Method and system for energy storage device participating transformer demand control
By using a hybrid prediction model combining LSTM and an improved slip algorithm, along with a three-dimensional control matrix, accurate prediction and dynamic adjustment of energy storage devices in transformer demand control are achieved. This solves the problems of accuracy and applicability of demand control, reduces electricity costs, extends equipment life, and adapts to complex operating conditions.
Patent Information
- Application Number
- CN202610263937.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-03-05
- Publication Date
- 2026-05-29
AI Technical Summary
When existing energy storage devices are used in transformer demand control, they suffer from demand forecast lag, insufficient control accuracy, and limited applicability. They cannot adapt to complex operating conditions such as mixed loads and multi-transformer networks, resulting in equipment losses, capacity waste, and billing disputes.
A hybrid prediction model combining LSTM and an improved slip algorithm is adopted, and a three-dimensional control matrix of SOC-transformer load-voltage is constructed by combining dynamic demand calculation window and metering calibration factor. Through millisecond-level response and dynamic power adjustment, accurate demand prediction and control are achieved, and a multi-scenario adaptive extended control strategy is constructed.
It has achieved demand forecasting error control within 5%, reduced monthly maximum demand by 25%-40%, reduced basic electricity expenses, extended energy storage cycle life by 20%-30%, improved energy storage capacity utilization, adapted to stable operation under complex working conditions, and reduced operation and maintenance costs.
Smart Images
Figure CN122118819A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of transformer demand control, and particularly to a method and system for a storage device to participate in transformer demand control. Background Art
[0002] In the power system's power transmission and distribution links, as a core device, the stability of the operating load of a transformer is directly related to grid safety and power consumption economy. For industrial and commercial users and distribution network transformers that implement two-part electricity pricing, the maximum demand (the maximum average power value within a specified settlement period, usually with a 15-minute demand period) is a key indicator for basic electricity charge accounting. Short-term load peaks (such as motor and air compressor startups) are likely to push up the maximum demand, resulting in a sharp increase in the user's electricity cost.
[0003] Existing transformer demand control means have obvious limitations: traditional load shedding and production process adjustment methods are less flexible and are likely to affect normal production operations; reactive power compensation devices can only optimize the power factor and cannot effectively suppress active load peaks; on-load tap-changing transformers adjusting tap positions are likely to cause voltage instability, and there is insufficient adjustment margin when the load impact is large. With the development of energy storage technology, energy storage devices have become a new path for demand control due to their fast charge and discharge response capabilities. However, existing energy storage participation methods still have shortcomings: demand prediction lags behind load peak changes, and the control accuracy is insufficient; only relying on a single parameter of the energy storage remaining capacity (SOC) for regulation, without considering the transformer operating state and energy storage life characteristics, is likely to cause equipment loss or capacity waste; the control strategy has poor adaptability to the grid metering cycle, and the peak shaving effect is likely to cause billing disputes; and it is designed for single scenarios and cannot adapt to complex working conditions such as mixed loads and multi-transformer networking, with limited applicability.
[0004] Therefore, there is an urgent need for a method for a storage device to participate in transformer demand control that takes into account accuracy, coordination, compatibility, and scenario adaptability, solves the pain points of the existing technology, and achieves multiple goals of accurately suppressing the maximum demand, efficiently utilizing the energy storage, and stable operation of the transformer.
[0005] Therefore, it is necessary to provide a method and system for a storage device to participate in transformer demand control to solve the above technical problems. Summary of the Invention
[0006] The present invention provides a method and system for a storage device to participate in transformer demand control, which solves the problems mentioned in the above background art.
[0007] To solve the above technical problems, the method and system for a storage device to participate in transformer demand control provided by the present invention specifically include the following steps: Step 1: Start the control system and complete the self-test of each unit, synchronize the grid metering device to obtain the demand metering cycle and settlement cycle parameters; load the preset threshold and hybrid prediction model parameters, call the historical load and energy storage operation parameter sample library to complete the model preheating. The preset threshold includes the transformer safe demand threshold, the energy storage SOC safe range, the charging and discharging power limit, the starting power threshold and the voltage safe range of 95%-105% of the rated voltage. Step 2: Acquire the three-phase voltage and current signals of the transformer bus with a sampling period of ≤1s, convert them into real-time power data and transmit them to the control unit; adopt the LSTM and improved slip algorithm fusion model, combined with dynamic demand calculation window and metering calibration factor, to achieve accurate demand prediction consistent with power grid metering data; Step 3: Based on the predicted demand and real-time operating parameters of the energy storage, determine whether to activate energy storage regulation; if the preset threshold is not exceeded, instruct the energy storage to standby charge and replenish energy; if the threshold is exceeded, further verify the energy storage charging and discharging conditions; in case of failure, link auxiliary devices to ensure continuous demand control; if the parameters meet the standards, trigger the dynamic regulation strategy. Step 4: Construct a three-dimensional control matrix of SOC-transformer load-voltage, dynamically generate charging and discharging commands, adjust power and depth according to real-time load fluctuations, adapt to grid demand response commands, and stably control the average power within the demand cycle below the preset threshold. Step 5: Synchronize energy storage operation parameters, load data and control effects to the backend and grid metering devices in real time to ensure data traceability and measurable peak shaving effect; iteratively optimize control strategies and model parameters based on machine learning, review and adaptively adjust preset thresholds after the monthly settlement cycle to adapt to changes in load characteristics.
[0008] Preferably, in step two, the three-phase voltage and current electrical signals of the transformer bus are collected with a sampling period of no more than 1 second and converted into real-time power data. A hybrid prediction model combining LSTM and improved slip algorithm is adopted, and combined with dynamic demand calculation window and metering calibration factor to achieve accurate demand prediction, ensuring that the predicted demand is consistent with the grid metering data.
[0009] Preferably, in step two, the hybrid prediction model predicts the peak occurrence time, peak size, and duration of impact loads 0.5-1 minutes in advance, and corrects the prediction results in real time for stable loads using an improved slip algorithm, with the demand prediction error controlled within 5%. The dynamic demand calculation window is adaptively adjusted according to the load fluctuation frequency, with an adjustment range of 1-15 minutes. The window is narrowed when the load fluctuates drastically and widened when the load is stable.
[0010] Preferably, in step three, based on the predicted demand and the operating parameters of the energy storage unit, it is determined whether to activate energy storage regulation; if the predicted demand does not exceed the preset threshold, the energy storage unit is instructed to standby charge; if it exceeds the threshold, the charging and discharging conditions of the energy storage unit are further determined, and in the case of a fault, the auxiliary device is linked to ensure that the demand control is not interrupted, and if the parameters meet the standard, the dynamic regulation strategy of energy storage is triggered.
[0011] Preferably, in step four, a three-dimensional control matrix of SOC-transformer load-voltage is constructed to dynamically generate charging and discharging control commands. The charging and discharging power and depth are adjusted based on load fluctuations to adapt to the grid demand response commands and ensure that the average power within the demand cycle remains stable below the safety threshold.
[0012] Preferably, in step four, the three-dimensional control matrix generates control commands by real-time acquisition of energy storage SOC, transformer load rate, and bus voltage parameters. During periods of frequent peaks, the depth of charge and discharge (DOD) is controlled between 20% and 60%. At the same time, the charging termination SOC is optimized by combining the load forecast results of the next day, charging to 80% when there are many peaks and charging to 60% when there are few peaks.
[0013] Preferably, in step five, the energy storage operation parameters, load data, and demand control effects are synchronized in real time to the background management system and the power grid metering device. The control strategy is iteratively optimized based on machine learning algorithms, and the preset parameter thresholds are reviewed and adaptively adjusted after the monthly settlement cycle.
[0014] Preferably, the method further includes a multi-scenario adaptive extended control step: in the multi-transformer network scenario, a "main control-partition control" architecture is adopted, and the peak superposition period is predicted by load time sequence analysis, and the energy storage is scheduled to discharge centrally or switch services dynamically across transformers; in the mixed load scenario, the discharge phase is optimized by the load type identification algorithm to improve the energy storage response speed to within 20ms, and "pre-charge and discharge + real-time compensation" is adopted to offset the instantaneous peak.
[0015] The present invention also discloses a system for energy storage devices to participate in transformer demand control, including a parameter configuration unit, a data acquisition unit, a status analysis unit, a collaborative control unit, and a data synchronization unit.
[0016] Compared with related technologies, the method and system for energy storage devices to participate in transformer demand control provided by this invention have the following beneficial effects: A hybrid prediction model integrating LSTM and an improved slip algorithm, combined with a dynamic demand calculation window and metering calibration factor, strictly controls the demand prediction error to within 5%. Millisecond-level energy storage response and dynamic power adjustment can reduce the maximum monthly demand by 25%-40%, effectively reducing basic electricity costs. It is also compatible with the grid metering system, eliminating the risk of billing disputes. Through a three-dimensional control matrix of SOC-transformer load-voltage and dynamic control of charge and discharge depth, overcharging and discharging of energy storage is avoided. By combining load forecasting with optimized charging and discharging parameters, the energy storage cycle life can be extended by 20%-30%, while improving energy storage capacity utilization and reducing the operating cost of the energy storage system. The fault-tolerant mechanism, through multi-device linkage, ensures uninterrupted demand control during energy storage failures, adapting to the continuous production needs of industrial applications. Multi-link communication and parameter threshold control ensure stable operation of the system under complex conditions, without negatively impacting transformers and the power grid. The entire process is closed-loop automated control, requiring no manual intervention. Combined with machine learning, iterative optimization of strategies is achieved, adapting to dynamic changes in load characteristics, reducing operation and maintenance costs, and improving the practicality of the solution. Attached Figure Description
[0017] Figure 1 Flowchart provided for this invention; Figure 2 This is a system block diagram provided for the present invention. Detailed Implementation
[0018] The present invention will be further described below with reference to the accompanying drawings and embodiments. The method for energy storage devices to participate in transformer demand control specifically includes the following steps: Step 1: Start the control system and complete the self-test of each unit, synchronize the grid metering device to obtain the demand metering cycle and settlement cycle parameters; load the preset threshold and hybrid prediction model parameters, call the historical load and energy storage operation parameter sample library to complete the model preheating. The preset threshold includes the transformer safe demand threshold, the energy storage SOC safe range, the charging and discharging power limit, the starting power threshold and the voltage safe range of 95%-105% of the rated voltage. Step 2: Acquire the three-phase voltage and current signals of the transformer bus at a sampling period of ≤1s, convert them into real-time power data and transmit them to the control unit; adopt a hybrid prediction model that combines LSTM and improved slip algorithm, combined with dynamic demand calculation window and metering calibration factor, to achieve accurate demand prediction consistent with grid metering data; in Step 2, the three-phase voltage and current electrical signals of the transformer bus are acquired at a sampling period of no more than 1s and converted into real-time power data. A hybrid prediction model that combines LSTM and improved slip algorithm, combined with dynamic demand calculation window and metering calibration factor, is used to achieve accurate demand prediction, ensuring that the predicted demand is consistent with grid metering data.
[0019] The hybrid prediction model described in step two predicts the peak occurrence time, peak size, and duration of shock loads 0.5-1 minutes in advance, and corrects the prediction results in real time for stable loads using an improved slip algorithm, keeping the demand prediction error within 5%. The dynamic demand calculation window is adaptively adjusted according to the load fluctuation frequency, with an adjustment range of 1-15 minutes. The window is narrowed when the load fluctuates drastically and widened when the load is stable.
[0020] Step 3: Based on the predicted demand and real-time energy storage operating parameters, determine whether to activate energy storage regulation; if the predicted demand does not exceed the preset threshold, instruct the energy storage unit to standby charge; if the threshold is exceeded, further verify the energy storage charging and discharging conditions. In case of a fault, the auxiliary device is linked to ensure continuous demand control, and if the parameters meet the standards, the dynamic regulation strategy is triggered. In Step 3, based on the predicted demand and energy storage unit operating parameters, determine whether to activate energy storage regulation; if the predicted demand does not exceed the preset threshold, instruct the energy storage unit to standby charge; if the threshold is exceeded, further verify the energy storage unit charging and discharging conditions. In case of a fault, the auxiliary device is linked to ensure uninterrupted demand control, and if the parameters meet the standards, the dynamic regulation strategy for energy storage is triggered.
[0021] Step 4: Construct a three-dimensional control matrix of SOC-transformer load-voltage, dynamically generate charging and discharging commands, adjust power and depth according to real-time load fluctuations, adapt to grid demand response commands, and stabilize the average power within the demand cycle below the preset threshold.
[0022] The three-dimensional control matrix described in step four generates control commands by real-time acquisition of energy storage SOC, transformer load rate, and bus voltage parameters. During periods of frequent peaks, the depth of charge and discharge (DOD) is controlled between 20% and 60%. At the same time, the charging termination SOC is optimized by combining the load forecast results of the next day, charging to 80% when there are many peaks and charging to 60% when there are few peaks.
[0023] Step 5: Synchronize energy storage operation parameters, load data and control effects to the backend and grid metering devices in real time to ensure data traceability and measurable peak shaving effect; iteratively optimize control strategies and model parameters based on machine learning, review and adaptively adjust preset thresholds after the monthly settlement cycle to adapt to changes in load characteristics.
[0024] In step five, the energy storage operation parameters, load data, and demand control effects are synchronized in real time to the background management system and the grid metering device. The control strategy is iteratively optimized based on machine learning algorithms, and the preset parameter thresholds are adaptively adjusted after the monthly settlement cycle.
[0025] It also includes multi-scenario adaptive extended control steps: In the multi-transformer network scenario, a "main control-zone control" architecture is adopted, and the peak superposition period is predicted by load time sequence analysis, and the energy storage is scheduled to discharge in a concentrated manner or switch services across transformers dynamically; In the mixed load scenario, the discharge phase is optimized by load type identification algorithm to improve the energy storage response speed to within 20ms, and "pre-charge and discharge + real-time compensation" is used to offset instantaneous peaks.
[0026] The present invention also discloses a system for energy storage devices to participate in transformer demand control, including a parameter configuration unit, a data acquisition unit, a status analysis unit, a collaborative control unit, and a data synchronization unit. Example
[0027] Application scenario: A chemical company has configured a single 1000kVA industrial transformer and uses a maximum demand billing method. The historical monthly maximum demand is about 950kW. During the production process, the start-up of motors and air compressors generates frequent load peaks. The demand control target is to reduce the maximum demand to below 850kW, while ensuring stable power supply for precision production and avoiding production interruptions.
[0028] Basic configuration: The load monitoring unit uses a 0.2S-level high-precision current transformer and voltage sensor with a sampling period of 1s; the energy storage unit is a 200kWh lithium iron phosphate battery system with a safe SOC range of 20%-80%, a maximum charge / discharge power of 200kW, a response time of ≤20ms, and is equipped with a liquid-cooled BMS system; the energy conversion unit is a 200kW bidirectional converter with a passive filter and a harmonic suppression rate of ≥95%; the control unit uses an industrial-grade PLC with a built-in hybrid prediction model and a three-dimensional control matrix; the communication unit adopts a dual-link redundant design of RS485 and Ethernet to connect the enterprise energy management system and the grid metering device.
[0029] Implementation process: 1. Initialization phase: Start the system and complete the self-test of each unit. The grid metering cycle is 15 minutes. The preset transformer safe demand threshold is 850kW. The voltage safety range is 95%-105% of the rated voltage. Load the historical load data of the past 30 days and the energy storage operation sample library to complete the preheating of the hybrid prediction model.
[0030] 2. Prediction Phase: The load monitoring unit collects bus voltage and current signals in real time, converts them into real-time power data, and transmits them to the PLC; the hybrid prediction model predicts that motor startup will cause load spikes during a certain demand cycle, predicting a maximum demand of 920kW (exceeding the trigger threshold of 850kW×1.05=892.5kW), and the energy storage unit's current SOC=75%, single cell voltage 3.4V, and cell temperature 32℃ are all within the standard range, triggering the energy storage dynamic control strategy.
[0031] 3. Control Phase: The PLC generates control commands through a three-dimensional control matrix, instructing the energy storage unit to discharge at a power of 100kW, adjusting the power in real time to match load fluctuations, and maintaining the average power at around 840kW within the demand cycle; during the discharge process, the DOD is controlled at 50% to avoid deep discharge, while the bus voltage is monitored to be stable at 98% of the rated voltage, meeting safety requirements.
[0032] 4. Optimization phase: When the load drops below 700kW, the energy storage unit is instructed to charge at 150kW until the SOC reaches 80% and then stops. The prediction model parameters are iteratively optimized daily through machine learning. After monthly review, the transformer safe demand threshold is fine-tuned to 840kW to adapt to changes in load characteristics.
[0033] Implementation Results: After one month of trial operation, the maximum monthly demand of the transformer was reduced to 830kW, basic electricity expenses were reduced by 22%, the energy storage charging and discharging efficiency reached over 90%, the energy storage cycle life was increased by 25% compared to the conventional operation mode, and there was no impact on the company's normal production or the power quality of the power grid. The peak shaving effect was recognized by the power grid department, and there were no billing disputes. Example
[0034] Application Scenario: An industrial park is equipped with three 800kVA distribution transformers to supply power to the chemical workshop, assembly workshop and office building respectively. The load peaks of each transformer have time-series superposition characteristics (the peak period for the chemical workshop and assembly workshop is from 9 to 10 am). They share a 500kWh energy storage system. It is necessary to realize cluster demand control and efficient energy storage scheduling, while supporting grid demand response services.
[0035] Implementation Process: A "main control-regional control" architecture is adopted. The main control unit coordinates global demand data and energy storage scheduling, while three regional control units correspond to three transformers respectively, collecting local load, voltage, and demand data. After system initialization, the main control unit loads preset thresholds, calls historical load data to complete model preheating, and synchronizes with the grid metering cycle of 15 minutes. The preset safe demand threshold for a single transformer is 700kW.
[0036] At 8:50 AM, the hybrid forecasting model predicted that the peak demand from the chemical plant and the assembly plant would overlap between 9:00 and 10:00 AM, with the maximum demand expected to reach 1500kW (exceeding the combined threshold of the two transformers, 1400kW × 1.05 = 1470kW). The main control unit instructed the energy storage unit to enter a standby state. During the peak overlap period at 9:00 AM, the energy storage unit was instructed to discharge at a concentrated power of 300kW to offset the overlap peak and keep the total demand of the two transformers below 1450kW. After 10:00 AM, the load of the chemical plant dropped below the threshold, but the assembly plant still had a separate peak (predicted demand of 720kW). The main control unit dispatched the energy storage unit to the assembly plant to discharge at a power of 80kW to reduce the peak. During the off-peak period at night (00:00-06:00), based on the peak forecast results for the next day, the energy storage unit was charged to SOC=70% to meet the load demand for the next day. During the month, the grid received three peak reduction instructions, all of which automatically reduced the demand control threshold to 680kW, increased the energy storage discharge capacity, and obtained additional grid peak shaving benefits.
[0037] Implementation results: The park's overall monthly maximum demand has decreased by 25%, the energy storage utilization rate has increased by 42% compared to the single transformer scenario, and the park saves millions of yuan in electricity costs every year. At the same time, additional revenue is generated through grid demand response services, achieving a win-win situation for demand control and economic benefits.
[0038] Compared with related technologies, the method and system for energy storage devices to participate in transformer demand control provided by this invention have the following beneficial effects: A hybrid prediction model integrating LSTM and an improved slip algorithm, coupled with a dynamic demand calculation window and metering calibration factor, strictly controls demand prediction errors to within 5%. Millisecond-level energy storage response and dynamic power adjustment can reduce monthly maximum demand by 25%-40%, effectively reducing basic electricity costs. It is also compatible with the grid metering system, eliminating the risk of billing disputes. Through a three-dimensional control matrix of SOC-transformer load-voltage and dynamic control of charge and discharge depth, overcharging and discharging of energy storage are avoided. Combining load forecasting to optimize charge and discharge parameters can extend the energy storage cycle life by 20%-30%, while improving energy storage capacity utilization and reducing the operating cost of the energy storage system. The fault-coordinated fault-tolerance mechanism ensures uninterrupted demand control during energy storage failures through multi-device linkage, adapting to industrial-grade continuous production needs. Multi-link communication and parameter threshold control ensure stable operation of the system under complex conditions without negatively impacting transformers and the grid. The entire process is closed-loop automated control without manual intervention. Combined with machine learning, iterative optimization of strategies is achieved to adapt to dynamic changes in load characteristics, reducing operation and maintenance costs and improving the practicality of the solution.
[0039] The solution of this application has been described in detail above with reference to the accompanying drawings. In the above embodiments, the descriptions of each embodiment have different emphases; for parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments. Those skilled in the art should also understand that the actions and modules involved in the specification are not necessarily essential to this application. Furthermore, it is understood that the steps in the method of this application's embodiments can be adjusted, combined, and deleted according to actual needs, and the modules in the device of this application's embodiments can be combined, divided, and deleted according to actual needs.
[0040] Furthermore, the method according to this application can also be implemented as a computer program or computer program product, which includes computer program code instructions for performing some or all of the steps in the method described above.
[0041] Alternatively, this application may be implemented as a non-transitory machine-readable storage medium (or computer-readable storage medium, or machine-readable storage medium) storing executable code (or computer program, or computer instruction code) thereon, which, when executed by a processor of an electronic device (or electronic device, server, etc.), causes the processor to perform part or all of the steps of the methods described above according to this application.
[0042] Those skilled in the art will also understand that the various exemplary logic blocks, modules, circuits, and algorithm steps described in connection with the present application can be implemented as electronic hardware, computer software, or a combination of both.
[0043] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems and methods according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.
[0044] The various embodiments of this application have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.
[0045] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.
Claims
1. A method for energy storage devices to participate in transformer demand control, characterized in that, Specifically, the following steps are included: Step 1: Start the control system and complete the self-test of each unit, and synchronize the power grid metering device to obtain the demand metering cycle and settlement cycle parameters; Load preset thresholds and hybrid prediction model parameters, and call the historical load and energy storage operation parameter sample library to complete model preheating. The preset thresholds include transformer safe demand threshold, energy storage SOC safe range, charging and discharging power limit, starting power threshold and voltage safe range of 95%-105% of rated voltage. Step 2: Acquire the three-phase voltage and current signals of the transformer bus with a sampling period of ≤1s, convert them into real-time power data and transmit them to the control unit; adopt the LSTM and improved slip algorithm fusion model, combined with dynamic demand calculation window and metering calibration factor, to achieve accurate demand prediction consistent with power grid metering data; Step 3: Based on the predicted demand and real-time energy storage operating parameters, determine whether to activate energy storage regulation; if the preset threshold is not exceeded, instruct the energy storage to standby charge and replenish energy. If the threshold is exceeded, the energy storage charging and discharging conditions will be further checked. In case of a fault, the auxiliary device will be linked to ensure continuous demand control. If the parameters meet the standard, the dynamic control strategy will be triggered. Step 4: Construct a three-dimensional control matrix of SOC-transformer load-voltage, dynamically generate charging and discharging commands, adjust power and depth according to real-time load fluctuations, adapt to grid demand response commands, and stably control the average power within the demand cycle below the preset threshold. Step 5: Synchronize energy storage operation parameters, load data and control effects to the backend and grid metering devices in real time to ensure data traceability and measurable peak shaving effect; iteratively optimize control strategies and model parameters based on machine learning, review and adaptively adjust preset thresholds after the monthly settlement cycle to adapt to changes in load characteristics.
2. The method for energy storage devices participating in transformer demand control according to claim 1, characterized in that, In step two, the three-phase voltage and current electrical signals of the transformer bus are collected with a sampling period of no more than 1 second and converted into real-time power data. A hybrid prediction model combining LSTM and improved slip algorithm is adopted, and combined with dynamic demand calculation window and metering calibration factor to achieve accurate demand prediction, ensuring that the predicted demand is consistent with the grid metering data.
3. The method for energy storage devices participating in transformer demand control according to claim 1, characterized in that, In step two, the hybrid prediction model predicts the peak occurrence time, peak size, and duration of shock loads 0.5-1 minutes in advance, and corrects the prediction results in real time for stable loads using an improved slip algorithm, keeping the demand prediction error within 5%. The dynamic demand calculation window is adaptively adjusted according to the load fluctuation frequency, with an adjustment range of 1-15 minutes. The window is narrowed when the load fluctuates drastically and widened when the load is stable.
4. The method for energy storage devices participating in transformer demand control according to claim 1, characterized in that, In step three, based on the predicted demand and the operating parameters of the energy storage unit, it is determined whether to activate energy storage regulation. If the predicted demand does not exceed the preset threshold, the energy storage unit is instructed to standby charge. If it exceeds the threshold, the charging and discharging conditions of the energy storage unit are further determined. In the event of a fault, the auxiliary device is linked to ensure that the demand control is not interrupted. If the parameters meet the standard, the dynamic regulation strategy of energy storage is triggered.
5. The method for energy storage devices participating in transformer demand control according to claim 1, characterized in that, In step four, a three-dimensional control matrix of SOC-transformer load-voltage is constructed to dynamically generate charging and discharging control commands. Based on load fluctuations, the charging and discharging power and depth are adjusted to adapt to the grid demand response commands, so as to stabilize the average power below the safety threshold within the demand cycle.
6. The method for energy storage devices participating in transformer demand control according to claim 1, characterized in that, In step four, the three-dimensional control matrix generates control commands by real-time acquisition of energy storage SOC, transformer load rate, and bus voltage parameters. During periods of frequent peaks, the depth of charge and discharge (DOD) is controlled between 20% and 60%. At the same time, the charging termination SOC is optimized by combining the load forecast results of the next day. When there are many peaks, the charge is increased to 80%, and when there are fewer peaks, the charge is increased to 60%.
7. The method and system for energy storage devices participating in transformer demand control according to claim 1, characterized in that, In step five, the energy storage operation parameters, load data, and demand control effects are synchronized in real time to the background management system and the power grid metering device. The control strategy is iteratively optimized based on machine learning algorithms, and the preset parameter thresholds are adaptively adjusted after the monthly settlement cycle.
8. The method for energy storage devices participating in transformer demand control according to claim 1, characterized in that, The system also includes multi-scenario adaptive extended control steps: In the multi-transformer network scenario, a "main control-partition control" architecture is adopted, and the peak overlap period is predicted by load time sequence analysis, and the energy storage is scheduled to discharge centrally or switch services dynamically across transformers; In the mixed load scenario, the discharge phase is optimized by load type identification algorithm to improve the energy storage response speed to within 20ms, and "pre-charge and discharge + real-time compensation" is used to offset instantaneous peaks.
9. A system for energy storage devices participating in transformer demand control, characterized in that, It includes a parameter configuration unit, a data acquisition unit, a status analysis unit, a collaborative control unit, and a data synchronization unit.