001] This application relates to the field of power management technology, and in particular to an intelligent management method, system and storage medium for uninterruptible power supplies. Background Technology

[0002] An uninterruptible power supply (UPS) is a device that provides continuous and stable power to a load through batteries and power electronic control technology. Its core function is to ensure the continuity of power supply, seamlessly switching to a backup power source in the event of a power outage, voltage fluctuation, or frequency anomaly, thus preventing equipment downtime.

[0003] To maximize the application effect of uninterruptible power supplies (UPS), various management methods have been proposed in existing technologies. For example, Chinese patent document CN118054547A discloses a management method, system, device, and storage medium for UPS devices. This method selects one UPS unit from N UPS units in the UPS device as the current power supply unit and supplies power to the virtualization system. During the power supply process, when the remaining power of each UPS unit is lower than the i-th power range, all virtual machines in the virtualization system at the i-th importance level are shut down. When the remaining power of each UPS unit is in the K-th power range, the system virtual machine, data virtual machine, and resource management virtual machine are shut down, and the host of the virtualization system is controlled to enter standby mode. This effectively improves the working time of the virtualization system.

[0004] However, the above technical solutions only consider runtime and do not optimize the dynamic matching mechanism between load efficiency and power supply, thus resulting in low power utilization and insufficient guarantee of critical business loads. Summary of the Invention

[0005] To address the problems mentioned in the background section, this application provides an intelligent management method, system, and storage medium for uninterruptible power supplies.

[0006] To achieve the aforementioned objectives, this invention proposes an intelligent management method for uninterruptible power supplies, comprising:

[0007] Establish a power supply network, which includes the power supply end and the load end;

[0008] The actual power of the load is updated at the first interval. When the power supply starts, the load fluctuation rate at the power outage time is obtained based on the prediction model. The actual power is corrected based on the load fluctuation rate to obtain the required power.

[0009] Start at least one uninterruptible power supply based on the required power to ensure that the load utilization at the power supply end is within the standard range;

[0010] Define the uninterruptible power supply that is started as the first power supply. At every second time interval, obtain the remaining power and power decay rate of the first power supply. Based on the remaining power and power decay rate, identify the high-risk power supply.

[0011] If there is an idle uninterruptible power supply (UPS) at the power supply end, then connect the idle UPS to replace the high-risk power supply. If there is no idle UPS at the power supply end, then shut down the load devices at the load end in sequence.

[0012] After the mains power is restored, the charging priority is determined based on the battery health model, and the battery is charged to the uninterruptible power supply at the power supply end based on the charging priority.

[0013] Furthermore, shutting down the load device at the load end includes the following steps:

[0014] The remaining power supply duration of each first power source is calculated based on the remaining power and power decay rate. After a first number of second durations, the first number of remaining power supply durations are organized into a dataset corresponding to each first power source. A first function about the remaining power supply duration is generated based on the dataset. The offline time point of each first power source is predicted based on the first function. Load change data of the power supply end is generated based on the offline time point.

[0015] Set a critical value, obtain the power consumption of each load device every third time interval, determine the shutdown time point based on load change data and power consumption, and shut down the load devices in sequence after the shutdown time point so that the load utilization rate of the power supply end is lower than the critical value.

[0016] Further, determining the shutdown time includes the following steps:

[0017] The system identifies the first time point when the load utilization exceeds the critical value, assigns an importance level to each load device, defines load devices that are still running before the first time point as devices to be shut down, selects multiple device groups from the devices to be shut down, each device group must include at least one load device, shuts down all load devices in a device group to bring the load utilization below the critical value, selects a shutdown group from the device groups, selects an action time point before the first time point, uses the action time point as the shutdown time point of the shutdown group, and shuts down the load devices included in the shutdown group in sequence after the shutdown time point is reached.

[0018] After determining the shutdown group corresponding to the first time point, predict the second time point when the load utilization exceeds the critical value, and determine the shutdown group corresponding to the second time point and the corresponding action time point, until all load devices are completely shut down or the mains power is restored.

[0019] Further, selecting the shutdown group within the device group includes the following steps:

[0020] The average grade of each equipment group is calculated based on the importance level. The ideal shutdown time of each load device in the equipment group is obtained. The ideal shutdown time is the time taken for the load device to completely shut down after receiving the shutdown command. The ideal shutdown times in the equipment group are sorted from largest to smallest. The difference between adjacent ideal shutdown times is obtained. The evaluation value is obtained based on the average grade and the difference. The equipment group with the highest evaluation value is the shutdown group.

[0021] Furthermore, sequentially shutting down the load devices included in the shutdown group after reaching the shutdown time point includes the following steps:

[0022] Before the shutdown time point, a verification time point is set. After the verification time point is reached, the offset of the remaining power supply time of the high-risk power source before the verification time point is calculated. If the offset is greater than the first threshold, the load devices in the shutdown group are shut down sequentially after the calculation is completed. Otherwise, the load devices in the shutdown group are shut down sequentially after the shutdown time point is reached.

[0023] Furthermore, calculating the offset and volatility involves the following steps:

[0024] Establish a coordinate system with the remaining power as the horizontal axis and the remaining power supply duration as the vertical axis. Based on the first function, plot the curve of the remaining power supply duration of the high-risk power source as the remaining power. Locate the third time point when the remaining power is obtained. Plot multiple coordinate points with the remaining power at the third time point as the horizontal axis and the remaining power supply duration as the vertical axis. Calculate the straight-line distance between each coordinate point and the curve and use the sum of the straight-line distances as the offset.

[0025] Furthermore, starting at least one uninterruptible power supply based on power demand includes the following steps:

[0026] Obtain the number of charging and discharging cycles for each uninterruptible power supply (UPS), calculate the average number of charging and discharging cycles, sort the UPSs from smallest to largest based on the average number of cycles, and select the top N UPSs as the UPS to start based on the required power during a power outage.

[0027] Furthermore, the prediction model is an LSTM time series prediction model.

[0028] This application also provides an intelligent management system for uninterruptible power supplies (UPS), which implements the aforementioned intelligent management method for UPS. The system includes:

[0029] The data acquisition module establishes a power supply network, which includes a power supply end and a load end. The front-end module updates the actual power of the load end at the first time interval. When the power supply end starts to supply power, the load fluctuation rate at the power outage time point is obtained based on the prediction model. The actual power is corrected based on the load fluctuation rate to obtain the required power.

[0030] The startup module starts at least one uninterruptible power supply based on the required power to ensure that the load utilization of the power supply is within the standard range.

[0031] The monitoring module defines the uninterruptible power supply that is started as the first power supply. Every second time interval, it acquires the remaining power and power decay rate of the first power supply and determines the high-risk power supply based on the remaining power and power decay rate.

[0032] Adjust the module; if there is an idle uninterruptible power supply (UPS) at the power supply end, connect the idle UPS to replace the high-risk power supply; if there is no idle UPS at the power supply end, shut down the load devices at the load end in sequence.

[0033] After the mains power is restored, the charging module determines the charging priority based on the battery health model and charges the uninterruptible power supply in the power supply end according to the charging priority.

[0034] This application also provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the intelligent management method for uninterruptible power supply as described above.

[0035] The beneficial effects of this invention are as follows:

[0036] This invention first establishes a dynamic correction mechanism based on real-time power calibration and load fluctuation prediction to ensure that the power supply start-up and shutdown strategy accurately adapts to load demands and maintains the optimal power supply efficiency range. Then, relying on dual-parameter monitoring of remaining power and attenuation rate to trigger intelligent switching of redundant power supplies or load hierarchical management, it constructs a closed loop for ensuring the continuity of critical business operations. Finally, combining a battery health assessment model and off-peak charging control strategy, it achieves a sustainable balance between charging safety and equipment lifespan. In summary, this invention, through the deep coupling of hardware redundancy architecture, state monitoring algorithms, and resource scheduling logic, achieves a triple improvement in power supply efficiency, system reliability, and equipment lifecycle management, ultimately achieving end-to-end management without relying on complex models. Attached Figure Description

[0037] Figure 1 This is a schematic diagram of an intelligent management method for an uninterruptible power supply according to this application;

[0038] Figure 2 This is a schematic diagram for calculating the deviation value in this application. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0040] It is understood that the terms "first," "second," etc., used in this application may be used herein to describe various elements, but unless otherwise specified, these elements are not limited by these terms. These terms are used only to distinguish one element from another. For example, without departing from the scope of this application, a first script may be referred to as a second script, and similarly, a second script may be referred to as a first script.

[0041] like Figure 1 As shown, an intelligent management method for uninterruptible power supplies includes:

[0042] S1: Establish a power supply network, which includes the power supply end and the load end.

[0043] Specifically, the power supply side includes a combination of multiple UPS units connected in parallel via a power distribution bus and equipped with lithium iron phosphate battery packs; the load side includes multiple load devices, such as multiple servers, which are connected to the power supply system separately through hierarchical protection circuits.

[0044] S2: Update the actual power of the load at the first interval. When the power supply starts, obtain the load fluctuation rate at the time of power outage based on the prediction model, and correct the actual power based on the load fluctuation rate to obtain the required power.

[0045] The prediction model is an LSTM time series prediction model.

[0046] The actual power is the sum of the actual power of all load devices at the load end. For industrial scenarios, the first duration can be set to 10 seconds, and for commercial scenarios, the first duration can be set to 30 seconds. During data acquisition, the current of each load device is collected in real time through Hall sensors, and the actual power is calculated by combining voltage sampling and moving average filtering. When the power supply starts supplying power, it means that the mains power has failed. The corresponding time point when the power fails is recorded as the power outage time point, and the load fluctuation rate at the power outage time point is predicted based on the LSTM model.

[0047] Load volatility specifically refers to the load fluctuation within a first time period. When the load volatility is large, a larger correction coefficient is required, and when the load volatility is low, a smaller correction coefficient is required. The LSTM prediction model is trained based on 12 months of historical data, which includes load volatility, equipment codes, dates, and environmental variables for multiple first time periods. Environmental variables include weather, temperature output, and real-world event scheduling, such as business activity arrangements.

[0048] The load fluctuation rate is set to 1% to 100% and divided into 5 ranges, each with a corresponding correction factor. The correction factor is then multiplied by the actual power to obtain the required power. The correction factor is between 1 and 1.5. For example, if the load fluctuation rate at the predicted power outage time is 50%, the corresponding correction factor is 1.2, and the actual power is 1.2 times the measured required power. In particular, the actual power here is the actual power obtained from the last measurement before the power outage time.

[0049] S3: Start at least one uninterruptible power supply based on the required power to ensure that the load utilization of the power supply is within the standard range.

[0050] When the load is between 50% and 75%, the uninterruptible power supply (UPS) can generally operate in its optimal efficiency range. Therefore, the number of UPS units to be started is determined based on the required power. For example, when the required power is 4.2kW, two 3kW rated power supplies can be selected and combined to provide a 6kW output, so that the load utilization rate reaches 70%. The startup sequence is determined by a sorting algorithm based on the number of charge and discharge cycles.

[0051] Specifically, the number of charging and discharging cycles for each uninterruptible power supply (UPS) is obtained, the average number of charging and discharging cycles is calculated, and the UPSs are sorted from smallest to largest based on the average number of cycles. When a power outage occurs, the top N UPSs in the sorted list are selected as the UPS to be started based on the required power.

[0052] For example, if four power supplies are selected and their average charge / discharge cycles are 20 / 30 / 50 / 80, and they are arranged in descending order, the first two power supplies with 20 and 30 cycles should be selected for startup.

[0053] S4: Define the uninterruptible power supply that is started as the first power supply. Every second time interval, obtain the remaining power and power decay rate of the first power supply, and determine the high-risk power supply based on the remaining power and power decay rate.

[0054] Assuming the uninterruptible power supply (UPS) can sustain power for 60 minutes, the second duration is set to 5 minutes. Data is collected every 5 minutes by the battery management system. Based on a voltage-capacity mapping table, the remaining power of each primary power source is obtained. For example, when the voltage drops to 51V, the remaining power is 40%. The power decay rate is determined by calculating the difference in power change between adjacent cycles. For instance, if the previous cycle's power was 50% and the current cycle's is 38%, the decay rate is 12 / 5 = 2.4% / min. If a primary power source simultaneously meets the conditions of remaining power ≤ 20% and decay rate ≥ 2% / min, it is classified as a high-risk power source; or if the remaining power is > 20% but the decay rate is ≥ 5% / min, it is also marked as a high-risk power source.

[0055] S5: If there is an idle uninterruptible power supply (UPS) at the power supply end, connect the idle UPS to replace the high-risk power supply. If there is no idle UPS at the power supply end, shut down the load devices at the load end in sequence.

[0056] If power supply A needs to be shut down and an idle power supply B is available, the system will automatically activate power supply B and seamlessly switch to and isolate power supply A within 10 seconds via the power distribution unit, ensuring zero power interruption at the load end. If no idle power supply is available, the load devices will be shut down sequentially according to a preset priority. For example, low-priority loads will be shut down first within 5 seconds, followed by medium-priority devices within 15 seconds, and finally high-priority loads. However, this shutdown sequence is not used in this embodiment; details will be provided later.

[0057] S6: After the mains power is restored, the charging priority is determined based on the battery health model, and the battery is charged to the uninterruptible power supply in the power supply end based on the charging priority.

[0058] After mains power is restored, the battery health model calculates the health score of each power source based on factors such as charge / discharge cycles, remaining cycle life, internal resistance change rate, voltage consistency, and temperature deviation. This battery health model is based on existing technology. For example, power source C has accumulated 300 charge / discharge cycles, has 60% remaining cycle life, internal resistance increased by 15% compared to factory settings, voltage fluctuation standard deviation of 0.8V, and recent highest temperature deviation of 4℃. After weighted calculation, it achieves a health score of 68. Subsequently, the system generates a charging queue in descending order of health score and performs staggered charging of the uninterruptible power supplies (UPS) based on the charging sequence. For example, the UPS with the lowest health score is charged first, and then the UPS with the next highest health score is charged 20 seconds later. This differentiated charging control effectively reduces the impact risk to the battery when mains power is restored.

[0059] This invention achieves real-time calibration of demand power by periodically updating the actual power data at the load end and dynamically correcting the load fluctuation trend at the moment of power outage using a predictive model. This mechanism ensures that the number of uninterruptible power supplies (UPS) activated is precisely matched with load demand, maintaining the load utilization rate at the power supply end within the standard range as much as possible in the initial stage. Subsequently, based on a dual monitoring mechanism of remaining power supply capacity and power decay rate, it proactively identifies nodes with power supply stability risks. Through intelligent switching strategies for redundant power supplies or tiered shutdown logic for load devices, critical loads are prioritized for operation when power resources are limited. Finally, during the mains power restoration phase, a differentiated charging priority strategy is generated through a battery health assessment model, combined with a peak-shaving charging control mechanism, to reduce the system impact when multiple power supplies charge in parallel, thereby extending the overall lifespan of the battery pack.

[0060] This solution uses a dynamic matching mechanism of real-time power calibration and load fluctuation prediction to accurately adjust the power output and predict load change trends. Combined with intelligent switching of redundant power supplies, it ensures that the power supply efficiency is always in the optimal range.

[0061] This embodiment includes the following steps to shut down the load device on the load side:

[0062] The remaining power supply duration of each first power source is calculated based on the remaining power and power decay rate. After a first number of second durations, the first number of remaining power supply durations are organized into a dataset corresponding to each first power source. A first function about the remaining power supply duration is generated based on the dataset. The offline time point of each first power source is predicted based on the first function. Load change data of the power supply end is generated based on the offline time point.

[0063] This embodiment calculates the remaining power supply duration based on the following steps: First, the offline power level of the first power source is set, for example, it needs to be offline when the power level is 5%. During the calculation, the remaining power level of the first power source is obtained, such as 29%, and the current power decay rate is, for example, 2.4% / min. Then the remaining power supply duration is (29-5)% / 2.4%=10min. The first quantity is set to 4. After four 5-minute intervals after the power is cut off, that is, after 20 minutes, the remaining power supply duration of each first power source will be collected at four time points. The four remaining power supply durations are organized into a dataset. For example, the dataset of the first power source A includes 30, 24, 20, and 15. A linear function is used to fit it to obtain the corresponding first function. The first function is the correlation function between the power level and the remaining power supply duration.

[0064] When predicting the offline time, a coordinate system is established with battery level as the horizontal axis and remaining power supply time as the vertical axis. A first function is then plotted in this first coordinate system, and the difference in the vertical axis between the current battery level and the offline battery level is determined. This difference represents the time required for the current battery level to reach the offline level. Adding the current time to the calculated required time yields the offline time of the first power supply. Next, load change data is calculated. This load change data is the predicted data, referring to the change in load utilization after the offline time, without shutting down the load equipment. For example, the load change data might include: after the first power supply A goes offline, if the load equipment is not shut down, the load utilization will increase from 50% to 70%; subsequently, after the first power supply B goes offline, if the load equipment is not shut down, the load utilization will increase from 70% to 90%.

[0065] Set a critical value, obtain the power consumption of each load device every third time interval, determine the shutdown time point based on load change data and power consumption, and shut down the load devices in sequence after the shutdown time point so that the load utilization rate of the power supply end is lower than the critical value.

[0066] The critical value is set to 80%, but in other embodiments, it can also be set to 75%, 70%, etc. After obtaining the load change data, if it is determined that the load utilization rate will exceed the critical value after the first power supply B goes offline, a shutdown time point is determined based on the shutdown time of the first power supply B. After the shutdown time point, at least one load device is selected to be shut down, so that the overall load utilization rate of the power supply end is still lower than the critical value after the first power supply B goes offline. For example, by shutting down load device A, the load utilization rate of the power supply end is 72% after the first power supply B goes offline, which does not exceed the critical value.

[0067] This embodiment determines the shutdown time point through the following steps:

[0068] The system identifies the first point in time when load utilization exceeds a critical value, assigns an importance level to each load device, defines load devices still running before the first point in time as devices to be shut down, selects multiple device groups from the devices to be shut down, each device group must include at least one load device, shuts down all load devices in a device group to bring the load utilization below the critical value, selects a shutdown group from the device groups, selects an action time point before the first point in time, uses the action time point as the shutdown time point of the shutdown group, and shuts down the load devices included in the shutdown group in sequence after the shutdown time point is reached.

[0069] As described above, the first time point is determined by the offline time and load change data of each first power supply. Before this, an importance level is first set for each load device. The importance level is set manually, and there are three levels: level 1, level 2, and level 3. The smaller the value of the importance level, the more important the load device. After determining when the first time point occurs, the load devices that are still running before the first time point are defined as devices to be shut down, such as devices 1, 2, and 3. In this embodiment, the device group is generated in the following way: First, the uninterruptible power supply to be shut down is determined based on the first time point. The load utilization rate of the power supply end after the power supply to be shut down is calculated. The power reduction is calculated based on the difference between the load utilization rate after shutdown and the critical value. For example, after shutdown, the difference between the load utilization rate and the critical threshold is 10%. By reducing the power at the load end by 3kW, the load utilization rate can be reduced by 10%, so 3kW is the power reduction.

[0070] Next, devices with power consumption less than 3kW that need to be shut down are first removed. Then, the remaining devices to be shut down are traversed and combined to obtain multiple initial groups. If devices 1, 2, and 3 need to be shut down, the generated initial groups include (1), (2), (3), (1,2), (2,3), (1,3), and (1,2,3), totaling 6 groups. Then, the actual power of the devices to be shut down in the initial groups is summed to obtain the total power. The initial groups with a total power greater than or equal to the power reduction power are selected as device groups. Through calculation, the initial groups (1,2), (2,3), (1,3), and (1,2,3) are retained. The above four initial groups are then used as device groups. Finally, the best group is selected as the shutdown group.

[0071] After identifying the shutdown group, obtain the ideal shutdown time for each device to be shut down in the shutdown group. The ideal shutdown time is the time taken for the device to be shut down from receiving the shutdown command to complete shutdown. Obtain the ideal shutdown time with the largest value, multiply it by 2 to make it a redundant time, and set the action time point before the first time point and at the interval of the redundant time. For example, if the ideal shutdown time with the largest value is 2 minutes, it becomes 4 minutes after being multiplied by 2. If the first time point is 20:04, then the action time point is 20:00, and 20:00 is taken as the shutdown time point.

[0072] After determining the shutdown group corresponding to the first time point, predict the second time point when the load utilization exceeds the critical value, and determine the shutdown group corresponding to the second time point and the corresponding action time point, until all load devices are completely shut down or the mains power is restored.

[0073] During forecasting, it is first necessary to assume the load change data at the power supply end after the shutdown group is shut down. Based on this load change data, a second time point is determined. The determination methods for the three parameters—the second time point, the corresponding shutdown group, and the action time point—are the same as those for the first time point, and will not be described here. This process is repeated until all load devices are shut down, or mains power is restored.

[0074] In this embodiment, selecting the shutdown group from the device group includes the following steps:

[0075] The average grade of each equipment group is calculated based on the importance level. The ideal shutdown time of each load device in the equipment group is obtained. The ideal shutdown time is the time taken for the load device to completely shut down after receiving the shutdown command. The ideal shutdown times in the equipment group are sorted from largest to smallest. The difference between adjacent ideal shutdown times is obtained. The evaluation value is obtained based on the average grade and the difference. The equipment group with the highest evaluation value is the shutdown group.

[0076] For the initial group (1, 2), the importance level of load device 1 and load device 2 is both 1, so the average importance level is 1. For the initial group (2, 3), the importance level of load device 3 is 3, so the average importance level is 2. If the initial group includes only one load device, the difference is uniformly set to 3. For the initial group (1, 2, 3), the ideal shutdown time of load device 1 is 3 minutes, the shutdown time of load device 2 is 1 minute, and the ideal shutdown time of load device 3 is 2 minutes. After sorting, two differences are obtained, both of which are 1 minute. In particular, since the sorting has been performed from largest to smallest, the calculated differences are all positive values. The purpose of this step is to select the initial groups with lower average importance levels and the ability to perform tiered shutdown as shutdown groups. To this end, this embodiment sums the reciprocal of the average importance level with the difference to obtain the evaluation value. For example, if the average importance level of the initial group (1, 2, 3) is 5 / 3≈1.67, then the evaluation value is 1 / 1.67+1+1≈2.6. A higher rating indicates a higher average grade in the initial grouping and a greater difference in ideal shutdown time between adjacent load devices.

[0077] By shutting down the server in stages, the load on the power supply can be gradually reduced, thereby extending its lifespan.

[0078] This embodiment involves the following steps to sequentially shut down the load devices included in the shutdown group after the shutdown time point is reached:

[0079] Before the shutdown time point, a verification time point is set. After the verification time point is reached, the offset of the remaining power supply time of the high-risk power source before the verification time point is calculated. If the offset is greater than the first threshold, the load devices in the shutdown group are shut down sequentially after the calculation is completed. Otherwise, the load devices in the shutdown group are shut down sequentially after the shutdown time point is reached.

[0080] A verification time point can be set 2 minutes before the shutdown time. Upon reaching the verification time point, the system calculates the offset of the remaining power supply duration of the high-risk power source between the power outage time and the verification time point. A larger offset indicates a more non-linear decrease in the remaining power supply duration. Strictly shutting down load devices according to the shutdown time point may pose the following risks: the high-risk power source may be taken offline before the load devices are fully shut down, potentially causing excessive load on the power supply side initially and power outages before the load devices have completely shut down later. Therefore, if the offset is greater than the first threshold, the load devices in the shutdown group should be shut down directly at the verification time point; otherwise, the load devices should be shut down only upon reaching the actual shutdown time point.

[0081] This embodiment calculates the offset and volatility using the following steps:

[0082] Establish a coordinate system with the remaining power as the horizontal axis and the remaining power supply duration as the vertical axis. Based on the first function, plot the curve of the remaining power supply duration of the high-risk power source as the remaining power. Locate the third time point when the remaining power is obtained. Plot multiple coordinate points with the remaining power at the third time point as the horizontal axis and the remaining power supply duration as the vertical axis. Calculate the straight-line distance between each coordinate point and the curve and use the sum of the straight-line distances as the offset.

[0083] The following reference Figure 2 The method for calculating the offset is introduced. First, a coordinate system is established, where the horizontal axis represents the remaining power of the high-risk power source, and the vertical axis represents the remaining power supply duration. As described earlier, the first function relating the two has been obtained. Then, based on this first function, a curve showing the relationship between the remaining power and the remaining power supply duration is plotted. Figure 2 As shown by L in the diagram. Before the verification time point, the remaining power supply duration is calculated each time the remaining power is obtained. Here, the time point when the remaining power is obtained is defined as the third time point. If the remaining power is obtained 10 times before the verification time point, there will be 10 third time points, which will be plotted as 10 coordinate points in the coordinate system. The straight-line distance from each coordinate point to the change curve is obtained, for example, Figure 2 The straight-line distance between the coordinate point r and the curve is d. Finally, the straight-line distances obtained from each coordinate point are summed to obtain the offset.

[0084] This application also provides an intelligent management system for an uninterruptible power supply (UPS), which implements the above-described method. The system includes:

[0085] The data acquisition module establishes a power supply network, which includes a power supply end and a load end. The front-end module updates the actual power of the load end at the first time interval. When the power supply end starts supplying power, the load fluctuation rate at the time of power outage is obtained based on the prediction model. The actual power is then corrected based on the load fluctuation rate to obtain the required power.

[0086] The startup module starts at least one uninterruptible power supply based on the required power to ensure that the load utilization of the power supply is within the standard range.

[0087] The monitoring module defines the uninterruptible power supply that is started as the first power supply. Every second time interval, it acquires the remaining power and power decay rate of the first power supply and identifies high-risk power supplies based on the remaining power and power decay rate.

[0088] The adjustment module will connect an idle uninterruptible power supply (UPS) to replace the high-risk power supply if there is one available at the power supply end. If there is no available UPS at the power supply end, the load devices at the load end will be shut down sequentially.

[0089] After the mains power is restored, the charging module determines the charging priority based on the battery health model and charges the uninterruptible power supply in the power supply end according to the charging priority.

[0090] This application also provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the method described above.

[0091] It should be understood that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments have been described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0092] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An intelligent management method for uninterruptible power supplies, characterized in that, include: Establish a power supply network, which includes the power supply end and the load end; The actual power of the load is updated at the first interval. When the power supply starts, the load fluctuation rate at the time of power outage is obtained based on the prediction model. The actual power is corrected based on the load fluctuation rate to obtain the required power. Specifically, the load fluctuation rate is set to 1% to 100% and divided into 5 ranges, each range corresponding to a correction coefficient. Then, the correction coefficient is multiplied by the actual power to obtain the required power. Start at least one uninterruptible power supply based on the required power to ensure that the load utilization at the power supply end is within the standard range; Define the uninterruptible power supply that is started as the first power supply. At every second time interval, obtain the remaining power and power decay rate of the first power supply. Based on the remaining power and power decay rate, identify the high-risk power supply. If there is an idle uninterruptible power supply (UPS) at the power supply end, the idle UPS is brought online to replace the high-risk power supply. If there is no idle UPS at the power supply end, the load devices at the load end are turned off in sequence. Specifically, the remaining power supply duration of each first power supply is calculated based on the remaining power and power decay rate. After a first number of second durations, the first number of remaining power supply durations are organized into a dataset corresponding to each first power supply. A first function about the remaining power supply duration is generated based on the dataset. The offline time point of each first power supply is predicted based on the first function. Load change data of the power supply end is generated based on the offline time point. Set a critical value, obtain the power consumption of each load device every third time interval, determine the shutdown time point based on load change data and power consumption, and shut down the load devices in sequence after the shutdown time point so that the load utilization rate of the power supply end is lower than the critical value. After the mains power is restored, the charging priority is determined based on the battery health model, and the battery is charged to the uninterruptible power supply at the power supply end based on the charging priority.

2. The intelligent management method for uninterruptible power supplies according to claim 1, characterized in that, Determining the shutdown time includes the following steps: The system identifies the first time point when the load utilization exceeds the critical value, assigns an importance level to each load device, defines load devices that are still running before the first time point as devices to be shut down, selects multiple device groups from the devices to be shut down, each device group must include at least one load device, shuts down all load devices in a device group to bring the load utilization below the critical value, selects a shutdown group from the device groups, selects an action time point before the first time point, uses the action time point as the shutdown time point of the shutdown group, and shuts down the load devices included in the shutdown group in sequence after the shutdown time point is reached. After determining the shutdown group corresponding to the first time point, predict the second time point when the load utilization exceeds the critical value, and determine the shutdown group corresponding to the second time point and the corresponding action time point, until all load devices are completely shut down or the mains power is restored.

3. The intelligent management method for uninterruptible power supplies according to claim 2, characterized in that, Selecting the shutdown group in the device group includes the following steps: The average grade of each equipment group is calculated based on the importance level. The ideal shutdown time of each load device in the equipment group is obtained. The ideal shutdown time is the time taken for the load device to completely shut down after receiving the shutdown command. The ideal shutdown times in the equipment group are sorted from largest to smallest. The difference between adjacent ideal shutdown times is obtained. The evaluation value is obtained based on the average grade and the difference. The equipment group with the highest evaluation value is the shutdown group.

4. The intelligent management method for uninterruptible power supplies according to claim 3, characterized in that, The steps for sequentially shutting down the load devices included in the shutdown group after the shutdown time point are reached are as follows: Before the shutdown time point, a verification time point is set. After the verification time point is reached, the offset of the remaining power supply time of the high-risk power source before the verification time point is calculated. If the offset is greater than the first threshold, the load devices in the shutdown group are shut down sequentially after the calculation is completed. Otherwise, the load devices in the shutdown group are shut down sequentially after the shutdown time point is reached.

5. The intelligent management method for uninterruptible power supplies according to claim 4, characterized in that, Calculating offset and volatility involves the following steps: Establish a coordinate system with the remaining power as the horizontal axis and the remaining power supply duration as the vertical axis. Based on the first function, plot the curve of the remaining power supply duration of the high-risk power source as the remaining power. Locate the third time point when the remaining power is obtained. Plot multiple coordinate points with the remaining power at the third time point as the horizontal axis and the remaining power supply duration as the vertical axis. Calculate the straight-line distance between each coordinate point and the curve and use the sum of the straight-line distances as the offset.

6. The intelligent management method for uninterruptible power supplies according to claim 1, characterized in that, Starting at least one uninterruptible power supply based on power demand includes the following steps: Obtain the number of charging and discharging cycles for each uninterruptible power supply (UPS), calculate the average number of charging and discharging cycles, sort the UPSs from smallest to largest based on the average number of cycles, and select the top N UPSs as the UPS to start based on the required power during a power outage.

7. The intelligent management method for uninterruptible power supplies according to claim 1, characterized in that, The prediction model is an LSTM time series prediction model.

8. An intelligent management system for an uninterruptible power supply (UPS), used to implement the intelligent management method for the UPS as described in any one of claims 1-7, characterized in that, include: The data acquisition module establishes a power supply network, which includes a power supply end and a load end. The front-end module updates the actual power of the load end at the first time interval. When the power supply end starts supplying power, the load fluctuation rate at the time of power outage is obtained based on the prediction model. The actual power is corrected based on the load fluctuation rate to obtain the required power. Specifically, the load fluctuation rate is set to 1% to 100% and divided into 5 ranges, each range corresponding to a correction coefficient. Then, the correction coefficient is multiplied by the actual power to obtain the required power. The startup module starts at least one uninterruptible power supply (UPS) based on the required power to ensure that the load utilization rate of the power supply is within the standard range; the monitoring module defines the started UPS as the first power supply, acquires the remaining power and power decay rate of the first power supply at second intervals, and identifies high-risk power supplies based on the remaining power and power decay rate. The adjustment module, if there is an idle uninterruptible power supply (UPS) in the power supply end, will connect the idle UPS to replace the high-risk power supply. If there is no idle UPS in the power supply end, the load devices in the load end will be shut down in sequence. Specifically, the remaining power supply duration of each first power supply is calculated based on the remaining power and power decay rate. After a first number of second durations, the first number of remaining power supply durations are organized into a dataset corresponding to each first power supply. A first function about the remaining power supply duration is generated based on the dataset. The offline time point of each first power supply is predicted based on the first function. Load change data of the power supply end is generated based on the offline time point. Set a critical value, obtain the power consumption of each load device every third time interval, determine the shutdown time point based on load change data and power consumption, and shut down the load devices in sequence after the shutdown time point so that the load utilization rate of the power supply end is lower than the critical value. After the mains power is restored, the charging module determines the charging priority based on the battery health model and charges the uninterruptible power supply in the power supply end according to the charging priority.

9. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the intelligent management method for uninterruptible power supply as described in any one of claims 1-7.

Images