User side energy storage operation adjusting method based on load power prediction

Abstract

The invention relates to the technical field of electrochemical energy storage application, and discloses a load power prediction-based user side energy storage operation regulation method, which comprises the following steps of: acquiring load power and energy storage power in a sampling period, calculating a load prediction value at the end of a next period according to the initial load power and the end load power in the period, establishing a historical load extreme value database, and executing an asymmetric condition updating rule, namely, performing online correction on the database according to different comparison conditions in a charging mode and a discharging mode; an adjusting instruction is generated according to the capacity constraint of the transformer in the charging mode, and an adjusting instruction is generated according to the anti-countercurrent threshold constraint in the discharging mode. Preventive measures for two different operation risks of charging overload and discharging countercurrent are unified in a continuous control process with online self-learning capability according to system operation state switching adjustment reference and boundary conditions, and the economical efficiency and the safety of operation are both considered.

Description

Technical Field

[0001] This invention relates to a user-side energy storage operation regulation method based on load power prediction, belonging to the field of electrochemical energy storage application technology. Background Technology

[0002] Currently, configuring energy storage systems for peak shaving and valley filling to manage electricity costs has become a common technical approach. By utilizing the peak-valley electricity price difference of the power grid, charging during off-peak hours and discharging during peak hours, users can reduce their electricity expenses. This model has theoretical economic value and has been widely deployed and applied.

[0003] However, when this energy storage operation strategy based on the ideal electricity price model is placed in the real and ever-changing load environment on the user side, a long-standing contradiction emerges: there is a trade-off between the economic benefits of system operation and its safety. The fluctuation of user-side load power is instantaneous and unpredictable. When the energy storage system charges during off-peak hours, its charging power will be superimposed on the user's own load power, which will bring additional load pressure to the transformer. If not controlled, it will easily lead to transformer overload, thereby triggering high demand charges and increasing operating costs. During peak hours, if the user load suddenly drops significantly, the energy storage system may feed power back to the grid. Once this happens, it will not only trigger the company's power outage protection and cause production interruption, but also bring safety hazards and economic losses.

[0004] To avoid the aforementioned risks, a common improvement approach in engineering practice is to pre-set a fixed, conservative safety boundary for energy storage operation. However, this is essentially a compromise at the expense of economic efficiency and does not solve the problem. Another seemingly more direct approach is to focus on improving the accuracy of load power prediction, hoping to guide the charging and discharging of energy storage through more accurate predictions. However, this approach also faces difficulties in user-side scenarios. On the one hand, complex prediction models place high demands on the computing power and cost of energy storage controllers. On the other hand, and more importantly, no prediction model can completely avoid sudden changes and random disturbances in the load. A model that relies on prediction accuracy... The safety of energy storage systems is at risk when facing low-probability extreme operating conditions. Specifically, existing technologies have the following shortcomings: 1. Energy storage systems face completely different safety risks in charging and discharging operations, but their control strategies lack corresponding asymmetric risk management mechanisms; 2. The system's safety boundaries are usually set as a set of static parameters, which cannot dynamically adapt to long-term changes in user electricity consumption behavior, leading to a decrease in effectiveness over time; 3. The pursuit of operational economy and the guarantee of operational safety are separated into two mutually restrictive goals, lacking a control architecture that can inherently unify the two. Therefore, how to construct an energy storage operation regulation method that can proactively adjust operating power based on future load trend judgments, while also ensuring that its safety boundaries are strictly constrained by historical operating experience and possess dynamic self-correction capabilities, thereby integrating the optimization of operational efficiency and the guarantee of operational safety into a unified closed-loop control process, becomes the technical problem to be solved by this invention. Summary of the Invention

[0005] This invention provides a user-side energy storage operation regulation method based on load power prediction. Its main purpose is to solve the problem that the existing energy storage control methods have rigid safety boundaries and lack of specificity for risk states, which leads to a mutual constraint between operational economy and safety.

[0006] To achieve the above objectives, this invention provides a user-side energy storage operation regulation method based on load power prediction. This method establishes a set of online self-correcting operation boundary rules, and includes:

[0007] Within a sampling period, the starting load power, ending load power, and energy storage power of that sampling period are collected.

[0008] Based on the starting load power and the ending load power, calculate the load forecast value at the end of the next cycle;

[0009] Establish a historical load extreme value database and perform an asymmetric condition update step: if the current operating mode is charging and the load prediction value is greater than the corresponding historical maximum value in the historical load extreme value database, then update the load prediction value to the new historical maximum value; and if the current operating mode is discharging and the load prediction value is less than the corresponding historical minimum value in the historical load extreme value database, then update the load prediction value to the new historical minimum value; if the current operating mode is charging, calculate the power gap, and generate an adjustment command to reduce the energy storage power based on the condition that the power gap is negative; if the current operating mode is discharging, calculate the predicted transformer net power, and generate an adjustment command to reduce the energy storage power based on the condition that the predicted transformer net power is less than a positive anti-reverse current threshold.

[0010] Preferably, the step of calculating the load forecast value at the end of the next cycle, in charging mode, specifically involves: obtaining a trend forecast value by adding the ending load power to a power change determined by the difference between the ending load power and the starting load power; comparing this trend forecast value with the historical maximum value in the historical load extreme value database; and taking the larger of the two values ​​as the load forecast value at the end of the next cycle. This calculation process is defined by the following formula: ,in, This is the load forecast value at the end of the next cycle. This is the historical maximum value in the historical load extreme value database. To end the load power, This represents the initial load power.

[0011] Preferably, the power deficit is calculated by performing the following computational steps: ,in, For power gap, This refers to the rated capacity of the transformer. For a safety factor less than 1, To end the load power, Energy storage capacity; the regulation command used to reduce energy storage capacity includes a value equal to the power deficit. The amount of power regulation.

[0012] Preferably, the load forecast value at the end of the next cycle is obtained by adding the ending load power to a power change calculated from the difference between the ending load power and the starting load power.

[0013] Preferably, the predicted net power of the transformer is obtained by subtracting the energy storage power from the predicted load value, and then subtracting the positive anti-reverse current threshold; the positive anti-reverse current threshold is a constant power value set according to the power grid safety regulations.

[0014] Preferably, the historical load extreme value database has a data structure that is a set with timestamps as index keys and load power extreme values ​​as storage values. When performing the asymmetric conditional update step, the old load power extreme values ​​indexed by the current timestamp are replaced with new load prediction values.

[0015] Preferred safety factor The value of is dynamically determined based on the transformer's health status assessment data, or it is a fixed value set according to the operation scheduling strategy.

[0016] Preferably, the adjustment command is generated by an energy storage controller, and the energy storage controller sends the adjustment command to an energy storage converter, which then performs the adjustment of the energy storage power.

[0017] Preferably, the execution of the asymmetric condition update step for calculating the load prediction value and the generation of adjustment instructions are a set of data processing and instruction generation operations executed sequentially by a controller at the end of each sampling period.

[0018] Compared with the prior art, the beneficial effects of the present invention are:

[0019] 1. This method establishes an asymmetric risk response mechanism. During the charging period of the energy storage system, the short-term trend prediction value of the load power is compared with the historical power extreme value, and the larger value is taken as the adjustment benchmark. Power control is carried out with the transformer capacity margin as the boundary. During the discharging period, the smaller value is taken as the adjustment benchmark, and power control is carried out with the grid zero power exchange point as the boundary. This operation mode of switching the adjustment benchmark and boundary conditions according to the system operating status makes the prevention measures against the two different types of operating risks, namely charging overload and discharging reverse current, unified in a continuous control process.

[0020] 2. This method inherently binds the maintenance of the historical load database with each power prediction and adjustment action. In each adjustment cycle, when the load power deduced from the real-time operating conditions exceeds the boundary recorded in the historical database, the database will automatically and instantly update itself with new extreme value data. This process enables the system's safety boundary to continuously track the long-term evolution of users' actual electricity consumption behavior, avoiding the problem of fixed parameters failing due to equipment additions or reductions or changes in production plans, and giving the entire adjustment system long-term self-adaptive capability.

[0021] 3. By coupling three different time-scale adjustment methods—short-term trend prediction, dynamic historical extreme value constraint, and fixed safety margin—a hierarchical operation adjustment structure is formed. Trend prediction serves to maximize energy storage benefits, dynamic historical extreme values ​​constitute an adaptive safety baseline, and the fixed safety margin provides ultimate absolute protection. The three work together to ensure that when the system is pursuing economical operation, the adjustment of its operating power is always limited to a dynamically shrinking and absolutely reliable safety domain, thus taking into account both the economy of operation and the safety of operation. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the implementation environment of the user-side energy storage operation regulation system of the present invention;

[0023] Figure 2 This is a curve showing the anti-backflow regulation effect of the present invention under the condition of sudden load drop;

[0024] Figure 3 This is a flowchart of the closed-loop control logic of the adaptive adjustment method of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] This invention provides a user-side energy storage operation regulation method based on load power prediction. This method is deployed in a user-side energy storage system consisting of a load meter and an energy storage meter connected to an energy storage controller, and one or more energy storage converters scheduled by the controller. It aims to address the potential conflict between the random fluctuations of user-side load and the preset charging and discharging strategies of the energy storage system. By establishing a set of online self-correcting operating boundary rules, it coordinates operational economy and safety within a closed-loop scheduling process. The method mainly includes steps such as acquiring power data within a preset sampling period, calculating the load prediction value at the end of the next period based on the acquired data, performing asymmetric condition updates to dynamically correct a historical load extreme value database, and generating and executing power regulation commands based on different safety constraints under both charging and discharging operating modes.

[0027] In a specific application scenario, this method is applied to a system with... The manufacturing plant for electrochemical energy storage systems uses a unit with a rated capacity for The transformer draws power from the grid. To mitigate the risk of transformer overload or power backflow caused by load power fluctuations due to production plan adjustments and the start-up and shutdown of large equipment, the energy storage controller in the energy storage system is configured to execute this regulation method. First, the basic parameters and data structure of this method need to be initialized and defined. One step is to establish a historical load extreme value database. The data structure of this database is set as a set with timestamps as index keys and the historical maximum load power and historical minimum load power corresponding to that timestamp as storage values. To ensure the validity of the initial data, the initialization of this database is completed through a deterministic calibration process: before the energy storage system is put into operation, the load meters are used to continuously measure the factory's load. The load power for each working day is recorded, with a sampling frequency of [frequency missing]. Subsequently, with Using minutes as a time unit, calculate the maximum and minimum load power within each time unit, and then associate these extreme values ​​with the corresponding timestamps, for example, Monday. They are stored together in the database, thus providing an initial operating boundary for each scheduling cycle based on actual historical operating conditions.

[0028] After the system is put into operation, the energy storage controller uses a fixed sampling period. This adjustment method is executed cyclically. The determination of this sampling period is based on frequency analysis of historical load data from users, aiming to balance the sensitivity of the adjustment with the stability of the system. For example, if the analysis shows that the main fluctuation period of the factory's load power is concentrated in... Instant Between seconds, the sampling period is... Can be set to The sampling frequency is set to a specific value per second to ensure the controller can capture significant power change trends while avoiding data redundancy caused by excessively high sampling frequencies; in each sampling cycle... At the start and end times, the energy storage controller collects the starting load power through the load meter. With the power of the end load Simultaneously, the current energy storage power is collected through the energy storage side meter. , here The value is positive in charging mode and negative in discharging mode. It should be noted that, to ensure the reliability of the data link, industrial Ethernet is used as the preferred communication path between the controller and the meter, and a system based on... The network's wireless communication link serves as a backup path. If the controller fails to receive data from the preferred path within a preset time, it will automatically switch to the backup path, thereby ensuring the stable acquisition of core input data.

[0029] After acquiring the power data, the system enters an asymmetric condition update phase based on load power prediction and historical extreme value database. This phase is introduced because static safety boundaries cannot adapt to long-term changes in user electricity consumption behavior. Therefore, this method establishes a mechanism that dynamically binds short-term trend prediction with long-term historical experience. First, the controller updates the load power based on the initial load power... With the power of the end load One way to calculate the load forecast at the end of the next cycle is by measuring the power output of the final load. With a power change within that period The predicted load value is obtained by adding the determined trend vectors. In other words, the system performs an asymmetric condition update step based on the current energy storage operation mode. If the current operation mode is charging, the controller will compare the predicted load value with the historical maximum value indexed from the current timestamp in the historical load extreme value database. The two values ​​are compared, and the larger value is taken as the load power value referenced for the next cycle adjustment. The calculation process can be defined by the following formula: Furthermore, if the predicted load value is greater than the historical maximum value This means a new historical load peak has appeared. The controller will use this load prediction value to replace the old historical maximum value in the database. Correspondingly, if the current operating mode is discharge, the system will focus on the risk of load decline. In this case, the load prediction value will be compared with the corresponding historical minimum value in the database. The system compares the predicted load value with the historical minimum value, and the controller only updates the predicted value to the new historical minimum value when the predicted load value is less than the historical minimum value. This asymmetric setting associates the database update behavior with the different sources of risk in the two modes, allowing the system's security boundary to track the evolution of user electricity consumption behavior.

[0030] After completing load forecasting and database updates, the controller enters the differentiated power regulation command generation stage based on the current operating mode. When the system is in charging mode, the main operational risk is transformer overload. To mitigate this risk, the system generates regulation commands by calculating the power gap. This is achieved by performing the following computational steps: ,in, This refers to the rated capacity of the transformer. For a less than The safety factor can be dynamically determined based on the transformer's health status assessment data, or set to a fixed value based on the operation and scheduling strategy, for example, set to 1. ; The load power at the end of the sampling period, i.e. ; The current energy storage charging power; if the calculated power gap A positive value or zero indicates that the transformer capacity has a margin, and the energy storage system can continue to operate at the current power; if A negative value indicates an overload risk. In this case, the controller will generate an adjustment command to reduce the energy storage power. This command includes a value equal to the power deficit. The power regulation amount is determined, and the command is sent to the energy storage converter for execution; for example, in , Under the condition that the current load power is terminated for Energy storage charging power for The total load of the transformer is greater than the safety threshold The calculated power gap Therefore, the controller will send a command to the energy storage converter to reduce the charging power. to This keeps the transformer load within safe limits.

[0031] When the system switches to discharge mode, the main operational risk becomes reverse power transmission to the grid caused by a sudden drop in user load. To address this risk, the benchmark and boundary conditions of the regulation mechanism also switch accordingly. At this time, the system calculates and predicts the net power of the transformer to determine whether a regulation command needs to be generated. First, the controller calculates the load forecast value at the end of the next cycle using the aforementioned method. Subsequently, load forecast values ​​were used. Subtract the discharge power of the energy storage system (denoted here as) Then subtract a constant positive anti-reverse current threshold set according to power grid safety regulations. ,For example This allows the system to obtain the predicted net power of the transformer. If the predicted net power is greater than or equal to zero, it indicates that the system is still in a state of net power purchase from the grid and is operating safely. Conversely, if the value is less than zero, it indicates a risk of power backflow, and the controller will then generate an adjustment command to reduce the energy storage discharge power. For example, if the predicted load value... for Current discharge power of the energy storage system for Anti-backflow threshold for The predicted net power of the transformer is If this value is less than zero, there is a risk of backfeeding; therefore, the controller will instruct the energy storage converter to reduce the discharge power, and the new target discharge power value will be set to no higher than [the specified value]. and The difference, i.e., not higher than This maintains the power state at the transformer cut-off point within a safe area that appears as net load from the grid side. The entire method involves the execution of the load prediction calculation asymmetric condition update step and the generation of adjustment commands. These are a set of data processing and command generation operations executed sequentially by the energy storage controller at the end of each sampling period, thus forming a continuous, closed-loop adjustment process with mode-adaptive capabilities.

[0032] Example 1: This example is a specific application of a user-side energy storage operation regulation method based on load power prediction in a specific industrial scenario. All technical steps and parameter calibrations follow the procedures disclosed in the specific implementation method. The scenario is a large data center, which uses a rated capacity... for A dedicated transformer is connected to the regional power grid and a set of The user-side energy storage system is used for electricity cost management, and some of the server clusters in the data center handle online transaction processing, which requires a high degree of power supply continuity.

[0033] During a nighttime period when grid off-peak prices occur, the energy storage system is in charging mode, to... Operating at the required power, the data center's basic load remains stable. At this time, the total load of the transformer is At a safe level Set as Within the operating range; during this period, maintenance personnel began powering on a group of newly deployed server racks in batches, causing the total load on the data center to increase slowly over an hour; during this process, the energy storage controller... The sampling period is seconds. Continuous operation, collecting the power of the terminated load in each cycle. All are higher than the starting load power Based on this, the system calculates a continuously increasing power change trend. Since the system is in charging mode, its asymmetric condition update step is activated. The controller compares the increased load prediction value calculated in each round with the historical maximum value corresponding to the timestamp in the historical load extreme value database. By comparison, as the on-site load continues to rise, the predicted load value begins to exceed the historical extreme value for the same period, triggering an online update of the database. This causes the system's security boundary to dynamically rise in accordance with changes in on-site operating conditions. In other words, when the total data center load grows to... At that time, the charging power of the energy storage system is still [missing information]. Total load reached At this point, the load forecast value for the next cycle calculated by the controller has reached [a certain value] due to the superposition of the growth trend and the updated historical extreme values. According to the power gap calculation formula ,Right now The system determines that if the current charging power is maintained, the transformer load will exceed [a certain threshold] in the next cycle. Upon reaching the safety threshold, the controller immediately generates and issues a power adjustment command before the actual overload occurs, reducing the energy storage charging power. to Subsequently, as the load further increases, the charging power is gradually reduced, so that the total load of the transformer is always kept below the safe threshold, avoiding additional operating costs caused by demand charges.

[0034] In the afternoon of another peak electricity price period, the energy storage system switched to discharge mode to... The power is supplied to the data center to reduce the power draw from the grid during peak hours, when the data center load is... The net power of the transformer is At a certain moment, the regional power grid experienced a prolonged [damage / disruption] due to an upstream line fault. A voltage dip caused some server power modules within the data center to momentarily disconnect from the load as a self-protection mechanism, resulting in a fluctuation in the total load of the data center within a few seconds. sudden drop The deep valley; for an energy storage system using fixed power discharge, when the load suddenly drops... At that time, its The discharge power will exceed the load demand, resulting in... The power is fed back to the grid, which may trigger the backfeed protection of the main substation; when applying this method, the operation of the regulation mechanism is as follows: in the first... Within a second sampling period, the controller detected a significant negative power change trend, and the system immediately calculated the load forecast for the next period. Far below the current discharge power Therefore, the predicted net power of the transformer The calculation result is a value much smaller than the positive backflow prevention threshold. (set to) A negative value for ) indicates that power backflow is imminent; before this risk materializes, the controller has generated an adjustment command based on the prediction, instructing the energy storage converter to reduce the discharge power from sharply reduced to a new level not higher than The safety target value is achieved by completing this series of actions within a single control cycle. Its response speed is faster than the action delay of the backfeed protection relay in the main substation of the data center, thereby suppressing the occurrence of power backfeed at the source and ensuring the continuity of power supply to the data center. This method asymmetrically couples short-term trend prediction capability with long-term historical boundary memory and adjusts the benchmark according to the operation mode, enabling it to make proactive power adjustments under different risk scenarios. Ultimately, without sacrificing economic benefits, it improves the system's operational stability under complex processes.

[0035] Example 2: To objectively verify the actual operating effect of the regulation method of the present invention in dealing with high-frequency and large-amplitude load fluctuations, a hardware-in-the-loop simulation test platform was built and comparative tests were performed. The aim was to quantitatively verify the performance difference between the present method and the traditional fixed-power energy storage control strategy in maintaining the stability of key grid connection point parameters. The test platform consists of a real-time digital simulator and an actual energy storage controller running the regulation method of the present invention. The simulator deploys a grid model corresponding to the application scenario of the specific implementation, including a rated capacity for A transformer model and a The energy storage system model was constructed. The load data used in the experiment originated from a set of actual power curves collected from the electric arc furnace workshop of a special steel plant over a period of 24 hours with a resolution of 1 second, used to simulate the industrial application environment. Two operating groups were set up: an experimental group using the adjustment method of this invention and a control group using the traditional fixed power control method. To ensure the effectiveness of the comparison, both experimental groups shared the same hardware platform and load curves. Key parameters were set as follows: the target values ​​for the initial charging power of the energy storage system during off-peak hours and the initial discharging power during peak hours were both set to [value missing]. Sampling period of the experimental group Set as The value of seconds is determined by balancing the speed of the adjustment response with the data processing load. Through spectral analysis of the input load data, its main energy is concentrated in... Therefore A sampling period of one second is sufficient to capture critical fluctuation information; transformer safety factor Set as That is, the maximum allowed operating load is Positive backflow prevention threshold Set as The control logic for the control group is to use [the following] during the trough period: Constant power charging, during peak hours Constant power discharge, only when the transformer load is detected to exceed Or net power less than Only then will a protective shutdown be triggered.

[0036] In a typical impact load scenario during off-peak charging hours, the system states of the two groups exhibit different characteristics. At any given time, the workshop's basic load is Both energy storage systems are based on Power charging, total transformer load is The load factor was 87.5%; to During this period, the load increased dramatically due to changes in production conditions. Time to reach At this time, the energy storage system of the control group still maintained The charging power caused the total load on the transformer to reach [a certain value]. The load factor climbed to 94.0%, exceeding its... The safe operating boundary; in contrast, the controller of the test group is at... The increasing load trend was detected within the sampling period, and the predicted load for the next period was calculated to cause the total load to exceed the limit. Therefore, its power deficit... The calculation result was a negative value, and the controller then... to An adjustment command was executed during the cycle, reducing the charging power from Actively downgraded to This makes in At any given time, even though the external load is the same However, the total load on the transformer is suppressed to The load rate remained at The data from this process show that the experimental group's adjustment method, through its predictive mechanism, achieved proactive avoidance of overload risks.

[0037] During peak discharge periods, when the load suddenly drops significantly due to production breaks, different results were observed between the two control methods. sudden drop Under the operating conditions described above, the energy storage system in the control group, due to the lag in its control logic, experienced a decrease in load as the load decreased. It lasted for several seconds afterward. Power discharge caused a discharge at the transformer stop. The reverse power supply triggered the grid backflow protection in the simulation environment; while the controller of the test group predicted in the first sampling period of the load drop that the net power would soon be less than the set positive anti-reverse current threshold. And based on the load forecast value With threshold The difference will reduce the discharge power from Quickly downgraded to This ensures that the transformer's net power is always maintained at In the above, no power backflow occurred. The adjustment logic in the discharge mode maintains a minimum positive power flow as the boundary, thereby actively avoiding the formation of reverse flow. The test results show that, compared with the traditional control method that uses fixed boundary conditions, the adjustment method based on load power prediction of this invention can maintain key grid-connected parameters such as transformer load rate and net power at the cut-off point within the preset safe range under high dynamic load scenarios on the user side through online prediction and asymmetric adjustment.

[0038] Example 3: This example combines Figures 1 to 3 This document describes a user-side energy storage operation regulation method based on load power prediction, as follows: Figure 1 As shown, the power supply to bus section I is provided by the upstream power grid via a transformer. Various loads and energy storage systems, which serve as regulating units, are connected to the bus. In the figure, the energy storage controller is the core. It collects electrical data from the load meter C1 on the load side and the energy storage meter C2 on the energy storage side through sampling lines. Based on its internal calculation and decision-making logic, it sends commands to the power conversion system PCS of one or more energy storage systems through control lines to regulate the charging and discharging behavior between the system and the battery, thereby achieving the operation regulation of the user-side power grid.

[0039] like Figure 2 As shown, the horizontal axis represents time in seconds (s), and the vertical axis represents power in kilowatts (kW). In this operating condition, the load power (kW) curve experienced a sharp drop from approximately 7500kW to below 1000kW between 10s and 25s. To prevent power backflow, the energy storage discharge power (kW) curve, under the control of the controller, rapidly decreased in line with the decreasing trend of the load power. The transformer net power (kW) curve, which represents the power difference between the two, was always maintained above a constant positive anti-reverse current threshold curve, thus ensuring the safe operation of the system.

[0040] like Figure 3As shown, the process begins with the power data acquisition step, which obtains the system status within a sampling period. It then proceeds to the step of calculating the load prediction value for the next period. This step combines short-term trends and historical experience, and interacts with a historical load extreme value database to perform asymmetric condition updates. This enables online self-correction of the dynamic safety boundary provided by the database. Next, based on the result of judging the operating mode, the system enters either the charging mode adjustment or the discharging mode adjustment sub-process. The former uses transformer capacity constraints to prevent overload risks, while the latter uses anti-reverse current threshold constraints to prevent backfeed risks. If a risk is determined in either mode, an adjustment command is generated and the power adjustment amount is calculated. Finally, the energy storage converter executes the adjustment, adjusting the energy storage power to the safe target value. Through the feedback path shown by the dashed arrow, the entire process enters the next sampling period, forming a continuous closed-loop adjustment process.

[0041] Example 4: This example illustrates the systematic calibration procedure for the core operating parameters of the adjustment method when it is first deployed at a new user-side site. This procedure is used to determine the sampling period of the energy storage controller based on the user's specific electrical load characteristics and equipment status. With transformer safety factor In a calibration scenario, the energy storage system was installed in a daily chemical product manufacturing company. Its main load consisted of multiple mixer filling production lines and a constant-temperature reaction vessel, exhibiting a load characteristic of both periodic and random variations. Before the system was put into operation, a sampling cycle was first executed. The process involves taking as input a historical load power data sequence collected continuously from the enterprise's electricity meters for one month, with a resolution of 1 second. The first step is to perform a Fast Fourier Transform on the data sequence to obtain a spectral distribution map of the load power fluctuations. The second step is to integrate the spectrum from low to high frequencies to determine a frequency point that covers 95% of the total spectral energy, denoted as the characteristic frequency. The third step, based on the Nyquist sampling theorem, is to adjust the system's sampling frequency to capture these fluctuations without distortion. Set to no less than twice the characteristic frequency, i.e. Ultimately, the sampling period It is determined to be the reciprocal of the sampling frequency, i.e. Applying this process to the company's load data, the characteristic frequencies derived from its spectrum analysis... for Therefore, the sampling frequency Set to be no less than To optimize computing resources while meeting technical requirements, the sampling period The final calibration value was determined to be The time is specified in seconds, and this parameter is written into the energy storage controller as the time base for all subsequent control logic.

[0042] After determining the sampling period Then, the transformer safety factor was determined. During the calibration phase, the setting of this parameter aims to correlate the transformer's theoretical capacity with its current actual health condition, thereby establishing a dynamic safety boundary that more closely reflects the actual load-bearing capacity of the equipment. This calibration procedure integrates multiple measurable data reflecting the transformer's health status and calculates it using a weighted loss reduction model. Value; the baseline value of this model. Set as This represents the proportion of safe capacity available to a transformer under ideal conditions, and is subsequently adjusted based on the following quantifiable deductions: First, transformer service life deductions. For transformers with a service life exceeding 15 years, a reduction value is set. Second, operating temperature loss item. By reading the top-level oil temperature data of the transformer oil temperature online monitoring system If the historical peak temperature exceeds Set its reduction value to Third, internal fault gas loss item According to the latest dissolved gas analysis report in oil, if the key gas characteristic of severe overheating or discharge failure is acetylene (… The concentration of ) is greater than Then set its reduction value to The final safety factor The following formula can be used to calculate: This calibration procedure was applied to a machine that had been in operation for 18 years and had a historical record for the highest oil temperature. Furthermore, the DGA report indicates that for transformers with zero acetylene concentration, the impairment items are determined as follows: , and Its final safety factor Calculated as By executing the above two procedures, the core parameters of the energy storage controller are ensured. and The settings are all based on the analysis and calculation of objective data from a specific site. This series of operations constitutes a standardized process for the system during the on-site commissioning phase. The final output is an energy storage control system that has completed the initial parameter optimization for the user, laying the foundation for the stable operation of subsequent adjustment methods.

[0043] Example 5: This example describes the built-in operation protection procedures of the regulation method to cope with the loss of critical sensor data or extreme boundary conditions. The core input on which this regulation method relies is the power data of the load meter. In order to cope with the condition of data interruption, the energy storage controller integrates a set of hierarchical fault handling logic. When the controller fails to receive load power data from the preferred communication path within a sampling period, it will use the valid data of the previous period and historical trends for a short-term extrapolation to maintain the continuity of control. If data cannot be obtained through the preferred path within three consecutive sampling periods, the system will automatically switch to the backup wireless communication link for data acquisition. If the backup link still cannot establish effective communication within 60 seconds after activation, the system determines that the load sensor or its link has suffered a continuous failure. At this time, in order to avoid continuing to adjust the power without real-time load reference, the controller will automatically execute a safety lockout procedure, that is, regardless of whether it is currently in charging or discharging mode, the power of the energy storage system will be linearly reduced to zero within 30 seconds, and an alarm will be issued to the background monitoring system until the sensor data is recovered.

[0044] The basic logic of this adjustment method also applies to extreme boundary situations where the user-side load experiences a sudden and complete disconnection. In a specific operating condition, when the energy storage system... During power discharge, the total load on the user side is reduced due to the tripping of its main incoming circuit breaker. It instantly dropped to near zero; at the end of the first sampling period immediately following, the controller collected the power of the terminated load. If the value is zero, the system immediately executes its standard discharge regulation judgment procedure, which involves calculating the predicted net power of the transformer and comparing it with the positive anti-reverse current threshold. Comparison; due to load forecast values Under this operating condition, the calculated predicted net power of the transformer also approaches zero. Approximately The value is less than a value that is set to Positive anti-backflow threshold Therefore, the system's adjustment logic determines that there is a risk of power backfeeding and automatically generates an adjustment command to lower the target discharge power of the energy storage converter to no higher than [the target value]. That is, it is close to the level of zero kilowatts. The entire response process is completed within a single control cycle, thus utilizing its inherent regulation mechanism to keep the system within safe operating boundaries even under extreme conditions.

[0045] Example 6: To ensure that the operation of the adjustment method matches the power grid interface protection characteristics of a specific site, after completing the initial parameter calibration, a field parameter fine-tuning procedure needs to be executed. This procedure first involves adjusting the positive anti-reverse current threshold. The settings include: during a low-load period at night, shutting down non-essential production loads on the user side, and retaining only a small amount of data. Stable base load Subsequently, the energy storage system was placed in discharge mode, and its discharge power was adjusted via background commands. The rate of increase is slow and linear, starting from zero. Simultaneously, a power quality analyzer with an accuracy class of 0.2S is used to monitor the net exchange power at the high-voltage side of the transformer. The instant the net exchange power at the switch point first flips from a positive to a negative value, the discharge power value of the energy storage system at that moment is immediately recorded and denoted as the reverse flow point power. After three repeated tests, the measured values ​​were... Stable at Ultimately, to allow for a safety margin, the positive backflow prevention threshold was set. Set as ,in, For one The fixed margin value is thus obtained. The final setting value is .

[0046] After completion After setting the parameters, the robustness of the regulation method under nonlinear oscillating load impacts was verified; in the energy storage system... During power discharge, a power unit on the user side is started. The large air compressor generates an oscillation process lasting several seconds with multiple power fluctuations during startup. This causes the load forecast based on linear extrapolation to deviate from the actual load value within a single sampling period. During this process, by monitoring the internal data of the energy storage controller, it was observed that although the load forecast... The load fluctuates around the actual load, but the energy storage power adjustment command is also adjusted accordingly in each sampling cycle. The system performs rapid, small-amplitude bidirectional corrections. When the predicted value is too high, the discharge power is slightly reduced, and when the predicted value in the next cycle is too low due to the change in the base value in the previous moment, the discharge power is slightly increased. The response of the entire system does not show continuous unidirectional divergence or oscillating amplification of power regulation due to short-term mismatch of the prediction model. The net power at the transformer switch is always maintained above the set positive anti-reverse current threshold. This result shows that the overall stability of this regulation method comes from the combined effect of its high-frequency closed-loop feedback correction mechanism and the boundary constraints provided by the historical extreme value database.

[0047] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0048] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A user-side energy storage operation regulation method based on load power prediction, characterized in that, The method includes: Within a sampling period, the starting load power, ending load power, and energy storage power of that sampling period are collected. Based on the starting load power and the ending load power, calculate the load forecast value at the end of the next cycle; Establish a historical load extreme value database and perform an asymmetric condition update step: if the current operating mode is charging and the load prediction value is greater than the corresponding historical maximum value in the historical load extreme value database, then update the load prediction value to the new historical maximum value; and if the current operating mode is discharging and the load prediction value is less than the corresponding historical minimum value in the historical load extreme value database, then update the load prediction value to the new historical minimum value; if the current operating mode is charging, calculate the power gap, and generate an adjustment command to reduce the energy storage power based on the condition that the power gap is negative; if the current operating mode is discharging, calculate the predicted transformer net power, and generate an adjustment command to reduce the energy storage power based on the condition that the predicted transformer net power is less than a positive anti-reverse current threshold.

2. The user-side energy storage operation regulation method based on load power prediction according to claim 1, characterized in that, The steps for calculating the load forecast at the end of the next cycle, in charging mode, are as follows: A trend forecast is obtained by adding the ending load power to a power change determined by the difference between the ending load power and the starting load power. This trend forecast is then compared with the historical maximum value in the historical load extreme value database, and the larger of the two values ​​is taken as the load forecast at the end of the next cycle. This calculation process is defined by the following formula: ,in, This is the load forecast value at the end of the next cycle. This is the historical maximum value in the historical load extreme value database. To end the load power, This represents the initial load power.

3. The user-side energy storage operation regulation method based on load power prediction according to claim 1, characterized in that, The power deficit is calculated by performing the following computational steps: ,in, For power gap, The rated capacity of the transformer. For a safety factor less than 1, To end the load power, Energy storage capacity; the regulation command used to reduce energy storage capacity includes a value equal to the power deficit. The amount of power regulation.

4. The user-side energy storage operation regulation method based on load power prediction according to claim 1, characterized in that, The predicted load value at the end of the next cycle is obtained by adding the power of the ending load to a power change calculated from the difference between the ending load power and the starting load power.

5. The user-side energy storage operation regulation method based on load power prediction according to claim 1, characterized in that, The predicted net power of the transformer is obtained by subtracting the energy storage power from the predicted load value, and then subtracting the positive reverse current prevention threshold; the positive reverse current prevention threshold is a constant power value set according to the power grid safety regulations.

6. The user-side energy storage operation regulation method based on load power prediction according to claim 1, characterized in that, The historical load extreme value database is a collection of data structures that uses timestamps as index keys and load power extreme values ​​as storage values. When performing the asymmetric conditional update step, the old load power extreme value indexed by the current timestamp is replaced with the new load prediction value.

7. The user-side energy storage operation regulation method based on load power prediction according to claim 3, characterized in that, Safety factor The value of is dynamically determined based on the transformer's health status assessment data, or it is a fixed value set according to the operation scheduling strategy.

8. The user-side energy storage operation regulation method based on load power prediction according to claim 1, characterized in that, The regulation command is generated by an energy storage controller, which then sends the regulation command to an energy storage converter, which performs the adjustment of the energy storage power.

9. The user-side energy storage operation regulation method based on load power prediction according to claim 1, characterized in that, The calculation of load forecast values, the execution of asymmetric condition update steps, and the generation of adjustment instructions are a set of data processing and instruction generation operations that are sequentially executed by a controller at the end of each sampling period.

Images