A smart grid energy optimization management method and system

Abstract

The application provides a kind of smart grid energy optimization management method and system, it is related to data processing technical field, the method includes: step 1, synchronously collects photovoltaic output fluctuation data, user load demand data and power grid frequency state data, to obtain original monitoring data set;Step 2, original monitoring data set is mapped to space-time characteristics, extracts multi-dimensional feature vector, and determines the principal feature axis system based on covariance analysis, calculates the dynamic coupling degree between feature axis system to divide control domain;Select time series key feature quantity in control domain, construct load response state transition path, and deduce user behavior correction coefficient accordingly, realize load dynamic balance through multi-period rolling strategy, generate load distribution scheme.The application realizes the balance optimization of smart grid energy supply and demand by real-time dynamic regulation of power output, generating incentive electricity price signal and linking user demand response.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a smart grid energy optimization management method and system. Background Technology

[0002] In smart grid energy management in residential communities, traditional technologies have some shortcomings. Traditional technologies are insufficient in terms of load forecasting accuracy, mostly relying on linear extrapolation from electricity consumption data. They cannot fully consider the dynamic changes in residents' living habits and the impact of unforeseen factors. For example, on a weekday in summer, traditional methods predict that the electricity load of the community between 6 pm and 8 pm will be 800 kilowatts based on data from the same period in the past. However, in reality, there is a sudden and sustained high temperature during that week, and residents turn on their air conditioners early after returning home. At the same time, some families use various electrical appliances at home because their children are on summer vacation early, resulting in an actual electricity load of 1,000 kilowatts during that period. The predicted value deviates significantly from the actual value. This deviation may cause the power grid to experience power shortages during peak electricity consumption periods due to insufficient preparation, affecting residents' normal electricity use.

[0003] In addition, traditional technologies lack flexibility in the regulation of distributed energy. For distributed energy in communities, such as solar photovoltaic panels and small energy storage devices, traditional management methods often adopt fixed access and dispatch modes, which are difficult to dynamically adjust according to real-time energy output and residents' electricity demand. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a smart grid energy optimization management method and system, which improves the stability of power grid supply and energy utilization efficiency, and reduces electricity costs and resource waste through load forecasting and dynamic distributed energy regulation.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows: Firstly, a smart grid energy optimization management method, the method comprising: Step 1: Simultaneously collect photovoltaic power output fluctuation data, user load demand data, and grid frequency status data to obtain the raw monitoring dataset; Step 2: Perform spatiotemporal feature mapping on the original monitoring dataset, extract multidimensional feature vectors, determine the dominant feature axis system based on covariance analysis, calculate the dynamic coupling degree between feature axes system to divide the control domain; select key time-series feature quantities within the control domain, construct the load response state transition path, derive the user behavior correction coefficient accordingly, achieve dynamic load balance through a multi-period rolling strategy, and generate a load allocation scheme. Step 3: Based on the load allocation scheme, achieve supply and demand balance by modifying the power supply strategy and pricing strategy, and obtain power dispatch instructions that match supply and demand. Step 4: Based on power dispatch instructions, and considering the fluctuation characteristics of photovoltaic output, dynamically change the energy storage charging and discharging instructions in combination with probability distribution constraints; Step 5: Based on the energy storage charging and discharging instructions, the distributed computing unit coordinates the information exchange between photovoltaic, energy storage and load units to generate a power allocation scheme; Step 6: Based on the power allocation scheme, construct a multi-type power supply coordination process, evaluate the power output ratio and coordination benefits according to dynamic contribution, and form a low-carbon scheduling strategy. Step 7: Based on the low-carbon dispatch strategy, the low-carbon dispatch strategy is periodically adjusted through the grid frequency deviation characteristics to generate an incentive electricity price signal and feed it back to the user response process.

[0006] Secondly, a smart grid energy optimization management system includes: The data acquisition module is used to simultaneously collect photovoltaic power output fluctuation data, user load demand data, and power grid frequency status data to obtain the raw monitoring dataset; The analysis and regulation module is used to perform spatiotemporal feature mapping on the original monitoring dataset, extract multidimensional feature vectors, determine the dominant feature axis system based on covariance analysis, calculate the dynamic coupling degree between feature axes system to divide the control domain; select key time-series feature quantities within the control domain, construct load response state transition paths, derive user behavior correction coefficients accordingly, achieve dynamic load balance through multi-period rolling strategy, and generate load allocation schemes. The balance dispatch module is used to achieve supply and demand balance by modifying the power supply strategy and pricing strategy according to the load allocation plan, and to obtain power dispatch instructions that match supply and demand. The energy storage control module is used to dynamically change the energy storage charging and discharging commands based on power dispatch instructions, taking into account the fluctuation characteristics of photovoltaic output and combining probability distribution constraints. The coordination and allocation module is used to coordinate the information exchange between photovoltaic, energy storage and load units through the distributed computing unit according to the energy storage charging and discharging instructions, and generate a power allocation scheme. The collaborative scheduling module is used to construct a collaborative process for multiple types of power sources based on the power allocation scheme, evaluate the power output ratio and collaborative benefits according to the dynamic contribution, and form a low-carbon scheduling strategy. The frequency feedback module is used to periodically adjust the low-carbon dispatch strategy based on the grid frequency deviation characteristics, generate incentive price signals, and feed them back to the user response process.

[0007] Thirdly, a computing device, comprising: One or more processors; A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method.

[0008] Fourthly, a computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0009] The above-described solution of the present invention has at least the following beneficial effects: In the load forecasting stage, an adaptive learning mechanism is constructed by integrating multi-source information such as real-time meteorological data, residents' dynamic electricity consumption behavior tags, and holiday patterns to improve the dynamic accuracy of forecasts. Compared with traditional methods, it can quickly respond to the impact of sudden factors (such as extreme weather and temporary community activities) on electricity load, allowing sufficient adjustment time for grid dispatch and effectively avoiding power supply gaps caused by the lag in traditional forecasts. In terms of distributed energy utilization, relying on intelligent collaborative control algorithms, real-time matching of photovoltaic, energy storage, and residents' electricity demand is achieved. When the power generation of community photovoltaic panels fluctuates with changes in sunshine, the power distribution can be dynamically adjusted, and energy storage can be automatically called to supplement the grid power supply. In terms of power supply reliability, an intelligent early warning mechanism is constructed by real-time monitoring of grid line load, equipment status, and residents' electricity consumption trends. When an area's line is about to be overloaded, it can automatically guide residents in that area to use high-energy-consuming appliances during off-peak hours, while allocating energy storage equipment for temporary supplementary power, keeping the line load within a safe range, reducing the probability of power outages caused by overload under traditional management models, and improving the continuity of residents' electricity use. Attached Figure Description

[0010] Figure 1 This is a flowchart illustrating a smart grid energy optimization management method provided by an embodiment of the present invention.

[0011] Figure 2 This is a schematic diagram of a smart grid energy optimization management system provided by an embodiment of the present invention. Detailed Implementation

[0012] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0013] like Figure 1 As shown, an embodiment of the present invention proposes a smart grid energy optimization management method, the method comprising the following steps: Step 1: Simultaneously collect photovoltaic power output fluctuation data, user load demand data, and grid frequency status data to obtain the raw monitoring dataset; Step 2: Perform spatiotemporal feature mapping on the original monitoring dataset, extract multidimensional feature vectors, determine the dominant feature axis system based on covariance analysis, calculate the dynamic coupling degree between feature axes system to divide the control domain; select key time-series feature quantities within the control domain, construct the load response state transition path, derive the user behavior correction coefficient accordingly, achieve dynamic load balance through a multi-period rolling strategy, and generate a load allocation scheme. Step 3: Based on the load allocation scheme, achieve supply and demand balance by modifying the power supply strategy and pricing strategy, and obtain power dispatch instructions that match supply and demand. Step 4: Based on power dispatch instructions, and considering the fluctuation characteristics of photovoltaic output, dynamically change the energy storage charging and discharging instructions in combination with probability distribution constraints; Step 5: Based on the energy storage charging and discharging instructions, the distributed computing unit coordinates the information exchange between photovoltaic, energy storage and load units to generate a power allocation scheme; Step 6: Based on the power allocation scheme, construct a multi-type power supply coordination process, evaluate the power output ratio and coordination benefits according to dynamic contribution, and form a low-carbon scheduling strategy. Step 7: Based on the low-carbon dispatch strategy, the low-carbon dispatch strategy is periodically adjusted through the grid frequency deviation characteristics to generate an incentive electricity price signal and feed it back to the user response process.

[0014] In this embodiment of the invention, by simultaneously collecting multi-dimensional raw monitoring data, it is helpful to more accurately grasp the actual situation of photovoltaic output, user load, and grid status. In terms of load regulation, combining phased regulation to achieve instantaneous load balance can effectively cope with dynamic changes in user electricity consumption behavior, improve the rationality and flexibility of load allocation, and reduce supply and demand imbalance during peak electricity consumption periods. By correcting power supply and pricing strategies to achieve supply and demand balance and generating precise power dispatch instructions, power supply can better match user demand, avoid oversupply or undersupply, and improve the stability of grid operation. Dynamically changing energy storage charging and discharging instructions in response to photovoltaic output fluctuations can give full play to the regulatory role of energy storage devices, reduce the impact of unstable photovoltaic output on the grid, and enhance the photovoltaic energy... By leveraging distributed computing units to coordinate information exchange among various units and generate power allocation schemes, the system promotes efficient collaboration among photovoltaic, energy storage, and load units, optimizes the flow and distribution of energy among units, constructs multi-type power source collaboration processes, and dynamically evaluates and allocates output ratios and benefits. The resulting low-carbon dispatch strategy helps promote the full utilization of clean energy, reduce carbon emissions, and achieve green and low-carbon operation. Based on the periodic adjustment of the dispatch strategy according to the grid frequency deviation and the generation of incentive electricity price signals to feed back to users, it can guide users to rationally adjust their electricity consumption behavior, further promote the balance of grid supply and demand, and enhance users' enthusiasm for participating in grid optimization management through price incentives, forming a virtuous cycle of electricity consumption, and comprehensively improving the energy optimization management efficiency and low-carbon environmental protection level of the smart grid.

[0015] In a preferred embodiment of the present invention, step 1 above, which involves simultaneously collecting photovoltaic power output fluctuation data, user load demand data, and grid frequency status data to obtain the raw monitoring dataset, may include: In this embodiment of the invention, the photovoltaic side determines the photovoltaic power stations (such as distributed rooftop photovoltaics and centralized photovoltaic power stations) and key monitoring points (photovoltaic array combiner boxes and inverter output terminals) that need to be collected; the user side delineates the load coverage area (such as residential community distribution substations, industrial park main incoming lines, and commercial complex distribution rooms); and the power grid side selects key frequency monitoring nodes (such as regional power grid main substation busbars, load center tie lines, and frequency stability sensitive area monitoring points) to ensure that the three types of data cover power grid operation-related scenarios.

[0016] On the photovoltaic side, photovoltaic power sensors are installed (to monitor the output power of the modules in real time) and inverter data acquisition terminals are installed (to record active / reactive output and operating status) to ensure that the equipment supports wide-range power monitoring (to adapt to power fluctuations caused by changes in sunlight). On the user side, smart meters (with second-level data recording function) and load monitors (to distinguish between active and reactive loads) are deployed to adapt to the electricity consumption characteristics of different user types (residential, industrial, and commercial). On the grid side, high-precision frequency transmitters (measurement range 45-55Hz) and synchronous phasor measurement devices (PMU, supporting high-frequency sampling) are configured to ensure the real-time performance and accuracy of frequency data.

[0017] The photovoltaic power sensor is calibrated with a standard power source, and the error is controlled within ±1%; the smart meter is calibrated by a metering and testing institution, and the error is ≤ ±0.5%; the frequency transmitter is calibrated with a standard frequency signal source, and the resolution reaches 0.001Hz. The data units are unified, with photovoltaic output and user load in "kW" (which can be expanded to "MW" for large-capacity scenarios), and grid frequency in "Hz", to avoid data misalignment caused by unit differences.

[0018] Establish a unified clock synchronization system to provide time synchronization for all data acquisition devices via BeiDou / GPS satellite clocks, ensuring that the device time error is ≤10 milliseconds; set the data acquisition cycle according to monitoring needs (e.g., 1 second / time for high-frequency monitoring, 5 minutes / time for regular monitoring), and clarify the start and end times of data acquisition (e.g., continuous 72-hour real-time monitoring, typical day monitoring in a specific season), thus forming a fixed data acquisition rhythm.

[0019] The photovoltaic power sensor captures the power fluctuations of the photovoltaic module in real time due to changes in light intensity and temperature, and reads the DC side power value at set intervals. The inverter terminal synchronously collects the AC side output active power, records the dynamic changes of power over time (such as the power peak at noon on sunny days and intermittent fluctuations on cloudy days), and removes abnormal data caused by equipment failure (such as inverter protection shutdown), communication interruption (data transmission failure) or extreme weather (such as invalid values ​​caused by sudden power drop due to sandstorms), retaining a continuous and true power fluctuation sequence.

[0020] Smart meters record users' real-time power consumption according to a set cycle. User-side monitoring instruments summarize the load of each user in the area (such as the total load of 3 buildings in a community or the real-time load of a production line in a factory) to form total user demand data. Residential load data excludes value jumps caused by meter malfunctions (such as instantaneous power jumping from 2kW to 100kW), and industrial load data excludes invalid data caused by the momentary surge of equipment starting and stopping (such as short-term super-high power when a motor starts). This ensures that the data reflects the smooth fluctuations of users' actual electricity demand (such as the load characteristics of morning and evening peak hours).

[0021] The frequency transmitter and PMU device monitor the grid operating frequency in real time, and record the instantaneous frequency value (such as 50.02Hz, 49.98Hz) at set intervals. It filters out high-frequency noise caused by electromagnetic interference (such as frequency spikes of more than ±1Hz) and abnormal values ​​caused by equipment sampling errors (such as continuous data deviating from the normal range of 50Hz±0.2Hz), ensuring that the recorded frequency data accurately reflects the grid supply and demand balance (such as frequency dips caused by sudden load increases and frequency spikes caused by sudden increases in photovoltaic output).

[0022] Based on a unified clock synchronization system, a unique timestamp is added to each piece of photovoltaic power output data, user load data, and grid frequency data (formatted as "year-month-day hour:minute:second,millisecond", e.g., 2025-07-18). (09:30:00.123) Ensure that the timestamps of the three types of data at the same acquisition time (e.g., 09:30:00.123) are completely consistent, eliminating time misalignment caused by equipment clock deviation (e.g., photovoltaic data timestamp leading load data by 50 milliseconds). Check whether each time point has a valid photovoltaic output value, user load value, and grid frequency value simultaneously. If there is missing data at a certain time point (e.g., only photovoltaic data is missing) or the timestamp does not match (e.g., the load data timestamp is 09:30:00.124, which differs from other data by 1 millisecond), then delete all data at that time point to ensure that the three types of data at each time point in the data sequence are completely corresponding, with no isolated or misaligned records. Each record contains four core fields: "acquisition timestamp", "real-time photovoltaic output value", "total user load demand value", and "instantaneous grid frequency value". Eliminate the differences in the original data format of different devices (e.g., photovoltaic data contains inverter status codes, load data contains voltage values ​​and other redundant information), and retain only the core values ​​to form a basic data sequence arranged continuously in chronological order.

[0023] The total number of valid data entries within the collection period is counted, and the data integrity rate is calculated as: number of valid data entries ÷ theoretical total number of data entries (set period × number of collections) × 100%. If the integrity rate is greater than or equal to the preset threshold (e.g., 95%), the data collection is considered qualified. If the integrity rate is insufficient, the cause is investigated (e.g., equipment offline, communication interruption), and data for the missing period is collected in a targeted manner until the standard is met. The standardized data sequence that has passed the integrity assessment is integrated in chronological order to form an original dataset containing all valid monitoring records. The dataset must completely cover all synchronous data points within the collection period, and each record must clearly correspond to the photovoltaic output fluctuation, user load demand, and grid frequency status at a specific time.

[0024] In a preferred embodiment of the present invention, step 2 above, which involves performing spatiotemporal feature mapping on the original monitoring dataset, extracting multidimensional feature vectors, determining the dominant feature axis system based on covariance analysis, and calculating the dynamic coupling degree between feature axes to divide the control domain, may include: Step 220: Align the photovoltaic output, load demand and grid frequency data in the original monitoring dataset by timestamp, and map them into a multi-dimensional feature vector set with time dimension through a spatiotemporal correlation matrix; Step 221: Perform covariance matrix decomposition on the multidimensional feature vector set, and identify orthogonal feature vectors with prominent variance contribution rates through principal component analysis to form the dominant feature axis system; Step 222: Calculate the rate of change of the included angle and the ratio of projected energy between the dominant feature axes based on the sliding time window, and fuse them to generate a dynamic coupling index; Step 223: Set a dynamic threshold based on the gradient distribution of the dynamic coupling degree index, cluster the continuous and stable coupling degree regions into independent control domains, and divide the independent control domains with similar dynamic characteristics.

[0025] In this embodiment of the invention, photovoltaic power output data, load demand data, and grid frequency data are filtered from a database or file storing raw monitoring data. Photovoltaic power output data refers to the actual electrical energy output of the photovoltaic power station at different times, load demand data refers to the electrical energy demand of power users at different times, and grid frequency data refers to the frequency of the grid system at different times. The timestamp information corresponding to each data record is checked. The timestamp is accurate to the smallest unit of data collection, such as seconds or minutes. Data with the same timestamp among the three types of data (photovoltaic power output, load demand, and grid frequency) are grouped together. For data with non-overlapping timestamps, if the interval between adjacent timestamps is small, linear interpolation is used to supplement the data at intermediate times to ensure that the three types of data correspond one-to-one in the time dimension, forming three sets of data sequences arranged in chronological order.

[0026] A two-dimensional spatiotemporal correlation matrix is ​​constructed, with chronological order as the rows and photovoltaic output, load demand, and grid frequency as the columns. Each element in the matrix is ​​the specific value of a certain type of data at a corresponding time point. For example, the element in the 3rd row and 1st column is the photovoltaic output value at the 3rd time point. From the spatiotemporal correlation matrix, the elements of each row are extracted in chronological order to form a vector. Each vector contains the three values ​​of photovoltaic output, load demand, and grid frequency at that time point. Since these vectors are arranged in chronological order, the entire set becomes a multidimensional feature vector set with a time dimension, and each vector carries a corresponding time identifier.

[0027] Arrange the multidimensional feature vector set obtained in step 220 in order, and clarify the feature type corresponding to each value in each vector, that is, which value corresponds to photovoltaic output, which corresponds to load demand, and which corresponds to grid frequency; calculate the covariance between different features in the multidimensional feature vector set. For the two features of photovoltaic output and load demand, calculate the average of the product of their numerical deviations at all time points. That is, first calculate the deviation between photovoltaic output and the average photovoltaic output at each time point, and the deviation between load demand and the average load demand at that time point. Multiply these two deviations, and then average the product over all time points to obtain the covariance between the two. Calculate the covariance between photovoltaic output and grid frequency, and between load demand and grid frequency in the same way. These covariance values ​​constitute the covariance matrix.

[0028] The covariance matrix is ​​decomposed, resulting in a series of eigenvalues ​​and corresponding eigenvectors. The eigenvalues ​​are sorted in descending order, and each eigenvalue represents the variance explained by the corresponding eigenvector. The proportion of each eigenvalue to the sum of all eigenvalues ​​is calculated, and this proportion is the variance contribution rate. The eigenvectors with the highest variance contribution rates and the highest cumulative variance contribution rates (e.g., above 85%) are selected. These eigenvectors are orthogonal to each other (i.e., the angle between them is 90 degrees, and they are independent of each other). Combining these selected orthogonal eigenvectors forms the dominant eigenaxis system, with each eigenvector representing an eigenaxis.

[0029] Set a fixed-length time window, for example, containing 10 consecutive time points. Move this window along the time axis from the starting point of the data, moving one time point at a time to obtain multiple consecutive time window datasets. Within each sliding time window, take two feature axes from the dominant feature axis system and calculate the initial angle between these two feature axes within the window, i.e., calculate the angle using their direction vectors. Then calculate the angle at the end of the window. Subtract the initial angle from the ending angle to obtain the change in angle. Divide the change in angle by the time length of the window to obtain the rate of change of angle. Perform this calculation for all pairwise feature axis combinations.

[0030] For each dominant feature axis within a sliding time window, the data of the multidimensional feature vectors concentrated within that window are projected onto each dominant feature axis. The sum of squares of all projected values ​​on each feature axis is calculated, and this sum of squares is the projected energy of that feature axis within the window. The ratio of the projected energy of any two feature axes is calculated to obtain the projected energy ratio. This calculation is performed for all pairwise feature axis combinations. For each sliding time window, all calculated angle change rates and projected energy ratios are comprehensively processed. For example, these indicators are first normalized to ensure they are within the same numerical range. Then, different weights are assigned according to the importance of each indicator. The weighted indicators are summed to obtain the dynamic coupling index of that window. As the window slides, a series of dynamic coupling indices that change over time are generated.

[0031] Calculate the difference in dynamic coupling index between two adjacent sliding time windows. This difference reflects the rate of change of the coupling index, i.e., the gradient. Arrange the gradient values ​​of all adjacent windows in chronological order to form a gradient distribution. Analyze the gradient distribution to identify regions with smaller gradient values ​​(indicating slow changes in coupling) and regions with larger gradient values ​​(indicating drastic changes in coupling). Based on the statistical characteristics of the gradient distribution, such as the average value and standard deviation of the gradient, set a dynamic threshold. When the gradient value is less than or equal to the threshold, the coupling is considered to be in a stable state; when the gradient value is greater than the threshold, the coupling is considered to be in a state of drastic change.

[0032] Based on the dynamic threshold, continuous time windows are found in the dynamic coupling degree index sequence. The gradient values ​​of these windows are all less than or equal to the dynamic threshold, which are regions with continuous and stable coupling degree. These continuous and stable regions are grouped into one category, and each category is a preliminary clustering result. Each continuous and stable region obtained by clustering is analyzed to determine whether the dynamic coupling degree index in these regions has similar numerical ranges and changing trends, ensuring that the dynamic characteristics in each region are similar. These clustered regions are determined as independent control domains, thereby dividing independent control domains with similar dynamic characteristics.

[0033] Data is aligned by timestamp and multidimensional feature vector sets are generated, ensuring accurate time-dimensional correspondence between photovoltaic output, load demand, and grid frequency data. This reduces errors caused by data asynchrony. By identifying dominant feature axes through covariance matrix decomposition and principal component analysis, the key features that contribute the most to variance can be extracted from complex multidimensional data. Redundant information is eliminated, allowing the analysis to focus on the main factors affecting the system's dynamic characteristics. By calculating the dynamic coupling degree index based on a sliding time window and combining the angle change rate and projected energy ratio, the dynamic correlation changes between the dominant characteristic axes can be captured in real time, comprehensively reflecting the coupling degree of the system in different time windows, and providing a dynamic quantitative basis for the division of control domains. Dynamic thresholds are set according to the gradient distribution of the dynamic coupling degree index, and independent control domains are clustered, so that the divided control domains have similar dynamic characteristics. The dynamic changes of the system within each control domain are relatively stable, facilitating the formulation of precise control strategies for different control domains, improving the effectiveness and targeting of power grid control, and ensuring the stable operation of the power grid system. The entire process, through sliding time windows and dynamic thresholds, can adapt to the dynamic changes in data such as photovoltaic output and load demand in the power grid system, making the division of control domains real-time and dynamic, enabling timely responses to changes in system operating status, and improving the adaptive control capability of the power grid.

[0034] In another preferred embodiment of the present invention, key time-series features are selected within the control domain to construct a load response state transition path, and user behavior correction coefficients are derived accordingly. A multi-period rolling strategy is then used to achieve dynamic load balancing and generate a load allocation scheme, which may include: Step 2220: Within the defined control domain, select key time-series features that characterize the load change trend, and construct a load response state transition path based on the key time-series features. Step 2221: Based on the load response state transition path, calculate the deviation of user electricity consumption behavior, and generate user behavior correction coefficient based on the calculated deviation of user electricity consumption behavior. Step 2222: Adopt a multi-period rolling strategy, combined with user behavior correction coefficient, to adjust the load allocation weight. A dynamic load balance is achieved through continuous adjustment of the load allocation weight, and the final load allocation scheme is generated when the dynamic load balance is achieved.

[0035] In this embodiment of the invention, load-related data for all time points within a predefined independent control domain are extracted, including load demand data and corresponding timestamp information, to ensure that the data covers the complete time range of the control domain. The changes in load demand data over time are analyzed, and the load change at different time intervals is calculated, such as the load difference between adjacent hours and adjacent days. The peak occurrence time, trough occurrence time, and rate of change of load changes are observed, and features that can clearly reflect the load increase, decrease trend, and stable state are selected, such as the daily average load fluctuation amplitude, the duration of the load peak, and the slope of the load change. These features constitute key time-series features.

[0036] Based on the numerical range of key time-series characteristics, the load response state is divided into different types, such as low load stable state, load rising state, high load stable state, and load falling state. Each state corresponds to a specific numerical range of key time-series characteristics. For example, when the slope of the load change is positive and within a certain range, it is classified as a load rising state. The changes in load state are sorted out in chronological order, and the load state corresponding to each time point is recorded. The number of times and specific time points of transition from one state to another are counted, such as the specific time of transition from low load stable state to load rising state. These states are connected in chronological order to form a continuous load response state transition path, and the duration and transition time of each state are marked on the path.

[0037] Based on the load response state transition path under the same or similar conditions, and combined with the typical electricity consumption patterns of users within the control domain, a standard user electricity consumption behavior curve is determined. This curve reflects the load state transition pattern and corresponding load value changes under normal conditions. Actual user electricity consumption behavior data is extracted from the load response state transition path of the current control domain, including the actual load value and corresponding state transition at each time point. At the same time point, the difference between the actual electricity consumption behavior data and the standard electricity consumption behavior curve is compared, and the difference between the actual load value and the standard load value at each time point is calculated. The absolute values ​​of these differences are accumulated and then divided by the total time length to obtain the average deviation value. Simultaneously, the number of times the actual state transition is inconsistent with the standard state transition is counted, and combined with the time length of each inconsistency, the user electricity consumption behavior deviation is calculated comprehensively. The larger the deviation value, the greater the difference between the actual electricity consumption behavior and the standard behavior.

[0038] The baseline value of the correction coefficient under standard conditions is set to 1. Based on the calculated deviation, the baseline value is adjusted. If the deviation is 0, the correction coefficient remains at 1. If a deviation exists, when the actual load is higher than the standard load, the correction coefficient is less than 1, and the larger the deviation, the smaller the correction coefficient, in order to reduce the user's allocation weight and guide their actual load back to the standard level. When the actual load is lower than the standard load, the correction coefficient is greater than 1, and the larger the deviation, the larger the correction coefficient, in order to increase the user's allocation weight and encourage them to increase electricity consumption during low-load periods. In this way, the deviation is converted into a corresponding user behavior correction coefficient.

[0039] The total time range of the control domain is divided into multiple consecutive time periods at fixed intervals, such as each time period being 1 hour, forming multiple rolling time windows. Based on factors such as the load allocation ratio of different users or power consumption areas within the control domain and the importance of power demand, an initial load allocation weight is set for each user or power consumption area. The sum of the initial weights is 1. In the first time period, the user behavior correction coefficient is multiplied by the initial load allocation weight to obtain the preliminary adjustment weight for that time period. The difference between the actual load and the expected load in that time period is checked. If a difference exists, the preliminary adjustment weight is fine-tuned according to the magnitude of the difference to make the load allocation more in line with actual demand. When entering the next time period, the weight adjusted in the previous time period is used as the initial weight, and the adjustment is made again in combination with the user behavior correction coefficient for that time period. This process is repeated. The load allocation weight is continuously adjusted in each time period based on the adjustment result of the previous time period and the current correction coefficient to ensure that the load allocation in each time period can adapt to changes in actual power consumption behavior.

[0040] After each period of adjustment, the difference between the actual load and the system power supply capacity is calculated. When this difference is within the preset allowable range and remains stable for multiple consecutive periods, it is determined that a dynamic load balance state has been achieved. When a dynamic load balance state is achieved, the load allocation weight of each user or power consumption area is recorded. Based on these weights, the specific load value that should be allocated to each user or area is calculated. This information is then compiled into the final load allocation scheme, which clearly defines the load allocation situation and corresponding adjustment basis for each period.

[0041] By screening key time-series features and constructing state transition paths, the system can accurately capture load changes and state transition patterns at different times, gain a deeper understanding of load dynamics, calculate user electricity behavior deviations and generate correction coefficients, quantifying the difference between actual and standard electricity behavior. This allows staff to intuitively understand user behavior deviations and provides quantitative indicators for targeted adjustments. A multi-period rolling strategy is employed to adjust load allocation weights, combined with user behavior correction coefficients, enabling real-time responses to changes in user electricity behavior. This ensures that load allocation weights can be flexibly adjusted according to actual conditions, avoiding the problem of fixed allocation methods being unable to adapt to dynamic changes. Continuous adjustment of load allocation weights can promptly balance the difference between actual load and power supply capacity, maintaining a dynamic load balance in the power grid system at different times. This reduces grid instability caused by excessive load fluctuations, ensuring the safe and reliable operation of the power grid. The final load allocation scheme is based on dynamic equilibrium, fully considering users' actual electricity behavior and load change characteristics, making load allocation more reasonable and efficient, improving the utilization efficiency of electrical resources, and reducing power grid operating costs.

[0042] In a preferred embodiment of the present invention, step 3 above, which involves achieving supply-demand balance by modifying the power supply strategy and pricing strategy according to the load allocation scheme, and obtaining a power dispatch instruction that matches supply and demand, may include: Step 330: Based on the load allocation scheme, extract time-based power demand data and available capacity data of power generation units, calculate the supply-demand difference value for each time period, and generate a supply-demand difference dataset; Step 331: Based on the supply and demand difference dataset, revise the power supply strategy and generate a power generation unit output adjustment plan that includes the power generation unit identifier, adjustment period, and output adjustment amount. Step 332: Based on the supply and demand difference dataset, adjust the pricing strategy and generate a dynamic electricity pricing mechanism parameter table including electricity price time period partitions and floating coefficients; Step 333: Based on the power generation unit output adjustment plan, update the power generation unit output time series data sequence to generate a corrected power supply time series dataset; Step 334: Calculate the adjustment amount of user electricity consumption behavior based on the dynamic electricity pricing mechanism parameter table, and generate the corrected user demand time series dataset; Step 335: Align and compare the power supply time series dataset with the user demand time series dataset to generate a supply-demand balance status verification result. Step 336: Based on the supply and demand balance verification results, extract the power supply data under the balance state and generate power dispatch instructions including dispatch timestamps and output values.

[0043] In this embodiment of the invention, when extracting time-based power demand data based on the load allocation scheme, the load allocation ratio of each power consumption area in different time periods is found from the load allocation scheme. Combined with the total power consumption of each area, the specific power demand of each area in each time period is calculated. Then, the demand of all areas in the same time period is added together to obtain the total power demand data for that time period. After being sorted in chronological order, time-based power demand data is formed. When extracting the available capacity data of power generation units, the rated capacity, current operating status and maintenance plan of each power generation unit are collected. The actual operable capacity is determined according to the operating status. After deducting the unavailable capacity during maintenance periods, the available capacity of each power generation unit in each time period is obtained. The available capacity of all power generation units in the same time period is summarized to form the available capacity data of power generation units. When calculating the supply and demand difference value for each time period, the total power demand data for each time period is subtracted from the total available capacity of the power generation units in that time period. If the result is positive, it means that the demand is greater than the supply, and the difference value is positive. If the result is negative, it means that the supply is greater than the demand, and the difference value is negative. The supply and demand difference values ​​for all time periods are arranged in chronological order to generate a supply and demand difference dataset.

[0044] When revising the power supply strategy based on the supply-demand difference dataset, first check the supply-demand difference value for each time period. When the difference value is positive for a certain time period, power generation output needs to be increased; when the difference value is negative, power generation output needs to be reduced. For the time period that requires increased output, select the currently available power generation units that have not reached their maximum output from the available capacity data of power generation units, sort them according to their adjustment response speed, select the power generation units with the fastest response speed, and allocate the output adjustment amount to each selected power generation unit. Based on the maximum adjustable output of the unit and the total supply-demand difference value for the time period, allocate the adjustment amount proportionally to ensure that the sum of the adjustment amounts of each unit equals the supply-demand difference value. Determine the identifier of each power generation unit, the specific time period that requires output adjustment, and the corresponding output adjustment amount, and generate a power generation unit output adjustment plan after processing.

[0045] When adjusting pricing strategies based on supply and demand difference datasets, the day is divided into multiple electricity price period partitions, typically peak, flat, and off-peak periods. This division is based on the magnitude of the difference values ​​between different periods in the supply and demand difference dataset. Periods with large positive differences are designated as peak periods, those with near-zero differences as flat periods, and those with negative differences as off-peak periods. A floating coefficient is set for each electricity price period partition. During peak periods, to suppress demand, the floating coefficient is greater than 1, and the larger the difference value, the larger the floating coefficient. For example, if the difference value exceeds 10% of total demand, the floating coefficient is set to 1.5. During flat periods, the floating coefficient is set to 1, maintaining the benchmark electricity price. During off-peak periods, to stimulate demand, the floating coefficient is less than 1, and the larger the absolute value of the difference, the smaller the floating coefficient. For example, if the difference value is negative and the absolute value exceeds 5% of total demand, the floating coefficient is set to 0.8. The electricity price period partitions and their corresponding floating coefficients are compiled into a table to generate a dynamic electricity pricing mechanism parameter table.

[0046] When updating the power generation unit output time series data sequence based on the power generation unit output adjustment plan, the original power generation unit output time series data, i.e., the planned output value for each time period, is first obtained. According to the power generation unit output adjustment plan, the corresponding adjustment time period and output adjustment amount for each power generation unit are found. In the original output time series data, the output value of the adjustment time period is modified. The adjustment amount is added for the time period that needs to increase output, and the adjustment amount is subtracted for the time period that needs to decrease output. After the modification is completed, the output data of each power generation unit is reorganized in chronological order. Then, the output data of all power generation units in each time period are summarized to generate the corrected power supply time series dataset.

[0047] When calculating the adjustment amount for user electricity consumption behavior based on the dynamic electricity pricing mechanism parameter table, the benchmark electricity price for each time period is first determined. Then, combined with the floating coefficient for each time period in the dynamic electricity pricing mechanism parameter table, the actual electricity price for each time period is calculated. The actual electricity price equals the benchmark electricity price multiplied by the floating coefficient. By analyzing the data, the user's response pattern to electricity price changes is determined. Generally, when the electricity price increases during peak hours, user electricity demand decreases; when the electricity price decreases during off-peak hours, user electricity demand increases. Based on the difference between the actual electricity price and the benchmark electricity price, and the user response pattern, the adjustment amount for user electricity consumption behavior in each time period is estimated. For example, if the electricity price increases by 20% during peak hours, user electricity demand may decrease by 10%. The corresponding adjustment amount is subtracted or added to the original time-series electricity demand data to obtain the corrected user demand data for each time period. The corrected user demand time-series dataset is then generated by organizing the data in chronological order.

[0048] When aligning and comparing the power supply time-series dataset with the user demand time-series dataset, ensure that the time divisions of the two datasets are completely consistent, with each time period corresponding to the previous one. Compare the power supply data and user demand data for each time period, calculate the difference between the two, and when the absolute value of the difference is less than a preset balance threshold (usually 5% of the demand data for that time period), it indicates that the supply and demand for that time period are balanced; when the absolute value of the difference exceeds the threshold, it indicates that the balance has not been achieved. Record the comparison results for each time period, and after summarizing, generate a supply and demand balance status verification result, clarifying which time periods are balanced, which time periods are unbalanced, and the specific difference for unbalanced periods.

[0049] Based on the power supply and demand balance verification results, extract the power supply data under the balance state, filter the power generation unit output data corresponding to the supply and demand balance period from the power supply time series dataset, including the specific output value of each power generation unit in that period, determine the scheduling timestamp corresponding to each balance state, i.e. the start and end time of that period; organize the power generation unit identifier, scheduling timestamp and corresponding output value into the form of instructions, clearly specify the output requirements of each power generation unit within a specific time, and generate power dispatch instructions that match supply and demand.

[0050] By accurately extracting time-specific electricity demand data and available capacity data of power generation units and calculating the supply-demand difference, the specific situation of electricity supply and demand in each time period can be clearly understood. The supply-demand difference dataset intuitively presents the gap or surplus between supply and demand, enabling power dispatchers to formulate targeted adjustment measures and avoid resource waste or exacerbation of supply-demand imbalance caused by blind adjustments. Based on the supply-demand difference dataset, the power supply strategy is revised and a power generation unit output adjustment plan is generated, realizing the optimal allocation of power generation resources. By prioritizing power generation units with fast response speeds and rationally allocating output adjustment amounts, the supply-demand difference can be quickly made up to ensure... The timely supply of electricity meets demand. Clearly defined generation unit identifiers, adjustment periods, and output adjustment amounts make the operation and adjustment of generation units more operational, improving the flexibility and reliability of electricity supply. Based on the supply and demand difference dataset, the pricing strategy is modified to generate a dynamic electricity price mechanism parameter table. The price lever is used to regulate user electricity consumption behavior. Electricity prices are increased during peak hours to curb excessive demand, and electricity prices are reduced during off-peak hours to stimulate electricity consumption, achieving peak shaving and valley filling of user electricity demand. Reasonable electricity price period zoning and floating coefficient settings guide users to rationally arrange their electricity consumption time, optimize the allocation efficiency of power resources, and reduce large fluctuations in electricity supply and demand.

[0051] In a preferred embodiment of the present invention, step 4 above, based on power dispatch instructions and considering the fluctuation characteristics of photovoltaic output, dynamically changes the energy storage charging and discharging instructions in combination with probability distribution constraints, and may include: Step 440: Based on the power dispatch instructions, extract the dispatch period and planned output value of the photovoltaic unit to form a photovoltaic output plan data sequence; Step 441: Based on the photovoltaic power output plan data sequence and combined with the photovoltaic power output database, calculate the deviation rate between the actual power output and the planned power output for each time period under the same meteorological conditions, and generate a statistical table characterizing the photovoltaic fluctuation characteristics. Step 442: Based on the deviation rate distribution in the statistical table characterizing photovoltaic fluctuation characteristics, determine the output fluctuation range under the predetermined confidence interval, form a set of energy storage charging and discharging constraint rules, and collect the actual output data of photovoltaic units in real time, compare it with the photovoltaic output plan data sequence time by time, and generate a real-time fluctuation deviation sequence. Step 443: Input the real-time fluctuation deviation sequence into the energy storage charging and discharging constraint rule set for compliance judgment, so as to obtain the charging and discharging action trigger flag and the corresponding power correction amount; Step 444: Generate a dynamic energy storage instruction set including energy storage unit identifier, charging / discharging time point and power limit based on the charging / discharging action trigger flag and power correction amount.

[0052] In this embodiment of the invention, when extracting the scheduling period and planned output value of photovoltaic units based on power dispatch instructions, the content related to photovoltaic units is first filtered out from the power dispatch instructions to find the specific time range in which each photovoltaic unit is dispatched. For example, the scheduling period of photovoltaic unit A is 8:00-16:00. Then, the planned output value of the photovoltaic unit for each hour within the scheduling period is determined in the dispatch instructions, such as 200kW planned output from 8:00-9:00 and 250kW planned output from 9:00-10:00. The scheduling period of each photovoltaic unit is divided into multiple sub-periods by hour, and the corresponding planned output value is matched one-to-one with the sub-periods. After arranging them in chronological order, a photovoltaic output planning data sequence is formed.

[0053] When calculating the deviation rate based on the photovoltaic power output plan data sequence and the photovoltaic power output database, the actual power output records of the photovoltaic unit in the same historical period under the same meteorological conditions (such as the same light intensity, temperature, and weather conditions) are first searched in the photovoltaic power output database. For example, if the current planned power output is 200kW, 10 actual power output records for the same period under the same meteorological conditions are found in the database, which are 190kW, 210kW, 185kW, etc. The difference between each actual power output record and the planned power output value is calculated. Then, the difference is divided by the planned power output value to obtain the deviation rate of each record. All deviation rates for each period are statistically analyzed, and the maximum, minimum, and average values ​​of the deviation rates, as well as the frequency of occurrence of different deviation rate intervals, are recorded. After sorting, a statistical table characterizing the photovoltaic fluctuation characteristics is generated.

[0054] When determining the power output fluctuation range based on the deviation rate distribution in the statistical table, first calculate the probability of each deviation rate appearing in the statistical table, and select a predetermined confidence interval, such as a 95% confidence interval. That is, find the range that covers 95% of the deviation rate data. For example, if 95% of the data falls between -15% and 10% among all deviation rates, then the power output fluctuation range under this confidence interval is -15% to 10% of the planned power output value. Based on this fluctuation range, formulate energy storage charging and discharging constraint rules, such as triggering energy storage discharge when the actual power output is 15% lower than the planned power output, and triggering energy storage charging when it is 10% higher than the planned power output, etc., forming a set of energy storage charging and discharging constraint rules. When collecting the actual power output data of the photovoltaic unit in real time, record the actual power output value every 15 minutes. Subtract the actual power output value of each time period from the planned power output value of the corresponding time period in the photovoltaic power output planned data sequence to obtain the real-time fluctuation deviation for that time period. Arrange these deviation values ​​in chronological order to generate a real-time fluctuation deviation sequence.

[0055] When inputting the real-time fluctuation deviation sequence into the constraint rule set for compliance judgment, each deviation value in the real-time fluctuation deviation sequence is checked one by one to see if it meets the energy storage charging and discharging constraint rules. When the real-time fluctuation deviation value of a certain period is less than 15% of the planned output, it meets the discharge triggering rule and is determined to trigger a discharge action, generating a charging and discharging action triggering flag of "discharge". When the deviation value is higher than 10% of the planned output, it meets the charging triggering rule and the triggering flag is "charging". When the deviation value is within the normal range, the triggering flag is "no action". When calculating the power correction amount, if discharge is triggered, the correction amount is the difference between the planned output value and the actual output value, that is, the energy storage needs to release the power of this difference to make up for the shortfall. If charging is triggered, the correction amount is the difference between the actual output value and the planned output value, that is, the energy storage needs to absorb the power of this difference to store the excess power.

[0056] When generating a dynamic energy storage instruction set based on the charge / discharge action trigger flags and power correction values, the identifiers of the energy storage units participating in the charge / discharge process are first determined. Typically, the energy storage unit closest to the photovoltaic unit and whose current available capacity meets the charge / discharge requirements is selected. The charge / discharge time point for each energy storage unit is then determined, i.e., the start time of the period triggering the charge / discharge action. If discharge is triggered during a certain period, the time point is the start time of that period. Power limits are determined based on the power correction value. During discharge, the power limit does not exceed the correction value and the maximum discharge power of the energy storage unit. During charging, the power limit does not exceed the correction value and the maximum charging power of the energy storage unit. The energy storage unit identifiers, charge / discharge time points, and corresponding power limits are then organized into instruction form to generate the dynamic energy storage instruction set.

[0057] By extracting the scheduling periods and planned output values ​​of photovoltaic (PV) units to form a data sequence, the expected output targets and time ranges of PV units were clarified. This data sequence makes the planned PV output clearly visible, facilitating comparison of actual output with the planned output, and laying the foundation for generating energy storage charging and discharging commands. Combining the data to calculate the deviation rate and generate statistical tables, a comprehensive understanding of the fluctuation patterns of PV output under the same meteorological conditions was achieved, including the magnitude and frequency of fluctuations. These statistical data provide a basis for determining a reasonable output fluctuation range, improving the scientific nature of the rules. Based on the output fluctuation range and constraint rule set determined by the deviation rate distribution, clear triggering conditions are provided for energy storage charging and discharging actions, avoiding blind charging and discharging of energy storage. The real-time fluctuation deviation sequence can promptly reflect the actual fluctuation of PV output, enabling the energy storage system to respond based on real-time data. To ensure the timeliness and accuracy of charging and discharging operations and effectively mitigate photovoltaic power output fluctuations, the system generates trigger flags and power correction values ​​through compliance checks. This provides a clear basis for the execution of energy storage charging and discharging operations. The trigger flags clearly indicate the type of action the energy storage should perform, and the power correction values ​​specify the exact power level of charging and discharging. This ensures that the energy storage system can accurately compensate for photovoltaic power output gaps or absorb excess power, improving the utilization efficiency of the energy storage system. The generated dynamic energy storage instruction set clearly defines the identification of the energy storage unit, the charging and discharging time points, and the power limits, enabling the energy storage unit to execute charging and discharging operations in an orderly manner according to the instructions. The dynamic instruction set can be adjusted according to real-time fluctuations in photovoltaic power output, ensuring coordinated operation between the energy storage system and the photovoltaic unit. This effectively mitigates photovoltaic power output fluctuations, improves the stability and reliability of the power system, and guarantees a continuous and stable power supply.

[0058] In a preferred embodiment of the present invention, step 5 above, which generates a power allocation scheme by coordinating information interaction between photovoltaic, energy storage, and load units through a distributed computing unit based on energy storage charging and discharging commands, may include: Step 550: Based on the energy storage charging and discharging commands, the distributed computing unit is used to collect in parallel the real-time output data of the photovoltaic unit, the current state of charge and charging and discharging capacity data of the energy storage unit, and the real-time power demand data of the load unit. Step 551: Perform timestamp synchronization and consistency verification on the real-time power demand data, coordinate the information interaction cycle between photovoltaic units, energy storage units and load units, and generate a coordinated dataset for synchronizing the status of each unit. Step 552: Based on the coordinated dataset, dynamically integrate the photovoltaic predicted output time series data, the dispatchable capacity boundary of the energy storage unit, and the load demand fluctuation characteristics to calculate the final power flow distribution that satisfies the power balance constraint. Step 553: Based on the final power flow distribution, generate a power allocation scheme including the power allocation ratio of each unit, timing control commands, and power adjustment thresholds.

[0059] In this embodiment of the invention, when data is collected in parallel using a distributed computing unit based on energy storage charging and discharging commands, the energy storage charging and discharging commands are first decomposed into collection tasks for each unit and sent to the corresponding collection nodes. For photovoltaic units, the collection nodes acquire real-time output data every 5 minutes, measure the current actual power generation through dedicated sensors, and record the timestamp. For energy storage units, the collection nodes acquire the current state of charge in real time, and calculate the remaining percentage of power by monitoring parameters such as battery voltage and current, combined with the battery. At the same time, charging and discharging capacity data are collected, including the current maximum charging power and maximum discharging power, which depends on factors such as battery health and temperature. For load units, the collection nodes collect real-time power demand data every 3 minutes, and use smart meters to count the total power consumption of all electrical devices in the area. Each collection node synchronously transmits the collected data back to the distributed computing unit to form a multi-source data set.

[0060] When synchronizing real-time power demand data with timestamps, the timestamp format of each unit's data is first identified and converted into a standard time format. Then, an interpolation algorithm is used to align data from different sampling frequencies. For example, the data from the load unit every 3 minutes and the data from the photovoltaic unit every 5 minutes are aligned to the same time point. During consistency verification, the logical rationality of the data from each unit at the same time point is checked. For example, when the photovoltaic output data shows a high value, it is checked whether the load demand is within a reasonable range. If the load demand is abnormally low and the energy storage charge status is normal, there may be data anomalies. When coordinating the information interaction cycle, a unified information interaction cycle is determined based on the rate of change of each unit's data. For example, the interaction cycle is set to 5 minutes to ensure that each unit can update its data in each cycle. Through these operations, status data containing photovoltaic, energy storage, and load units within a unified time period is generated, forming a coordinated dataset.

[0061] When dynamically integrating multi-source data based on a coordinated dataset, the first step is to acquire the time-series data of photovoltaic (PV) predicted output. This is based on weather forecasts and data predictions of PV unit output over a future period, combined with the current state of charge (SOC) and charging / discharging capabilities of energy storage units, to determine the dispatchable capacity boundaries. For example, when the SOC is 80%, the dispatchable discharge capacity is 30% of the total capacity, and the dispatchable charging capacity is 20% of the total capacity. Next, the load demand fluctuation characteristics are analyzed, identifying peak and off-peak periods and fluctuation amplitudes. When calculating the final power flow distribution that satisfies power balance constraints, the principle that total power generation (PV output + energy storage discharge) equals total power demand (load demand + energy storage charging) is followed. When PV output is sufficient, load demand is prioritized, and the remaining power is stored in energy storage. When PV output is insufficient, the gap is filled by energy storage discharge. By repeatedly adjusting the power allocation of each unit, the power balance constraints are ensured at every time point, ultimately determining the direction and magnitude of power flow between units.

[0062] When generating a power allocation scheme based on the final power flow distribution, the power allocation ratio of each unit is first calculated. For photovoltaic units, the allocation ratio is calculated based on their real-time output and total power generation. For energy storage units, the ratio is calculated based on charging and discharging power and total power generation or demand. For load units, the ratio is calculated based on their demand power and total demand power. When determining the timing control instructions, the power control requirements of each unit in different time periods are specified. For example, during the peak photovoltaic output period from 8:00 to 10:00, the energy storage unit charges, and the charging power is controlled according to the calculated ratio. During the peak load period from 18:00 to 20:00, the energy storage unit discharges, and the discharging power is controlled according to the calculated ratio. A power adjustment threshold is set. When the power fluctuation of a unit exceeds the threshold, the power allocation scheme is recalculated and adjusted. For example, when the photovoltaic output fluctuation exceeds 15% of its predicted value, the scheme adjustment process is initiated. These contents are compiled into a document to form a complete power allocation scheme.

[0063] By using distributed computing units to collect data from each unit in parallel, data acquisition efficiency is improved, ensuring real-time data performance. The data acquisition frequency of different units is optimized according to their characteristics, accurately capturing state changes of each unit. Simultaneously, the distributed acquisition method reduces the risk of single-point failures, improving system reliability. Timestamp synchronization and consistency verification ensure time alignment and logical rationality of data from each unit, avoiding control errors caused by data inconsistency. A unified information exchange cycle allows units to exchange information under the same time base, improving system synergy. The generation of coordinated datasets enables the fusion of multi-source data into an organic whole. Dynamically integrating photovoltaic forecasting, energy storage dispatchable capacity, and load fluctuation characteristics allows power flow calculation to fully consider changes in various factors, improving the accuracy and adaptability of calculation results. By satisfying power balance constraints, stable operation of the power system is ensured, avoiding power surplus or shortage, and improving energy utilization efficiency. Clear power allocation ratios provide a quantitative basis for the power output or consumption of each unit, ensuring rational energy allocation. Timing control commands provide specific operational guidance for each unit, enabling operation according to predetermined plans, improving the overall performance and reliability of the power system.

[0064] In a preferred embodiment of the present invention, step 6 above, based on the power allocation scheme, constructs a multi-type power supply coordination process, evaluates the power output ratio and coordination benefits according to dynamic contribution, and forms a low-carbon scheduling strategy, which may include: Step 660: Based on the power allocation scheme, construct the collaborative operation process of photovoltaic, energy storage and backup power, and generate the output constraints and collaborative interaction rules of each power source. Step 661: In the collaborative operation process, the real-time power generation of photovoltaic, the charging and discharging rate of energy storage and the start-up delay time of backup power are collected. Combined with the preset carbon emission intensity benchmark library, the output capacity, carbon emission intensity and response speed level of each power source are obtained. Step 662: Calculate the dynamic contribution coefficient of each power source based on the output capacity quantification value, carbon emission intensity value and response speed level, and allocate the rated output ratio of each power source based on the dynamic contribution weight. Step 663: Calculate the photovoltaic absorption increment, energy storage peak-shaving benefits, and backup power capacity compensation based on the rated output ratio, generate a collaborative revenue distribution scheme, and integrate it with the rated output ratio to generate a low-carbon dispatch strategy.

[0065] In this embodiment of the invention, when constructing the coordinated operation process of photovoltaic, energy storage, and backup power, the basic operating boundaries of the three must first be clarified. For photovoltaic power, the upper limit of maximum output and the lower limit of minimum output under different light intensities need to be determined based on its installed capacity, component type, and operating data, serving as output constraints for photovoltaics. For energy storage devices, the maximum charging rate and maximum discharging rate are determined based on their battery capacity, charging and discharging efficiency, and safe operating temperature range. Simultaneously, upper and lower limits of the energy storage SOC (State of Charge) are set to avoid overcharging and over-discharging, forming charging and discharging constraints for energy storage. Regarding backup power, based on its equipment model, The fuel type and operating power range determine the minimum stable output and maximum output limits after startup, as well as the continuous operating time limit, serving as output constraints for the backup power supply. In terms of collaborative interaction rules, when the photovoltaic output exceeds the load demand, the excess electricity is preferentially stored in the energy storage device, at which point the energy storage enters a charging state, and the charging rate does not exceed its maximum charging rate. When the photovoltaic output is lower than the load demand, the energy storage is first called to discharge to make up the gap. If the energy storage discharges to the lower limit and still cannot meet the demand, the backup power supply is started. After the backup power supply is started, it must reach the rated output within a specified time, and during the startup process, it must avoid conflicts with the output of photovoltaic and energy storage.

[0066] In the collaborative operation process, when collecting photovoltaic power generation data in real time, metering devices installed on the photovoltaic array record the actual power generation value every 5 minutes. After collecting data for multiple time periods, the average value is taken as the real-time power generation. For the energy storage charging and discharging rate, the energy storage control system monitors the charging and discharging current and voltage in real time. The instantaneous charging and discharging power is obtained according to the power calculation formula, and then converted into a charging and discharging rate percentage, i.e., the ratio of the current charging and discharging power to the rated power, based on the rated power. The start-up delay time of the backup power supply is obtained by recording the time from issuing the start-up command to the power supply reaching the minimum stable output through multiple actual start-up tests. The average value of multiple tests is taken as the start-up delay time. When setting up a carbon emission intensity benchmark library, various types of solar energy data are collected. Carbon emission data for photovoltaic modules, different types of energy storage batteries (such as lithium batteries and lead-acid batteries), and various backup power sources (such as gas generators and diesel generators) throughout their entire life cycle, including carbon emissions during production, transportation, operation, and disposal, are divided by their total power generation or total operating time to obtain a benchmark value for carbon emission intensity per unit output, which is then stored in the database. When obtaining the quantified value of the output capacity of each power source, the real-time photovoltaic power generation is compared with the maximum possible power generation during that period to obtain the photovoltaic output capacity ratio; the current chargeable and dischargeable power of energy storage is compared with its rated power to obtain the energy storage output capacity ratio; and the actual output of backup power sources after the start-up delay time is compared with their rated output to obtain the backup power output capacity ratio. These ratios are the quantified values ​​of output capacity. The carbon emission intensity value is directly retrieved from the carbon emission intensity benchmark library for the corresponding power type. If the fuel type or equipment parameters change during actual operation, it is recalculated and updated based on the actual amount of fuel consumed and the carbon emission factor. In terms of response speed level, photovoltaic has the fastest response speed and is set as Level 1; energy storage has a shorter time from receiving the charge / discharge command to reaching the target power and is set as Level 2; backup power has the slowest response speed due to the start-up delay and is set as Level 3. The smaller the level value, the faster the response speed.

[0067] When calculating the dynamic contribution coefficient of each power source, first determine the weights of the output capacity quantification value, carbon emission intensity value, and response speed level. Based on the system's power requirements, if current emphasis is placed on power supply stability, the weight of the output capacity quantification value is set to 40%, the carbon emission intensity value to 30%, and the response speed level to 30%. If low carbon emissions are emphasized, the weight of the carbon emission intensity value can be increased to 40%. For the output capacity quantification value, its percentage value is directly used as the score for that item. The lower the carbon emission intensity value, the higher the score. The score for that item is obtained by dividing the benchmark carbon emission intensity by the actual carbon emission intensity value. The response speed level is 100 points for level one, 80 points for level two, and 60 points for level three. Then, multiply each score by its corresponding weight, and add the products together. The sum is the dynamic contribution coefficient of each power source. When allocating the rated output ratio based on the dynamic contribution weight, first calculate the sum of the dynamic contribution coefficients of all power sources, and then divide the dynamic contribution coefficient of each power source by the sum to obtain the proportion of that power source in the total output. This proportion is the rated output ratio of that power source.

[0068] When calculating the incremental photovoltaic (PV) power consumption, the actual PV power generation at the rated output ratio is compared with the amount of curtailed PV when this strategy is not adopted. The difference between the two is the incremental PV power consumption. The calculation of energy storage peak-shaving benefits involves statistically analyzing the electricity replenished by energy storage discharge during peak load periods and the excess electricity absorbed by energy storage during off-peak periods. The difference in load fluctuation before and after peak-shaving is multiplied by the unit peak-shaving benefit coefficient to obtain the energy storage peak-shaving benefit. Regarding backup power capacity compensation, based on the actual start-up time and output of backup power sources within the dispatch cycle, combined with their capacity cost and maintenance cost, the calculation is based on the proportion of the actual provided backup capacity to the total backup demand. That is, backup power capacity compensation equals the total backup capacity cost multiplied by its capacity ratio. When generating a collaborative revenue distribution scheme, the total revenue from the incremental PV power consumption, energy storage peak-shaving benefits, and backup power capacity compensation is allocated according to the dynamic contribution coefficient ratio of each power source. Power sources with higher dynamic contribution coefficients receive a higher proportion of the revenue. Finally, the rated output ratio and collaborative revenue distribution scheme are integrated to determine the specific output values, revenue distribution amounts, and corresponding operation instructions for each power source at different times, forming a complete low-carbon dispatch strategy.

[0069] By collecting detailed real-time operational data from photovoltaic, energy storage, and backup power sources, and combining this data with a pre-defined carbon emission intensity benchmark library, the output capacity, carbon emission intensity, and response speed of each power source are precisely quantified. This provides solid data support for the calculation of the dynamic contribution coefficient. The rated output ratio is allocated based on dynamic contribution weights, avoiding the limitations of traditional fixed-ratio allocation. This allows for dynamic adjustments based on the actual operating status of the power sources, making the output allocation more aligned with the actual needs of the system. During the calculation process, carbon emission intensity is incorporated as a key indicator into the dynamic contribution coefficient calculation, giving power sources with low carbon emission intensity (such as photovoltaic) a higher weight in the output allocation and prioritizing them for more power. This provides an opportunity for photovoltaic power generation. At the same time, by calculating the incremental photovoltaic power consumption, the consumption rate of clean energy such as photovoltaics is improved, the curtailment of photovoltaic power is reduced, and the dependence on high-carbon emission backup power sources is reduced. By calculating the peak-shaving benefits of energy storage, the role of energy storage in smoothing load fluctuations is fully utilized, reducing energy waste caused by excessive load peak-valley differences and improving energy utilization efficiency. The generation of a collaborative benefit distribution scheme rationally allocates the incremental photovoltaic power consumption, the peak-shaving benefits of energy storage, and the capacity compensation of backup power sources, so that each power source can obtain corresponding benefits in collaborative operation, stimulating the enthusiasm of each power source to participate in collaborative operation and improving the economic benefits of the entire power system.

[0070] In a preferred embodiment of the present invention, step 7 above, based on a low-carbon dispatch strategy, periodically adjusts the low-carbon dispatch strategy through the grid frequency deviation characteristics to generate an incentive-based electricity price signal and feeds it back to the user response process, and may include: Step 770: Execute grid dispatch based on low-carbon dispatch strategy, collect grid frequency measurement values ​​in real time, and calculate the instantaneous deviation between the frequency measurement values ​​and the reference frequency. Step 771: Based on the preset frequency fluctuation level threshold, the instantaneous deviation is classified into positive offset fluctuation, negative offset fluctuation and steady state, and the power grid frequency fluctuation mode is identified. Step 772: When the frequency fluctuation mode is positive offset fluctuation, reduce the upper limit of photovoltaic output by the first preset ratio and start energy storage discharge; when it is negative offset fluctuation, increase the backup power output ratio by the second preset ratio and start energy storage charging; when it is in a stable state, maintain the current low-carbon dispatch strategy. Step 773: Based on the adjusted output threshold and power supply status, calculate the real-time supply and demand balance of the power grid and the change in carbon emission intensity, and generate an incentive-based electricity price signal including time-of-use electricity price tiers and carbon credit subsidy parameters. Step 774: Input the incentive electricity price signal into the user-side demand response device to trigger the load transfer response operation of the user's electrical equipment.

[0071] In this embodiment of the invention, during the grid dispatching process based on the low-carbon dispatching strategy, multiple frequency measuring instruments deployed in the grid continuously collect grid frequency data at fixed time intervals (e.g., every second), obtaining a series of frequency measurement values. Then, each collected frequency measurement value is compared with the grid's set reference frequency (usually 50Hz). The difference between each frequency measurement value and the reference frequency is the instantaneous deviation at that moment. For example, if a frequency measurement value is 50.2Hz and the reference frequency is 50Hz, then the instantaneous deviation is 0.2Hz; if the measurement value is 49.7Hz, the instantaneous deviation is -0.3Hz.

[0072] First, different frequency fluctuation level thresholds are preset. For example, a threshold of positive offset fluctuation is defined as an instantaneous deviation greater than 0.2Hz, a threshold of negative offset fluctuation is defined as a deviation less than -0.2Hz, and a threshold of stable state is defined as a deviation between -0.2Hz and 0.2Hz. Then, each instantaneous deviation calculated in step 770 is compared with these preset thresholds. If the instantaneous deviation is greater than the threshold of positive offset fluctuation, it is classified as positive offset fluctuation; if the instantaneous deviation is less than the threshold of negative offset fluctuation, it is classified as negative offset fluctuation; if the instantaneous deviation is within the threshold range of stable state, it is classified as stable state. Through this classification method, the frequency fluctuation mode of the current power grid is identified.

[0073] When the frequency fluctuation pattern is determined to be a positive offset fluctuation, first determine the upper limit of the current photovoltaic output, and then calculate the output reduction value according to the first preset ratio (e.g., 20%). Multiply the current photovoltaic output upper limit by (1 - the first preset ratio) to obtain the adjusted photovoltaic output upper limit. In this way, the photovoltaic output upper limit is reduced. At the same time, check the current status of the energy storage device. After confirming that it has the conditions for discharge, start the energy storage discharge operation to release the stored electrical energy. When the frequency fluctuation pattern is a negative offset fluctuation, first determine the total output of the backup power supply, and calculate the increase in the backup power supply output according to the second preset ratio (e.g., 15%). Multiply the total backup power supply output by (1 + the second preset ratio) to obtain the adjusted backup power supply output. In this way, the backup power supply output ratio is increased. At the same time, check the current capacity and charging conditions of the energy storage device, and start the energy storage charging operation. When it is in a stable state, no adjustments are made to the current photovoltaic output upper limit, backup power supply output ratio, or energy storage charging and discharging status, and the existing low-carbon dispatch strategy remains unchanged.

[0074] After obtaining the adjusted photovoltaic output ceiling, backup power output ratio, and energy storage charging and discharging status, the current total power supply of the grid is first calculated, including the total output of photovoltaic, backup power, and energy storage discharge. Then, the current total electricity consumption of the grid, i.e., the total electricity demand on the user side, is calculated. The total power supply is subtracted from the total electricity consumption, and the difference is compared with the total electricity consumption to calculate the supply-demand balance ratio, thereby determining the real-time supply-demand balance of the grid. Simultaneously, carbon emission data of each power source is collected. Based on the adjusted output, the total carbon emission change before and after the adjustment is calculated. Combined with the change in total power supply, the change in carbon emission intensity is obtained. Based on the real-time supply-demand balance and the change in carbon emission intensity, different electricity price standards are set for different time periods, such as higher prices during peak hours and lower prices during off-peak hours, forming a time-of-use pricing tier. At the same time, based on the reduction in carbon emission intensity and the potential emission reduction contribution of users, the calculation standard and subsidy amount for carbon credits are determined, generating carbon credit subsidy parameters. The time-of-use pricing tier and carbon credit subsidy parameters are combined to form an incentive-based electricity price signal.

[0075] The generated incentive electricity price signal is transmitted to the user-side demand response device. After receiving the signal, the device analyzes the time-of-use electricity price tiers and carbon credit subsidy parameters in the signal. Based on the analysis results, it identifies the user's electrical equipment that can be load-shifted, such as washing machines and water heaters. Then, combined with the high and low electricity price periods, it triggers these devices to increase electricity consumption during low electricity price periods and reduce electricity consumption during high electricity price periods, thereby achieving reasonable adjustment of the user's electricity load.

[0076] By continuously collecting grid frequency measurements and calculating instantaneous deviations, subtle changes in grid frequency can be detected in a timely manner. Accurate identification of frequency fluctuation patterns based on preset thresholds helps to quickly maintain the grid frequency within a stable range, ensuring the safe and stable operation of the grid. Differentiated adjustment measures can be taken for different frequency fluctuation patterns. During positive offset fluctuations, the upper limit of photovoltaic output is reduced and energy storage discharge is initiated; during negative offset fluctuations, backup power output is increased and energy storage charging is initiated, achieving dynamic balance of power output. Simultaneously, the rational use of energy storage charging and discharging functions improves energy utilization efficiency and reduces energy waste. By calculating the real-time supply and demand balance of the grid and changes in carbon emission intensity, the grid's supply and demand status and carbon emission situation can be grasped in a timely manner. Based on this, an incentive-based electricity price signal can guide the power supply side to adjust output, reducing the use of high-carbon emission power sources and lowering the overall carbon emission intensity.

[0077] like Figure 2 As shown, embodiments of the present invention also provide a smart grid energy optimization management system, comprising: The data acquisition module is used to simultaneously collect photovoltaic power output fluctuation data, user load demand data, and power grid frequency status data to obtain the raw monitoring dataset; The analysis and regulation module is used to perform spatiotemporal feature mapping on the original monitoring dataset, extract multidimensional feature vectors, determine the dominant feature axis system based on covariance analysis, calculate the dynamic coupling degree between feature axes system to divide the control domain; select key time-series feature quantities within the control domain, construct load response state transition paths, derive user behavior correction coefficients accordingly, achieve dynamic load balance through multi-period rolling strategy, and generate load allocation schemes. The balance dispatch module is used to achieve supply and demand balance by modifying the power supply strategy and pricing strategy according to the load allocation plan, and to obtain power dispatch instructions that match supply and demand. The energy storage control module is used to dynamically change the energy storage charging and discharging commands based on power dispatch instructions, taking into account the fluctuation characteristics of photovoltaic output and combining probability distribution constraints. The coordination and allocation module is used to coordinate the information exchange between photovoltaic, energy storage and load units through the distributed computing unit according to the energy storage charging and discharging instructions, and generate a power allocation scheme. The collaborative scheduling module is used to construct a collaborative process for multiple types of power sources based on the power allocation scheme, evaluate the power output ratio and collaborative benefits according to the dynamic contribution, and form a low-carbon scheduling strategy. The frequency feedback module is used to periodically adjust the low-carbon dispatch strategy based on the grid frequency deviation characteristics, generate incentive price signals, and feed them back to the user response process.

[0078] It should be noted that this system is a system corresponding to the above method. All implementation methods in the above method embodiments are applicable to this embodiment and can achieve the same technical effect.

[0079] Embodiments of the present invention also provide a computing device, including: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0080] Embodiments of the present invention also provide a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0081] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A smart grid energy optimization management method, characterized in that, The method includes: Step 1: Simultaneously collect photovoltaic power output fluctuation data, user load demand data, and grid frequency status data to obtain the raw monitoring dataset; Step 2: Align the photovoltaic output, load demand, and grid frequency data in the original monitoring dataset by timestamp, and map them into a multi-dimensional feature vector set with a time dimension through a spatiotemporal correlation matrix. Perform covariance matrix decomposition on the multi-dimensional feature vector set, and identify orthogonal feature vectors with prominent variance contribution rates through principal component analysis to form the dominant feature axis system. Calculate the angle change rate and projected energy ratio between the dominant feature axes based on a sliding time window, and fuse them to generate a dynamic coupling degree index. Specifically, this includes: setting a fixed-length time window, sliding it along the time axis in unit steps to obtain multiple continuous window datasets; within each sliding window, for any two dominant feature axes, calculate the angle between the start and end times of the window, and divide the angle change by the time window length to obtain the angle change rate; simultaneously, project the multi-dimensional feature vectors within the corresponding window onto each dominant feature axis, calculate the sum of squares of the projected values ​​of each axis as the projected energy, and then take the ratio of the projected energies of any two feature axes as the projected energy. Projected energy ratio; all calculated angle change rates and projected energy ratios are normalized and assigned different weights according to their importance. The weighted index values ​​are summed to obtain the dynamic coupling degree index for the corresponding sliding time window. A dynamic threshold is set according to the gradient distribution of the dynamic coupling degree index, and the continuous and stable coupling degree region is clustered into independent control domains, dividing independent control domains with similar dynamic characteristics. Within the divided control domains, time-series key feature quantities characterizing the load change trend are selected, and load response state transition paths are constructed based on the time-series key feature quantities. Based on the load response state transition paths, the deviation degree of user electricity consumption behavior is calculated, and user behavior correction coefficients are generated based on the calculated user electricity consumption behavior deviation degree. A multi-period rolling strategy is adopted, combined with the user behavior correction coefficients, to adjust the load allocation weights. Through continuous adjustment of the load allocation weights, a dynamic load balance state is achieved, and the final load allocation scheme is generated when the dynamic load balance state is achieved. Step 3: Based on the load allocation scheme, achieve supply and demand balance by modifying the power supply strategy and pricing strategy, and obtain power dispatch instructions that match supply and demand. Step 4: Based on power dispatch instructions, and considering the fluctuation characteristics of photovoltaic output, dynamically change the energy storage charging and discharging instructions in combination with probability distribution constraints; Step 5: Based on the energy storage charging and discharging instructions, the distributed computing unit coordinates the information exchange between photovoltaic, energy storage and load units to generate a power allocation scheme; Step 6: Based on the power allocation scheme, construct a multi-type power supply coordination process, evaluate the power output ratio and coordination benefits according to dynamic contribution, and form a low-carbon scheduling strategy. Step 7: Based on the low-carbon dispatch strategy, the low-carbon dispatch strategy is periodically adjusted through the grid frequency deviation characteristics to generate an incentive-based electricity price signal and feed it back to the user response process.

2. The smart grid energy optimization management method according to claim 1, characterized in that, Based on the load allocation scheme, supply and demand balance is achieved by adjusting the power supply strategy and pricing strategy, resulting in power dispatch instructions that match supply and demand, including: Based on the load allocation scheme, time-based power demand data and available capacity data of power generation units are extracted, and the supply-demand difference value for each time period is calculated to generate a supply-demand difference dataset. Based on the supply and demand difference dataset, the power supply strategy is revised to generate a power generation unit output adjustment plan that includes the power generation unit identifier, adjustment period, and output adjustment amount. Based on the supply and demand difference dataset, the pricing strategy is revised to generate a dynamic electricity pricing mechanism parameter table that includes electricity price time period partitioning and floating coefficients; Based on the power generation unit output adjustment plan, update the power generation unit output time series data sequence to generate a corrected power supply time series dataset. Based on the dynamic electricity pricing mechanism parameter table, calculate the adjustment amount of user electricity consumption behavior and generate the corrected user demand time series dataset. The power supply time series dataset and the user demand time series dataset are aligned and compared to generate a supply and demand balance status verification result. Based on the verification results of the supply and demand balance, the power supply data under the balance state is extracted, and power dispatch instructions including dispatch timestamps and output values ​​are generated.

3. The smart grid energy optimization management method according to claim 2, characterized in that, Based on power dispatch instructions, and considering the fluctuation characteristics of photovoltaic power output, the energy storage charging and discharging instructions are dynamically changed in combination with probability distribution constraints, including: Based on power dispatch instructions, the dispatch period and planned output value of photovoltaic units are extracted to form a photovoltaic output plan data sequence; Based on the photovoltaic power output plan data sequence and combined with the photovoltaic power output database, the deviation rate between the actual power output and the planned power output in each time period under the same meteorological conditions is calculated, and a statistical table characterizing the photovoltaic fluctuation characteristics is generated. Based on the deviation rate distribution in the statistical table characterizing photovoltaic fluctuations, the output fluctuation range under the predetermined confidence interval is determined, forming a set of energy storage charging and discharging constraint rules. The actual output data of the photovoltaic unit is collected in real time and compared with the photovoltaic output plan data sequence time by time to generate a real-time fluctuation deviation sequence. The real-time fluctuation deviation sequence is input into the energy storage charging and discharging constraint rule set for compliance judgment, so as to obtain the charging and discharging action trigger flag and the corresponding power correction amount; Based on the charging / discharging action trigger flag and power correction amount, a dynamic energy storage instruction set is generated, including energy storage unit identifier, charging / discharging time point and power limit.

4. The smart grid energy optimization management method according to claim 3, characterized in that, Based on the energy storage charging and discharging commands, a power allocation scheme is generated by coordinating information exchange between photovoltaic, energy storage, and load units through a distributed computing unit, including: Based on energy storage charging and discharging commands, the distributed computing unit is used to collect real-time output data of photovoltaic units, current state of charge and charging / discharging capacity data of energy storage units, and real-time power demand data of load units in parallel. The system performs timestamp synchronization and consistency verification on real-time power demand data, coordinates the information exchange cycle between photovoltaic units, energy storage units and load units, and generates a coordinated dataset for synchronizing the status of each unit. Based on the coordinated dataset, the photovoltaic predicted output time series data, the dispatchable capacity boundary of energy storage units and the load demand fluctuation characteristics are dynamically integrated to calculate the final power flow distribution that satisfies the power balance constraint. Based on the final power flow distribution, a power allocation scheme is generated, which includes the power allocation ratio of each unit, timing control commands, and power adjustment thresholds.

5. The smart grid energy optimization management method according to claim 4, characterized in that, Based on power allocation schemes, a multi-type power source coordination process is constructed. The power output ratio and coordination benefits are evaluated according to dynamic contributions, forming a low-carbon dispatch strategy, including: Based on the power allocation scheme, a collaborative operation process for photovoltaic, energy storage and backup power is constructed, and output constraints and collaborative interaction rules for each power source are generated. In the collaborative operation process, the real-time power generation of photovoltaic, the charging and discharging rate of energy storage and the start-up delay time of backup power are collected. Combined with the preset carbon emission intensity benchmark library, the output capacity, carbon emission intensity and response speed level of each power source are obtained. Based on the output capacity quantification value, carbon emission intensity value and response speed level, calculate the dynamic contribution coefficient of each power source, and allocate the rated output ratio of each power source based on the dynamic contribution weight. Based on the rated output ratio, the incremental photovoltaic consumption, energy storage peak-shaving benefits, and backup power capacity compensation are calculated to generate a collaborative revenue distribution scheme, which is then integrated with the rated output ratio to generate a low-carbon dispatch strategy.

6. The smart grid energy optimization management method according to claim 5, characterized in that, Based on a low-carbon dispatch strategy, the strategy is periodically adjusted according to the grid frequency deviation characteristics to generate incentive-based electricity price signals and feed them back to the user response process, including: Power grid dispatch is performed based on a low-carbon dispatch strategy, and power grid frequency measurements are collected in real time to calculate the instantaneous deviation between the frequency measurements and the reference frequency. Based on the preset frequency fluctuation level threshold, the instantaneous deviation is classified into positive offset fluctuation, negative offset fluctuation and steady state, and the power grid frequency fluctuation mode is identified. When the frequency fluctuation mode is positive offset fluctuation, the photovoltaic output limit is reduced by the first preset ratio and the energy storage discharge is started; when it is negative offset fluctuation, the backup power output ratio is increased by the second preset ratio and the energy storage charging is started; when it is in a stable state, the current low-carbon dispatch strategy is maintained. Based on the adjusted output threshold and power status, the real-time supply and demand balance of the power grid and the change in carbon emission intensity are calculated, and an incentive-based electricity price signal including time-of-use electricity price tiers and carbon credit subsidy parameters is generated. The incentive electricity price signal is input into the user-side demand response device to trigger the load transfer response operation of the user's electrical equipment.

7. A smart grid energy optimization management system, wherein the system implements the method as described in any one of claims 1 to 6, characterized in that, include: The data acquisition module is used to simultaneously collect photovoltaic power output fluctuation data, user load demand data, and power grid frequency status data to obtain the raw monitoring dataset; The analysis and regulation module is used to perform spatiotemporal feature mapping on the original monitoring dataset, extract multidimensional feature vectors, determine the dominant feature axis system based on covariance analysis, calculate the dynamic coupling degree between feature axes system to divide the control domain; select key time-series feature quantities within the control domain, construct load response state transition paths, derive user behavior correction coefficients accordingly, achieve dynamic load balance through multi-period rolling strategy, and generate load allocation schemes. The balance dispatch module is used to achieve supply and demand balance by modifying the power supply strategy and pricing strategy according to the load allocation plan, and to obtain power dispatch instructions that match supply and demand. The energy storage control module is used to dynamically change the energy storage charging and discharging commands based on power dispatch instructions, taking into account the fluctuation characteristics of photovoltaic output and combining probability distribution constraints. The coordination and allocation module is used to coordinate the information exchange between photovoltaic, energy storage and load units through the distributed computing unit according to the energy storage charging and discharging instructions, and generate a power allocation scheme. The collaborative scheduling module is used to construct a collaborative process for multiple types of power sources based on the power allocation scheme, evaluate the power output ratio and collaborative benefits according to the dynamic contribution, and form a low-carbon scheduling strategy. The frequency feedback module is used to periodically adjust the low-carbon dispatch strategy based on the grid frequency deviation characteristics, generate incentive price signals, and feed them back to the user response process.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the method as described in any one of claims 1 to 6.

Images