System standby demand assessment method considering supply and demand random fluctuation and power grid security constraint

Abstract

The invention discloses a system reserve demand assessment method considering supply and demand random fluctuation and power grid security constraints, and relates to the technical field of market reserve demand prediction, the method comprises the steps of obtaining scene construction data and assessment data, generating a prediction scene and a fluctuation scene through a prediction model obtained by training the scene construction data, based on the first optimization model and the corresponding constraint condition, performing optimization solution on the first optimization model by using the scene data of the prediction scene to obtain a first decision variable; based on a second optimization model and a constraint condition of the second optimization model, performing network security constraint verification on the first decision variable by using scene data of the fluctuation scene, and obtaining a second decision variable of the second optimization model; if the first decision variable and the second decision variable meet a preset verification passing condition, determining the standby demand to be verified as a target system standby demand; the problem that in actual scheduling, standby application can not be called is solved, and the economical efficiency and reliability of the system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of market reserve demand prediction, and more particularly, to a system reserve demand evaluation method considering random fluctuations in supply and demand and grid security constraints. Background Art

[0002] As new power systems transition to a high proportion of renewable energy, the random fluctuations in both the source and load sides pose unprecedented challenges to the assessment of system reserve demand. Traditional methods have significant limitations in dealing with such complex scenarios, specifically:

[0003] 1. Fluctuations on both the source and load sides increase uncertainty. Fluctuations on the power supply side: The output of renewable energy sources such as wind power and photovoltaics is highly dependent on meteorological conditions, and has strong intermittent and anti-peak characteristics. Fluctuations on the load side: The large-scale access of new loads such as electric vehicle charging and demand response, coupled with the frequent occurrence of extreme weather events, has significantly increased load forecasting errors.

[0004] 2. Grid security constraints limit the availability of backup resources; reserve deployment is hindered: Traditional backup demand assessment methods may formulate system backup capacity based on a fixed backup capacity ratio or historical data. Although this meets the total system demand, it is limited by grid security constraints such as line transmission capacity and section stability limits. In actual scheduling, the dilemma of "having backup resources but being unable to deploy them" may occur.

[0005] In the current existing technology, the backup demand assessment basically adopts the prediction method, and does not adopt the optimization method to finely consider the random fluctuations on both the source and load sides and the grid security constraints that limit the backup availability, affecting the system economy and reliability. Patent document CN110580538A discloses a method for predicting power system backup demand. The method first determines the power system's preliminary backup demand based on thermal power, photovoltaic, and wind power data and user load data. It then determines comprehensive user satisfaction based on user electricity satisfaction and payment satisfaction of different user types. Finally, the predicted power system backup demand is determined based on the preliminary backup demand and comprehensive user satisfaction. This patent fails to consider the uncertainty caused by random fluctuations in both the source and load sides, nor the potential for backup generation failures. Patent document CN117559431A discloses a method and device for assessing power grid ramping backup demand based on scheduling plan timing differences. This method uses minute-by-minute ramping increments to perform minute-by-minute average ramping uncertainty analysis, determining a ramping regulation capacity demand curve under normal operating conditions. This is combined with the emergency backup demand to determine a ramping regulation capacity demand curve under emergency conditions. Finally, the ramping regulation capacity demands under normal and emergency conditions are compared, with the larger value taken as the system ramping regulation capacity demand curve. While this patent focuses on ramping demand, it still fails to consider the uncertainty caused by random fluctuations in both the source and load sides, nor the potential for backup generation failures. Summary of the Invention

[0006] The purpose of the present invention is to provide a system backup demand assessment method that takes into account random fluctuations in supply and demand and grid security constraints. An optimization method is used to refine the fluctuation scenarios and grid security constraints that need to be considered for random fluctuations on both the source and load sides, so as to solve the problem of "there is backup but it cannot be called" that may occur in actual scheduling, and improve the system economy and reliability.

[0007] The above technical objectives of the present invention are achieved through the following technical solutions:

[0008] This application provides a system reserve demand assessment method considering random fluctuations in supply and demand and grid security constraints, including the following specific steps:

[0009] Obtain scenario construction data and backup demand optimization evaluation data, and preprocess the scenario construction data and evaluation data. The preprocessing includes data cleaning, standardization, and screening.

[0010] The pre-processed scenario construction data is used to train the prediction model, and the prediction model is used to generate prediction scenarios and fluctuation scenarios. The fluctuation scenarios include typical scenarios and extreme scenarios.

[0011] A first optimization model is established using the evaluation data, and based on the first optimization model and constraints of the first optimization model, the first optimization model is optimized and solved using scenario data of the prediction scenario to obtain a first decision variable of the first optimization model, where the decision variable includes the output of the first unit and the standby demand to be verified;

[0012] A second optimization model is established using the evaluation data, and based on the second optimization model and the constraints of the second optimization model, the first decision variable is verified for cybersecurity constraints using scenario data of the fluctuation scenario, and a second decision variable of the second optimization model is obtained, where the second decision variable includes the output of the second unit;

[0013] If the output of the first unit and the output of the second unit meet the preset verification pass condition, the standby demand to be verified is determined as the target system standby demand, and the target system standby demand includes a positive standby demand and a negative standby demand.

[0014] On the basis of the above technical solution, the present invention can also be improved as follows.

[0015] Furthermore, the above method further includes:

[0016] Calculate the evaluation index of the prediction scenario and / or fluctuation scenario, and adjust the model parameters of the prediction model based on the evaluation index until the evaluation index meets the preset conditions.

[0017] Furthermore, the above evaluation indicators include coverage, average bandwidth, daily load rate and daily fluctuation rate.

[0018] Furthermore, the above coverage is specifically:

[0019]

[0020] The average bandwidth is:

[0021]

[0022] Where, CR α represents coverage, N is the number of evaluation data, I(·) is the indicative function, and is 1 when the condition is met; x i Indicates the actual value, Indicates the corresponding x i The confidence interval of the prediction scenario 1-α, Represent the upper and lower limits of the confidence interval, which are the quantiles of 2 / α or 1-2 / α of the predicted scenario; ΔP i α Indicates the width of the confidence interval;

[0023] The specific daily load rate is: A = B / C

[0024] The specific daily volatility is: D = E / F

[0025] Where A is the daily load rate, B is the daily maximum load, C is the daily average load, D is the daily fluctuation rate, E is the daily load standard deviation, and F is the daily load mean.

[0026] Furthermore, the above optimization and solution of the first optimization model is specifically as follows:

[0027]

[0028] Where N is the total number of units; T is the total number of time periods considered. If 96 time periods are considered in one day, T is 96; P i,t represents the output of unit i in period t under the forecast scenario; C i,t (P i,t ), are the operating cost and startup cost of unit i in period t under the forecast scenario, where the unit operating cost C i,t (P i,t ) is the power generation cost or a multi-segment linear function related to the output ranges declared by the unit and the corresponding energy prices; PR i,t NR i,t are the positive reserve demand and negative reserve demand of unit i in period t under the forecast scenario; are the prices of positive and negative reserves provided by unit i in time period t under the forecast scenario; M s is the penalty factor for relaxing the network power flow constraints used for optimization; are the forward and reverse tidal flow relaxation variables of section s in time period t under the prediction scenario.

[0029] Furthermore, the verification of the network security constraint on the first decision variable includes: optimizing and solving the second optimization model, specifically:

[0030]

[0031] Where NS is the total number of sections, are the positive relaxation and negative relaxation of the tidal current at section S in period t under the fluctuation scenario k, respectively, and M is the penalty coefficient.

[0032] In a second aspect, the present application provides a system backup demand assessment system that considers random fluctuations in supply and demand and grid security constraints, which is applied to the system backup demand assessment method that considers random fluctuations in supply and demand and grid security constraints in any one of the first aspects, including:

[0033] The data processing module is used to obtain scenario construction data and evaluation data for backup demand optimization, and preprocess the scenario construction data and evaluation data. The preprocessing includes data cleaning, standardization and screening.

[0034] The model training module is used to train the prediction model using pre-processed scenario construction data, and generate prediction scenarios and fluctuation scenarios through the prediction model. The fluctuation scenarios include typical scenarios and extreme scenarios.

[0035] a model solving module, configured to establish a first optimization model using the evaluation data, and optimize and solve the first optimization model based on the first optimization model and the constraints of the first optimization model using the scenario data of the prediction scenario to obtain a first decision variable of the first optimization model, wherein the decision variable includes the output of the first unit and the standby demand to be verified;

[0036] a demand verification module, configured to establish a second optimization model using the evaluation data, and verify the network security constraints of the first decision variable using scenario data of the fluctuation scenario based on the second optimization model and the constraints of the second optimization model, and obtain a second decision variable of the second optimization model, the second decision variable including the output of the second unit;

[0037] The demand determination module is used to determine the standby demand to be verified as the target system standby demand if the output of the first unit and the output of the second unit meet the preset verification pass conditions. The target system standby demand includes positive standby demand and negative standby demand.

[0038] Furthermore, the model training module also includes:

[0039] Calculate the evaluation index of the prediction scenario and / or fluctuation scenario, and adjust the model parameters of the prediction model based on the evaluation index until the evaluation index meets the preset conditions.

[0040] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements any one of the methods in the first aspect when executing the computer program.

[0041] In a fourth aspect, the present application provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions enable a computer to execute any one of the methods in the first aspect.

[0042] Compared with the prior art, the present invention has at least the following beneficial effects:

[0043] In this application, a system backup demand assessment optimization model considering random fluctuations on both sides of supply and demand and grid security constraints adopts a two-stage model, namely a first optimization model and a second optimization model. Stage 1 adopts the first optimization model: a prediction scenario is used to determine the positive and negative backup demands of the system and / or partition, without considering the unit backup call process; Stage 2 adopts the second optimization model: the fluctuation scenario verifies whether the positive and negative backup demands of the system and / or partition determined by the prediction scenario can pass the safety check, considering the unit backup call; an optimization method is used to refine the positive and negative backup demands of the system and / or partition, taking into account the fluctuation scenarios and grid security constraints to be considered for random fluctuations on both sides of source and load, which can solve the problem of "having backup but unable to call" that may occur in actual scheduling, minimize the system operating cost, improve the absorption of new energy, and thereby improve the system reliability and economy. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0045] Figure 1 A flowchart of an evaluation method according to an embodiment of the present invention;

[0046] Figure 2 Optimization flow chart of the first optimization model and the second optimization model in an embodiment of the present invention;

[0047] Figure 3 A connection diagram of an evaluation system according to an embodiment of the present invention;

[0048] Figure 4 Schematic diagram of the connection of electronic equipment in an embodiment of the present invention. DETAILED DESCRIPTION

[0049] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0050] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but rather merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort shall fall within the scope of protection of the present invention.

[0051] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings.

[0052] In the description of the embodiments of the present invention, "a plurality of" means at least two.

[0053] Example 1: To address the issues of increased uncertainty due to fluctuations in both the source and load sides, and limited backup availability due to grid security constraints, this example provides a system backup demand assessment method that considers random fluctuations in supply and demand and grid security constraints, including the following specific steps:

[0054] S1. Obtain scenario construction data and evaluation data for backup demand optimization, and preprocess the scenario construction data and evaluation data. The preprocessing includes data cleaning, standardization, and screening.

[0055] The data required for forecasting scenarios and fluctuation scenarios include:

[0056] New energy data: daily wind power and photovoltaic output data for n years in history (data collected every hour or every 15 minutes).

[0057] Historical load data: daily load curves for n years, including load categories such as industrial, commercial, and residential; records of sudden load changes during extreme weather such as cold waves and heat waves; and temporary load surges caused by large-scale events or emergencies.

[0058] New load data: Daily electric vehicle charging behavior patterns (charging time, power demand) over n years of history.

[0059] Meteorological data: radiation, wind speed, temperature, precipitation.

[0060] Furthermore, the data required by the backup demand optimization model (the first optimization model and the second optimization model) include:

[0061] Conventional unit data: unit region, rated capacity, minimum technical output, ramp rate, startup cost, power generation cost curve or typical quotation curve.

[0062] New energy data: the region to which the new energy belongs, and the predicted output curve (every hour or every 15 minutes).

[0063] Grid model data: parameters such as node-line connection relationship, line impedance, and section stability limit.

[0064] Others: interconnection line power, busbar load.

[0065] Specifically, the above-mentioned data cleaning time-sensitive parts include the elimination of abnormal values (such as mutation values), identifying and removing abnormal data by setting thresholds or based on statistical methods (such as the 3σ criterion); interpolation methods (such as linear interpolation, spline interpolation) or filling in missing data based on the correlation of neighboring nodes to ensure data integrity.

[0066] The above standardization process can realize the load L, new energy output P new The data are normalized according to the annual maximum value or installed capacity, and the formula can be expressed as follows:

[0067]

[0068] Where, L max is the annual maximum load, P cap This step aims to eliminate the magnitude differences in multi-year data and ensure the consistency of subsequent meteorological-power mapping relationships.

[0069] The above-mentioned index screening can be used to screen out wind speed and radiation at sites with strong correlations with renewable energy output in the region based on the geographical concentration of renewable energy installed capacity. Similarly, the temperature at sites with strong correlations with load can be screened out. Alternatively, the Pearson correlation coefficient of historical data can be used to screen out strongly correlated sites. The correlation threshold can be defined as:

[0070]

[0071] Among them, α is the correlation threshold, which needs to be given according to the specific regional conditions. For example, the correlation of wind and solar power generally selects meteorological stations with a correlation of 0.85 or above. For areas with a high proportion of temperature-sensitive loads, the temperature-load correlation threshold is above 0.8.

[0072] S2. Use the preprocessed scenario construction data to train the prediction model, and generate prediction scenarios and fluctuation scenarios through the prediction model. The fluctuation scenarios include typical scenarios and extreme scenarios.

[0073] The prediction model includes meteorological prediction model, load prediction model and new energy power prediction model. Specifically, at different time points in a day, the actual meteorological (wind speed V, radiation R, temperature T) is established in relation to the predicted meteorological (wind speed radiation Temperature ), which is actually equivalent to modeling the error random variable in the sense of probability operation. It can be assumed that: Where X = (v, P, T) T , are random vectors of actual weather and forecast weather respectively, are the conditional mean vector and covariance matrix, respectively, which are functions of the forecast meteorological random vector and are estimated through the sample set.

[0074] Furthermore, massive multi-scenario generation can be performed: given the forecast weather value x for the next day, A large number of scenarios are extracted to represent the entire space of meteorological conditions for the next day; the synchronous iterative elimination method is used to reduce the massive meteorological scenarios to a small number of typical scenarios, retaining representative scenarios while significantly reducing computational complexity.

[0075] The load forecasting model mentioned above can be used to predict the load based on the temperature T, date code (such as weekdays, weekends, holidays), and recent load L in the meteorological data. recent , build a short-term load forecasting model, the model kernel can choose XGBoost or neural network and other algorithm ideas, output the predicted load The model can adopt the more mature structure currently, which will not be described here.

[0076] The above-mentioned new energy power prediction model can be constructed based on the wind speed V and / or radiation R in the meteorological data, and also uses XGBoost or neural network as the model kernel to output the predicted power The mapping relationship between input and output variables can be expressed in the following form: Similarly, the model can adopt the more mature structure currently, which will not be elaborated here.

[0077] Furthermore, the above-mentioned prediction model can generate multiple prediction scenarios (i.e., scenarios including meteorological data, load data, and new energy function data, etc.); and the fluctuation scenarios include typical scenarios and extreme scenarios. Among them, the generation of typical scenarios can be done by inputting typical meteorological data into the trained meteorological-power mapping model (i.e., load prediction model and new energy power prediction model) to obtain typical load scenarios and typical new energy power scenarios respectively;

[0078] Furthermore, the definition and generation of extreme scenarios can be as follows: define several extreme combination scenarios, such as high-risk scenarios such as "cold wave: wind and solar power shutdown + load increase", "Spring Festival: new energy boom + load drop", and screen out meteorological scenarios corresponding to extreme conditions (such as low temperature extremes, high radiation extremes, etc.) from massive scenarios, and input them into the meteorological-power mapping model to obtain extremely high-risk load and new energy power scenarios; the mathematical description of extreme scenarios needs to set appropriate boundary conditions according to regional specificity, such as considering 5 days as a weather process based on data-driven, determining the characteristic vector describing its meteorology, and using clustering to screen out a certain proportion of scenarios that are far away from the class center as extreme scenarios.

[0079] Furthermore, massive meteorological data can be input into the meteorological-power mapping model to obtain massive scenarios of load, power and meteorology, and then the confidence interval (such as the fluctuation range under 90% confidence level) can be calculated. The specific calculation formula is as follows:

[0080] [L low ,L high ]=Quantile(L,[0.05,0.95]),[P low ,P high ]=Quantile(P,[0.05,0.95]).

[0081] Specifically, a two-level evaluation system can be designed to measure the effectiveness of the generated scenarios and adjust the model parameters based on the evaluation results; the first level is risk economic evaluation, with the main indicators being coverage and average bandwidth; the second level is physical property evaluation, such as the difference measurement of characteristic indicators such as load rate and volatility; that is: calculate the evaluation indicators of the prediction scenarios and / or fluctuation scenarios, and adjust the model parameters of the prediction model based on the evaluation indicators until the evaluation indicators meet the preset conditions.

[0082] Among them, the coverage rate CR of the scene or scene confidence interval α ,mean interval width (MPIW) MIW α Commonly used as an evaluation metric to measure the performance of scene prediction. α It indicates the ratio of the actual value falling within the confidence interval of the predicted scene 1-α (α is a small positive value). The closer the value is to 1-α, the better the scene quality is, that is, the higher the fit between the predicted scene and the actual data is. α Indicates the overall width of the confidence interval of the scenario. When CR α When the same, MIW α The narrower the better, and the more conducive it is to the economic efficiency of risk control.

[0083] Optionally, the above coverage is specifically:

[0084]

[0085] The average bandwidth is:

[0086]

[0087] Where, CR α represents coverage, N is the number of evaluation data, I(·) is the indicative function, and is 1 when the condition is met; x i Indicates the actual value, Indicates the corresponding x i The confidence interval of the prediction scenario 1-α, Represent the upper and lower limits of the confidence interval, which are the quantiles of 2 / α or 1-2 / α of the predicted scenario; ΔP i α Indicates the width of the confidence interval.

[0088] Furthermore, the above daily load rate is specifically: A = B / C

[0089] The specific daily volatility is: D = E / F

[0090] Where A is the daily load rate, B is the daily maximum load, C is the daily average load, D is the daily fluctuation rate, E is the daily load standard deviation, and F is the daily load mean.

[0091] In the above, the meteorological uncertainty modeling and massive scenario generation method based on the Gaussian probability model includes conditional probability modeling, scenario generation, and scenario reduction. This method applies the multivariate conditional probability model to the probabilistic modeling of forecast meteorological error distribution, especially the multivariate modeling of temperature, wind speed, and radiation, providing the uncertainty meteorological boundary under the influence of meteorological coupling, laying the foundation for the meteorological background interpretation of the source-load boundary.

[0092] At the same time, the extreme scenario generation and new energy load risk boundary quantification methods, including extreme combination definition, risk boundary quantification, and post-scenario evaluation, apply cluster analysis to the identification and screening of extreme weather that affects source loads; apply the idea of chaotic perturbation system, regard the weather-power function mapping relationship as a complex system, and obtain power fluctuation scenarios by inputting multiple meteorological disturbance scenarios; calculate its probability derivatives based on the scenario, such as confidence intervals, typical scenarios, and extreme scenarios, and conduct model evaluation and iteration through mainstream post-scenario evaluation methods.

[0093] Specifically, the system backup demand evaluation model proposed in this application considering random fluctuations on both sides of supply and demand and grid security constraints adopts a two-stage model (i.e., the first optimization model and the second optimization model below). The optimization process of the first optimization model and the second optimization model for evaluating the system backup demand considering random fluctuations on both sides of supply and demand and grid security constraints is as follows: Figure 2 As shown, that is:

[0094] Phase 1 is achieved through the first optimization model: the prediction scenario is used to determine the positive and negative reserve requirements of the system and / or partition, without considering the unit reserve call process.

[0095] Phase 2 is implemented through the second optimization model: using the fluctuation scenario to verify whether the positive and negative standby demands of the system and / or partition determined by the prediction scenario can pass the safety check, considering the unit standby call; further explained in steps S3-S4 below.

[0096] S3. Use the evaluation data to establish a first optimization model, and based on the first optimization model and the constraints of the first optimization model, use the scenario data of the prediction scenario to optimize and solve the first optimization model to obtain the first decision variable of the first optimization model, where the decision variable includes the output of the first unit and the standby demand to be verified.

[0097] In the first optimization model, the decision variables may include: unit start and stop, unit output, unit positive standby, unit negative standby, zone positive standby demand, zone negative standby demand, system positive standby demand, and system negative standby demand; the optimization objective is to minimize the system power generation cost, startup cost, positive standby capacity cost, and negative standby capacity cost, which can be expressed as:

[0098]

[0099] Where N is the total number of units; T is the total number of time periods considered. If 96 time periods are considered in one day, T is 96; P i,t represents the output of unit i in period t under the forecast scenario; C i,t (P i,t ), are the operating cost and startup cost of unit i in period t under the forecast scenario, where the unit operating cost C i,t (P i,t ) is the power generation cost or a multi-segment linear function related to the output ranges declared by the unit and the corresponding energy prices; PR i,t NR i,t are the positive reserve demand and negative reserve demand of unit i in period t under the forecast scenario; are the prices of positive and negative reserves provided by unit i in time period t under the forecast scenario; M s is the penalty factor for relaxing the network power flow constraints used for optimization; are the forward and reverse tidal flow relaxation variables of section s in time period t under the prediction scenario.

[0100] Furthermore, the constraints of the first optimization model include the following:

[0101] System load balancing constraints: Among them, P i,t T represents the output of unit i in period t, j,t represents the planned power of tie line j in period t (input is positive, output is negative), NT is the total number of tie lines, D t is the system load during period t.

[0102] System positive and negative standby constraints:

[0103]

[0104] Indicates the system positive reserve demand variable, Represents the system negative spare demand variable.

[0105] Partition positive and negative spare constraints:

[0106]

[0107] Represents partition R n The positive spare demand variable, Represents partition R n Negative spare demand variable.

[0108] Constraints on positive and negative spare capacity of units:

[0109]

[0110] PR i,t It represents the positive reserve that unit i can provide during period t, NR i,t It represents the negative reserve that unit i can provide during period t.

[0111] Upper and lower limit constraints of unit output:

[0112]

[0113] It represents the adjustable upper limit of unit i in period t, Indicates the adjustable lower limit of unit i in period t, α i,t is the start and stop status of unit i in period t, α i,t 1 means the output of unit i in period t is not equal to 0, α i,t A value of 1 indicates that the output of unit i during period t is equal to 0.

[0114] Thermal power ramping constraints:

[0115] When the unit is climbing up or down a slope, it must meet the climbing rate requirements. The climbing constraint can be described as:

[0116]

[0117] Where ΔP i U is the maximum ramp rate of unit i, ΔP i D is the maximum ramp-down rate of unit i.

[0118] The unit output limit is determined by several factors:

[0119] When the unit is in normal operation, the unit's output range is ΔP i U , ΔP i D Decide;

[0120] When the unit is in the start period, the unit's output range is determined by the unit's allowable start rate (here )Decide;

[0121] When the unit is in shutdown period, the unit output range is determined by the unit's allowable shutdown rate (here )Decide.

[0122] Minimum continuous start and stop time constraints for thermal power plants:

[0123] Due to the physical properties of thermal power units and actual operating requirements, thermal power units are required to meet the minimum continuous start / stop time. The minimum continuous start / stop time constraint can be described as:

[0124]

[0125]

[0126] Among them, T U 、T D The minimum continuous start time and minimum continuous stop time of the unit; is the continuous start-up time and continuous shutdown time of unit i in period t.

[0127] The unit power constraint can be described as:

[0128]

[0129] Where T represents the total number of time periods considered; T0 is the length of a time period in the planning cycle. If 96 time periods are considered per day, each time period is 15 minutes, that is, T0 = 0.25 (hours); are the maximum and minimum power of unit i respectively.

[0130] Network constraints, considering the power flow constraints of key sections, can be described as:

[0131]

[0132] in, are the tidal flow transmission limits of section s respectively; G is the forward and reverse generator output power transfer distribution factor of the node where unit i is located on section s; s-j G is the generator output power transfer distribution factor of the node where the tie line j is located on section s; s-m is the generator output power transfer distribution factor for node k to section s. are the forward and reverse tidal flow relaxation variables of section s, respectively.

[0133] S4. Use the evaluation data to establish a second optimization model, and based on the second optimization model and the constraints of the second optimization model, use the scenario data of the fluctuation scenario to verify the network security constraints of the first decision variable, and obtain the second decision variable of the second optimization model, the second decision variable including the output of the second unit.

[0134] In the second optimization model, the decision variable is the unit output, and the optimization objective of the model is to minimize the relaxation of the network security constraint, which can be expressed as:

[0135] Where NS is the total number of sections, are the positive relaxation and negative relaxation of the tidal current at section S in period t under the fluctuation scenario k, respectively, and M is the penalty coefficient.

[0136] Furthermore, the constraints of the second optimization model may include the following:

[0137] System load balancing constraints:

[0138] in, is the output variable of unit i in period t under the fluctuation scenario k in stage 2, is the system load of the fluctuation scenario k in period t in stage 2, T j,t Same data as Phase 1.

[0139] Upper and lower limit constraints of unit output:

[0140] is the output variable of unit i in period t under the fluctuation scenario k in stage 2, represents the output value of unit i in period t in stage 1, Indicates the negative reserve value that unit i can provide in period t in stage 1, It represents the positive reserve value that unit i in stage 1 can provide during period t.

[0141] Thermal power ramping constraint: When the unit is ramping up or down, it must meet the ramp rate requirements. The ramping constraint can be described as:

[0142]

[0143] Indicates the start and stop status of unit i in period t in stage 1, Same data as Phase 1.

[0144] The unit power constraint can be described as: Same data as Phase 1.

[0145] Network constraints, considering the power flow constraints of key sections, can be described as:

[0146]

[0147] in, G s-i , G s-j , G s-m Same data as Phase 1. are the forward and reverse flow relaxation variables of section s in the fluctuation scenario k, respectively.

[0148] Among them, the bus load of fluctuation scenario k in period t is The processing method is:

[0149]

[0150] Furthermore, the two-stage model coordination and convergence conditions are:

[0151] 1) Convergence conditions

[0152]

[0153] ε is the convergence threshold, which should be a smaller positive number according to the actual situation.

[0154] If the cross-section limit in stage 2 is greater than the convergence threshold, a new term needs to be added to the objective function in stage 1. The following constraints need to be added to stage 1 for iterative optimization until the cross-section limit in stage 2 meets the convergence threshold.

[0155] 2) Add new constraints

[0156]

[0157] is the tidal current of section S at k in the mth iteration of period t in the scenario of stage 2, is the unit output of scenario k in period t in the m+1th iteration, is the unit output value of scenario k in stage 2 at time t, where:

[0158]

[0159] are the unit output variable, positive reserve variable, and negative reserve variable of the forecast scenario to be optimized in stage 1 of the m+1th iteration, where:

[0160]

[0161] The load balancing constraints that need to be satisfied in the fluctuation scenario k.

[0162] Furthermore, the first optimization model in the system backup demand assessment considering random fluctuations on both the supply and demand sides and grid security constraints is a mixed integer programming problem, and the second optimization model is a linear programming problem. Both can be solved directly using mature commercial software such as CPLEX or GUROBI, without any restrictions here.

[0163] Furthermore, after the solution, the positive and negative reserves of the system can be substituted into the constraints for verification to ensure that the constraints in the proposed system reserve demand assessment model considering random fluctuations on both the supply and demand sides and the grid security constraints can be met. The operating costs, new energy consumption conditions, and section exceeding limits of the fluctuation scenarios under the system reserve capacity based on a fixed reserve capacity ratio or historical data are compared to evaluate the performance of the obtained reserve demand scheme. The actual effects of the scheme in saving system operating costs, improving new energy consumption, and solving the problem of "having reserve but unable to call" that may occur in scheduling are analyzed to verify the effectiveness of the optimization solution.

[0164] S5. If the output of the first unit and the output of the second unit meet the preset verification pass condition, the standby demand to be verified is determined as the target system standby demand, and the target system standby demand includes a positive standby demand and a negative standby demand.

[0165] Among them, the system backup demand evaluation optimization model considering random fluctuations on both sides of supply and demand and grid security constraints adopts a two-stage model, namely the first optimization model and the second optimization model. Stage 1 adopts the first optimization model: the prediction scenario is used to determine the positive and negative backup demands of the system and / or partition, without considering the unit backup call process; Stage 2 adopts the second optimization model: the fluctuation scenario verifies whether the positive and negative backup demands of the system and / or partition determined by the prediction scenario can pass the safety check, considering the unit backup call; the optimization method is used to refine the positive and negative backup demands of the system and / or partition, taking into account the fluctuation scenarios and grid security constraints to be considered for random fluctuations on both sides of source and load, which can solve the problem of "having backup but unable to call" that may occur in actual scheduling, minimize the system operating cost, improve the consumption of new energy, and thus improve the system reliability and economy.

[0166] Example 2: This embodiment of the present application provides a system backup demand assessment system considering random fluctuations in supply and demand and grid security constraints, which is applied to the system backup demand assessment method considering random fluctuations in supply and demand and grid security constraints in Example 1, such as Figure 3 Shown, including:

[0167] The data processing module is used to obtain scenario construction data and evaluation data for backup demand optimization, and preprocess the scenario construction data and evaluation data. The preprocessing includes data cleaning, standardization and screening.

[0168] The model training module is used to train the prediction model using pre-processed scenario construction data, and generate prediction scenarios and fluctuation scenarios through the prediction model. The fluctuation scenarios include typical scenarios and extreme scenarios.

[0169] a model solving module, configured to establish a first optimization model using the evaluation data, and optimize and solve the first optimization model based on the first optimization model and the constraints of the first optimization model using the scenario data of the prediction scenario to obtain a first decision variable of the first optimization model, wherein the decision variable includes the output of the first unit and the standby demand to be verified;

[0170] a demand verification module, configured to establish a second optimization model using the evaluation data, and verify the network security constraints of the first decision variable using scenario data of the fluctuation scenario based on the second optimization model and the constraints of the second optimization model, and obtain a second decision variable of the second optimization model, the second decision variable including the output of the second unit;

[0171] The demand determination module is used to determine the standby demand to be verified as the target system standby demand if the output of the first unit and the output of the second unit meet the preset verification pass conditions. The target system standby demand includes positive standby demand and negative standby demand.

[0172] Furthermore, the model training module also includes:

[0173] Calculate the evaluation index of the prediction scenario and / or fluctuation scenario, and adjust the model parameters of the prediction model based on the evaluation index until the evaluation index meets the preset conditions.

[0174] Example 3: This embodiment of the present application provides an electronic device, such as Figure 4 As shown, it includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method of embodiment 1 is implemented.

[0175] Example 4: The embodiment of the present application provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions enable a computer to execute the method of Example 1.

[0176] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A system backup demand assessment method considering random fluctuations in supply and demand and grid security constraints, characterized by: The specific steps include: Acquire scenario construction data and backup demand optimization evaluation data, and preprocess the scenario construction data and the evaluation data, wherein the preprocessing includes data cleaning, normalization, and screening; Using the pre-processed scenario construction data to train a prediction model, and generating prediction scenarios and fluctuation scenarios through the prediction model, wherein the fluctuation scenarios include typical scenarios and extreme scenarios; Establishing a first optimization model using the evaluation data, and optimizing and solving the first optimization model based on the first optimization model and constraints of the first optimization model using the scenario data of the prediction scenario to obtain first decision variables of the first optimization model, wherein the decision variables include the output of the first unit and the standby demand to be verified; establishing a second optimization model using the evaluation data, and verifying the network security constraints of the first decision variable using the scenario data of the fluctuation scenario based on the second optimization model and the constraints of the second optimization model, and obtaining a second decision variable of the second optimization model, wherein the second decision variable includes the output of the second unit; If the output of the first unit and the output of the second unit meet the preset verification pass condition, the standby demand to be verified is determined as the target system standby demand, and the target system standby demand includes a positive standby demand and a negative standby demand.

2. The system backup demand assessment method considering random fluctuations in supply and demand and grid security constraints according to claim 1 is characterized in that: The method further comprises: Calculate evaluation indicators of the prediction scenario and / or fluctuation scenario, and adjust model parameters of the prediction model based on the evaluation indicators until the evaluation indicators meet preset conditions.

3. The system backup demand assessment method considering random fluctuations in supply and demand and grid security constraints according to claim 2 is characterized in that: The evaluation indicators include coverage, average bandwidth, daily load rate and daily fluctuation rate.

4. The system backup demand assessment method considering random fluctuations in supply and demand and grid security constraints according to claim 3 is characterized in that: The coverage is specifically: The average bandwidth is specifically: Where, CR α represents coverage, N is the number of evaluation data, I(·) is the indicative function, and is 1 when the condition is met; x i Indicates the actual value, Indicates the corresponding x i The confidence interval of the prediction scenario 1-α, Represent the upper and lower limits of the confidence interval, which are the quantiles of 2 / α or 1-2 / α of the predicted scenario; ΔP i α Indicates the width of the confidence interval; The daily load rate is specifically: A = B / C The daily volatility is specifically: D = E / F Where A is the daily load rate, B is the daily maximum load, C is the daily average load, D is the daily fluctuation rate, E is the daily load standard deviation, and F is the daily load mean.

5. The system backup demand assessment method considering random fluctuations in supply and demand and grid security constraints according to claim 1 is characterized in that: The first optimization model is optimized and solved, specifically: Where N is the total number of units; T is the total number of time periods considered. If 96 time periods are considered in one day, T is 96; P i,t represents the output of unit i in period t under the forecast scenario; C i,t (P i,t ), are the operating cost and startup cost of unit i in period t under the forecast scenario, where the unit operating cost C i,t (P i,t ) is the power generation cost or a multi-segment linear function related to the output ranges declared by the unit and the corresponding energy prices; PR i,t NR i,t are the positive reserve demand and negative reserve demand of unit i in period t under the forecast scenario; are the prices of positive and negative reserves provided by unit i in time period t under the forecast scenario; M s is the penalty factor for relaxing the network power flow constraints used for optimization; are the forward and reverse tidal flow relaxation variables of section s in time period t under the prediction scenario.

6. The system backup demand assessment method considering random fluctuations in supply and demand and grid security constraints according to claim 1 is characterized in that: Verifying the network security constraint of the first decision variable includes optimizing and solving the second optimization model, specifically: Where NS is the total number of sections, are the positive relaxation and negative relaxation of the tidal current at section S in period t under the fluctuation scenario k, respectively, and M is the penalty coefficient.

7. A system backup demand assessment system considering random fluctuations in supply and demand and grid security constraints, applied to the system backup demand assessment method considering random fluctuations in supply and demand and grid security constraints as claimed in any one of claims 1 to 6, characterized in that: include: A data processing module, configured to obtain scenario construction data and evaluation data for optimization of standby demand, and preprocess the scenario construction data and the evaluation data, wherein the preprocessing includes data cleaning, normalization, and screening; A model training module is used to train a prediction model using the pre-processed scenario construction data, and generate prediction scenarios and fluctuation scenarios through the prediction model, wherein the fluctuation scenarios include typical scenarios and extreme scenarios; a model solving module, configured to establish a first optimization model using the evaluation data, and optimize and solve the first optimization model based on the first optimization model and constraints of the first optimization model using the scenario data of the prediction scenario to obtain a first decision variable of the first optimization model, wherein the decision variable includes the output of the first unit and the standby demand to be verified; a demand verification module, configured to establish a second optimization model using the evaluation data, and verify the network security constraints of the first decision variable using the scenario data of the fluctuation scenario based on the second optimization model and the constraints of the second optimization model, and obtain a second decision variable of the second optimization model, wherein the second decision variable includes the output of the second unit; The demand determination module is used to determine the standby demand to be verified as the target system standby demand if the output of the first unit and the output of the second unit meet the preset verification pass condition. The target system standby demand includes a positive standby demand and a negative standby demand.

8. The system backup demand assessment system considering random fluctuations in supply and demand and grid security constraints according to claim 7 is characterized in that: The model training module also includes: Calculate evaluation indicators of the prediction scenario and / or fluctuation scenario, and adjust model parameters of the prediction model based on the evaluation indicators until the evaluation indicators meet preset conditions.

9. An electronic device, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the method according to any one of claims 1 to 6 is implemented when the processor executes the computer program.

10. A non-transitory computer-readable storage medium, characterized in that The non-transitory computer-readable storage medium stores computer instructions, which enable a computer to execute the method of any one of claims 1 to 6.

Images