A new energy power grid power generation power prediction method based on large model analysis

Abstract

The application discloses a new energy power grid power generation power prediction method based on a large model analysis, relates to the technical field of power grid control, and comprises the following steps: collecting the actual active power of a grid-connected point, the dispatching limit upper limit, the DC bus voltage, the grid-connected point voltage, the feeder current, the energy storage chargeable power and the equipment available capacity, identifying the limit release preannouncement and the resource enhancement preannouncement through a large model; when the difference between the actual active power and the dispatching limit upper limit is not greater than the limit release following error limit, and the DC bus voltage is in a high regulation zone, determining a limit release pressure lag period, extracting the active release amount and the DC voltage drop amount before and after the natural upward adjustment of the limit upper limit; determining the residual pressure retention value and the predicted increased power generation amount according to the limit release pressure lag time value, the active release amount and the DC voltage drop amount; generating the receivable power generation power prediction value and the energy storage charging, reactive power absorption and active power limitation input according to the predicted increased power generation amount, the grid-connected point voltage, the feeder current and the energy storage chargeable power, and identifying the hidden increased power generation risk.

Description

Technical Field

[0001] This invention relates to the field of power grid control technology, and in particular to a method for predicting the power generation of new energy power grids based on large model analysis. Background Technology

[0002] With the increasing integration of distributed photovoltaic, decentralized wind power, and energy storage devices, the operation of new energy power grids is gradually shifting from centralized dispatch to distributed collaborative control. Conventional power generation forecasting methods typically combine historical actual power generation, weather forecasts, dispatch-limited power generation information, and grid-connected point operating volumes to generate a curve of available power generation or grid-connected power for future periods. This curve is then used by the dispatch center to arrange active power allocation, energy storage charging and discharging, feeder safety checks, and short-term operation plans.

[0003] In scenarios where power generation is restricted, the actual active power generated on the AC side is often flattened by the upper limit of the dispatch restriction. Conventional forecasts cannot identify the resource capacity that is still suppressed before the restriction is lifted from the stable power curve. At the same time, the forecast curves mostly remain at the power value level, making it difficult to directly form distributed control input relationships such as energy storage charging, reactive power absorption, and active power limitation. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a new energy power generation prediction method based on large model analysis to solve the problems of insufficient identification of hidden additional generation risks before the lifting of power generation restrictions and the inability of prediction results to directly serve distributed control in existing technologies.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] This invention provides a method for predicting the power generation of a new energy power grid based on large-scale model analysis. The method includes: collecting the actual active power generated at the new energy grid connection point, the dispatch limit for power generation, the inverter DC bus voltage, the grid connection point voltage, the feeder current, the rechargeable power of energy storage, and the available capacity of equipment; and identifying the lifting of power generation restrictions and resource enhancement warnings from dispatch limit texts, lifting notices, and meteorological enhancement warning texts using a large-scale power model. When the difference between the actual active power generated and the dispatch limit for power generation is not greater than the limit for power generation following error limit, and the inverter DC bus voltage is in the high DC voltage regulation zone, the power generation lag period is determined, and the method is applied during the power generation lag period. The active power release and DC voltage reduction are extracted before and after the upper limit of the power generation restriction is naturally adjusted from the first upper limit to the second upper limit. The residual voltage reserve value of the power generation restriction is determined based on the time value of the power generation restriction lag period, the active power release, and the DC voltage reduction. When the residual voltage reserve value of the power generation restriction reaches the residual voltage reserve limit, the predicted increase in power generation is determined according to the available capacity of the equipment and the adjusted actual active power. The predicted value of the power generation that the grid can accept is generated based on the predicted increase in power generation, the grid connection point voltage, the feeder current, and the rechargeable power of the energy storage. The energy storage charging input, reactive power absorption input, and active power restriction input are generated according to the energy storage charging capacity, the voltage proximity state, and the feeder current proximity state.

[0008] As a preferred embodiment of the new energy grid power generation prediction method based on large model analysis described in this invention, the step of collecting the actual active power generated by the new energy grid connection point, the dispatch limit, the inverter DC bus voltage, the grid connection point voltage, the feeder current, the rechargeable power of energy storage, and the available capacity of equipment includes: acquiring the actual active power generated, the dispatch limit, the inverter DC bus voltage, the grid connection point voltage, the feeder current, the rechargeable power of energy storage, and the available capacity of equipment according to the same prediction sampling period; and associating the data belonging to the same prediction sampling period according to the new energy grid connection point number and the sampling time to form grid connection point prediction sampling data.

[0009] As a preferred embodiment of the new energy power generation prediction method based on large model analysis described in this invention, the specific steps of identifying the power restriction lifting forecast and resource enhancement forecast from dispatch restriction texts, power restriction lifting notices, and meteorological enhancement warning texts using a power sector large model are as follows: The dispatch restriction texts, power restriction lifting notices, and meteorological enhancement warning texts are organized into unified prompt data and input into the power sector large model; the power sector large model outputs the power restriction lifting forecast, resource enhancement forecast, the relevant new energy grid connection points, the effective time period, the allowed power field after lifting, and the latest power restriction upper limit field; when the output content of the power sector large model lacks the new energy grid connection point number or the effective time period, or when both the allowed power field after lifting and the latest power restriction upper limit field are missing, the corresponding output content is deleted so that the corresponding output content does not participate in subsequent power generation prediction; the unified prompt data includes the new energy grid connection point number, the current dispatch restriction upper limit, the start time of the power restriction, the planned time of the power restriction lifting, the meteorological enhancement type, the start time of the meteorological enhancement, and the end time of the meteorological enhancement.

[0010] As a preferred embodiment of the new energy power generation prediction method based on large model analysis described in this invention, the determination of the power lag period includes: when the difference between the actual generated active power and the upper limit of the dispatched power lag is not greater than the power lag following error limit, and the inverter DC bus voltage is not lower than the lower limit of the DC voltage high adjustment zone, the prediction sampling period is determined as the power lag sampling period; when two or more consecutive power lag sampling periods are located within the same dispatched power lag period, the time period composed of two or more consecutive power lag sampling periods is determined as the power lag period; the ratio of the cumulative duration of the inverter DC bus voltage in the DC voltage high adjustment zone within the power lag period to the total duration of the same power lag period is determined as the power lag time value.

[0011] As a preferred embodiment of the new energy grid power generation prediction method based on large model analysis described in this invention, the step of extracting the active power release and DC voltage reduction before and after the natural increase of the dispatch limit on the upper limit from the first upper limit to the second upper limit during the power restriction lag period includes: reading the records of changes in the dispatch limit on the upper limit during the power restriction lag period, and using the effective period of the power restriction lifting announcement as a time constraint, filtering out the upward adjustment process of the dispatch limit on the upper limit that occurs before the effective period of the power restriction lifting announcement; sorting the records of the dispatch limit on the upper limit of the same new energy grid connection point by time according to the effective time of the upper limit on the upper limit, and determining the upper limit power value in the later dispatch limit on the upper limit record when it is greater than the previous one. When determining the upper limit power value in the upper limit record of the dispatching power limit, the upper limit power value in the previous dispatching power limit record is taken as the first upper limit, and the upper limit power value in the next dispatching power limit record is taken as the second upper limit. The complete prediction sampling period before the end of the first upper limit is taken as the pre-adjustment sampling period, and the complete prediction sampling period after the start of the second upper limit is taken as the post-adjustment sampling period. The natural upward adjustment of active power release is determined based on the actual active power generated within the pre-adjustment sampling period and the post-adjustment sampling period, and the natural upward adjustment of DC voltage reduction is determined based on the inverter DC bus voltage within the pre-adjustment sampling period and the post-adjustment sampling period.

[0012] As a preferred embodiment of the new energy power generation prediction method based on large model analysis described in this invention, the determination of the residual pressure retention value for power generation restriction includes: when the naturally increased active power release reaches the effective release limit of natural increase, the current scheduling power generation restriction upper limit increase process is determined as an effective natural increase process; when there are multiple effective natural increase processes within the same power generation restriction lag period, the effective natural increase process with the shortest time interval is selected according to the time interval between the end time of the adjusted sampling period of the effective natural increase process and the start time of the effective period of the power generation restriction lifting announcement; the residual pressure retention value for power generation restriction is determined based on the power generation restriction lag period value, the naturally increased DC voltage reduction amount, the effective limit of DC voltage reduction, the naturally increased active power release amount, and the effective release limit of natural increase.

[0013] As a preferred embodiment of the new energy power grid power generation prediction method based on large model analysis described in this invention, the determination of the predicted increase in power generation includes: when there is an effective natural upward adjustment process, if the residual pressure reserve value of the power restriction reaches the residual pressure reserve limit, a rapid increase judgment is generated, and the predicted increase in power generation is determined based on the residual pressure reserve value of the power restriction, the available capacity of the equipment, and the adjusted actual active power; when there is an effective natural upward adjustment process and the residual pressure reserve value of the power restriction has not reached the residual pressure reserve limit, a normal recovery judgment is generated, and the predicted increase in power generation is recorded as zero; when there is no effective natural upward adjustment process and the time value of the power restriction lag reaches the effective lag limit, it is first determined whether the power restriction lifting announcement is in effect. Within the specified time period; if the issuance restriction lifting forecast is within the valid time period, a conservative issuance increase judgment is generated, and the predicted issuance increase is determined based on the issuance restriction lag time value, lag effective limit, equipment available capacity, and current actual active power; if the issuance restriction lifting forecast is not within the valid time period, it is then determined whether the resource enhancement forecast is within the valid time period. If the resource enhancement forecast is within the valid time period, a conservative issuance increase judgment is generated, and the predicted issuance increase is determined based on the issuance restriction lag time value, lag effective limit, equipment available capacity, and current actual active power; if both the issuance restriction lifting forecast and the resource enhancement forecast are not within the valid time period, a normal prediction judgment is generated, and the predicted issuance increase is recorded as zero.

[0014] As a preferred embodiment of the new energy grid power generation prediction method based on large model analysis described in this invention, the generation of the grid-acceptable power generation prediction value includes: during historical normal grid connection periods, selecting sample pairs where the grid connection point voltage does not exceed the limit and the active power change reaches the active power change screening lower limit, and using the median value of the ratio of the grid connection point voltage change to the active power change of each sample pair as the voltage active power sensitive quantity; during historical normal grid connection periods, selecting sample pairs where the feeder current does not exceed the limit and the active power change of the new energy grid connection point within the feeder reaches the feeder active power change screening lower limit, and using the median value of the ratio of the feeder current change to the active power change of each sample pair as the current active power sensitive quantity; using the voltage active power sensitive quantity to calculate the voltage-side increase power corresponding to the grid connection point voltage being lower than the voltage upper limit, and using the current active power sensitive quantity to calculate the current-side increase power corresponding to the feeder current being lower than the feeder allowable current, forming the voltage allowable power boundary and the current allowable power boundary used to constrain the predicted increase power.

[0015] As a preferred embodiment of the new energy grid power generation prediction method based on large model analysis described in this invention, the step of generating the grid-acceptable power generation prediction value further includes: when the predicted increase in power generation is not greater than the rechargeable power of energy storage, using the predicted increase in power generation as the energy storage charging input; when the predicted increase in power generation is greater than the rechargeable power of energy storage, using the rechargeable power of energy storage as the energy storage charging input; subtracting the energy storage charging input from the resource-available power generation prediction value to obtain the grid-connected power after energy storage absorption; sequentially limiting the grid-connected power after energy storage absorption using the voltage allowable power boundary, the current allowable power boundary, and the dispatch allowable power boundary to obtain the grid-acceptable power generation prediction value; when the grid-connected power after energy storage absorption is greater than the grid-acceptable power generation prediction value, using the difference between the two as the active power limit input; when the grid-connected power after energy storage absorption is not greater than the grid-acceptable power generation prediction value, recording the active power limit input as zero.

[0016] As a preferred embodiment of the new energy grid power generation prediction method based on large model analysis described in this invention, the generation of energy storage charging input, reactive power absorption input, and active power limiting input includes: when the rapid power increase judgment is valid and the energy storage charging power is not less than the predicted power increase, the entire predicted power increase is converted into energy storage charging input, and the reactive power absorption input and active power limiting input are recorded as zero; when the rapid power increase judgment is valid, the energy storage charging power is less than the predicted power increase, and the grid connection point voltage enters a state close to the upper voltage limit, the energy storage charging power is converted into energy storage charging input, the inverter's available reactive power absorption capacity is converted into reactive power absorption input, and the active power portion that still exceeds the predicted power generation value that the grid can accept after energy storage absorption is converted into... Active power limiting input; when the rapid increase judgment is valid, the energy storage charging power is less than the predicted increase, the grid connection point voltage has not entered the state close to the upper voltage limit and the feeder current has entered the state close to the feeder allowable current, the energy storage charging power is converted into energy storage charging input, the reactive power absorption input is recorded as zero, and the active power that still exceeds the predicted value of the grid's acceptable power generation after energy storage absorption is converted into active power limiting input; when the normal recovery judgment or the normal prediction judgment is valid, the predicted increase, energy storage charging input, reactive power absorption input and active power limiting input are all recorded as zero, and the grid's acceptable power generation prediction value is output according to the minimum value among the current actual active power, voltage allowable power boundary, current allowable power boundary and dispatch allowable power boundary.

[0017] The beneficial effects of this invention are as follows: By identifying the period of lag in power generation restrictions and the insufficient DC voltage reduction after the natural increase of the power generation limit, the hidden risk of rapid power generation increase can be detected in advance before the power generation restriction is lifted; by extracting only event fields from scheduling and meteorological texts in a large model and combining them with electrical quantities to calculate and predict the power generation increase, the power prediction process can be verified and black-box output can be avoided; by converting the predicted power generation increase into energy storage charging, reactive power absorption and active power limiting inputs, the prediction results can directly serve distributed control, reducing the risks of voltage rise, feeder overload and new energy curtailment. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of a new energy power generation prediction method based on large model analysis.

[0020] Figure 2 This is a schematic diagram for identifying hysteresis during power generation.

[0021] Figure 3 This is a schematic diagram for determining the natural upward adjustment of the pressure.

[0022] Figure 4 A schematic diagram for accepting forecast and control allocation. Detailed Implementation

[0023] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0024] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0025] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0026] Reference Figures 1-4This is one embodiment of the present invention, which provides a method for predicting the power generation of a new energy power grid based on large model analysis, including the following steps:

[0027] S1. Collect the actual active power generated by new energy grid connection points, dispatch limit, inverter DC bus voltage, grid connection point voltage, feeder current, energy storage chargeable power, and equipment available capacity. Then, through a large power sector model, identify the release warning and resource enhancement warning from dispatch limit texts, release limit lifting notices, and meteorological enhancement warning texts.

[0028] The system reads the actual active power generated in the grid-connected metering cabinet at the new energy grid-connected point, reads the dispatch limit on the dispatch command receiving end, reads the inverter DC bus voltage at the inverter DC side sampling end, reads the grid-connected point voltage at the grid-connected voltage sampling end, reads the feeder current at the feeder outgoing metering end, reads the energy storage chargeable power at the energy storage converter control end, reads the available capacity of the equipment and the available reactive power absorption capacity of the inverter in the new energy grid-connected point equipment status table, reads the voltage limit and the allowable current of the feeder in the feeder operation parameter table, and forms grid-connected point predicted sampling data according to the same predicted sampling period (e.g., 5 minutes).

[0029] Furthermore, the available capacity of the equipment is determined based on the rated capacity of the new energy grid connection point, the available capacity of the inverter, the available capacity of the wind turbine, the available capacity of the photovoltaic string, and the equipment maintenance capacity.

[0030] For distributed photovoltaic grid-connected points, the smaller of the available capacity of the inverter and the available capacity of the photovoltaic string is taken as the available capacity of the equipment; for distributed wind power grid-connected points, the available capacity of the wind turbine is taken as the available capacity of the equipment; for new energy grid-connected points with maintenance and shutdown equipment, the available capacity of the equipment is determined by deducting the capacity of the maintenance and shutdown equipment from the rated capacity of the new energy grid-connected point.

[0031] The rechargeable power of energy storage is determined based on the rated charging power of the energy storage, the rated capacity of the energy storage, the upper limit of the energy storage state of charge, the current energy storage state of charge, the predicted sampling period, and the allowable charging power of the energy storage management device. The expression is as follows:

[0032] ;

[0033] in, Indicates the rechargeable power of energy storage. Indicates the rated charging power of the energy storage. Indicates the rated capacity of energy storage. Indicates the upper limit of the energy storage state of charge. Indicates the current state of charge of the energy storage. Indicates the predicted sampling period. This indicates the allowed charging power of the energy storage management device.

[0034] It should be noted that the upper limit of the energy storage state of charge is determined by the energy storage operation procedure, and the value ranges from 0.85 to 0.95; when the current energy storage state of charge is not less than the upper limit of the energy storage state of charge, the energy storage rechargeable power is determined to be zero.

[0035] The upper limit of reactive power absorption by the inverter is determined based on the minimum power factor allowed by the grid connection agreement, the maximum reactive power regulation ratio allowed by the power grid company, and the rated apparent power capacity of the inverter. The expression for the upper limit of reactive power absorption by the inverter is:

[0036] ;

[0037] in, This indicates the upper limit of reactive power absorption by the inverter. This indicates the inverter's rated apparent power capacity. This indicates the minimum power factor allowed by the grid connection protocol. This indicates the maximum reactive power regulation ratio allowed by the power grid company.

[0038] The minimum power factor allowed by the grid connection agreement is between 0.90 and 0.99, and the maximum reactive power regulation ratio allowed by the power grid company is between 0.20 and 0.60.

[0039] The available reactive power absorption capacity of an inverter is determined based on the inverter's rated apparent power capacity, the actual active power generated during the current predicted sampling period, and the inverter's upper limit for reactive power absorption. The expression for the available reactive power absorption capacity of an inverter is:

[0040] ;

[0041] in, This indicates that the inverter has reactive power absorption capacity. This indicates the inverter's rated apparent power capacity. This represents the actual active power generated within the current prediction sampling period. This indicates the upper limit of reactive power absorption by the inverter.

[0042] Furthermore, voltage-sensitive quantities are determined by recording the voltage changes at the same new energy grid connection point before and after historical changes in active power.

[0043] Specifically, during historical normal grid connection periods, sampling pairs with active power changes not less than the active power change screening lower limit and grid connection point voltage not exceeding the limit are selected. The ratio of grid connection point voltage change to active power change for each sampling pair is calculated, and the median value of all ratios is determined as the voltage active power sensitive quantity.

[0044] It should be noted that the lower limit for active power variation screening is obtained during the commissioning and calibration phase of new energy grid-connected points. A stable operating period without dispatch restrictions, energy storage charging / discharging switching, inverter limits, or voltage alarms is selected. The absolute values ​​of actual active power variation in adjacent predicted sampling periods are read. These values ​​are then sorted from smallest to largest, and the highest quantile is selected. This highest quantile is then limited to the rated power limit of the new energy grid-connected point. The limited highest quantile is used as the lower limit for active power variation screening. The high-side quantile ratio ranges from 0.90 to 0.98, and the rated power limit range of the new energy grid-connected point is the power range corresponding to 0.5% to 3% of the rated power of the new energy grid-connected point.

[0045] Furthermore, the active current sensitive quantity is determined by recording the changes in feeder current before and after the historical active power changes of the same feeder.

[0046] Specifically, during historical normal grid connection periods, sampling pairs are selected where the change in active power at the new energy grid connection point within the feeder is not less than the lower limit of the feeder active power change screening and the feeder current has not exceeded the limit. The ratio of the feeder current change to the active power change for each sampling pair is calculated, and the median value of all ratios is determined as the current active power sensitive quantity.

[0047] It should be noted that the lower limit for filtering feeder active power changes is obtained during the feeder commissioning calibration phase. A stable operating period without feeder operation mode switching, protection current limiting actions, or feeder overload alarms is selected. The absolute value of the total active power change at the feeder's renewable energy grid connection point within adjacent prediction sampling periods is read. These absolute values ​​are sorted from smallest to largest, and the feeder's high-side quantile is taken. This high-side quantile is then limited to the feeder's renewable energy access capacity limit. The limited feeder high-side quantile is used as the lower limit for filtering feeder active power changes. The feeder high-side quantile ratio ranges from 0.90 to 0.98, and the feeder renewable energy access capacity limit ranges from 0.5% to 3% of the feeder's renewable energy access capacity, corresponding to a power range of [missing value].

[0048] Furthermore, before inputting dispatch restriction texts, restriction lifting notices, and enhanced meteorological warning texts into the power sector's large-scale model, the texts are first organized into unified prompt data.

[0049] The unified notification data includes the new energy grid connection point number, the current dispatch limit for power generation, the start time of the power generation limit, the planned time of lifting the power generation limit, the type of weather enhancement, the start time of the weather enhancement, and the end time of the weather enhancement.

[0050] The large-scale power generation model only outputs the forecast of the lifting of power restriction, the forecast of resource enhancement, the grid connection points of the new energy sources involved, the effective period, the allowed power after the lifting of the restriction, and the latest upper limit of the power restriction. It does not output the predicted value of power generation.

[0051] When the output of the large power model lacks the grid connection point number of new energy sources or the effective time period, the corresponding output content is deleted and will not participate in subsequent predictions; when both the allowed power field after the lifting of the restriction and the latest upper limit of power generation are missing, the corresponding output content is deleted and will not participate in subsequent predictions.

[0052] When the output of the large power model is missing the number of new energy grid connection points or the effective time period, the corresponding output is deleted and will not participate in subsequent predictions; when the allowed power field and the latest generation limit field are both missing after the lifting of the restriction, a field missing verification record is generated.

[0053] It should be noted that the large-scale power sector model adopts a decoding-based Transformer structure, which includes a text segmentation layer, a positional encoding layer, a multi-layer self-attention computation layer, a feedforward computation layer, a normalization layer, and a structured field output layer. The input content is unified prompt data compiled from dispatch restriction texts, restriction lifting notices, and meteorological enhancement warning texts. This unified prompt data includes the new energy grid connection point number, the current dispatch restriction limit, the restriction start time, the planned restriction lifting time, the meteorological enhancement type, the meteorological enhancement start time, and the meteorological enhancement end time. During training, historical dispatch restriction notices, restriction lifting notices, meteorological enhancement warnings, dispatch logs, and manual confirmation records are collected first. Each training text is labeled as a restriction lifting forecast, resource allocation forecast, or other related information. The model is trained by analyzing the source enhancement forecast, the relevant new energy grid connection points, the effective time period, the allowed power field after the lifting of the restriction, and the latest power generation limit field. The training text and field labels are then organized into sample pairs of input text and standard field output. Cross-entropy loss is used to train the large model to output the standard fields. After training, historical texts that were not used in the training are used for verification. When the extracted content of the new energy grid connection point number, effective time period, allowed power field after the lifting of the restriction, and latest power generation limit field is consistent with the manual annotation, the model parameters are retained. When the fields are missing, misaligned, or the output power generation prediction value is not generated, the corresponding samples are added to the retraining sample set for continued training. This ensures that the large model in the power field only undertakes the role of extracting dispatch text fields and identifying events, and does not directly generate power generation prediction values.

[0054] When the difference between the actual generated active power and the upper limit of dispatched generation limit is not greater than the generation limit following error limit, and the inverter DC bus voltage is not lower than the lower limit of the DC voltage high regulation zone, the corresponding prediction sampling period is determined as the generation limit lag sampling period; when two or more consecutive generation limit lag sampling periods are located within the same dispatched generation limit period, the time period composed of two or more consecutive generation limit lag sampling periods is determined as the generation limit lag period.

[0055] It should be noted that two or more consecutive sampling cycles of the power generation lag were used to eliminate the influence of single-cycle metering noise and communication delay, confirming that the actual active power limitation and the high regulation state of the DC bus have formed a continuous correlation.

[0056] It should be noted that the limit of the power generation restriction following error is determined by statistical analysis of the difference between the actual active power generated during the historical stable operation of the same new energy grid-connected point and the upper limit of the dispatched power generation restriction, with a value range of 0.5% to 3% of the rated power of the new energy grid-connected point; the lower limit of the DC voltage high regulation zone is determined by the inverter's normal maximum power tracking operation record, with a value range of 92% to 98% of the upper limit of the inverter's allowable DC bus voltage.

[0057] The ratio of the cumulative duration during which the inverter's DC bus voltage is in the high DC voltage regulation zone to the total duration of the same period is used as the lag time value for rationing.

[0058] S2. When the difference between the actual active power and the upper limit of dispatched power generation is not greater than the limit of power generation following error, and the inverter DC bus voltage is in the DC voltage high regulation zone, the power generation lag period is determined, and the active power release and DC voltage reduction are extracted before and after the upper limit of dispatched power generation is naturally adjusted from the first upper limit to the second upper limit during the power generation lag period.

[0059] During the period of restricted issuance, the record of changes in the upper limit of restricted issuance is read from the dispatch instruction receiving end, and the process of adjusting the upper limit of restricted issuance that occurs before the effective period of the notice of the lifting of restricted issuance is selected, with the effective period of the notice of the lifting of restricted issuance as the time constraint.

[0060] The scheduling emission limit records for the same renewable energy grid connection point are sorted by time according to the effective time of the emission limit. Two adjacent scheduling emission limit records are read. When the emission limit power value in the later scheduling emission limit record is greater than the emission limit power value in the earlier scheduling emission limit record, the emission limit power value in the earlier scheduling emission limit record is taken as the first emission limit, and the emission limit power value in the later scheduling emission limit record is taken as the second emission limit.

[0061] The first complete prediction sampling period before the end of the first issuance limit is taken as the pre-adjustment sampling period, and the first complete prediction sampling period after the start of the second issuance limit is taken as the post-adjustment sampling period.

[0062] The complete prediction sampling period is used to avoid communication delays and power follower transitions caused by the switching edge of scheduling instructions.

[0063] If there is a new record of a change in the upper limit of the scheduling limit within the sampling period before the adjustment, discard the current process of raising the upper limit of the scheduling limit; if there is a new record of a change in the upper limit of the scheduling limit within the sampling period after the adjustment, discard the current process of raising the upper limit of the scheduling limit.

[0064] Before adjustment, the actual active power sampling value and the inverter DC bus voltage sampling value are read during the sampling period. The actual active power sampling values ​​during the sampling period before adjustment are sorted from smallest to largest, and the median value is taken as the actual active power before adjustment. The DC bus voltage sampling values ​​during the sampling period before adjustment are sorted from smallest to largest, and the median value is taken as the DC bus voltage before adjustment. After adjustment, the actual active power sampling value and the inverter DC bus voltage sampling value are read during the sampling period after adjustment. The actual active power sampling values ​​during the sampling period after adjustment are sorted from smallest to largest, and the median value is taken as the actual active power after adjustment. The DC bus voltage sampling values ​​during the sampling period after adjustment are sorted from smallest to largest, and the median value is taken as the DC bus voltage after adjustment.

[0065] The difference between the adjusted actual active power and the original actual active power is taken as the naturally increased active power release.

[0066] When the adjusted actual active power is greater than the pre-adjustment actual active power, the naturally increased active power release is positive, indicating that the AC active power has been released after the second limit on power generation takes effect; when the adjusted actual active power is not greater than the pre-adjustment actual active power, the naturally increased active power release does not enter the effective natural increase process judgment.

[0067] The difference between the DC bus voltage before and after adjustment is used as the amount of DC voltage reduction to be naturally increased.

[0068] When the DC bus voltage before adjustment is greater than the DC bus voltage after adjustment, the natural DC voltage reduction amount is positive, indicating that the DC bus voltage has been reduced after the upper limit of the power generation is increased; when the DC bus voltage before adjustment is not greater than the DC bus voltage after adjustment, the natural DC voltage reduction amount is recorded as zero, indicating that no voltage reduction has occurred on the DC side.

[0069] When the naturally increased active power release reaches the effective release limit of the natural increase, the current scheduling limit increase process is recorded as an effective natural increase process. When there are multiple effective natural increase processes within the same limit delay period, they are sorted from shortest to longest according to the time interval between the end of the adjusted sampling period of the effective natural increase process and the start of the effective period of the limit release announcement. The effective natural increase process with the shortest time interval is selected and marked as valid. When there is no effective natural increase process within the same limit delay period, the invalid natural increase process is marked as valid.

[0070] It should be noted that the effective release limit for natural upward adjustment is written into the grid connection point operation parameter table during the commissioning calibration phase of the new energy grid connection point. During the commissioning calibration phase, historical operating periods with upward adjustment of the dispatch limit, no energy storage charging and discharging switching, no inverter limit, and no voltage alarm are selected. The actual active power generated before and after the upward adjustment of the dispatch limit is read, and the actual active power generated is sorted in ascending order of value. The release percentile value is then taken and limited to the rated power release limit of the new energy grid connection point. The release percentile value after the limit is taken as the effective release limit for natural upward adjustment. The value range of the release percentile ratio is 0.50 to 0.80, and the rated power release limit range of the new energy grid connection point is the power range corresponding to 2% to 10% of the rated power of the new energy grid connection point.

[0071] S3. Determine the residual pressure retention value of the power generation restriction period based on the time value of the power generation restriction period, the active power release, and the DC voltage reduction. When the residual pressure retention value of the power generation restriction reaches the residual pressure retention limit, determine the predicted increase in power generation based on the available capacity of the equipment and the adjusted actual active power.

[0072] When the effective flag for natural upward adjustment is established, the residual pressure retention value of the limited issuance is calculated using the following expression:

[0073] ;

[0074] in, This indicates the residual pressure reserve value for limited issuance. This indicates the time-limited delay value. This indicates a natural increase in DC voltage reduction. Indicates the effective limit of DC devoltage. This indicates a natural increase in the amount of active power released. This indicates that the effective release limit is naturally increased.

[0075] It should be noted that the effective limit of DC voltage reduction is written into the grid connection point operation parameter table during the commissioning calibration phase of the new energy grid connection point. During the commissioning calibration phase, a normal power release period without dispatching restrictions, energy storage charging and discharging switching, and inverter limits is selected. The DC bus voltage drop of the inverter during the normal power release period is read. The DC bus voltage drop of the inverter is sorted by value and the voltage reduction percentile value is taken. The voltage reduction percentile value is limited to the proportion of the upper limit of the inverter's allowable DC bus voltage. The voltage reduction percentile value after limitation is used as the effective limit of DC voltage reduction. The value range of the voltage reduction percentile ratio is 0.50 to 0.80, and the value range of the effective limit of DC voltage reduction is 0.5% to 3% of the upper limit of the inverter's allowable DC bus voltage.

[0076] When the power generation limit residual pressure reserve value reaches the residual pressure reserve limit, a rapid increase in power generation judgment is generated. Based on the power generation limit residual pressure reserve value, the available capacity of the equipment, and the adjusted actual active power, the predicted increase in power generation is calculated. The expression is as follows:

[0077] ;

[0078] in, This indicates the predicted increase in share issuance. This indicates the residual pressure reserve value for limited issuance. Indicates the available capacity of the equipment. This indicates the actual active power generated after adjustment.

[0079] When the residual pressure retention value of the limited issuance does not reach the residual pressure retention limit, a normal recovery judgment is generated, and the predicted issuance amount is recorded as zero.

[0080] Among them, the normal recovery judgment means that after the scheduling limit is naturally increased, the DC bus voltage has undergone sufficient voltage reduction, or the active power released by the natural increase is insufficient, and the predicted power is not increased according to the rapid increase scenario after the limit is completely lifted.

[0081] It should be noted that the residual pressure retention limit is written into the grid connection point operation parameter table during the commissioning calibration phase of the new energy grid connection point. During the commissioning calibration phase, samples whose actual active power generation after the lifting of historical power restrictions is higher than that of the previous sampling period in the continuous prediction sampling period are selected. The historical power restriction residual pressure retention value of each sample is calculated according to the formula for power restriction residual pressure retention value. The historical power restriction residual pressure retention values ​​are sorted by value and the residual pressure quantile value is taken. The residual pressure quantile value is limited to the residual pressure retention value range. The residual pressure quantile value after the limit is used as the residual pressure retention limit. The residual pressure quantile ratio ranges from 0.20 to 0.40, and the residual pressure retention limit ranges from 0.35 to 0.70.

[0082] Furthermore, when the natural upward adjustment invalidation flag is established, the time-limited issuance lag value reaches the effective lag limit, and at least one of the issuance restriction lifting announcement and resource enhancement announcement is within the effective period, a conservative issuance judgment is generated, and the lag value of the securities is calculated, expressed as:

[0083] ;

[0084] in, Indicates the value of delayed documents. This indicates the time-limited delay value. This indicates the effective limit of hysteresis.

[0085] It should be noted that the effective limit of lag voltage is written into the grid connection point operation parameter table during the commissioning calibration phase of the new energy grid connection point. During the commissioning calibration phase, samples are selected where the actual active power generated after the historical power restriction is lifted is higher than that of the previous sampling period in the continuous prediction sampling period. The power restriction lag voltage time value corresponding to each sample is read. The power restriction lag voltage time values ​​are sorted by value and the lag voltage quantile value is taken. The lag voltage quantile value is limited to the effective value range of lag voltage. The lag voltage quantile value after the limit is taken as the effective limit of lag voltage. The value range of the lag voltage quantile ratio is 0.20 to 0.40, and the value range of the effective limit of lag voltage is 0.50 to 0.80.

[0086] When the conservative assessment of increased share issuance is valid, the predicted issuance amount is calculated based on the depreciated value of stagnant documents, available equipment capacity, and current actual active power output. The expression is as follows:

[0087] ;

[0088] in, This indicates the predicted increase in share issuance. Indicates the value of delayed documents. Indicates the available capacity of the equipment. This indicates the current actual active power output.

[0089] When the invalidation flag for natural upward adjustment is established, and the time value of the issuance restriction lag does not reach the effective lag limit, a normal prediction judgment is generated, and the predicted issuance increase is recorded as zero; when the invalidation flag for natural upward adjustment is established, and the time value of the issuance restriction lag reaches the effective lag limit, but the issuance restriction lifting announcement and the resource enhancement announcement are not within the effective time period, a normal prediction judgment is generated, and the predicted issuance increase is recorded as zero.

[0090] Furthermore, after obtaining the predicted increase in power generation, the predicted resource availability is calculated based on the available equipment capacity, the current actual active power generation, and the predicted increase in power generation. The expression is as follows:

[0091] ;

[0092] in, This indicates the predicted value of available resources. Indicates the available capacity of the equipment. This indicates the current actual active power. This indicates the predicted increase in issuance.

[0093] S4. Generate the predicted power generation capacity that the grid can accept based on the predicted increase in power generation, grid connection point voltage, feeder current and energy storage charging power, and generate energy storage charging input, reactive power absorption input and active power limiting input according to energy storage charging capacity, voltage proximity state and feeder current proximity state.

[0094] When the voltage active power sensitivity is greater than zero, the voltage allowable power boundary is calculated using the following expression:

[0095] ;

[0096] in, Indicates the allowable power limit of the voltage. This indicates the current actual active power. Indicates the upper limit of voltage. Indicates the voltage at the grid connection point. This represents the voltage-sensitive active power quantity.

[0097] When the current active power sensitivity is greater than zero, the allowable power boundary of the current is calculated as follows:

[0098] ;

[0099] in, Indicates the allowable power boundary of the current. This indicates the current actual active power. Indicates the allowable current of the feeder. Indicates the feeder current. This represents the active power sensitive quantity of current.

[0100] When the difference between the upper voltage limit and the grid connection point voltage is not greater than the voltage approach limit, the grid connection point voltage is determined to be in the state of approaching the upper voltage limit; when the difference between the feeder allowable current and the feeder current is not greater than the current approach limit, the feeder current is determined to be in the state of approaching the feeder allowable current.

[0101] It should be noted that the voltage proximity limit is written into the grid connection point operation parameter table during the commissioning calibration phase of the new energy grid connection point. During the commissioning calibration phase, the voltage margin records before the historical voltage alarm are selected. The voltage margin records are sorted from smallest to largest value, and the lowest percentile value is taken. The lowest percentile value is limited to the range of the grid connection point's rated voltage. The lower percentile value after limiting is used as the voltage proximity limit. The value of the lowest percentile ratio ranges from 0.10 to 0.30. The range of the grid connection point's rated voltage proximity is the voltage corresponding to 0.5% to 2% of the grid connection point's rated voltage. The current approach limit is written into the feeder operation parameter table during the feeder commissioning calibration phase. During the commissioning calibration phase, the current margin records before the historical feeder overload alarm are selected. The current margin records are sorted by value from smallest to largest, and the feeder low-side percentile value is taken. The feeder low-side percentile value is limited to the feeder allowable current approach range. The feeder low-side percentile value after the limit is used as the current approach limit. The value range of the feeder low-side percentile ratio is 0.10 to 0.30. The feeder allowable current approach range is the current range corresponding to 3% to 10% of the feeder allowable current.

[0102] Furthermore, the energy storage charging input is calculated based on the predicted increase in power generation and the available charging power of the energy storage, expressed as follows:

[0103] ;

[0104] in, Indicates energy storage charging input. This indicates the predicted increase in share issuance. This indicates the rechargeable power of the energy storage.

[0105] The grid-connected power after energy storage absorption is calculated based on the predicted resource availability and the energy storage charging input, as expressed by the following expression:

[0106] ;

[0107] in, This indicates the grid-connected power after energy storage absorption. This indicates the predicted value of available resources. This indicates the energy storage charging input.

[0108] Furthermore, the predicted value of the power generation capacity that the power grid can accommodate is calculated, expressed as follows:

[0109] ;

[0110] in, This indicates the predicted power generation capacity that the power grid can accommodate. This indicates the grid-connected power after energy storage absorption. Indicates the allowable power limit of the voltage. Indicates the allowable power boundary of the current. This indicates the power boundary that is allowed during scheduling.

[0111] It should be noted that the dispatchable power boundary is obtained based on the post-removal permitted power field, the latest emission limit field, and the available equipment capacity. The latest emission limit field is the emission limit field with the latest effective time within the effective period for the same new energy grid connection point. When the post-removal permitted power field exists, the power value corresponding to the post-removal permitted power field is read, and the smaller value between the power value corresponding to the post-removal permitted power field and the available equipment capacity is used as the dispatchable power boundary. When the post-removal permitted power field does not exist but the latest emission limit field exists, the power value corresponding to the latest emission limit field is read, and the smaller value between the power value corresponding to the latest emission limit field and the available equipment capacity is used as the dispatchable power boundary. When neither the post-removal permitted power field nor the latest emission limit field exists, the current effective dispatch emission limit is used, and the smaller value between the current effective dispatch emission limit and the available equipment capacity is used as the dispatchable power boundary. At the same time, a field missing review record is generated.

[0112] Furthermore, the active power limit input is calculated, expressed as:

[0113] ;

[0114] in, Indicates active power limit input. This indicates the grid-connected power after energy storage absorption. This indicates the predicted power generation capacity that the power grid can accommodate.

[0115] When the rapid power generation judgment is valid, the available energy storage charging power is less than the predicted power generation, and the grid connection point voltage enters a state close to the upper voltage limit, the available reactive power absorption capacity of the inverter will be used as the reactive power absorption input.

[0116] When the rapid increase judgment is not established, or the grid connection point voltage does not enter the state close to the upper voltage limit, the reactive power absorption input is recorded as zero.

[0117] When the rapid increase in power generation is determined and the rechargeable power of the energy storage is not less than the predicted increase in power generation, the predicted increase in power generation is converted into energy storage charging input, the output power is the predicted value of the power generation that the grid can accept, and the reactive power absorption input and active power limitation input are recorded as zero.

[0118] When the rapid increase in power generation is confirmed, the available charging power of the energy storage is less than the predicted increase in power generation, and the voltage at the grid connection point approaches the upper voltage limit, the available charging power of the energy storage is converted into the energy storage charging input, the available reactive power absorption capacity of the inverter is converted into the reactive power absorption input, and the active power that still exceeds the predicted value of the power generation that the grid can accept after energy storage absorption is converted into the active power limiting input.

[0119] When the rapid increase in power generation is determined, the rechargeable power of energy storage is less than the predicted increase, the grid connection voltage has not entered the state close to the upper voltage limit, and the feeder current enters the state close to the allowable feeder current, the rechargeable power of energy storage is converted into energy storage charging input, the reactive power absorption input is recorded as zero, and the active power that still exceeds the predicted value of the power generation that the grid can accept after energy storage absorption is converted into active power limiting input.

[0120] When the rapid increase in power generation is determined, the available power of energy storage is less than the predicted increase, the grid connection voltage has not entered the state close to the upper voltage limit, and the feeder current has not entered the state close to the feeder allowable current, the available power of energy storage is converted into energy storage charging input, and the output is based on the predicted value of the power generation that the grid can accept. The active power limit input is not immediately issued. The remaining part of the predicted increase in power generation exceeding the available power of energy storage is written into the active power limit reserve input and recalculated in the next prediction sampling period.

[0121] When the conservative power generation increase judgment is valid, the output is based on the predicted power generation capacity that the grid can accept. When the energy storage charging power is not less than the predicted power generation increase, the predicted power generation increase is converted into energy storage charging input. When the energy storage charging power is less than the predicted power generation increase and the feeder current enters a state close to the feeder allowable current, the active power portion that still exceeds the predicted power generation capacity that the grid can accept after energy storage absorption is converted into active power limiting input. When the grid connection point voltage enters a state close to the upper voltage limit, only reactive power absorption reserve input is recorded, and reactive power absorption input is not given priority.

[0122] When the normal recovery judgment or the normal prediction judgment is valid, the predicted increase in power generation, energy storage charging input, reactive power absorption input, and active power limitation input are all recorded as zero. The predicted power generation value that the grid can accept is output according to the minimum value among the current actual active power, voltage allowable power boundary, current allowable power boundary, and dispatch allowable power boundary.

[0123] In summary, this invention identifies hidden rapid power generation risks before the lifting of power generation restrictions by recognizing periods of lag in power generation and insufficient DC voltage reduction after the natural increase of the power generation limit. By extracting only event fields from scheduling and meteorological texts using a large model and combining them with electrical quantities to calculate and predict the power generation increase, the power prediction process is verifiable and avoids black-box output. By converting the predicted power generation increase into inputs for energy storage charging, reactive power absorption, and active power limiting, the prediction results directly serve distributed control, reducing the risks of voltage rise, feeder overload, and renewable energy curtailment.

[0124] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for predicting the power generation of new energy power grids based on large-scale model analysis, characterized in that, include: The system collects the actual active power generated by new energy grid-connected points, the upper limit of dispatching power restriction, the DC bus voltage of the inverter, the voltage at the grid connection point, the feeder current, the rechargeable power of energy storage, and the available capacity of equipment. It also uses a large power sector model to identify power restriction lifting forecasts and resource enhancement forecasts from dispatching power restriction texts, power restriction lifting notices, and meteorological enhancement early warning texts. When the difference between the actual active power generated and the upper limit of dispatched power generation is not greater than the limit of power generation following error, and the inverter DC bus voltage is in the DC voltage high regulation zone, the power generation lag period is determined, and the active power release and DC voltage reduction are extracted before and after the upper limit of dispatched power generation is naturally adjusted from the first upper limit to the second upper limit during the power generation lag period. The residual voltage retention value of the power generation restriction period is determined based on the time value of the power generation restriction period, the active power release, and the DC voltage reduction. When the residual voltage retention value of the power generation restriction reaches the residual voltage retention limit, the predicted increase in power generation is determined according to the available capacity of the equipment and the adjusted actual active power. Based on the predicted increase in power generation, grid connection point voltage, feeder current, and energy storage rechargeable power, a predicted value of the power generation that the grid can accept is generated. Then, based on the energy storage rechargeable capacity, voltage near-state, and feeder current near-state, energy storage charging input, reactive power absorption input, and active power limiting input are generated.

2. The method for predicting the power generation of new energy power grids based on large model analysis as described in claim 1, characterized in that, The data collected include the actual active power generated at the new energy grid connection point, the upper limit of dispatched power generation, the inverter DC bus voltage, the grid connection point voltage, the feeder current, the rechargeable power of energy storage, and the available capacity of the equipment. The actual active power, dispatch limit, inverter DC bus voltage, grid connection point voltage, feeder current, energy storage chargeable power, and equipment available capacity are obtained separately according to the same prediction sampling period. Data belonging to the same prediction sampling period are associated with the new energy grid connection point number and sampling time to form grid connection point prediction sampling data.

3. The method for predicting the power generation of new energy power grids based on large model analysis as described in claim 1, characterized in that, The specific steps for identifying the release restriction lifting forecast and resource enhancement forecast from dispatch restriction texts, release restriction lifting notices, and meteorological enhancement warning texts using a large power sector model are as follows: After compiling the dispatch restriction texts, restriction lifting notices, and enhanced meteorological warning texts into unified prompt data, they are input into the large power sector model. The large-scale power model outputs forecasts of the lifting of power restrictions, resource enhancement, grid connection points of the new energy sources involved, effective time periods, allowed power fields after lifting restrictions, and the latest upper limit of power restrictions. When the output of the large power model lacks the number of new energy grid connection points or the effective time period, or when the allowed power field and the latest limit on power generation are both missing after the restriction is lifted, the corresponding output content is deleted so that the corresponding output content does not participate in the subsequent power generation prediction. The unified notification data includes the new energy grid connection point number, the current dispatch limit for power generation, the start time of the power generation limit, the planned time for lifting the power generation limit, the type of weather enhancement, the start time of the weather enhancement, and the end time of the weather enhancement.

4. The method for predicting the power generation of new energy power grids based on large model analysis as described in claim 2, characterized in that, The determination of the time period for limiting power generation includes: When the difference between the actual active power generated and the upper limit of the dispatch limit is not greater than the limit of the limit following error, and the inverter DC bus voltage is not lower than the lower limit of the DC voltage high regulation zone, the prediction sampling period is determined as the limit hysteresis sampling period. When two or more consecutive limit-up lag sampling periods are located within the same scheduling limit-up period, the time period consisting of the two or more consecutive limit-up lag sampling periods is determined as the limit-up lag period. The ratio of the cumulative duration during which the inverter's DC bus voltage is in the high DC voltage regulation zone to the total duration of the same period is determined as the lag time value during the rationing period.

5. The method for predicting the power generation of new energy power grids based on large model analysis as described in claim 3 or 4, characterized in that, The extraction of active power release and DC voltage reduction before and after the power generation limit is naturally adjusted from the first power generation limit to the second power generation limit during the power generation restriction lag period includes: During the period of stagnation in the restricted issuance, read the record of changes in the upper limit of the restricted issuance, and use the effective period of the issuance restriction lifting announcement as a time constraint to filter the process of adjusting the upper limit of the restricted issuance that occurred before the effective period of the issuance restriction lifting announcement. The scheduling power limit records for the same new energy grid connection point are sorted by time according to the effective time of the power limit. When the power limit power value in the later scheduling power limit record is greater than the power limit power value in the previous scheduling power limit record, the power limit power value in the previous scheduling power limit record is taken as the first power limit and the power limit power value in the later scheduling power limit record is taken as the second power limit. The first complete prediction sampling period before the end of the first issuance limit is taken as the pre-adjustment sampling period, and the first complete prediction sampling period after the start of the second issuance limit is taken as the post-adjustment sampling period. The natural upward adjustment of active power release is determined based on the actual active power generated during the sampling period before and after the adjustment, and the natural upward adjustment of DC voltage reduction is determined based on the inverter DC bus voltage during the sampling period before and after the adjustment.

6. The method for predicting the power generation of new energy power grids based on large model analysis as described in claim 5, characterized in that, The determination of the residual pressure retention value for limited power generation includes: When the naturally increased active power release reaches the effective release limit of natural increase, the current scheduling limit increase process is determined as an effective natural increase process. When multiple effective natural upward adjustment processes exist within the same period of restricted power generation, they are sorted from shortest to longest according to the time interval between the end of the adjusted sampling period of the effective natural upward adjustment process and the start of the effective period of the power generation restriction lifting announcement, and the effective natural upward adjustment process with the shortest time interval is selected. The residual voltage retention value of the power generation restriction is determined based on the time value of the power generation restriction lag, the amount of DC voltage reduction by natural adjustment, the effective limit of DC voltage reduction, the amount of active power released by natural adjustment, and the effective release limit of natural adjustment.

7. The method for predicting the power generation of new energy power grids based on large model analysis as described in claim 6, characterized in that, The determination of the predicted increase in issuance includes: When there is an effective natural upward adjustment process, if the residual pressure retention value of the limited power generation reaches the residual pressure retention limit, a rapid increase in power generation judgment is generated, and the predicted increase in power generation is determined based on the residual pressure retention value of the limited power generation, the available capacity of the equipment, and the actual active power generation after adjustment. When there is an effective natural upward adjustment process and the remaining pressure value of the limited issuance does not reach the remaining pressure limit, a normal recovery judgment is generated and the predicted issuance amount is recorded as zero. When there is no effective natural upward adjustment process and the time value of the restricted issuance lag reaches the effective lag limit, first determine whether the announcement of the lifting of the restricted issuance is within the effective period. If the announcement of lifting the power restriction is within the effective period, a conservative increase in power issuance judgment is generated, and the predicted increase in power issuance is determined based on the time value of the power restriction lag, the effective limit of the lag, the available capacity of the equipment, and the current actual active power. If the announcement of lifting the power restriction is not within the effective period, then it is determined whether the announcement of resource enhancement is within the effective period. If the announcement of resource enhancement is within the effective period, a conservative power increase judgment is generated, and the predicted power increase is determined based on the power restriction lag time value, lag effective limit, available equipment capacity and current actual active power. If the announcement of lifting the issuance restriction is not within the valid period and the announcement of resource enhancement is not within the valid period, a normal prediction judgment is generated, and the predicted issuance amount is recorded as zero.

8. The method for predicting the power generation of new energy power grids based on large model analysis as described in claim 7, characterized in that, The generated power generation capacity forecast for the power grid includes: During historical normal grid connection periods, sample pairs that have not exceeded the limit at the grid connection point voltage and whose active power change has reached the lower limit of active power change screening are selected. The median value of the ratio of the grid connection point voltage change to the active power change of each sample pair is used as the voltage active power sensitive quantity. During the normal grid connection period in history, the sampling pairs that have not exceeded the limit of feeder current and whose active power change of new energy grid connection point in feeder reaches the lower limit of feeder active power change screening are selected. The median value of the ratio of feeder current change to active power change of each sampling pair is used as the current active power sensitive quantity. The voltage active power sensitivity is used to calculate the voltage-side power increase corresponding to the voltage limit at the grid connection point, and the current active power sensitivity is used to calculate the current-side power increase corresponding to the feeder current limit, forming the voltage-side allowable power boundary and the current-side allowable power boundary used to constrain the predicted power increase.

9. The method for predicting the power generation of new energy power grids based on large model analysis as described in claim 8, characterized in that, The generated power grid capacity forecast also includes: When the predicted increase in power generation is not greater than the rechargeable power of the energy storage, the predicted increase in power generation will be used as the energy storage charging input. When the predicted increase in power generation exceeds the available charging power of energy storage, the available charging power of energy storage will be used as the energy storage charging input. Subtract the energy storage charging input from the predicted resource availability value to obtain the grid-connected power after energy storage absorption. The grid-connected power absorbed by the energy storage is successively limited by the voltage allowable power boundary, the current allowable power boundary, and the dispatch allowable power boundary to obtain the predicted value of the power generation that the grid can accept. When the grid-connected power absorbed by energy storage is greater than the predicted value of the power generation that the grid can accept, the difference between the two will be used as the active power limit input. When the grid-connected power absorbed by the energy storage is not greater than the predicted value of the power generation that the grid can accept, the active power limit input will be recorded as zero.

10. The method for predicting the power generation of new energy power grids based on large model analysis as described in claim 7 or 9, characterized in that, The generation of energy storage charging input, reactive power absorption input, and active power limiting input include: When the rapid increase in power generation is determined and the energy storage charging power is not less than the predicted increase in power generation, the predicted increase in power generation is converted into energy storage charging input, and the reactive power absorption input and active power limiting input are recorded as zero. When the rapid increase in power generation is confirmed, the available power of energy storage is less than the predicted increase in power generation, and the voltage at the grid connection point is close to the upper limit of voltage, the available power of energy storage is converted into energy storage charging input, the available reactive power absorption capacity of the inverter is converted into reactive power absorption input, and the active power that still exceeds the predicted value of the power generation that the grid can accept after energy storage absorption is converted into active power limiting input. When the rapid increase in power generation is determined, the rechargeable power of energy storage is less than the predicted increase in power generation, the grid connection point voltage has not entered the state close to the upper voltage limit and the feeder current has entered the state close to the feeder allowable current, the rechargeable power of energy storage is converted into energy storage charging input, the reactive power absorption input is recorded as zero, and the active power that still exceeds the predicted value of the power generation that the grid can accept after energy storage absorption is converted into active power limiting input. When the normal recovery judgment or normal prediction judgment is valid, the predicted increase in power generation, energy storage charging input, reactive power absorption input, and active power limitation input are all recorded as zero, and the predicted power generation value that the grid can accept is output according to the minimum value among the current actual active power, voltage allowable power boundary, current allowable power boundary, and dispatch allowable power boundary.

Images