Power grid checking and regulation strategy generation method based on user response load

Abstract

The application discloses a power grid checking and regulation strategy generation method based on user response load, constructs a two-dimensional classification standard based on photovoltaic access attributes and response elasticity, classifies users in layers according to the two-dimensional classification standard, and obtains high-priority response users, medium-priority response users and low-priority response users; constructs a checking model of three dimensions of photovoltaic output matching checking, equipment safety checking and response effectiveness checking, checks the current operation state of the power grid by using the checking model, and judges whether the current operation state of the power grid passes the checking; based on the user priority result and the checking result obtained by the multi-dimensional power grid checking step, an initial regulation instruction is generated, and the execution effect of the initial regulation instruction is checked in real time, if the feedback checking fails, the regulation instruction is adjusted, the feedback checking and instruction adjustment process are repeated, and the power grid operation state passes the checking.

Description

Technical Field

[0001] This invention belongs to the field of power system load regulation and grid security verification technology, and relates to a method for generating grid verification and regulation strategies based on user response load. Background Technology

[0002] In recent years, the scale of distributed photovoltaic (PV) power grid integration has experienced explosive growth. Meanwhile, demand-side response (DSR), as a key means of enhancing grid regulation capabilities, can effectively mitigate the impact of distributed PV power output fluctuations on the grid's supply-demand balance by guiding users to stagger and avoid peak electricity consumption. The synergy between these two approaches has become a core direction for distribution network operation and management.

[0003] However, current distributed photovoltaic and user response coordinated control technologies still face several technical bottlenecks, which restrict the safety, economy, and efficiency of power grid operation. Specific problems are as follows:

[0004] First, distributed photovoltaic data is highly volatile and intermittent, user response load data is time-series and random, and grid equipment data requires real-time performance and high reliability. Existing technologies mostly perform independent preprocessing for single types of data, without considering the correlation between the three types of data.

[0005] Second, existing user response priority classifications often rely on a single dimension: either based solely on user response capacity, ignoring photovoltaic (PV) grid connection attributes, or based solely on PV grid connection scale, without considering response resilience. This single-dimensional classification leads to a mismatch between user priorities and the actual grid control needs, resulting in wasted response resources or control failures.

[0006] Third, existing power grid verification often focuses on a single objective: either only verifying equipment safety without considering the matching degree between photovoltaic output and user load response; or only focusing on the matching of photovoltaic output while ignoring the verification of response effectiveness. In addition, some technologies do not integrate the verification of these three aspects, resulting in verification results that cannot fully reflect the actual operating status of the power grid and are prone to overlooking potential safety risks or scenarios where control is ineffective.

[0007] Fourth, existing control systems mostly adopt a one-way process from command generation to execution: after generating control commands based on initial data, there is no real-time feedback and verification of the command execution effect, making it impossible to detect and adjust commands in a timely manner; although some technologies have feedback links, the adjustment logic does not combine the dynamic changes of user priority and verification results to optimize commands, making it difficult for the power grid operation status to continuously meet the requirements of safety and efficiency, and even causing secondary fluctuations.

[0008] In summary, current distributed photovoltaic and user response coordinated control technologies have significant shortcomings in areas such as multi-source data collaborative preprocessing, dual-dimensional user classification, multi-dimensional verification model construction, and dynamic command optimization, making it difficult to meet the grid's requirements for precise scheduling and safe operation under a high proportion of distributed photovoltaic access. Summary of the Invention

[0009] To address the problems existing in the background technology, this invention proposes a grid verification and control strategy generation method based on user response load. The aim is to propose a multi-dimensional verification and closed-loop control strategy generation method that integrates the dynamic characteristics of user response load, distributed photovoltaic power output simulation and grid equipment status, which is applicable to the safe operation and load optimization control of regional power grids.

[0010] The first aspect of the present invention provides a method for generating power grid verification and control strategies based on user response load, comprising:

[0011] Preprocess the collected distributed photovoltaic data, user response load data, and grid equipment data;

[0012] A two-dimensional classification standard is constructed based on photovoltaic access attributes and response elasticity. Users are then classified into high-priority response users, medium-priority response users, and low-priority response users according to the two-dimensional classification standard.

[0013] A verification model is constructed with three dimensions: photovoltaic output matching verification, equipment safety verification, and response effectiveness verification. The verification model is used to verify the current operating status of the power grid and determine whether the current operating status of the power grid passes the verification.

[0014] Based on the user priority results and verification results, an initial control command is generated, and the execution effect of the initial control command is verified in real time. If the verification fails, the control command is adjusted, and the process of verification and command adjustment is repeated until the power grid operation status passes the verification.

[0015] Optionally, the preprocessing of the collected distributed photovoltaic data, user response load data, and grid equipment data includes:

[0016] Outlier removal and missing data completion are performed using horizontal and vertical outlier removal rules. The horizontal outlier removal rule is as follows: if the load change at different times on the same day exceeds the reasonable threshold for daytime and the change is not caused by the alternation of day and night, then the load data is removed. The vertical outlier removal rule is as follows: if the load change at the same time on the same day exceeds the reasonable threshold for cross-day and the change is not caused by weather, then the load data is removed. The missing data completion uses linear interpolation.

[0017] Optionally, the method for determining the photovoltaic access attribute includes: if a user accesses distributed photovoltaic, then the user is marked as a photovoltaic user; if a user does not access distributed photovoltaic, then the user is marked as an ordinary user; the mark value of the photovoltaic user is 1, and the mark value of the ordinary user is 0.

[0018] Optionally, the response resilience is calculated by weighted average of historical response rate and adjustable load percentage, wherein the calculation formula is: Response Resilience = α × Historical Response Rate + β × Adjustable Load Percentage; α is the weighting coefficient of historical response rate, β is the weighting coefficient of adjustable load percentage, and the values ​​of α and β are calibrated according to the characteristics of users in the region; the historical response rate = number of successful responses in the past / total number of commands received by the user; the adjustable load percentage = peak adjustable load of the user / peak total load of the user.

[0019] Optionally, users are stratified according to the dual-dimensional classification standard to obtain high-priority response users, medium-priority response users, and low-priority response users, including: if a user's photovoltaic access attribute is marked as 1 and the response elasticity is ≥0.6, then the user is a high-priority response user; if a user's photovoltaic access attribute is marked as 0 and 0.3≤response elasticity<0.6, then the user is a medium-priority response user; if a user's response elasticity is ≤0.3, then the user is a low-priority response user.

[0020] Optionally, if the photovoltaic output matching verification, equipment safety verification, and response effectiveness verification all pass, the current operating state of the power grid is determined to be safe, and the current control command is maintained; if any dimension of the photovoltaic output matching verification, equipment safety verification, and response effectiveness verification fails, the current operating state of the power grid is determined to be unsafe, and the command adjustment process is triggered.

[0021] Optionally, the step of generating initial control instructions based on user priority results and verification results using a control priority allocation method specifically includes: allocating control priorities by combining user type priorities, equipment safety margins, and photovoltaic matching deviations; wherein, user type priorities are ranked from high to low as high priority responders, medium priority responders, and low priority responders; the lower the equipment safety margin, the higher the control priority of the corresponding region's users; the greater the photovoltaic matching deviation, the higher the control priority of the corresponding region's users.

[0022] Optionally, the initial control instruction generation logic is as follows: control instructions are first issued to high-priority response users; during periods of high photovoltaic power generation, instructions to increase adjustable load are issued to high-priority response users; during periods of equipment overload, instructions to reduce adjustable load are issued to high-priority response users.

[0023] Optionally, the execution effect of the initial control command is evaluated in real time, including: if the response load of a high-priority responder cannot meet the grid control requirements, then the energy storage device is invoked to coordinate with the adjustable load to supplement the response load.

[0024] Optionally, the step of adjusting the control command if the feedback verification fails, and repeating the feedback verification and command adjustment process until the grid operation status passes the verification, includes: after the control command is issued, collecting actual user response load data, actual photovoltaic output data, and actual grid equipment load power data within a preset time; re-verifying the collected data using the verification model; if the re-verification fails, adjusting the control command of the medium-priority response user, collecting data again, and re-verifying until the grid operation status passes the verification.

[0025] Compared with the prior art, the present invention has the following beneficial effects:

[0026] This invention provides a method for generating power grid verification and control strategies based on user response load. The method involves preprocessing the collected distributed photovoltaic (PV) data, user response load data, and power grid equipment data; constructing a two-dimensional classification standard based on PV access attributes and response elasticity; classifying users into high-priority, medium-priority, and low-priority response users according to this standard; constructing a three-dimensional verification model encompassing PV output matching verification, equipment safety verification, and response effectiveness verification; using this model to verify the current operating state of the power grid and determine if it passes the verification; generating initial control commands based on user priority results and the verification results obtained from the multi-dimensional power grid verification steps; and performing real-time feedback verification on the execution effect of the initial control commands. If the feedback verification fails, the control commands are adjusted, and the feedback verification and command adjustment process is repeated until the power grid operating state passes the verification. This invention constructs a multi-dimensional power grid verification model that includes photovoltaic output matching verification, equipment safety verification, and response effectiveness verification. Combined with dynamic safety margin calculation and real-time feedback verification mechanism, it breaks through the limitations of existing single-dimensional verification, which can greatly improve the accuracy of equipment overload early warning, effectively reduce line tripping and other faults caused by photovoltaic output backfeed or equipment overload, and reduce the risk of power grid operation safety. Attached Figure Description

[0027] Figure 1 This is a flowchart of a method for generating power grid verification and control strategies based on user response load in one embodiment of the present invention. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] In one embodiment, such as Figure 1 As shown, a method for generating power grid verification and control strategies based on user response load is provided, and this method is applied to... Figure 1 Taking China as an example, the following specific steps will be used:

[0030] S10: Preprocess the collected distributed photovoltaic data, user response load data, and grid equipment data.

[0031] Specifically, in the data acquisition and preprocessing process, the collected distributed photovoltaic (PV) data undergoes preprocessing. This involves standardizing the data by removing outliers and completing missing data, addressing the grid topology data, gridded irradiance data, PV predicted output data, and PV actual output data. The scope of distributed PV data acquisition includes: grid topology data relating to stations, lines, transformers, and distributed PV systems to clarify the grid connection location of PV; gridded irradiance data to reflect the illumination conditions in different areas; PV predicted output data, i.e., the pre-estimated PV power generation, obtained through data integration and model calculation based on grid topology data and meteorological data, combined with PV module characteristics; and PV actual output data, i.e., the actual electricity generated by the PV equipment. The time granularity of distributed PV data acquisition is uniformly set according to preset threshold time points to ensure data timeliness and consistency. These preset threshold time points are pre-defined acquisition intervals based on specific needs.

[0032] The collection of user response load data includes total user load data, i.e., the total electricity consumption of users per unit time; adjustable user load data, specifically including load data of industrial and commercial chiller units, load data of residential energy storage devices, and other controllable load information; and historical user response command execution data, including the number of times users have successfully responded to commands and the total number of commands received by users, used to assess user responsiveness. This type of data is collected at preset time points to ensure temporal consistency with distributed photovoltaic data.

[0033] The data collected from power grid equipment includes main transformers and lines. Specifically, the data includes the rated load power data of both, i.e., the maximum allowable load power of the equipment; and the actual load power data, i.e. the power that the equipment is currently actually carrying, providing basic parameters for equipment safety verification.

[0034] The preprocessing process includes two stages: outlier removal and missing data completion. Outlier removal employs a dual-rule approach: horizontal outlier removal is defined as follows: if the load variation at different times within the same day exceeds a reasonable intraday threshold, and this variation is not caused by the day-night cycle, then the load data is considered an outlier and removed. Vertical outlier removal is defined as follows: if the load variation at the same time period on different dates exceeds a reasonable cross-day threshold, and this variation is not caused by weather factors, then the load data is considered an outlier and removed. Missing data completion uses linear interpolation to calculate a reasonable value for the missing time period using valid data from adjacent time periods, ensuring data integrity.

[0035] In one embodiment, preprocessing distributed photovoltaic data, user response load data, and grid equipment data achieves three main benefits: First, it improves data accuracy by eliminating outlier data through horizontal and vertical rules, preventing deviations in subsequent response elasticity calculations and deviation rate calculations. Second, it ensures data integrity by using linear interpolation to fill in missing data and maintaining the collection of the three types of data according to preset time points to ensure continuity of collection granularity. Third, it provides reliable data support for subsequent user hierarchical classification, multi-dimensional grid verification, and closed-loop control strategy generation modules, ensuring accurate user classification, accurate verification results, and scientific control commands.

[0036] S20: Construct a two-dimensional classification standard based on photovoltaic access attributes and response elasticity, and classify users into high-priority response users, medium-priority response users and low-priority response users according to the two-dimensional classification standard.

[0037] Specifically, in the user stratification and classification process, the dual-dimensional classification standard consists of photovoltaic (PV) access attributes and response resilience, which together form a complete user classification basis. The determination of PV access attributes is based on whether the user has access to distributed PV: if the user has access to distributed PV, they are marked as a PV user with a value of 1; if the user has not accessed distributed PV, they are marked as a regular user with a value of 0. This attribute directly reflects the relationship between the user and the distributed PV system, providing a foundation for subsequently matching PV output control needs.

[0038] Response resilience, another core dimension, is calculated as a weighted average of historical response rate and adjustable load percentage. The formula is: Response Resilience = α × Historical Response Rate + β × Adjustable Load Percentage. Here, the historical response rate is the ratio of the number of successful historical responses to the total number of commands received by the user, used to quantify the user's ability to execute past control commands; the adjustable load percentage is the ratio of the user's peak adjustable load to the user's total peak load, used to measure the proportion of the user's load that can be controlled; α is the weighting coefficient for the historical response rate, and β is the weighting coefficient for the adjustable load percentage. The specific values ​​of these two coefficients need to be calibrated according to the characteristics of users within the region, such as the differences between industrial and commercial users and residential users, to ensure that the response resilience calculation results are compatible with actual control needs.

[0039] Based on the above dual-dimensional classification criteria, users are divided into three categories: If a user's photovoltaic (PV) access attribute is marked as 1 and their response elasticity is ≥0.6, then the user is a high-priority response user. This type of user possesses both PV access characteristics and strong response capabilities, making them suitable for priority participation in PV output and consumption-related regulation. If a user's PV access attribute is marked as 0 and their response elasticity is ≤0.3 <0.6, then the user is a medium-priority response user. This type of user, although not connected to PV, possesses moderate response capabilities and can supplement the high-priority user's regulation capabilities when they are insufficient. If a user's response elasticity is ≤0.3, then regardless of their PV access attribute, they are determined to be a low-priority response user. This type of user has weak response capabilities and is only invoked in extreme scenarios. Through this hierarchical classification method, precise matching between user response potential and grid regulation needs can be achieved.

[0040] In this embodiment, the present invention constructs a two-dimensional classification standard based on photovoltaic access attributes and response elasticity. Users are classified into high-priority response users, medium-priority response users, and low-priority response users according to the two-dimensional classification standard. This ensures that the photovoltaic output consumption demand and user response capability can be accurately matched. High-priority response users can participate in photovoltaic output absorption first, improve photovoltaic utilization rate, and avoid "one-size-fits-all" regulation. It also reduces the forced regulation of low-priority response users, thus balancing grid security and user satisfaction.

[0041] S30: Construct a verification model with three dimensions: photovoltaic output matching verification, equipment safety verification, and response effectiveness verification. Use the verification model to verify the current operating status of the power grid and determine whether the current operating status of the power grid passes the verification.

[0042] Specifically, in the multi-dimensional power grid verification process, it is necessary to first construct a photovoltaic output matching verification model, an equipment safety verification model, and a response effectiveness verification model, and then use the three models in a coordinated manner to complete the verification of the current operating status of the power grid. Finally, the verification result is judged based on unified rules.

[0043] The photovoltaic output matching verification model is constructed with the deviation rate between the actual photovoltaic output and the user response load as the core evaluation indicator. The deviation rate is calculated as follows: Deviation rate = |Actual photovoltaic output - User response load| / Actual photovoltaic output × 100%. The verification judgment logic of the model is as follows: If the calculated deviation rate is ≤15%, the photovoltaic output matching verification is deemed to have passed, indicating that the absorption effect of the user response load on the photovoltaic output meets the requirements; if the deviation rate is >15%, the photovoltaic output matching verification is deemed to have failed, meaning that the matching degree between the user response load and the photovoltaic output is insufficient, and there is a risk of photovoltaic output waste or backfeed.

[0044] The equipment safety verification model is constructed with the safety margin of power grid equipment as the core evaluation indicator. Power grid equipment specifically includes main transformers and lines. The safety margin is calculated as follows: Safety Margin = (Rated Load Power of Equipment - Actual Load Power of Equipment - User Response Load) / Rated Load Power of Equipment × 100%. The verification judgment logic of this model is as follows: If the calculated safety margin is ≥ 5%, the equipment safety verification is deemed to have passed, indicating that the current load of the power grid equipment has not exceeded the safe range and there is no risk of overload; if the safety margin is < 5%, the equipment safety verification is deemed to have failed, indicating that the load of the power grid equipment has approached or exceeded the safe threshold and needs to be reduced through regulation.

[0045] The response effectiveness verification model is constructed with the user response compliance rate as the core evaluation indicator. The user response compliance rate is calculated as follows: User response compliance rate = Number of users whose response load deviation from command load is ≤10% / Total number of users receiving commands × 100%. Here, response load refers to the actual load adjustment by the user after receiving the control command issued by the power grid; command load refers to the load adjustment target issued by the power grid to the user according to the control requirements; and the deviation between response load and command load is the numerical difference between the two.

[0046] The deviation between response load and command load is the core evaluation criterion for response effectiveness verification. Specifically, it involves calculating the proportion of users with a deviation ≤10% from the command load to the total number of users receiving commands. If the user response compliance rate is ≥80%, the response effectiveness verification passes; otherwise, it fails. Quantitative evaluation of the deviation between response load and command load allows for accurate assessment of the actual execution effect of control commands by users, providing a direct basis for adjusting commands in subsequent closed-loop control and preventing ineffective execution of control commands. The verification model's judgment logic is as follows: if the calculated user response compliance rate is ≥80%, the response effectiveness verification passes, indicating that most users receiving commands can complete the load adjustment as required; if the user response compliance rate is <80%, the response effectiveness verification fails, indicating that some users have not effectively executed control commands, and the command issuance strategy needs to be optimized.

[0047] When verifying the current operating status of the power grid using the above three-dimensional verification models, it is necessary to calculate the deviation rate through the photovoltaic output matching verification model, the safety margin through the equipment safety verification model, and the user response compliance rate through the response effectiveness verification model. Then, a comprehensive judgment is made based on the verification judgment rules: if the photovoltaic output matching verification, equipment safety verification, and response effectiveness verification all pass, the current operating status of the power grid is determined to be safe, and the current control command is maintained; if any one of the three dimensions fails, the current operating status of the power grid is determined to be unsafe, and the command adjustment process in the closed-loop control strategy generation step needs to be triggered.

[0048] In this embodiment, the present invention constructs a verification model encompassing three dimensions: photovoltaic output matching verification, equipment safety verification, and response effectiveness verification. This verification model is used to verify the current operating state of the power grid and determine whether the current operating state passes the verification. The photovoltaic output matching verification of the present invention can avoid photovoltaic output waste or backfeed, improving photovoltaic absorption efficiency. The equipment safety verification of the present invention can promptly identify overload risks of main transformers and lines, ensuring the safety of power grid equipment. The response effectiveness verification of the present invention can ensure that user responses meet standards and avoid invalid control commands. The combination of these three aspects can comprehensively assess the power grid status, avoiding omissions in single-dimensional verification and ensuring the safe operation of the power grid.

[0049] Optionally, training the photovoltaic output matching verification model requires selecting historical data covering different seasonal weather conditions and electricity consumption periods, including: actual photovoltaic output data, user response load data, and data time granularity consistent with actual collection. Data should be collected at preset threshold time points, and must include at least 1000 complete photovoltaic output-response load corresponding samples to ensure coverage of various scenarios such as high photovoltaic generation, regular output, and low output. Feature engineering: Derived features based on actual photovoltaic output and user response load as core features include: the ratio of the difference in actual photovoltaic output between adjacent time periods to the output of the previous period; the ratio of the difference in user response load between adjacent time periods to the load of the previous period; and date types, such as weekday or holiday time period labels, morning peak, noon peak, and evening peak, as auxiliary features to improve the model's adaptability to different scenarios. Training Parameter Settings: The model employs a judgment logic based on the deviation rate threshold for training. Core training parameters include: an initial reasonable deviation rate threshold of 15% as the training baseline; normalization using minmax normalization; mapping photovoltaic output and response load data to a 0-1 interval; and an abnormal sample filtering threshold to remove extreme abnormal samples with a deviation rate exceeding 50% to avoid affecting model threshold calibration. Training Process: The preprocessed training data is divided into a training set and a validation set in a 7:3 ratio to maximize the consistency between the model's judgment results and the actual safe operation of the power grid. The deviation rate threshold is iteratively optimized and adjusted to achieve this goal. If the proportion of photovoltaic output backfeeding faults caused by deviation rate judgments in the validation set exceeds 1%, the threshold is reduced by 0.5%. If the proportion of samples requiring unnecessary adjustments exceeds 5% due to an excessively high threshold, the threshold is increased by 0.5%. The iteration count is set to 100 times to determine a stable deviation rate. The default judgment threshold is 15%, which can be dynamically calibrated according to the regional photovoltaic grid connection scale.

[0050] Optionally, training the equipment safety verification model requires collecting historical operating data of the main transformer lines within the target power grid, including: rated load power, actual load power, and user response load data. It also requires recording equipment overload fault history, load power at the time of fault occurrence, response load, and operating duration. The data volume must cover at least 50 main transformers and 100 lines, spanning six consecutive months of operating data to ensure diversity in equipment type and operating scenarios. Feature engineering: Core features include actual load power, user response load, and rated load power. Derived features include load rate and load growth rate. Equipment operating years and maintenance record tags are also incorporated as auxiliary features to adapt to the safety threshold requirements of equipment in different health states. Training parameter settings: Key training parameters include an initial safety margin benchmark threshold of 5%; an initial load rate warning interval of 80%-90%, where above 90% is considered a high-risk interval; a training iteration step size of 0.2% for adjusting the safety margin threshold; and a regularization parameter of 0.01 to avoid model overfitting. Training Process: A supervised learning model was adopted, with training data divided into training and testing sets in an 8:2 ratio. The accuracy of equipment overload fault prediction and the false alarm rate were used as dual evaluation metrics. If the overload fault prediction accuracy on the testing set was below 95%, the safety margin threshold was reduced by 0.2%; if the false alarm rate was above 3%, the safety margin threshold was increased by 0.2%. The model parameters were optimized using gradient descent, and iterative training was conducted until the evaluation metrics showed no significant change after 10 consecutive iterations. The final safety margin threshold and related parameters were then determined.

[0051] Optionally, training the response effectiveness verification model requires training data preparation: collecting historical user control command records, actual response load data, and execution data of different command types covering high, medium, and low priority users. The data must include user ID, command load, response load, response duration, and command execution time period, with a sample size of no less than 5000 records. User command execution records must ensure coverage of different user types and command scenarios. Feature engineering: Core features are command load, response load, and the deviation between response load and command load. Derived features include: response compliance indicator (deviation ≤10% marked as 1, otherwise 0), user priority label, command type label (increasing load is 1, decreasing load is 0), and time period label. Historical user response rate and adjustable load percentage are also included as auxiliary features to improve the model's adaptability to user response capabilities. Training parameter settings: Core training parameters include: an initial response compliance deviation threshold of 10%; an initial user response compliance rate threshold of 80%; data sampling weights assigning 1.2 times the weight to low-priority user samples to balance sample distribution; a training batch size of 32; and a learning rate of 0.001. Training Process: A training framework combining a binary classification model (determining whether a single user's response meets the standard) and a statistical model (calculating the user response compliance rate) is constructed. First, the binary classification model is trained using a training set to optimize the response compliance deviation threshold: if the model's accuracy for high-priority users' response compliance is below 90%, the deviation threshold is reduced by 0.5%; if the misclassification rate for low-priority users is above 8%, the deviation threshold is increased by 0.5%. Then, based on the optimized deviation threshold, the statistical model is trained to calibrate the user response compliance rate determination threshold, ensuring that the compliance rate determination results in the validation set match the actual grid control effect with a degree of greater than 92%. Finally, stable model parameters are determined.

[0052] S40: Generate initial control instructions based on user priority results and verification results.

[0053] Specifically, in the process of generating initial control instructions, the user priority results and verification results obtained from the user hierarchical classification steps should be used as dual core bases. First, the control priority allocation rules should be clarified, and then instructions should be generated in combination with the actual operation needs of the power grid.

[0054] First, user type priority serves as the basic allocation basis, with priority determined in a fixed order from highest to lowest: high-priority responders, medium-priority responders, and low-priority responders. High-priority responders correspond to users with a photovoltaic access attribute marker of 1 and a response elasticity ≥ 0.6; medium-priority responders correspond to users with a photovoltaic access attribute marker of 0 and a response elasticity of 0.3 ≤ response elasticity < 0.6; and low-priority responders correspond to users with a response elasticity ≤ 0.3. This fixed order ensures a basic match between user response capabilities and control requirements, avoiding misallocation of control resources due to user type confusion.

[0055] Secondly, equipment safety margin, as a key influencing factor, needs to be calculated using the formula: Equipment Safety Margin = (Rated Load Power of Equipment - Actual Load Power of Equipment - User Response Load) / Rated Load Power of Equipment × 100%. Then, the control priority of users in the corresponding area is adjusted based on the calculation results. For example, if the safety margin of the power grid equipment in a certain area is lower, it indicates that the current actual load of the equipment in that area is closer to the rated load power, and may even be close to overload. In this case, the control priority of all priority users in that area needs to be increased accordingly to reduce the equipment load by executing control commands earlier, thus avoiding safety risks caused by continuous high load operation. For instance, if the safety margin of the main transformer in a certain area is 3%, which is lower than the 5% verification threshold, the control priority of high-priority response users in that area must be higher than that of high-priority response users in another area with a safety margin of 6%.

[0056] Furthermore, photovoltaic (PV) matching deviation, another important influencing factor, needs to be calculated using the formula: PV matching deviation (deviation rate) = |Actual PV Output - User Response Load| / Actual PV Output × 100%. Then, the control priority of users in the corresponding area should be adjusted based on the calculation results. A larger PV matching deviation in a region indicates a worse match between the user response load and the actual PV output, potentially leading to insufficient PV output absorption and waste, or PV output being fed back, increasing equipment load. In this case, the control priority of all priority users in the region needs to be increased accordingly to adjust user response loads earlier and improve the matching effect between PV output and user load. For example, if the PV matching deviation rate in a region is 18%, higher than the 15% verification threshold, the control priority of medium-priority response users in that region should be higher than that of medium-priority response users in another region with a PV matching deviation rate of 12%.

[0057] Finally, the specific order of regulation is determined by comprehensively judging the above three factors: first, the basic level is divided according to the priority of user type; then, within the same user type priority, the order is further sorted according to the level of equipment safety margin and the magnitude of photovoltaic matching deviation. The lower the equipment safety margin and the larger the photovoltaic matching deviation, the higher the regulation order of the corresponding priority users in the region. For example, in region 1, a high-priority responder (high user type priority), with an equipment safety margin of 3% (low) and a photovoltaic matching deviation rate of 18% (large), and in region 2, a high-priority responder (high user type priority), with an equipment safety margin of 6% (high) and a photovoltaic matching deviation rate of 12% (small), the regulation order of the high-priority responder in region 1 takes precedence over that in region 2. If a medium-priority responder in region 3 has an equipment safety margin of 2% (low) and a photovoltaic matching deviation rate of 20% (large), its regulation order can take precedence over that of the high-priority responder in region 2. This ensures that regulation resources are tilted towards the regions and users that most urgently need adjustment, maximizing regulation efficiency.

[0058] Based on the determined control priorities and combined with the grid operation requirements reflected by the multi-dimensional grid verification results, initial control instructions are generated: If the multi-dimensional grid verification results show that the photovoltaic output is not being fully absorbed, and it is during a period of high photovoltaic generation, then an instruction to increase adjustable load should be issued to the highest priority responders with the highest control priority. By increasing the adjustable load scale of the high priority responders, the absorption capacity of the actual photovoltaic output is enhanced. If the multi-dimensional grid verification results show that the grid equipment load is too high, and it is during a period of equipment overload, then an instruction to reduce adjustable load should be issued to the highest priority responders with the highest control priority. By reducing the adjustable load scale of the high priority responders, the load pressure on the grid equipment is reduced.

[0059] If, during the generation of initial control commands, it is found that the adjustable load adjustment amount of high-priority response users cannot meet the grid control requirements, then energy storage devices and adjustable loads need to be called upon to respond in coordination. While issuing corresponding adjustable load adjustment commands to high-priority response users, the response load is supplemented through energy storage devices to ensure that the initial control commands can initially adapt to the current operation and adjustment requirements of the grid.

[0060] In this embodiment, an initial control command is generated based on the user priority result and the verification result obtained from the multi-dimensional power grid verification step. This invention, by prioritizing the use of high-priority responders, can accurately match photovoltaic power consumption or equipment load reduction needs, avoiding blind control. Furthermore, by combining the verification result with the generated command, it can specifically address issues such as poor photovoltaic matching and equipment overload, reducing ineffective control.

[0061] S50: Perform real-time feedback verification of the execution effect of the initial control command. If the feedback verification fails, adjust the control command and repeat the feedback verification and command adjustment process until the power grid operation status passes the verification.

[0062] Specifically, when verifying the real-time feedback of the execution effect of the initial control command, the work must be carried out in strict accordance with fixed time cycles and data collection requirements: After the control command is issued, three types of core data must be collected within a preset time: actual load response data of users, that is, the load amount actually adjusted by users after receiving the control command; actual output data of photovoltaics, that is, the actual power generation data of distributed photovoltaic equipment during the execution period of the control command; and actual load power data of grid equipment, that is, the power data currently actually carried by the main transformer and the line after the execution of the control command, to ensure that the collected data can fully reflect the actual execution effect of the initial control command.

[0063] After data collection is completed, the photovoltaic output matching verification model, equipment safety verification model, and response effectiveness verification model constructed in the multi-dimensional power grid verification steps need to be invoked to re-verify the collected user actual response load data, photovoltaic actual output data, and power grid equipment actual load power data: the photovoltaic output matching verification model is used to calculate the deviation rate between the actual photovoltaic output and the user actual response load to determine whether the photovoltaic output matching verification has passed; the equipment safety verification model is used to calculate the safety margin of the main transformer and the line to determine whether the equipment safety verification has passed; and the response effectiveness verification model is used to calculate the user response compliance rate to determine whether the response effectiveness verification has passed.

[0064] If the re-verification results show that the photovoltaic output matching verification, equipment safety verification, and response effectiveness verification all pass, the grid operation status is determined to meet safety requirements, and the current adjusted control instructions are maintained. If the re-verification results show that any dimension of the verification fails, the control instruction adjustment process needs to be initiated: the control instructions for medium-priority response users are adjusted first, and the adjustment direction is determined according to the dimension that failed the verification. If the photovoltaic output matching verification fails, an instruction to increase adjustable load is issued to medium-priority response users; if the equipment safety verification fails, an instruction to reduce adjustable load is issued to medium-priority response users; if the response effectiveness verification fails, the content of the instructions issued to medium-priority response users is optimized to improve the response compliance rate.

[0065] After adjusting the control commands for medium-priority response users, new actual user response load data, actual photovoltaic output data, and actual grid equipment load power data must be collected again within a preset time, and the above re-verification process must be repeated. If the re-verification still fails after adjusting the control commands for medium-priority response users, and the adjustable loads of high-priority and medium-priority response users have reached their adjustment limits, then energy storage devices are called in to supplement the response load. The energy storage devices adjust their own output to help meet the grid control requirements, and then data is collected again and re-verified. This feedback verification and command adjustment process is repeated until the photovoltaic output matching verification, equipment safety verification, and response effectiveness verification are passed, ensuring that the grid operation status meets safety standards.

[0066] In this embodiment, the execution effect of the initial control command is verified in real time. If the verification fails, the control command is adjusted, and the process of verification and command adjustment is repeated until the grid operation status passes the verification. This invention can promptly detect deficiencies in the initial control command, preventing the grid from being in an unsafe state for a long time due to command failure. At the same time, repeated verification and adjustment can adapt to real-time changes such as photovoltaic output fluctuations, enhancing the grid's ability to cope with dynamic operating conditions. Ultimately, it ensures that the grid achieves a balance between photovoltaic absorption, equipment safety, and effective response through multi-dimensional verification.

[0067] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for generating power grid verification and control strategies based on user response load, characterized in that, include: Preprocess the collected distributed photovoltaic data, user response load data, and grid equipment data; A two-dimensional classification standard is constructed based on photovoltaic access attributes and response resilience. Users are then stratified according to this standard to obtain high-priority response users, medium-priority response users, and low-priority response users. The photovoltaic access attributes are used to distinguish whether a user has access to distributed photovoltaic power. The response resilience is determined based on the user's historical response rate and adjustable load ratio. A verification model is constructed with three dimensions: photovoltaic output matching verification, equipment safety verification, and response effectiveness verification. The verification model is used to verify the current operating status of the power grid and determine whether the current operating status of the power grid passes the verification. Based on the user priority results and verification results, an initial control command is generated, and the execution effect of the initial control command is verified in real time. If the verification fails, the control command is adjusted, and the process of verification and command adjustment is repeated until the power grid operation status passes the verification.

2. The method for generating power grid verification and control strategies based on user response load according to claim 1, characterized in that, The preprocessing of the collected distributed photovoltaic data, user response load data, and grid equipment data includes: Outlier removal and missing data completion are performed using horizontal and vertical outlier removal rules. The rule for removing horizontal outliers is as follows: if the load change at different times on the same day exceeds the reasonable threshold during the day and the change is not caused by the alternation of day and night, then the load data will be removed. The vertical outlier removal rule is as follows: if the load change on the same day and at the same time exceeds the reasonable threshold for cross-day changes and the change is not caused by weather, then the load data is removed; the missing data completion uses linear interpolation. Using the preprocessed user response load data, photovoltaic output data, and grid equipment load data, the equipment safety margin and photovoltaic matching deviation are calculated. Based on the safety margin of the calculated equipment and the matching deviation of the photovoltaic system, the priority of regulation is comprehensively evaluated from two dimensions: the safety margin of the equipment and the matching deviation of the photovoltaic system.

3. The method for generating power grid verification and control strategies based on user response load according to claim 1, characterized in that, The method for determining the photovoltaic access attribute includes: if a user accesses distributed photovoltaic, they are marked as a photovoltaic user; if a user does not access distributed photovoltaic, they are marked as an ordinary user; the mark value of the photovoltaic user is 1, and the mark value of the ordinary user is 0.

4. The method for generating power grid verification and control strategies based on user response load according to claim 1, characterized in that, The response resilience is calculated by weighting the historical response rate and the adjustable load percentage, where the calculation formula is: Response Resilience = α × Historical Response Rate + β × Adjustable Load Percentage; α is the weighting coefficient of the historical response rate, and β is the weighting coefficient of the adjustable load percentage. The values ​​of α and β are calibrated according to the characteristics of users in the region; the historical response rate = number of successful responses in the user's history / total number of commands received by the user; the adjustable load percentage = peak adjustable load of the user / peak total load of the user.

5. The method for generating power grid verification and control strategies based on user response load according to claim 1, characterized in that, The process of classifying users into high-priority, medium-priority, and low-priority response users based on the dual-dimensional classification criteria includes: If a user's photovoltaic access attribute is marked as 1 and the response elasticity is ≥0.6, then the user is a high-priority response user; if a user's photovoltaic access attribute is marked as 0 and 0.3≤response elasticity<0.6, then the user is a medium-priority response user; if a user's response elasticity is ≤0.3, then the user is a low-priority response user.

6. The method for generating power grid verification and control strategies based on user response load according to claim 1, characterized in that, If the photovoltaic output matching verification, equipment safety verification, and response effectiveness verification all pass, the current operating state of the power grid is determined to be safe, and the current control command is maintained. If any dimension of the photovoltaic output matching verification, equipment safety verification, or response effectiveness verification fails, the current operating state of the power grid is determined to be unsafe, and the command adjustment process is triggered.

7. The method for generating power grid verification and control strategies based on user response load according to claim 1, characterized in that, Based on user priority results and verification results, initial control instructions are generated using a control priority allocation method. Specifically, this includes: allocating control priorities by combining user type priority, equipment safety margin, and photovoltaic matching deviation; where user type priority is from high to low as high priority responders, medium priority responders, and low priority responders; the lower the equipment safety margin, the higher the control priority of the corresponding user in the region; the greater the photovoltaic matching deviation, the higher the control priority of the corresponding user in the region.

8. The method for generating power grid verification and control strategies based on user response load according to claim 1, characterized in that, The initial control command generation logic includes: issuing control commands to high-priority response users first; issuing commands to increase adjustable load to high-priority response users during periods of high photovoltaic power generation; and issuing commands to reduce adjustable load to high-priority response users during periods of equipment overload.

9. The method for generating power grid verification and control strategies based on user response load according to claim 1, characterized in that, The execution effect of the initial control command is evaluated in real time, including: if the response load of high-priority responders cannot meet the grid control requirements, energy storage devices are invoked to coordinate with adjustable loads to supplement the response load.

10. The method for generating power grid verification and control strategies based on user response load according to claim 1, characterized in that, If the feedback verification fails, the control command will be adjusted, and the feedback verification and command adjustment process will be repeated until the power grid operating status passes the verification, including: After the control command is issued, the actual response load data of users, the actual output data of photovoltaic power, and the actual load power data of grid equipment are collected within a preset time. The collected data are then re-verified using the verification model. If the re-verification fails, the control command of the medium-priority response users is adjusted, and the data is collected and re-verified again until the grid operation status passes the verification.

Images