Optical storage and micro-grid collaborative control method

Abstract

This invention relates to the field of microgrid control technology, and particularly to a method for coordinated control of photovoltaic, energy storage, and charging systems with microgrids. The method includes: real-time acquisition of multi-source data; prediction of photovoltaic output; screening of temporary units; identification of energy storage units; identification of units with coordination anomalies; dynamic correction of power thresholds; and execution of power adjustment control. This invention organically couples multi-dimensional real-time parameters such as output power, state of charge (SCC), load demand, and user behavior to construct an adaptive closed loop: power fluctuation thresholds and charging request growth rates jointly trigger temporary unit screening, ensuring rapid response to sudden load changes; SCC and queuing time jointly determine energy storage unit scheduling, balancing the generation and demand sides; and a coordination index is generated by weighting reconnection frequency and power-load normalization to accurately identify coordination anomalies, effectively solving the problem of low response speed in coordinated control caused by static models and single-index judgment.

Description

Technical Field

[0001] This invention relates to the field of microgrid control technology, and in particular to a method for coordinated control of photovoltaic, energy storage, and charging systems with microgrids. Background Technology

[0002] With the large-scale deployment of distributed photovoltaic (PV) power and electric vehicle charging infrastructure, residential microgrids are rapidly evolving towards integrated and coordinated control of PV, energy storage, and charging. However, the inherent volatility of PV power generation, the charging and discharging limitations of energy storage systems, and the sudden increase in charging pile loads all pose serious challenges to grid stability. Therefore, there is an urgent need for a coordinated control strategy for PV, energy storage, charging, and microgrids to improve the microgrid's rapid response to fluctuations and sudden demands, as well as its overall dispatch efficiency.

[0003] Patent document CN117254526A discloses an energy collaborative optimization control method for an integrated photovoltaic-storage-charging-inspection microgrid station. The method includes: S1, collecting load data and photovoltaic power generation data within the integrated photovoltaic-storage-charging-inspection microgrid station; S2, establishing load models for different time periods based on historical transformer load data; S3, establishing photovoltaic system power generation models for different time periods based on historical power generation data; S4, comparing the transformer load model with the corresponding time period power generation model and the preset output power of the energy storage system; S5, switching different operating modes based on the comparison results, aggregated electricity price time, and real-time photovoltaic system power generation to adjust the operating states of the photovoltaic system, energy storage system, and power grid to meet load requirements.

[0004] Therefore, the energy collaborative optimization control method for the integrated photovoltaic-storage-charging-inspection microgrid has the following problems: Simply comparing the transformer load model with the photovoltaic power generation model lacks real-time response capability to the dynamic coupling between the two and sudden load changes, resulting in a disconnect between the load model and the power generation model; the model is only established based on a fixed time period, making it unable to adapt to sudden fluctuations in load and power generation, and difficult to adjust the model window in a timely manner; it directly adopts preset energy storage output without a dynamic adjustment strategy, making it impossible to flexibly allocate energy storage charging and discharging according to photovoltaic fluctuations or sudden load increases; different operating modes are only based on a few sets of comparison results and electricity price time period switching, lacking support for fine-grained strategies at the unit level or feeder level, making it difficult to achieve local optimization. Summary of the Invention

[0005] To address this, the present invention provides a method for coordinated control of photovoltaic, energy storage, charging, and microgrids, which overcomes the problem of low response speed of coordinated control caused by static models and single index judgment in the prior art through dynamic threshold adjustment driven by real-time multi-dimensional data.

[0006] To achieve the above objectives, the present invention provides a method for coordinated control of photovoltaic, energy storage, charging, and microgrids, comprising:

[0007] Real-time data collection includes the output power of each photovoltaic unit on each feeder in the photovoltaic array operating based on preset photovoltaic power generation, the status value of high current load switches in each feeder, the state of charge value of the energy storage system, the load demand at the building transformer end, the growth rate of charging requests on the target side, the average queuing time, and the frequency of interruption and reconnection.

[0008] The photovoltaic output forecast value is predicted based on all the output power, state of charge value, load demand, charging request growth rate, average queuing time, and interruption reconnection frequency within a preset historical period, as well as a preset prediction model.

[0009] Several temporary units are selected based on the output power, the preset total power fluctuation threshold, and the charging request growth rate.

[0010] Several energy storage units are determined based on the average queuing time, the output power of each temporary unit, and the state of charge value.

[0011] Several coordinated abnormal units are determined based on the interruption reconnection frequency, the output power of each energy storage unit, and the load demand.

[0012] The preset total power fluctuation threshold is adjusted according to the layout location and number of the coordinated abnormal unit in each feeder and the state value of the high current load switch to obtain the adjusted power fluctuation threshold.

[0013] The photovoltaic output prediction value is corrected based on the number of units of the coordinated abnormal unit that is re-determined using the adjusted power fluctuation threshold within the preset correction period, and the corrected prediction value is obtained.

[0014] Based on the corrected predicted value and the preset predicted value range, the coordinated anomaly unit executes a control command to adjust the preset photovoltaic power generation.

[0015] Furthermore, the process of selecting several temporary units based on the output power, the preset total power fluctuation threshold, and the charging request growth rate includes:

[0016] Calculate the sum of all the output powers to obtain the total output power;

[0017] Calculate the standard deviation of the total output power within the preset temporary duration to obtain the total power fluctuation value;

[0018] When the total power fluctuation value is greater than the preset total power fluctuation threshold and the charging request growth rate is greater than the preset standard growth rate, several temporary units are selected based on the output power and the total power fluctuation value.

[0019] Furthermore, based on the output power and the total power fluctuation value, several temporary units are selected, including:

[0020] Calculate the standard deviation of the total output power of each photovoltaic unit within the next preset temporary duration to obtain the output power fluctuation value;

[0021] Calculate the relative deviation between the output power fluctuation value and the total power fluctuation value to obtain a temporary judgment value;

[0022] When the temporary determination value is greater than the preset temporary determination threshold, the photovoltaic unit is determined to be the temporary unit, so as to filter out a number of temporary units.

[0023] Furthermore, the process of determining a plurality of energy storage units based on the average queuing time, the output power of each of the temporary units, and the state of charge value includes:

[0024] When the state of charge value is less than a preset state of charge threshold, the difference between the preset state of charge threshold and the state of charge value is calculated to obtain the charge difference.

[0025] When the charge difference is greater than a preset charge difference threshold, the average queuing time is less than a preset time threshold, and the output power is greater than a preset standard power, the temporary unit is determined to be the energy storage unit, thereby identifying several energy storage units.

[0026] When the charge difference is greater than a preset charge difference threshold and the average queuing time is greater than or equal to the preset time threshold, a number of energy storage units are determined based on the output power within a preset time period.

[0027] Furthermore, the process of determining a plurality of energy storage units based on the output power within a preset time period includes:

[0028] Calculate the standard deviation of the output power within a preset time period to obtain a determined power fluctuation value;

[0029] When the determined power fluctuation value is greater than a preset determined fluctuation threshold, the temporary unit is determined to be the energy storage unit, thereby identifying a number of energy storage units.

[0030] Furthermore, the process of determining several coordinated abnormal units based on the interruption reconnection frequency, the output power of each energy storage unit, and the load demand includes:

[0031] Obtain the interruption reconnection frequency within a preset coordination time period to obtain a reconnection frequency set;

[0032] Obtain the output power within the preset coordination time period to obtain an output power set;

[0033] Obtain the load demand within the preset coordination time period to obtain a load demand set;

[0034] The output power at each moment within the preset coordination time period is normalized by the maximum-minimum value based on the output power set to obtain the power normalization value;

[0035] The load demand at each moment within the preset coordination time period is normalized by the maximum-minimum value based on the load demand set to obtain the demand normalization value.

[0036] Calculate the product of the power normalization value and the preset power weight to obtain the first product; calculate the product of the demand normalization value and the preset demand weight to obtain the second product.

[0037] Calculate the sum of the first product and the second product to obtain the synergy index;

[0038] Obtain all the aforementioned synergistic indices to obtain a set of synergistic indices;

[0039] Calculate the correlation coefficient between the reconnection frequency set and the coordination index set to obtain the coordination anomaly degree;

[0040] When the degree of coordination anomaly is greater than the preset standard anomaly, the energy storage unit is determined to be the coordination anomaly unit, thereby identifying several coordination anomaly units.

[0041] Furthermore, the process of adjusting the preset total power fluctuation threshold based on the deployment location and number of the coordinated abnormality units within each feeder and the state value of the high-current load switch, to obtain the adjusted power fluctuation threshold, includes:

[0042] The placement positions of the coordinated abnormality units within each feeder are obtained to determine the relative distances between the coordinated abnormality units and the starting point of the feeder.

[0043] Obtain the feeder number to which each of the aforementioned collaborative anomaly units belongs, and count the number of collaborative anomaly units in each feeder based on the feeder number to obtain the number of units;

[0044] The relative distance at the current moment is normalized by a maximum-minimum value based on all the relative distances mentioned, resulting in several normalized distances;

[0045] Calculate the standard deviation of all the normalized distances to obtain the distribution concentration.

[0046] Calculate the concentration factor based on the distribution concentration and the number of units;

[0047] The current distribution concentration is normalized based on all the distribution concentrations within the preset adjustment period to obtain the normalized concentration.

[0048] The current number of units is normalized based on the total number of units within the preset adjustment period to obtain the normalized number.

[0049] The risk index is obtained by weighting and summing the preset concentration weight, the normalized concentration degree, the preset quantity weight, the normalized quantity, the preset state value weight, and the state value of the high current load switch.

[0050] When the risk index is greater than the preset risk index threshold, the preset total power fluctuation threshold is increased according to the relative deviation between the risk index and the preset risk index threshold and the preset adjustment coefficient to obtain the adjusted power fluctuation threshold.

[0051] Further, the process of correcting the photovoltaic output prediction value based on the number of the coordinated abnormal units re-determined using the adjusted power fluctuation threshold within a preset correction period, and obtaining the corrected prediction value, includes:

[0052] Calculate the standard deviation of the number of units within the preset correction period to obtain the correction quantity fluctuation value;

[0053] When the correction quantity fluctuation value is greater than the preset correction quantity fluctuation threshold, the photovoltaic output prediction value is increased according to the relative deviation between the correction quantity fluctuation value and the preset correction quantity fluctuation threshold and the preset correction coefficient to obtain the correction prediction value.

[0054] Furthermore, the process of executing a control command to adjust the preset photovoltaic power generation capacity of the coordinated anomaly unit based on the corrected predicted value and the preset predicted value range includes:

[0055] When the corrected predicted value is greater than the maximum value of the preset predicted value range, a control command to reduce the preset photovoltaic power generation is executed based on the relative deviation between the corrected predicted value and the maximum value of the preset predicted value range and the preset control coefficient.

[0056] When the corrected predicted value is less than the minimum value of the preset predicted value range, a control command to increase the photovoltaic power generation is executed based on the relative deviation between the minimum value of the preset predicted value range and the corrected predicted value, as well as the preset control coefficient.

[0057] Furthermore, the process of predicting the photovoltaic output value based on all the output power, the state of charge value, the load demand, the charging request growth rate, the average queuing time, the interruption reconnection frequency, and the preset prediction model within a preset historical time period includes:

[0058] All the aforementioned output power, state of charge value, charging pile load demand, target side charging request growth rate, average queuing time and interruption reconnection frequency are time series aligned, missing value imputed and normalized to construct a multi-dimensional input feature matrix.

[0059] The multi-dimensional input feature matrix is ​​input into the preset prediction model to predict the photovoltaic output value.

[0060] Compared with existing technologies, the beneficial effects of this invention are as follows: by organically coupling multi-dimensional real-time parameters such as output power, state of charge (SCC), load demand, and user behavior (request growth rate, queuing time, reconnection frequency), an adaptive closed loop is constructed in six major stages: prediction, screening, identification, threshold adjustment, prediction correction, and control execution. First, the power fluctuation threshold and the charging request growth rate jointly trigger temporary unit screening to ensure rapid response to sudden load changes. Second, the SCC and queuing time jointly determine the energy storage unit scheduling to balance the generation side and the demand side. Then, the reconnection frequency and power-load normalization weighted average generate a coordination index to accurately identify coordination anomalies. The distribution location and quantity of abnormal units at the feeder level are mapped to the threshold adjustment coefficient along with the load switch status to achieve regional risk perception. The secondary correction of the predicted value within the correction period compensates for model deviations, and finally, the photovoltaic power adjustment command is executed based on the correction error range. This significantly improves the system's stability, response speed, and energy utilization, effectively solving the problem of low response speed in coordinated control caused by static models and single-index judgment.

[0061] Furthermore, a screening mechanism based on a dual trigger logic of dynamic volatility detection and load growth signal identification organically combines the volatility of output power with demand-side load behavior (such as a surge in electric vehicle charging). The total power fluctuation value, as an important statistical indicator reflecting the stability of the photovoltaic system, forms the basis for subjective evaluation along with the output power; while the charging request growth rate reflects the dynamic energy demand of the target (user) side. Both factors jointly determine whether temporary intervention is needed. By comprehensively judging the "supply and demand fluctuation amplitude" and "future load pressure," not only can potential power supply instability be detected in advance, but it can also provide precise, localized, and controllable targets for subsequent energy storage scheduling, effectively improving the system's response speed and control accuracy to sudden load changes, and achieving more granular photovoltaic-storage-charging coordinated control.

[0062] Furthermore, by introducing a fluctuation judgment mechanism at the photovoltaic unit level, the system can identify the local fluctuation sources that have the greatest impact on system stability, even when overall power fluctuations are large and charging demand is rapidly increasing. The output power fluctuation value reflects the short-term instability of the unit itself, while the total power fluctuation value reflects the overall disturbance level of the system. The relative deviation constitutes a normalized comparison between the two, avoiding misjudgments at different power levels. This logic not only enhances the system's sensitivity to abnormal fluctuations but also improves the spatial resolution and strategy accuracy of photovoltaic-storage coordinated regulation, effectively supporting the prioritization of subsequent energy storage intervention resources.

[0063] Furthermore, by dynamically coupling the state of charge of the energy storage system, the charging demand pressure on the residential side (reflected by the average queuing time), and the current power generation capacity of the photovoltaic units, a logical "supply and demand linkage" mechanism is formed. On the one hand, the state of charge difference reflects the strength of energy storage demand, the average queuing time reflects the urgency of the immediate electricity load of residents, and the output power measures the availability of the photovoltaic units. The linkage of these three factors enables dynamic prioritization and flexible allocation of energy storage under the background of power generation fluctuations. When user demand is controllable, priority is given to charging the energy storage system, and when demand is tight, the power supply to residents is maintained. This improves the flexibility of photovoltaic output regulation and user satisfaction, while reducing the risk of local overload and load conflict.

[0064] Furthermore, by introducing a "determined power fluctuation value" to reflect the stability of photovoltaic output over a short period, and combining it with a "preset determined fluctuation threshold" as a screening criterion, units with drastic power fluctuations can be accurately identified as priority energy storage targets. This achieves linkage between unstable power regions and energy storage response mechanisms, preventing power fluctuations from disturbing the microgrid while fully utilizing the energy storage system for peak shaving and valley filling. The greater the change in output power, the higher the regulation value of energy storage; therefore, this strategy can dynamically match system stability with the optimized scheduling of energy storage resources, improving the overall flexibility and response speed of the system.

[0065] Furthermore, the correlation between charging interruption frequency and dynamic power-load coordination capability is used for assessment. Normalization and weighting are employed to achieve a unified measurement of parameters with different dimensions, avoiding the bias of relying on a single indicator. Output power reflects the energy supply capacity of the energy storage unit, load demand reflects the real-time load on the building's energy consumption side, and the ratio of these two measures the coordination capability. Interruption reconnection frequency reflects system stability and user experience. When the coordination index is high but interruptions are frequent, or when the coordination capability itself is strongly correlated with the interruption trend, it indicates potential fluctuations or imbalances in the energy storage response capability. This allows for accurate identification and isolation of units with abnormal coordination, improving intelligent diagnostic capabilities.

[0066] Furthermore, a risk assessment model is constructed by correlating three types of parameters: deployment location (spatial distribution), number of units (fault scale), and switch status (load risk), enabling dynamic adjustment of the system's power fluctuation tolerance range. Normalized concentration reflects whether coordinated abnormal units are clustered; a smaller standard deviation indicates a higher degree of concentration, potentially causing localized impact risks to a feeder. Normalized quantity reflects the absolute impact of the number of abnormal units. Load switch status values ​​reflect the current system's carrying capacity and impact response capability. The weighted average of these three factors generates a risk index, which comprehensively assesses the current system stability. This allows for reasonable adjustment of the power fluctuation threshold to enhance system resilience and emergency response capabilities, avoiding misjudgments or omissions due to fixed threshold settings, and improving the flexibility and security of coordinated scheduling.

[0067] Furthermore, by monitoring the fluctuations in the number of anomalous collaborative units, a dynamic correlation was established between anomalous behavior and photovoltaic output prediction. The underlying logic is as follows: the greater the fluctuation in the number of anomalous collaborative units, the greater the potential risk of output interference in the system, thus requiring an increase in the predicted value to allow for safety redundancy. A preset correction threshold for the number of anomalous units serves as a benchmark for stability judgment; the correction value reflects the degree of system disturbance, and the relative deviation between the two reflects the actual deviation magnitude. A preset correction coefficient is used to quantify the correction magnitude. These parameters together construct a data-driven prediction correction mechanism, which helps improve the accuracy of photovoltaic output prediction and the stability of system scheduling.

[0068] Furthermore, by using preset control coefficients and deviation calculations, the photovoltaic power generation can be dynamically adjusted based on the difference between the real-time corrected predicted value and the preset value range, thereby effectively achieving optimized power allocation. It can accurately control power increase and decrease, which not only improves the stability of system operation but also optimizes energy utilization efficiency.

[0069] Furthermore, time series alignment ensures the consistency of all parameters on the same time axis; missing value imputation ensures the integrity of the model input; normalization eliminates the bias caused by different units, providing balanced features for model learning; and the multi-dimensional feature matrix integrates the generation side (output power), energy storage side (SOC), load side (transformer demand, charging growth rate, queuing time) and stability indicators (reconnection frequency), enabling the prediction model to simultaneously capture the dynamic changes and mutual influences of multiple sources of supply, energy storage and demand at the underlying logic level, thereby significantly improving the accuracy and robustness of short-term photovoltaic output prediction. Attached Figure Description

[0070] Figure 1 This is a flowchart of the photovoltaic-storage-charging and microgrid coordinated control method in this embodiment;

[0071] Figure 2 This is a logic diagram for determining the temporary unit in this embodiment;

[0072] Figure 3 This is a logic diagram for determining the energy storage unit in this embodiment;

[0073] Figure 4 This is the logic diagram for determining the coordination anomaly unit in this embodiment. Detailed Implementation

[0074] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0075] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0076] Please see Figure 1 As shown, it is a flowchart of the photovoltaic-storage-charging and microgrid coordinated control method in this embodiment;

[0077] This embodiment provides a method for coordinated control of photovoltaic, energy storage, charging, and microgrids, including:

[0078] Real-time data collection includes the output power of each photovoltaic unit on each feeder in the photovoltaic array operating based on preset photovoltaic power generation, the status value of high current load switches in each feeder, the state of charge value of the energy storage system, the load demand at the building transformer end, the growth rate of charging requests on the target side, the average queuing time, and the frequency of interruption and reconnection.

[0079] The photovoltaic output forecast value is predicted based on all the output power, state of charge value, load demand, charging request growth rate, average queuing time, and interruption reconnection frequency within a preset historical period, as well as a preset prediction model.

[0080] Several temporary units are selected based on the output power, the preset total power fluctuation threshold, and the charging request growth rate.

[0081] Several energy storage units are determined based on the average queuing time, the output power of each temporary unit, and the state of charge value.

[0082] Several coordinated abnormal units are determined based on the interruption reconnection frequency, the output power of each energy storage unit, and the load demand.

[0083] The preset total power fluctuation threshold is adjusted according to the layout location and number of the coordinated abnormal unit in each feeder and the state value of the high current load switch to obtain the adjusted power fluctuation threshold.

[0084] The photovoltaic output prediction value is corrected based on the number of units of the coordinated abnormal unit that is re-determined using the adjusted power fluctuation threshold within the preset correction period, and the corrected prediction value is obtained.

[0085] Based on the corrected predicted value and the preset predicted value range, the coordinated anomaly unit executes a control command to adjust the preset photovoltaic power generation.

[0086] A microgrid refers to a power supply and management unit within a residential building, consisting of a photovoltaic power generation system, energy storage devices, load demand monitoring, and related control systems. The function of a microgrid is to integrate various distributed energy sources and dispatch power according to real-time demand to ensure a stable power supply. In this scenario, the microgrid, through the collaborative work of the photovoltaic-energy storage-charging system and other related modules, ensures that the power needs both inside and outside the building are met.

[0087] Photovoltaic arrays in residential buildings are responsible for converting solar energy into electricity, providing renewable power to the building. The number and layout of photovoltaic units are determined based on the building's roof area and sunlight conditions to maximize power generation efficiency. By monitoring the photovoltaic array's power output in real time, the microgrid can adjust the photovoltaic output according to the building's electricity demand, ensuring a stable power supply.

[0088] As the main channel for power transmission, feeders in residential buildings serve to transmit the power generated by photovoltaic arrays to transformers within the building, or to transmit power from microgrids to charging facilities (such as electric vehicle charging stations). The health and condition of feeders are crucial, affecting the power supply within the building and the operational efficiency of charging stations.

[0089] The energy storage system within the residential building primarily monitors battery charge levels to ensure that electricity demand matches the system's supply. During periods of insufficient sunlight or peak electricity demand, the system can provide additional power to prevent power shortages. Through real-time monitoring of the state of charge, the microgrid can optimize power dispatch, preventing batteries from being overcharged or over-discharged, thereby extending their lifespan.

[0090] The load demand at the building transformer end refers to the total power demand of all electrical equipment within the building, including lighting, air conditioning, elevators, electric water heaters, etc. By monitoring these load demands in real time through smart meters, the microgrid system can perform power dispatching based on actual power consumption, prioritizing the use of photovoltaic power generation, and dispatching energy storage systems to provide power when necessary, ensuring the stability of the building's power supply.

[0091] For electric vehicle charging stations within buildings, the target-side charging request growth rate, average queuing time, and interruption / reconnection frequency reflect the operating load of the charging stations and the charging demand of electric vehicles. The charging request growth rate refers to the rate of increase in electric vehicle charging requests over time, reflecting the changing trend of electric vehicle user charging demand. Especially with the increasing popularity of electric vehicles, charging demand may show a rapid growth trend. Average queuing time measures the utilization efficiency of the charging stations. When multiple electric vehicles need to charge simultaneously, the queuing time increases. By monitoring the queuing time in real time, photovoltaic arrays and energy storage systems can be rationally scheduled to reduce queuing time and improve charging efficiency. The interruption / reconnection frequency reflects the frequency of interruptions and reconnections caused by charging station malfunctions or external factors (such as grid fluctuations, insufficient battery power, etc.). Frequent interruptions and reconnections may indicate problems with the operation of the charging stations or unstable power supply, requiring power dispatching to optimize the charging process and ensure the continuity and stability of charging.

[0092] In this embodiment, based on a preset photovoltaic power generation operation strategy, key operating parameters are collected in real time through multi-level distributed sensing and communication methods: On each feeder branch of the photovoltaic array, current transformers (DC CTs) and voltage transformers (DC PTs) deployed on the DC side of each photovoltaic unit, combined with the power acquisition module built into the micro-inverter, accurately measure their output power and upload it via the Modbus protocol; digital load switch status acquisition devices are configured in the distribution cabinets of each feeder circuit to acquire the status values ​​of high-current load switches in real time (closed = 1, open = 0, fault = 2), and transmit them via an RS-485 link; the state of charge (SOC) value of the energy storage system is uploaded by the battery management system (BMS) to the energy management controller via the CAN bus in the form of periodic messages (e.g., every 10 seconds); the load demand at the building transformer end is measured by connecting to a three-phase smart meter conforming to the DL / T1490-2015 standard, and the total active power is measured and transmitted to the SCADA system via the Ethernet port; each smart charging pile in the target parking lot is connected to the charging station management system (CSMS), and the system is based on OCPP. The 1.6 protocol records connection requests, queuing events, and communication interruption logs for each charging pile, and provides real-time statistics on charging request growth rate, average queuing time, and interruption reconnection frequency. All collected data is ultimately aggregated by the RTU and uploaded to the upper-layer EMS system via fiber optic link according to the Modbus or IEC 61850 protocol. The data is then stored in the SCADA database at the second or minute level, enabling full-dimensional perception and dynamic control of the residential building microgrid system.

[0093] The preset photovoltaic power generation capacity refers to the target output power of the photovoltaic array that is pre-set based on historical operating efficiency, weather forecast data and user-side load characteristics when formulating the dispatch strategy. It depends on the predicted light intensity, photovoltaic module efficiency and electricity load trend, and is usually set between 40% and 90% of the rated power of the photovoltaic array. In this embodiment, it is set to 75%, which can provide a stable reference power target for microgrid operation.

[0094] The preset historical duration refers to the length of the historical data time window used for statistical and predictive analysis. It depends on the frequency of system load fluctuations and the data sampling period, and is usually set between 5 minutes and 1 hour. In this embodiment, it is set to 30 minutes to balance data response sensitivity and predictive stability.

[0095] The preset correction period refers to the time interval for dynamically correcting the predicted value during system operation. It depends on the frequency of environmental disturbances and the response rate of the scheduling system, and is usually set between 1 minute and 15 minutes. In this embodiment, it is set to 5 minutes, which can correct the photovoltaic output prediction value in a timely manner and improve control accuracy and system stability.

[0096] The preset prediction range refers to the upper and lower limits of the photovoltaic power generation prediction value. It depends on the rated power of the photovoltaic array and the error margin of the prediction model. It is usually set between 80% and 100%. In this embodiment, it is set to 85% to 95%, which can ensure that the photovoltaic output fluctuates within a reasonable tolerance and prevent the system from being overloaded or underloaded.

[0097] The preset prediction model in this embodiment is a multivariate time-series prediction model that integrates a Long Short-Term Memory (LSTM) network and a feature attention mechanism. It is specifically designed for short-term accurate prediction of photovoltaic output in a residential microgrid environment. This model fully considers the nonlinear variation characteristics of photovoltaic output and multiple influencing factors, and has good robustness and generalization ability, making it suitable for ultra-short-term prediction scenarios of 5-60 minutes.

[0098] I. Model Structure Composition

[0099] Input layer: Input variables include:

[0100] Output power sequence (P) of each photovoltaic unit;

[0101] Energy storage system SOC sequence (S);

[0102] Building transformer load demand sequence (L);

[0103] Charging request growth rate (R);

[0104] Average queuing time (Q);

[0105] Interruption reconnection frequency (F);

[0106] Each variable forms a sliding time window at fixed time intervals (e.g., every minute) according to a "preset historical duration" (e.g., 30 minutes), constituting a multivariate time series tensor.

[0107] Feature Attention Layer: This layer assigns different weights to multi-source input variables, automatically learns the influence of each parameter on photovoltaic output, and forms a weighted feature embedding representation to improve modeling accuracy. For example, during cloudy or rainy weather, the model may automatically increase the weights of load demand and energy storage SOC.

[0108] LSTM encoding layer: A two-layer LSTM unit is used to extract long-term dependency features in the time dimension to capture the changing trend of photovoltaic power output. This network has a memory gating mechanism, which can effectively preserve highly correlated time-series information and suppress noise interference.

[0109] Fully connected output layer (Dense Layer): The LSTM output feature vector is input into the fully connected layer to regress and predict the photovoltaic power output (unit: kW) at a future time.

[0110] II. Training Methods

[0111] Loss function: The mean squared error (MSE) is used as the loss function;

[0112] Optimizer: The Adam optimizer is used for gradient updates;

[0113] Training dataset: Select second-level running data from the past 30 days, and group them for enhancement according to solar radiation variation and load curve;

[0114] Training cycle: Set to 100 rounds or until the validation set error converges.

[0115] III. Model Deployment

[0116] Operating environment: Deployed on building-level EMS edge computing nodes (supports Python + TensorFlow or PyTorch);

[0117] Update frequency: Daily offline training, hourly hot updates;

[0118] Output frequency: Outputs the photovoltaic power output prediction value for the next 5 minutes every 1 minute.

[0119] IV. Model Advantages

[0120] Enhance the dynamic expression of different influencing factors through attention mechanisms;

[0121] LSTM structures enhance the ability to model complex time-varying changes;

[0122] It supports multi-source heterogeneous data input, adapting to the actual needs of photovoltaic, energy storage and charging coordinated scheduling in microgrids;

[0123] It can provide high-precision prediction support for downstream strategy modules, improving the reliability and economy of prediction-driven control response.

[0124] First, real-time data is collected on the output power of photovoltaic units on each feeder of the photovoltaic array, the status of high-current load switches, the state of charge of the energy storage system, the load demand at the building transformer end, as well as the charging request growth rate, average queuing time, and reconnection frequency on the target side. Then, these real-time parameters are aligned and normalized according to a preset historical timeframe and input into the prediction model to obtain a preliminary photovoltaic output prediction. Next, based on the output power, a preset total power fluctuation threshold, and the charging request growth rate, temporary units with significant fluctuations are selected. Then, usable energy storage units are determined based on the average queuing time, the output power of the temporary units, and the state of charge. Then, by combining the output power, load demand, and reconnection frequency of these energy storage units, collaborative abnormal units are identified. The total power fluctuation threshold is dynamically adjusted based on the distribution location, number of abnormal units, and their load switch status in each feeder. Within a preset correction period, the number of abnormal units is reassessed using the updated threshold, and the photovoltaic output prediction is corrected to generate a more realistic corrected prediction. Finally, when the corrected prediction exceeds a preset range, control commands to increase or decrease photovoltaic power generation are issued to the abnormal units, achieving closed-loop adaptive control of the photovoltaic-energy storage-charging system.

[0125] By organically coupling multi-dimensional real-time parameters such as output power, state of charge (SCC), load demand, and user behavior (request growth rate, queuing time, reconnection frequency), an adaptive closed loop is constructed in six stages: prediction, screening, identification, threshold adjustment, prediction correction, and control execution. First, the power fluctuation threshold and the charging request growth rate jointly trigger temporary unit screening to ensure rapid response to sudden load changes. Second, SCC and queuing time jointly determine energy storage unit scheduling to balance the generation and demand sides. Then, a coordination index is generated by weighting reconnection frequency and power-load normalization to accurately identify coordination anomalies. The distribution location and quantity of abnormal units at the feeder level are mapped to the threshold adjustment coefficient along with the load switch status to achieve regional risk perception. Secondary correction of the predicted value within the correction period compensates for model bias. Finally, photovoltaic power adjustment commands are executed based on the correction error range, which can significantly improve the system's stability, response speed, and energy utilization rate, effectively solving the problem of low response speed of coordinated control caused by static models and single-index judgment.

[0126] Specifically, the process of selecting several temporary units based on the output power, the preset total power fluctuation threshold, and the charging request growth rate includes:

[0127] Calculate the sum of all the output powers to obtain the total output power;

[0128] Calculate the standard deviation of the total output power within the preset temporary duration to obtain the total power fluctuation value;

[0129] When the total power fluctuation value is greater than the preset total power fluctuation threshold and the charging request growth rate is greater than the preset standard growth rate, several temporary units are selected based on the output power and the total power fluctuation value.

[0130] The preset standard growth rate refers to the baseline growth threshold of the number of charging requests per unit time on the target side. It depends on the changes in the average charging requests in the same period in history and the system load response capability. It is usually set between 5% and 20%. In this embodiment, it is set to 12%, which can identify the rapid growth trend of charging load in a timely manner and provide a basis for judgment on energy storage intervention under photovoltaic fluctuation conditions.

[0131] The total output power of the microgrid is obtained by summing the output power of each photovoltaic unit. The standard deviation of this total power is then calculated over a preset temporary period (e.g., 15 minutes) to measure the degree of photovoltaic output fluctuation. If this value exceeds a pre-set total power fluctuation threshold, and the growth rate of charging requests from the target side (e.g., residential building charging stations) exceeds the set standard growth rate, the system considers there to be a risk of "photovoltaic fluctuations coupled with load surges." In this case, the system prioritizes photovoltaic units with a greater impact on system stability, marking them as temporary units, and proceeding to the next stage of energy storage regulation or control intervention, based on the deviation of each photovoltaic unit's real-time output power from the total power fluctuation value.

[0132] By employing a screening mechanism built upon a dual-trigger logic of dynamic volatility detection and load growth signal identification, the volatility of output power is organically combined with demand-side load behavior (such as a surge in electric vehicle charging). The total power fluctuation value, as a crucial statistical indicator reflecting the stability of the photovoltaic system, forms the basis for subjective evaluation along with the output power; while the charging request growth rate reflects the dynamic energy demand of the target (user) side. Both factors jointly determine whether temporary intervention is necessary. By comprehensively judging the "amplitude of supply and demand fluctuations" and "future load pressure," potential power instability can be detected in advance, providing precise, localized, and controllable targets for subsequent energy storage scheduling. This effectively improves the system's response speed and control accuracy to sudden load changes, achieving more granular photovoltaic-storage-charging coordinated control.

[0133] Please continue reading. Figure 2 As shown, this is the determination logic diagram of the temporary unit in this embodiment;

[0134] Based on the output power and the total power fluctuation value, several temporary units are selected, including:

[0135] Calculate the standard deviation of the total output power of each photovoltaic unit within the next preset temporary duration to obtain the output power fluctuation value;

[0136] Calculate the relative deviation between the output power fluctuation value and the total power fluctuation value to obtain a temporary judgment value;

[0137] When the temporary determination value is greater than the preset temporary determination threshold, the photovoltaic unit is determined to be the temporary unit, so as to filter out a number of temporary units.

[0138] The preset temporary judgment threshold is a standard benchmark used to determine whether the output power fluctuation of a single photovoltaic unit is abnormal. It depends on the system's allowable local power fluctuation tolerance and historical operating stability data, and is usually set between 0.1 and 0.3. In this embodiment, it is set to 0.2. That is, when the output power fluctuation of a photovoltaic unit deviates from the overall fluctuation by more than 20%, it is regarded as a temporary abnormal unit. It can effectively filter normal fluctuations, accurately identify units that contribute significantly to system disturbances, and provide a basis for subsequent optimization of control strategies.

[0139] By refining the output power fluctuation behavior down to the individual photovoltaic (PV) unit, accurate identification of temporary units can be achieved. Specifically, this involves: firstly, calculating the standard deviation of the output power of each PV unit within the next preset temporary duration to obtain its output power fluctuation value; then, calculating the relative deviation between this value and the total power fluctuation value of the entire system to obtain the temporary judgment value for each PV unit; if this judgment value exceeds the set temporary judgment threshold, the PV unit is marked as a temporary unit for subsequent key monitoring and intervention in energy storage and load dispatching.

[0140] By introducing a fluctuation judgment mechanism at the photovoltaic unit level, the system can further identify the local fluctuation sources that have the greatest impact on system stability, even when overall power fluctuations are large and charging demand is rapidly increasing. The output power fluctuation value reflects the short-term instability of the unit itself, while the total power fluctuation value reflects the overall disturbance level of the system. The relative deviation constitutes a normalized comparison between the two, avoiding misjudgments at different power levels. This logic not only enhances the system's sensitivity to abnormal fluctuations but also improves the spatial resolution and strategy accuracy of photovoltaic-storage coordinated regulation, effectively supporting the prioritization of subsequent energy storage intervention resources.

[0141] Specifically, the process of determining a plurality of energy storage units based on the average queuing time, the output power of each of the temporary units, and the state of charge value includes:

[0142] When the state of charge value is less than a preset state of charge threshold, the difference between the preset state of charge threshold and the state of charge value is calculated to obtain the charge difference.

[0143] When the charge difference is greater than a preset charge difference threshold, the average queuing time is less than a preset time threshold, and the output power is greater than a preset standard power, the temporary unit is determined to be the energy storage unit, thereby identifying several energy storage units.

[0144] When the charge difference is greater than a preset charge difference threshold and the average queuing time is greater than or equal to the preset time threshold, a number of energy storage units are determined based on the output power within a preset time period.

[0145] The preset charge difference threshold depends on the operating characteristics of the energy storage system and the output power fluctuation of the photovoltaic unit. It is usually set between 10% and 20%. In this embodiment, it is set to 15%, which can ensure that the charging process is started in time when the energy storage system charge is lower than this threshold, so as to avoid system instability caused by low charge.

[0146] The preset duration threshold depends on the user's tolerance for power demand and the system response time, and is usually set between 3 and 10 minutes. In this embodiment, it is set to 5 minutes, which can reduce the user's queuing time and improve charging efficiency while ensuring system stability.

[0147] The preset standard power depends on the power generation capacity of the photovoltaic unit and the needs of the energy storage system. It is usually set between 30% and 50% of the maximum output power. In this embodiment, it is set to 40%, which can effectively balance the output power of the photovoltaic unit and the charging needs of the energy storage unit, and avoid unnecessary fluctuations caused by excessively high or low power.

[0148] The preset time limit refers to the time window used to calculate the output power fluctuation of the photovoltaic unit. It depends on the system's sensitivity to power fluctuation response and the energy storage response cycle. It is usually set between 5 minutes and 30 minutes. In this embodiment, it is set to 10 minutes, which can fully capture the power change trend while ensuring the real-time response, and is used to accurately identify energy storage adjustment needs.

[0149] From the selected temporary units, photovoltaic units suitable for energy storage regulation are further identified. The system first determines the state of charge (SOC) of the energy storage device corresponding to each temporary unit. When the SOC is lower than a preset SOC threshold, the difference between it and the threshold (charge difference) is calculated. If the difference exceeds the set charge difference threshold, it indicates that the energy storage system is in a low-power state and needs charging. At this time, if the average queuing time in the area where the unit is located is less than a preset queuing time threshold (i.e., the pressure on users to wait for charging services is relatively low), and the unit's output power is higher than the standard power value, then the unit is identified as a priority energy storage unit and marked as such. Conversely, if the queuing time is high (indicating that residents' charging needs are more urgent), the unit is not immediately designated as an energy storage unit. Instead, its output power level over a period of time is further considered to comprehensively determine whether it is suitable as a participating unit in energy storage regulation.

[0150] By dynamically coupling the state of charge (SBC) of the energy storage system, the charging demand pressure on the residential side (reflected by the average queuing time), and the current power generation capacity of the photovoltaic (PV) units, a logical "supply and demand linkage" mechanism is formed. On the one hand, the SBC difference reflects the strength of energy storage demand, the average queuing time reflects the urgency of the immediate electricity load of residents, and the output power measures the availability of the PV units. The linkage of these three factors enables dynamic prioritization and flexible allocation of energy storage under the background of power generation fluctuations. When user demand is controllable, priority is given to charging the energy storage system, and when demand is tight, the power supply to residents is maintained. This improves the flexibility of PV output regulation and user satisfaction, while reducing the risk of local overload and load conflicts.

[0151] Please continue reading. Figure 3 As shown, it is the determination logic diagram for determining the energy storage unit in this embodiment;

[0152] The process of determining a number of energy storage units based on the output power within a preset time period includes:

[0153] Calculate the standard deviation of the output power within a preset time period to obtain a determined power fluctuation value;

[0154] When the determined power fluctuation value is greater than a preset determined fluctuation threshold, the temporary unit is determined to be the energy storage unit, thereby identifying a number of energy storage units.

[0155] The preset fluctuation threshold is the standard deviation threshold used to determine whether the power fluctuation of the photovoltaic unit is significant. It depends on the system's tolerance to power fluctuation and the energy storage intervention strategy. It is usually set between 50W and 300W. In this embodiment, it is set to 150W, which can effectively identify photovoltaic units with significant power fluctuations and accurately trigger the energy storage unit to participate in regulation.

[0156] In determining the energy storage unit based on the output power within a preset time period, the system first performs statistical analysis on the output power data of all temporary units within that time period, calculates its standard deviation, and obtains the determined power fluctuation value for that period. Next, the system compares this fluctuation value with a preset determined fluctuation threshold. If the fluctuation value is greater than the threshold, it indicates that the corresponding photovoltaic unit's power output is unstable or exhibits abrupt changes during that time period. In this case, the photovoltaic unit is identified as an energy storage unit, thus enabling energy storage regulation intervention.

[0157] By introducing a "determined power fluctuation value" to reflect the stability of photovoltaic output over a short period, and combining it with a "preset determined fluctuation threshold" as a screening criterion, units with drastic power fluctuations can be accurately identified as priority energy storage targets. This achieves linkage between unstable power regions and energy storage response mechanisms, preventing power fluctuations from disturbing the microgrid while fully utilizing the energy storage system for peak shaving and valley filling. The greater the change in output power, the higher the regulation value of energy storage; therefore, this strategy can dynamically match system stability with the optimized scheduling of energy storage resources, improving the overall flexibility and response speed of the system.

[0158] Please continue reading. Figure 4 As shown, it is the determination logic diagram of the coordination anomaly determination unit in this embodiment;

[0159] The process of determining several coordinated abnormal units based on the interruption reconnection frequency, the output power of each energy storage unit, and the load demand includes:

[0160] Obtain the interruption reconnection frequency within a preset coordination time period to obtain a reconnection frequency set;

[0161] Obtain the output power within the preset coordination time period to obtain an output power set;

[0162] Obtain the load demand within the preset coordination time period to obtain a load demand set;

[0163] The output power at each moment within the preset coordination time period is normalized by the maximum-minimum value based on the output power set to obtain the power normalization value;

[0164] The load demand at each moment within the preset coordination time period is normalized by the maximum-minimum value based on the load demand set to obtain the demand normalization value.

[0165] Calculate the product of the power normalization value and the preset power weight to obtain the first product; calculate the product of the demand normalization value and the preset demand weight to obtain the second product.

[0166] Calculate the sum of the first product and the second product to obtain the synergy index;

[0167] Obtain all the aforementioned synergistic indices to obtain a set of synergistic indices;

[0168] Calculate the correlation coefficient between the reconnection frequency set and the coordination index set to obtain the coordination anomaly degree;

[0169] When the degree of coordination anomaly is greater than the preset standard anomaly, the energy storage unit is determined to be the coordination anomaly unit, thereby identifying several coordination anomaly units.

[0170] Maximum-minimum value normalization is a current technique and will not be elaborated upon here;

[0171] The preset coordination time refers to the time window used to evaluate the coordinated operation status of the energy storage unit. It depends on the response cycle of photovoltaic output and load demand, and is usually set between 10 minutes and 1 hour. In this embodiment, it is set to 30 minutes, which can accurately capture the dynamic changes in the coordinated response between the energy storage unit and the load.

[0172] The preset power weight is a coefficient used to measure the importance of the energy storage unit's output power in the calculation of the synergy index. It depends on the degree to which the energy storage response capability of the system depends on load regulation. It is usually set between 0.4 and 0.7. In this embodiment, it is set to 0.5, which can reasonably reflect the impact of energy storage output on the overall synergy capability.

[0173] The preset demand weight is a coefficient used to measure the importance of load demand in the calculation of the coordination index. It depends on the load-side adjustment priority and the real-time supply and demand balance pressure. It is usually set between 0.3 and 0.6. In this embodiment, it is set to 0.5, which can reflect the reference value of load changes in the determination of coordination anomalies.

[0174] The preset standard anomaly degree is a correlation coefficient threshold used to identify whether there is a coordination anomaly in the energy storage unit. It depends on the typical correlation characteristics between the frequency of interruptions and the coordination index in the historical operation data. It is usually set between 0.6 and 0.9. In this embodiment, it is set to 0.75, which can effectively screen out energy storage units that are potentially unstable or mismatched in operation.

[0175] Energy storage units exhibiting abnormal operating conditions are identified using a multi-dimensional data fusion method. First, within a preset coordination period, the frequency of interruptions and reconnections, output power, and load demand of each energy storage unit are acquired, forming frequency sets, output power sets, and load demand sets, respectively. Then, the output power set and load demand set are normalized to their maximum and minimum values, yielding corresponding normalized value sequences. These sequences are then multiplied by preset power and load demand weights, and the weighted sum is calculated to form a coordination index, reflecting the energy storage unit's contribution to supply and demand matching during that period. Finally, the Pearson correlation coefficient is calculated between the coordination index set and the interruption and reconnection frequency set to obtain the coordination anomaly degree. If this coordination anomaly degree exceeds a preset standard anomaly threshold, the energy storage unit is considered to exhibit abnormal behavior during operation and is thus classified as a coordination anomaly unit.

[0176] The correlation between charging interruption frequency and dynamic power-load coordination capability is used for assessment. Normalization and weighting are employed to achieve a unified measurement of parameters with different dimensions, avoiding the bias of relying on a single indicator. Output power reflects the energy supply capacity of the energy storage unit, load demand reflects the real-time load on the building's energy consumption side, and the ratio of these two measures the coordination capability. Interruption reconnection frequency reflects system stability and user experience. A high coordination index coupled with frequent interruptions, or a strong correlation between coordination capability and interruption trends, indicates potential fluctuations or imbalances in the energy storage response capability. This allows for accurate identification and isolation of units with abnormal coordination, enhancing intelligent diagnostic capabilities.

[0177] Specifically, the process of adjusting the preset total power fluctuation threshold based on the location and number of the coordinated abnormality units in each feeder and the state value of the high-current load switch, to obtain the adjusted power fluctuation threshold, includes:

[0178] The placement positions of the coordinated abnormality units within each feeder are obtained to determine the relative distances between the coordinated abnormality units and the starting point of the feeder.

[0179] Obtain the feeder number to which each of the aforementioned collaborative anomaly units belongs, and count the number of collaborative anomaly units in each feeder based on the feeder number to obtain the number of units;

[0180] The relative distance at the current moment is normalized by a maximum-minimum value based on all the relative distances mentioned, resulting in several normalized distances;

[0181] Calculate the standard deviation of all the normalized distances to obtain the distribution concentration.

[0182] Calculate the concentration factor based on the distribution concentration and the number of units;

[0183] The current distribution concentration is normalized based on all the distribution concentrations within the preset adjustment period to obtain the normalized concentration.

[0184] The current number of units is normalized based on the total number of units within the preset adjustment period to obtain the normalized number.

[0185] The risk index is obtained by weighting and summing the preset concentration weight, the normalized concentration degree, the preset quantity weight, the normalized quantity, the preset state value weight, and the high current load switch state value. Wherein, Q = w1×J + w2×S + w3×K, Q is the risk index, w1 is the preset concentration weight, J is the normalized concentration degree, w2 is the preset quantity weight, S is the normalized quantity, w3 is the preset state value weight, and K is the high current load switch state value.

[0186] When the risk index is greater than the preset risk index threshold, the preset total power fluctuation threshold is increased according to the relative deviation between the risk index and the preset risk index threshold and the preset adjustment coefficient to obtain the adjusted power fluctuation threshold, where T'=T×[1+e×(Q-Q0) / Q], T' is the adjusted power fluctuation threshold, T is the preset total power fluctuation threshold, e is the preset adjustment coefficient, and Q0 is the preset risk index threshold.

[0187] Feeder numbers are unique identifiers used to identify each feeder in a power distribution system. They depend on the distribution network topology and operation and maintenance requirements, and are typically set according to region, voltage level, or substation outgoing line sequence. This embodiment uses a numbering method based on substation outgoing line sequence, which clearly identifies the physical feeder path to which each coordinated anomaly unit belongs. This facilitates the statistical analysis of the distribution of anomaly units across different feeders, thereby supporting subsequent distribution concentration calculations and risk assessments.

[0188] The preset concentration weight is a weighting factor used to measure the degree of concentration of the distribution of collaborative abnormal units in the feeder on the impact of the risk index. It depends on the system's sensitivity to the influence of the deployment location distribution and is usually set between 0.2 and 0.5. In this embodiment, it is set to 0.35, which can reasonably reflect the importance of the deployment location in risk assessment.

[0189] The preset quantity weight is a weighting factor used to measure the contribution of the number of collaborative abnormal units to the overall risk index. It depends on the influence of the number of abnormal units on the stability of local fluctuations and is usually set between 0.3 and 0.6. In this embodiment, it is set to 0.45, which can accurately express the driving effect of unit density on system instability.

[0190] The preset state value weight is used to characterize the importance of the state value of the high current load switch on the system risk index. It depends on the contribution of load state fluctuations to power instability and is usually set between 0.1 and 0.4. In this embodiment, it is set to 0.2, which can appropriately reflect the impact of the actual load state on the risk level.

[0191] The preset risk index threshold is a boundary value used to determine whether the current system has entered a high-risk state. It depends on the system's tolerance to power fluctuations and reliability requirements. It is usually set between 0.5 and 0.8. In this embodiment, it is set to 0.65, which can enable timely identification and response to abnormal aggregation.

[0192] The preset adjustment coefficient is a proportional parameter used to adjust the power fluctuation threshold according to the relative deviation of the risk index. It depends on the system's response sensitivity to threshold changes and the adjustment range requirements. It is usually set between 1.1 and 1.5. In this embodiment, it is set to 1.2, which can dynamically enhance the power fluctuation threshold to adapt to system disturbances caused by abnormal unit density.

[0193] By obtaining the relative distance between the location of the coordinated abnormal unit within each feeder and the feeder starting point, and combining this with the corresponding feeder number to count the number of units, the standard deviation of the normalized distance is calculated to obtain the distribution concentration. The distribution concentration and the number of units are then normalized within a preset adjustment period (existing technology, not elaborated further) to obtain the normalized concentration and normalized number. Finally, the normalized concentration, normalized number, high-current load switch status value, and their respective preset weights are combined and weighted to calculate the risk index. When the risk index exceeds the preset risk index threshold, the preset total power fluctuation threshold is dynamically adjusted based on its relative deviation and preset adjustment coefficient, generating a new adjusted power fluctuation threshold.

[0194] A risk assessment model is constructed by correlating three parameters: deployment location (spatial distribution), number of units (fault scale), and switch status (load risk), enabling dynamic adjustment of the system's power fluctuation tolerance range. Normalized concentration reflects whether coordinated abnormal units are clustered; a smaller standard deviation indicates a higher degree of concentration, potentially posing a localized impact risk to a feeder. Normalized quantity reflects the absolute impact of the number of abnormal units. Load switch status reflects the current system's carrying capacity and impact response capability. The weighted average of these three parameters generates a risk index, which comprehensively assesses the current system stability. This allows for reasonable adjustment of the power fluctuation threshold to enhance system resilience and emergency response capabilities, avoiding misjudgments or omissions due to fixed threshold settings, and improving the flexibility and security of coordinated scheduling.

[0195] Specifically, the process of correcting the photovoltaic output prediction value based on the number of the coordinated abnormal units re-determined using the adjusted power fluctuation threshold within a preset correction period, and obtaining the corrected prediction value, includes:

[0196] Calculate the standard deviation of the number of units within the preset correction period to obtain the correction quantity fluctuation value;

[0197] When the corrected quantity fluctuation value is greater than the preset corrected quantity fluctuation threshold, the photovoltaic output prediction value is increased according to the relative deviation between the corrected quantity fluctuation value and the preset corrected quantity fluctuation threshold and the preset correction coefficient to obtain the corrected prediction value, where Y'=Y×[1+i×(L-L0) / L0], Y' is the corrected prediction value, Y is the photovoltaic output prediction value, i is the preset correction coefficient, L is the corrected quantity fluctuation value, and L0 is the preset corrected quantity fluctuation threshold.

[0198] The preset correction quantity fluctuation threshold is a benchmark value for judging whether the fluctuation of the number of abnormal collaborative units is abnormal. It depends on the sensitivity requirements of the system to quantity changes and is usually set between 1 and 5. In this embodiment, it is set to 3, which can effectively identify the period when the number of abnormal collaborative units in the system fluctuates greatly, providing a basis for prediction and correction.

[0199] The preset correction factor is a multiplier factor used to adjust the photovoltaic output prediction value according to the fluctuation deviation. It depends on the system's tolerance for risk redundancy and the prediction accuracy requirements. It is usually set between 0.1 and 1.0. In this embodiment, it is set to 0.5, which can moderately adjust the prediction value when the fluctuation is significant, thereby enhancing the robustness of system scheduling.

[0200] By statistically analyzing the number of cooperatively anomalous units re-determined using an adjusted power fluctuation threshold within a preset correction period, and calculating the standard deviation of the unit number during this period, a corrected quantity fluctuation value is obtained. When the corrected quantity fluctuation value exceeds the preset corrected quantity fluctuation threshold, it indicates that the number of cooperatively anomalous units fluctuates significantly, increasing the system instability. Therefore, based on the relative deviation between this fluctuation value and the threshold, and a preset correction coefficient, the photovoltaic output prediction value is adjusted upwards to obtain a corrected prediction value, thereby enhancing the robustness and adaptability of the prediction.

[0201] By monitoring the fluctuations in the number of anomalous collaborative units, a dynamic correlation was established between anomalous behavior and photovoltaic (PV) output prediction. The underlying logic is as follows: the greater the fluctuation in the number of anomalous collaborative units, the greater the potential risk of output interference in the system, thus requiring an increase in the predicted value to allow for safety redundancy. A preset correction threshold for the number of anomalous units serves as a benchmark for stability judgment. The correction value reflects the degree of system disturbance, and the relative deviation between the two reflects the actual deviation magnitude. A preset correction coefficient is used to quantify the correction magnitude. These parameters together construct a data-driven prediction correction mechanism, which helps improve the accuracy of PV output prediction and the stability of system scheduling.

[0202] Specifically, the process of executing a control command to adjust the preset photovoltaic power generation capacity of the coordinated anomaly unit based on the corrected predicted value and the preset predicted value range includes:

[0203] When the corrected predicted value is greater than the maximum value of the preset predicted value range, a control command to reduce the preset photovoltaic power generation is executed based on the relative deviation between the corrected predicted value and the maximum value of the preset predicted value range and the preset control coefficient, where F' = F × [1 - p × (U - Umax) / Umax], F' is the preset photovoltaic power generation after the control command is executed, F is the preset photovoltaic power generation, p is the preset control coefficient, Umax is the maximum value of the preset predicted value range, and U is the corrected predicted value;

[0204] When the corrected predicted value is less than the minimum value of the preset predicted value range, a control command to increase the preset photovoltaic power generation is executed based on the relative deviation between the minimum value of the preset predicted value range and the corrected predicted value, as well as the preset control coefficient, where F' = F × [1 + p × (Umin - U) / U], F' is the preset photovoltaic power generation after the control command is executed, and Umin is the minimum value of the preset predicted value range.

[0205] The preset control coefficient refers to the scaling factor used to adjust the photovoltaic power generation based on the power deviation. It depends on the system's response sensitivity to output fluctuations and safety redundancy requirements, and is usually set between 0.1 and 1.0. In this embodiment, it is set to 0.3, which can provide a moderate adjustment range when the power deviation is significant, balancing response speed and system stability.

[0206] First, the corrected photovoltaic (PV) power output forecast is compared with a preset forecast range. When the corrected forecast exceeds the maximum value of the preset range, the relative deviation is calculated and combined with a preset control coefficient to execute a control command that reduces PV power generation. When the corrected forecast is less than the minimum value of the preset range, a similar calculation method is used to execute a control command that increases PV power generation. This process adjusts the deviation between the corrected forecast and the preset range to ensure that PV power generation is controlled within a reasonable range, avoiding system overload or resource waste.

[0207] By using preset control coefficients and deviation calculations, the photovoltaic power generation can be dynamically adjusted based on the difference between the real-time corrected predicted value and the preset value range, thereby effectively achieving optimized power allocation. It can accurately control power increase and decrease, which not only improves the stability of system operation but also optimizes energy utilization efficiency.

[0208] Specifically, the process of predicting photovoltaic output based on all output power, state of charge value, load demand, charging request growth rate, average queuing time, and interruption reconnection frequency within a preset historical time period, as well as a preset prediction model, includes:

[0209] All the aforementioned output power, state of charge value, charging pile load demand, target side charging request growth rate, average queuing time and interruption reconnection frequency are time series aligned, missing value imputed and normalized to construct a multi-dimensional input feature matrix.

[0210] The multi-dimensional input feature matrix is ​​input into the preset prediction model to predict the photovoltaic output value.

[0211] First, the photovoltaic unit output power, energy storage SOC, building transformer load demand, charging request growth rate, average queuing time, and interruption reconnection frequency collected within a preset historical time period are aligned by timestamps. Then, missing values ​​in the data are interpolated or imputed forward. Finally, each variable is normalized to generate a multi-dimensional time series feature matrix with a unified scale. Subsequently, this feature matrix is ​​input into a preset multivariate time series prediction model to output the photovoltaic power output prediction value for future periods.

[0212] Time series alignment ensures consistency of parameters on the same time axis; missing value imputation ensures the integrity of model input; normalization eliminates bias caused by different units, providing balanced features for model learning; and the multi-dimensional feature matrix integrates generation side (output power), energy storage side (SOC), load side (transformer demand, charging growth rate, queuing time) and stability indicators (reconnection frequency), enabling the prediction model to simultaneously capture the dynamic changes and mutual influences of multiple sources of supply, energy storage and demand at the underlying logic level, thereby significantly improving the accuracy and robustness of short-term photovoltaic output prediction.

[0213] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for coordinated control of photovoltaic, energy storage, charging, and microgrid, characterized in that, include: Real-time data collection includes the output power of each photovoltaic unit on each feeder in the photovoltaic array operating based on preset photovoltaic power generation, the status value of high current load switches in each feeder, the state of charge value of the energy storage system, the load demand at the building transformer end, the growth rate of charging requests on the target side, the average queuing time, and the frequency of interruption and reconnection. The photovoltaic output forecast value is predicted based on all the output power, state of charge value, load demand, charging request growth rate, average queuing time, and interruption reconnection frequency within a preset historical period, as well as a preset prediction model. Several temporary units are selected based on the output power, the preset total power fluctuation threshold, and the charging request growth rate. Several energy storage units are determined based on the average queuing time, the output power of each temporary unit, and the state of charge value. Several coordinated abnormal units are determined based on the interruption reconnection frequency, the output power of each energy storage unit, and the load demand. The preset total power fluctuation threshold is adjusted according to the layout location and number of the coordinated abnormal unit in each feeder and the state value of the high current load switch to obtain the adjusted power fluctuation threshold. The photovoltaic output prediction value is corrected based on the number of units of the coordinated abnormal unit that is re-determined using the adjusted power fluctuation threshold within the preset correction period, and the corrected prediction value is obtained. Based on the corrected predicted value and the preset predicted value range, the coordinated anomaly unit executes a control command to adjust the preset photovoltaic power generation power. The process of correcting the photovoltaic output prediction value based on the number of the coordinated abnormal units re-determined using the adjusted power fluctuation threshold within a preset correction period, and obtaining the corrected prediction value, includes: Calculate the standard deviation of the number of units within the preset correction period to obtain the correction quantity fluctuation value; When the correction quantity fluctuation value is greater than the preset correction quantity fluctuation threshold, the photovoltaic output prediction value is increased according to the relative deviation between the correction quantity fluctuation value and the preset correction quantity fluctuation threshold and the preset correction coefficient to obtain the correction prediction value, where Y'=Y×[1+i×(L-L0) / L0], Y' is the correction prediction value, Y is the photovoltaic output prediction value, i is the preset correction coefficient, L is the correction quantity fluctuation value, and L0 is the preset correction quantity fluctuation threshold.

2. The method for coordinated control of photovoltaic energy storage and charging with microgrids according to claim 1, characterized in that, The process of selecting several temporary units based on the output power, the preset total power fluctuation threshold, and the charging request growth rate includes: Calculate the sum of all the output powers to obtain the total output power; Calculate the standard deviation of the total output power within the preset temporary duration to obtain the total power fluctuation value; When the total power fluctuation value is greater than the preset total power fluctuation threshold and the charging request growth rate is greater than the preset standard growth rate, several temporary units are selected based on the output power and the total power fluctuation value.

3. The method for coordinated control of photovoltaic energy storage and charging with microgrids according to claim 2, characterized in that, Based on the output power and the total power fluctuation value, several temporary units are selected, including: Calculate the standard deviation of the total output power of each photovoltaic unit within the next preset temporary duration to obtain the output power fluctuation value; Calculate the relative deviation between the output power fluctuation value and the total power fluctuation value to obtain a temporary judgment value; When the temporary determination value is greater than the preset temporary determination threshold, the photovoltaic unit is determined to be the temporary unit, so as to filter out a number of temporary units.

4. The method for coordinated control of photovoltaic energy storage and charging with microgrids according to claim 3, characterized in that, The process of determining a number of energy storage units based on the average queuing time, the output power of each temporary unit, and the state of charge value includes: When the state of charge value is less than a preset state of charge threshold, the difference between the preset state of charge threshold and the state of charge value is calculated to obtain the charge difference. When the charge difference is greater than a preset charge difference threshold, the average queuing time is less than a preset time threshold, and the output power is greater than a preset standard power, the temporary unit is determined to be the energy storage unit, thereby identifying several energy storage units. When the charge difference is greater than a preset charge difference threshold and the average queuing time is greater than or equal to the preset time threshold, a number of energy storage units are determined based on the output power within a preset time period.

5. The method for coordinated control of photovoltaic energy storage and charging with microgrids according to claim 4, characterized in that, The process of determining a number of energy storage units based on the output power within a preset time period includes: Calculate the standard deviation of the output power within a preset time period to obtain a determined power fluctuation value; When the determined power fluctuation value is greater than a preset determined fluctuation threshold, the temporary unit is determined to be the energy storage unit, thereby identifying a number of energy storage units.

6. The method for coordinated control of photovoltaic energy storage and charging with microgrids according to claim 5, characterized in that, The process of determining several coordinated abnormal units based on the interruption reconnection frequency, the output power of each energy storage unit, and the load demand includes: Obtain the interruption reconnection frequency within a preset coordination time period to obtain a reconnection frequency set; Obtain the output power within the preset coordination time period to obtain an output power set; Obtain the load demand within the preset coordination time period to obtain a load demand set; The output power at each moment within the preset coordination time period is normalized by the maximum-minimum value based on the output power set to obtain the power normalization value; The load demand at each moment within the preset coordination time period is normalized by the maximum-minimum value based on the load demand set to obtain the demand normalization value. Calculate the product of the power normalization value and the preset power weight to obtain the first product; calculate the product of the demand normalization value and the preset demand weight to obtain the second product. Calculate the sum of the first product and the second product to obtain the synergy index; Obtain all the aforementioned synergistic indices to obtain a set of synergistic indices; Calculate the correlation coefficient between the reconnection frequency set and the coordination index set to obtain the coordination anomaly degree; When the degree of coordination anomaly is greater than the preset standard anomaly, the energy storage unit is determined to be the coordination anomaly unit, thereby identifying several coordination anomaly units.

7. The method for coordinated control of photovoltaic, energy storage, charging, and microgrids according to claim 6, characterized in that, The process of adjusting the preset total power fluctuation threshold based on the location and number of the coordinated abnormality units in each feeder and the state value of the high-current load switch includes: The placement positions of the coordinated abnormality units within each feeder are obtained to determine the relative distances between the coordinated abnormality units and the starting point of the feeder. Obtain the feeder number to which each of the aforementioned collaborative anomaly units belongs, and count the number of collaborative anomaly units in each feeder based on the feeder number to obtain the number of units; The relative distance at the current moment is normalized by a maximum-minimum value based on all the relative distances mentioned, resulting in several normalized distances; Calculate the standard deviation of all the normalized distances to obtain the distribution concentration. Calculate the concentration factor based on the distribution concentration and the number of units; The current distribution concentration is normalized based on all the distribution concentrations within the preset adjustment period to obtain the normalized concentration. The current number of units is normalized based on the total number of units within the preset adjustment period to obtain the normalized number. The risk index is obtained by weighting and summing the preset concentration weight, the normalized concentration degree, the preset quantity weight, the normalized quantity, the preset state value weight, and the state value of the high current load switch. When the risk index is greater than the preset risk index threshold, the preset total power fluctuation threshold is increased according to the relative deviation between the risk index and the preset risk index threshold and the preset adjustment coefficient to obtain the adjusted power fluctuation threshold.

8. The method for coordinated control of photovoltaic, energy storage, charging, and microgrids according to claim 7, characterized in that, The process of executing a control command to adjust the preset photovoltaic power generation power of the coordinated anomaly unit based on the corrected predicted value and the preset predicted value range includes: When the corrected predicted value is greater than the maximum value of the preset predicted value range, a control command to reduce the preset photovoltaic power generation is executed based on the relative deviation between the corrected predicted value and the maximum value of the preset predicted value range and the preset control coefficient. When the corrected predicted value is less than the minimum value of the preset predicted value range, a control command to increase the photovoltaic power generation is executed based on the relative deviation between the minimum value of the preset predicted value range and the corrected predicted value, as well as the preset control coefficient.

9. The method for coordinated control of photovoltaic energy storage and charging with microgrids according to claim 8, characterized in that, The process of predicting photovoltaic output based on all output power, state of charge value, load demand, charging request growth rate, average queuing time, and interruption reconnection frequency within a preset historical time period, as well as a preset prediction model, includes: All the aforementioned output power, state of charge value, charging pile load demand, target side charging request growth rate, average queuing time and interruption reconnection frequency are time series aligned, missing value imputed and normalized to construct a multi-dimensional input feature matrix. The multi-dimensional input feature matrix is ​​input into the preset prediction model to predict the photovoltaic output value.

Images