Multi-type standby resource collaborative optimization method and device considering source load uncertainty

Abstract

The invention discloses a multi-type standby resource collaborative optimization method and device considering source-load uncertainty, and the method comprises the steps: generating a plurality of new energy and load prediction error sub-databases based on data clustering, generating a rhombic convex hull uncertainty set based on the sub-databases, generating a standby demand curve set of a limit scene, and generating a system operation standby constraint; constructing a multi-type standby resource collaborative optimization model and a multi-type standby resource scheduling operation model which take minimum day-ahead stage cost and intra-day stage cost as targets and take system operation standby constraints as constraint conditions; and solving the multi-type standby resource collaborative optimization model to obtain the standby demand capacity of the system at each moment and a corresponding configuration scheme of the multi-type standby resources. According to the method, accurate quantitative evaluation of the system operation standby capacity demand and optimal configuration of standby resources are realized, and the new energy consumption is effectively promoted and the scheduling operation economy is improved while the sufficient standby margin of the system is guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electrical engineering, and in particular to a multi-type reserve resource collaborative optimization method and device considering source-load uncertainty. BACKGROUND

[0002] Sustainable clean energy mainly including wind power and photovoltaic power has entered a period of rapid development. Related research shows that by 2035 and 2050, the installed capacity of power generation in China will reach 35.9 GW and 48.6 GW respectively, and the proportion of renewable energy power generation installed capacity will be 57% and 75% respectively. Wind power and photovoltaic power generation power are easily affected by meteorological factors and have the inherent characteristics of multi-time scale fluctuation. Compared with the traditional power system, the new type of power system has changed from unilateral uncertainty on the load side to bilateral uncertainty on the generation side and the load side. In order to meet the power balance demand caused by the prediction deviation of both sides of the system, a certain reserve capacity, especially the operating reserve capacity, must be maintained in the power system to avoid system operation accidents. Reasonable setting of reserve capacity is one of the important ways to ensure the economic and safe operation of the new type of power system. The existing extensive reserve management is easy to cause excessive or insufficient reserve, and insufficient system reserve capacity may lead to difficulty in balancing the power fluctuation of load and new energy during operation, and even threaten the safe and stable operation of the power system in emergency; excessive reserve capacity will cause the operating efficiency of part of the synchronous unit to be reduced, and the excessive start and stop of the thermal power unit due to the technical output limit will make it difficult for the system to effectively absorb new energy during the load valley period, resulting in wind and light curtailment and increasing the economic cost of system operation.

[0003] To solve the above problems, on the one hand, it is necessary to accurately evaluate the new energy / load prediction deviation and reasonably reserve the operating reserve capacity to effectively avoid insufficient reserve or redundant configuration, so as to maximize the promotion of new energy consumption while ensuring the adequacy of system reserve capacity; on the other hand, the configuration of multiple types of appropriate flexible reserve resources in the existing reserve system can effectively suppress the adverse effects of wind power and photovoltaic fluctuation on the system. The complementary of multiple types of flexible reserve resources not only can ensure the reliable and stable operation of the system, but also can realize the optimal allocation of resources, which is a key strategy to effectively improve the economic efficiency of the system.

[0004] In summary, as for the optimization configuration method of the operating reserve of the new type of power system under the condition of large-scale new energy grid connection, there are two problems that need further research: (1) how to consider the prediction uncertainty of new energy and load to accurately quantify the evaluation of the system operating reserve capacity demand, so as to maximize the promotion of new energy consumption while ensuring the power balance demand of the system. (2) How to consider the regulation capacity of multiple types of flexible reserve resources to realize the collaborative optimization configuration of the system operating reserve resources, so as to effectively improve the economic efficiency of the system. SUMMARY

[0005] The application aims to overcome the above-mentioned defects and problems existing in the prior art, and provide a multi-type reserve resource collaborative optimization method and device considering source load uncertainty, so as to realize accurate quantitative evaluation of system operation reserve capacity demand and optimization configuration of reserve resources, effectively promote new energy consumption, and improve dispatching operation economy while ensuring sufficient system reserve margin.

[0006] To achieve the above-mentioned purpose, the technical solution of the application is as follows:

[0007] In a first aspect, the application provides a multi-type reserve resource collaborative optimization method considering source load uncertainty, comprising:

[0008] Analyzing the prediction error distribution characteristics of new energy and load, generating a plurality of new energy and load prediction error sub-databases based on data clustering, generating a rhombus convex hull uncertainty set based on the sub-databases, extracting the vertices of the set to generate a reserve demand curve set of limit scenarios, and generating system operation reserve constraints according to the reserve demand capacity of the system at each time;

[0009] Constructing a multi-type reserve resource collaborative optimization model and a dispatching operation model of multi-type reserve resources, the multi-type reserve resource collaborative optimization model taking the minimum total cost as the target and the system operation reserve constraints as the constraint conditions, the total cost including day-ahead stage cost and intra-day stage cost, the day-ahead stage cost including generation cost, new energy curtailment cost and reserve cost, and the intra-day stage cost including new energy curtailment and load shedding penalty under each reserve limit scenario;

[0010] Solving the multi-type reserve resource collaborative optimization model to obtain the reserve demand capacity of the system at each time and the corresponding configuration scheme of the multi-type reserve resources.

[0011] Preferably, the generation method of the new energy and load prediction error sub-database is as follows:

[0012] Arranging the new energy prediction error data based on the new energy day-ahead prediction output value in size, sequentially selecting samples to form a new energy prediction error sub-database; wherein, is the number of new energy day-ahead prediction samples, is the number of new energy day-ahead prediction sample groups;

[0013] The K-means clustering algorithm is used to obtain the monthly grouping of load day-ahead prediction error, and the monthly grouping is subjected to K-means clustering based on the load day-ahead prediction value to obtain a load prediction error sub-database.

[0014] Preferably, the acquisition method of the rhombus convex hull uncertainty set comprises:

[0015] Let the collected historical scenarios be denoted as where, is the number of collected historical scenarios;

[0016] Solve a high-dimensional ellipsoid to enclose all historical scenarios, and then solve the high-dimensional ellipsoid parameters to solve the following optimization problem:

[0017] ;

[0018] In the formula, is a constant, representing the volume of the unit sphere in dimension n; is a positive definite matrix; is a translation parameter;

[0019] Select the convex set enclosed by the original high-dimensional ellipsoid of vertices as the initial uncertainty set, and perform orthogonal decomposition on the positive definite matrix :

[0020] ;

[0021] ;

[0022] In the formula, is a diagonal matrix; and are the eigenvalues of the diagonal matrix ; is a transformation matrix;

[0023] Rotate and translate the original high-dimensional ellipsoid , and the mathematical expression of the obtained high-dimensional ellipsoid is as follows:

[0024] ;

[0025] In the formula, is the high-dimensional ellipsoid equation obtained after rotation and translation; is the coordinate value of the random variable in the rotated coordinates; is an n*n matrix;

[0026] The vertex coordinates of the high-dimensional ellipsoid are as follows:

[0027] ;

[0028] In the formula, is the coordinate value of the first vertex of the high-dimensional ellipsoid ; The number of vertices. ;

[0029] The original high-dimensional ellipsoid can be obtained by using the inverse transformation of the above equation. The vertex equation is derived from the original high-dimensional ellipsoid. The equation of the rhombus convex hull formed by the vertices is shown below:

[0030] ;

[0031] In the formula, The equation for the rhombic convex hull enclosed by the vertices of a high-dimensional ellipsoid; Transformation matrix Element;

[0032] Introducing a magnification factor to enlarge and reduce the rhomboid convex hull, the corrected uncertain set is:

[0033] ;

[0034] ;

[0035] in, It is an uncertain set of rhomboid convex hull; This is the extreme scene obtained after scaling the convex hull. This is the magnification factor in convex hull scaling.

[0036] Preferably, the method for calculating the magnification factor in the convex hull scaling is as follows:

[0037] Collection of historical scenes Rotation and translation yield a new set of historical scenes ;

[0038] According to the convexity theory, if the new historical scenario If a surface can be covered by a rhombic convex hull enclosed by the vertices of a high-dimensional ellipsoid, then the following equation holds:

[0039] ;

[0040] ;

[0041] In the formula, The coefficients are positive.

[0042] for Given several historical scenarios, we establish the following optimization problem to determine the inclusion relationship between the historical scenarios and the convex hull:

[0043] ;

[0044] ;

[0045] In the formula, For the first One magnification factor; Positive coefficients With magnification The product;

[0046] The above inclusion relationship is obtained as follows The sequence formed by the magnification factors is the maximum value in the convex hull scaling. .

[0047] Preferably, the system's standby constraint is:

[0048] ; ;

[0049] In the formula, and for The system adjusts and lowers its reserve capacity during specific time periods. and These represent both increasing and decreasing the reserve capacity for each type of reserve resource. and These are the coefficient matrices representing the proportion of increased and decreased reserve capacity for each type of reserve resource.

[0050] Preferably, the objective function of the multi-type backup resource collaborative optimization model is:

[0051] ;

[0052] In the formula, , and Synchronous generator units exist Fuel costs, start-up costs, and shutdown costs during different time periods; This refers to the penalty coefficient for curtailment of renewable energy. For nodes The load shearing penalty coefficient; For new energy electric fields The power curtailment penalty coefficient; For new energy electric fields exist Maximum output during the time period; and They are respectively The total upward and downward adjustment of reserve costs for the time-period system; Probabilities for different extreme scenarios; For the scene Down Time period nodes cut load amount; for the scenario down period new energy station of the power.

[0053] Preferably, the dispatching operation model of the multiple types of reserve resources includes a synchronous unit operation model, a wind farm operation model, a photovoltaic power station operation model, an ultra-high voltage direct current transmission line operation model, a regional alternating current tie line operation model, an energy storage device operation model, and a demand side response load operation model.

[0054] In a second aspect, the present application provides a multiple types of reserve resources collaborative optimization device considering source and load uncertainty, which is used to implement the method described above, and the device comprises:

[0055] A reserve constraint acquisition module is configured to analyze the prediction error distribution characteristics of new energy and load, generate a plurality of new energy and load prediction error sub-databases based on data clustering, generate a rhombus convex hull uncertainty set based on the sub-databases, extract the vertices of the set to generate a reserve demand curve set of limit scenarios, and generate system operation reserve constraints according to the reserve demand capacity of the system at each time.

[0056] A model construction module is configured to construct a multiple types of reserve resources collaborative optimization model and a dispatching operation model of multiple types of reserve resources, wherein the multiple types of reserve resources collaborative optimization model takes the minimum total cost as the target and takes the system operation reserve constraints as the constraint conditions, the total cost includes a day-ahead stage cost and an intra-day stage cost, the day-ahead stage cost includes a generation cost, a new energy curtailment cost, and a reserve cost, and the intra-day stage cost includes new energy curtailment and cut load penalty under each reserve limit scenario.

[0057] A model solving module is configured to solve the multiple types of reserve resources collaborative optimization model to obtain the reserve demand capacity of the system at each time and the corresponding configuration scheme of the multiple types of reserve resources.

[0058] In a third aspect, the present application provides a multiple types of reserve resources collaborative optimization device considering source and load uncertainty, which comprises a memory and a processor.

[0059] The memory is configured to store computer program codes and transmit the computer program codes to the processor.

[0060] The processor is configured to execute the method described above according to the instructions in the computer program codes.

[0061] In a fourth aspect, the present application provides a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the method described above.

[0062] Compared with the prior art, the present application has the beneficial effects that:

[0063] In the method for collaborative optimization of multiple types of reserve resources considering source-load uncertainty provided by the present application, the bilateral prediction uncertainty of new energy and load and the regulation capacity of multiple types of reserve resources are considered, accurate quantitative evaluation of system operation reserve capacity demand and optimal allocation of reserve resources are realized, the system reserve margin is ensured to be sufficient, new energy consumption is effectively promoted, and dispatching operation economy is improved. BRIEF DESCRIPTION OF DRAWINGS

[0064] Figure 1 is a flow chart of the method for collaborative optimization of multiple types of reserve resources considering source-load uncertainty provided by the present application.

[0065] Figure 2 is a flow chart of the solution of the collaborative optimization model of multiple types of reserve resources provided in the embodiment of the present application.

[0066] Figure 3 is a schematic diagram of the modified IEEE28 node system provided in the embodiment of the present application.

[0067] Figure 4 is a schematic diagram of the system operation reserve demand evaluation and reserve resource allocation result based on the maximum load of the traditional fixed proportion system provided in the embodiment of the present application.

[0068] Figure 5 is a schematic diagram of the system operation reserve demand evaluation and reserve resource allocation result considering source-load uncertainty provided in the embodiment of the present application.

[0069] Figure 6 is a structural block diagram of the device for collaborative optimization of multiple types of reserve resources considering source-load uncertainty provided by the present application.

[0070] Figure 7 is a structural block diagram of the equipment for collaborative optimization of multiple types of reserve resources considering source-load uncertainty provided by the present application. DETAILED DESCRIPTION

[0071] The present application is further described in detail below in combination with the description of the drawings and specific embodiments.

[0072] Referring to Figure 1 , a method for collaborative optimization of multiple types of reserve resources considering source-load uncertainty, comprising:

[0073] S1, analyze the prediction error distribution characteristics of new energy and load, generate a plurality of different significant new energy and load prediction error sub-databases based on data clustering; generate a rhombus convex hull uncertainty set based on the sub-database, extract the set vertex to generate a backup demand curve set of the limit scenario, and generate system operation backup constraints according to the backup demand capacity of the system at each time;

[0074] S2, construct a multi-type backup resource collaborative optimization model and a multi-type backup resource scheduling and operation model, the multi-type backup resource collaborative optimization model taking the minimum total cost as the target and the system operation backup constraint as the constraint condition, the total cost including a day-ahead stage cost and an intra-day stage cost, the day-ahead stage cost including a generation cost, a new energy curtailment cost and a backup cost, and the intra-day stage cost including new energy curtailment and load shedding penalty under each backup limit scenario;

[0075] S3, solve the multi-type backup resource collaborative optimization model to obtain the backup demand capacity of the system at each time and the corresponding multi-type backup resource configuration scheme.

[0076] The application proposes a multi-type backup resource collaborative optimization method considering accurate description of source and load uncertainty, which aims to consider the prediction uncertainty of new energy and load and the regulation capacity of multi-type flexible backup resources, realize accurate quantitative evaluation of system operation backup capacity demand and optimization configuration of flexible backup resources, effectively promote new energy consumption and improve dispatching and operation economy while ensuring sufficient system backup margin.

[0077] Further, the application needs to obtain the parameters of each device in the power system. The devices in the power system include synchronous units, new energy units, energy storage, tie lines, power grids and electrical loads.

[0078] Specifically, the conventional parameters of the synchronous units, new energy units, energy storage, tie lines, power grids and electrical loads in the power system are collected, as well as the output curve of the new energy power station.

[0079] The conventional parameters of the synchronous units include: minimum unit output, maximum unit output, maximum up-ramp rate, maximum down-ramp rate, maximum start-up ramp rate, maximum shutdown ramp rate, minimum start-up time, and minimum shutdown time.

[0080] The conventional parameters of the new energy power station include: new energy power station capacity and node location of the new energy power station.

[0081] The conventional parameters of the energy storage include: energy storage capacity, energy storage rated power, and node location of the energy storage.

[0082] The conventional parameters of the electrical load include: node electrical load day-ahead prediction value and node electrical load demand response capacity.

[0083] The conventional parameters of the power grid include a transmission line maximum capacity and a power flow sensitivity matrix.

[0084] The new energy power station output curve is an electric load prediction curve and a maximum active power output curve of the new energy power station.

[0085] Further, the data-driven diamond convex hull uncertainty set construction method can effectively consider the correlation between multiple variables, and when applied to operation reserve demand evaluation, can effectively realize accurate description of uncertainty, but related prior art often has defects such as dependence on effectiveness of original data, and evaluation results lack theoretical support. In this regard, the present application proposes a source-load uncertainty set construction method considering new energy / load prediction error distribution characteristics; first, analyze the prediction error distribution characteristics of new energy and load, generate multiple differentiated significant new energy / load prediction error sub-databases based on data clustering, and according to the new energy and load day-ahead prediction value and the month in which it is located, select the most matched sub-database as the quantitative basis for reserve demand evaluation, thereby improving the effectiveness of data used for data-driven analysis. Then, generate a diamond convex hull uncertainty set based on the sub-database data, and based on a certain confidence, extract the vertices to generate an up / down reserve limit scenario set, accurately quantify system operation reserve demand, and finally generate system operation reserve constraints. The quantitative evaluation results obtained by the method have certain theoretical interpretability.

[0086] New energy and load prediction bias has obvious seasonal and time period distribution characteristics, and has obvious correlation with day-ahead prediction value. In order to improve the effectiveness of the data, based on the prediction error distribution characteristics of new energy and load, the following differentiated classification strategy is proposed:

[0087] New energy classification strategy based on day-ahead prediction output value: assuming that the number of new energy day-ahead prediction samples in the database is , the number of new energy day-ahead prediction sample groups is In order to avoid insufficient number of samples in some edge scenarios, an equal grouping strategy is adopted. That is, the new energy prediction error data is arranged in size based on the new energy day-ahead prediction output value, and samples are selected in sequence to form a new energy prediction error sub-database.

[0088] Load clustering strategy based on day-ahead prediction output value and seasonal and time period distribution characteristics: load day-ahead prediction bias has obvious correlation characteristics with months. The K-means clustering algorithm is used to obtain month grouping of load day-ahead prediction error, and K-means clustering is performed on month grouping based on load day-ahead prediction value, to obtain load prediction error sub-database under different load day-ahead prediction output levels.

[0089] The essence of K-means clustering algorithm is a data induction algorithm based on Euclidean distance, by dividing the given sample into clusters with similar characteristics , the similarity of samples in the cluster is as high as possible, and the difference between different samples is as large as possible; The specific steps are as follows:

[0090] Original data preprocessing and data normalization.

[0091] Select samples as initial clustering centers , wherein The value is obtained by the distribution characteristics of the load day-ahead prediction deviation and the month.

[0092] Calculate the Euclidean distance of each sample in the data to the first Clustering center, and induct it into the category with the smallest Euclidean distance.

[0093] ;

[0094] In the formula, The Euclidean distance of sample To clustering center .

[0095] After completing a new round of clustering, recalculate the clustering center:

[0096] ;

[0097] In the formula, The dimension of the first Classification (i.e. the number of samples).

[0098] Repeat the above steps until the termination condition is met (the number of iterations is reached, the clustering center changes less than the set value in the adjacent two iterations, etc.), and output the clustering result.

[0099] According to the new energy and load day-ahead prediction value of the day to be evaluated and the month it is in, select the most matched sub-database as the quantitative basis for subsequent standby demand assessment.

[0100] Further, for the matched sub-database, a data-driven uncertain set modeling method is proposed, which can fully consider the correlation between random variables, and adopts a rhombus convex hull representation form to facilitate analytical solution. The data-driven rhombus convex hull uncertain set Mathematical expression is as follows:

[0101] ;

[0102] Wherein, For representative scenarios extracted from historical data scenarios, they are defined as limit scenarios in the present application;

[0103] Specifically, the method for obtaining the rhombus convex hull uncertain set comprises:

[0104] (1) Collect the historical data as column vector form, wherein each group of historical data is a historical scenario, denoted as , wherein, is the number of collected historical scenarios.

[0105] (2) Assuming the following premise conditions: when the number of collected historical data is large enough, the historical prediction error data is representative of the prediction error data of the concerned period, that is, if there is a closed set that can completely cover the historical scenarios, the prediction error data of the concerned period is also in the closed set. By means of high-dimensional closed ellipsoid algorithm, a high-dimensional ellipsoid is solved to surround all historical scenarios, and then the high-dimensional ellipsoid parameters are solved to solve the following optimization problem:

[0106] ;

[0107] In the formula, is a constant, representing the volume of a unit sphere in dimension; is a positive definite matrix; is a translation parameter.

[0108] The above optimization is in the form of convex optimization, which is convenient for fast solution.

[0109] (3) Select the convex set surrounded by the vertices of the original high-dimensional ellipsoid as the initial uncertain set, and perform orthogonal decomposition on the positive definite matrix :

[0110] ;

[0111] ;

[0112] In the formula, is a diagonal matrix; and are eigenvalues of the diagonal matrix ; is a transformation matrix;

[0113] In order to obtain the vertices corresponding to the high-dimensional ellipsoid, the ellipsoid is rotated and translated to make its symmetry axis coincide with the coordinate axis. The original high-dimensional ellipsoid is rotated and translated to obtain the mathematical expression of the high-dimensional ellipsoid as follows:​

[0114] ;

[0115] wherein, is the high-dimensional ellipsoid equation after rotation and translation; is a random variable is the coordinate value in the coordinate after rotation; is an n*n matrix;

[0116] high-dimensional ellipsoid The vertex coordinates of the high-dimensional ellipsoid are:

[0117] ;

[0118] wherein, is the high-dimensional ellipsoid The coordinate value of the first vertex; is the number of vertices, ;

[0119] The vertex equation of the original high-dimensional ellipsoid can be obtained by inverse transformation of the above formula, and the rhombus convex hull equation surrounded by the vertices of the original high-dimensional ellipsoid is as follows:

[0120] ;

[0121] wherein, is the rhombus convex hull equation surrounded by the vertices of the high-dimensional ellipsoid; is an element of the transformation matrix .

[0122] (4) Correction of the rhombus convex hull and extraction of the limit scene. Since the rhombus convex hull surrounded by the selected high-dimensional ellipsoid vertices is not enough to cover all the historical scenes, an amplification factor needs to be introduced to enlarge and reduce the rhombus convex hull, and the corrected uncertain set is:

[0123] ;

[0124] ;

[0125] wherein, is the rhombus convex hull uncertain set; is the limit scene obtained after the convex hull is scaled; is the amplification factor in the convex hull scaling.

[0126] Further, the calculation method of the amplification factor in the convex hull scaling is:

[0127] The historical scene set is rotated and translated to obtain a new historical scene set ​​;

[0128] According to the convex optimization theory, if the new historical scenario can be covered by the rhombus convex hull formed by the vertices of the high-dimensional ellipsoid, the following formula is established:

[0129] ;

[0130] ;

[0131] In the formula, is a positive coefficient;

[0132] If there is a historical scenario outside the rhombus convex hull, the convex hull needs to be enlarged, and let be the enlargement factor, for historical scenarios, the following optimization problem is established to determine the inclusion relationship between the historical scenario and the convex hull:

[0133] ;

[0134] ;

[0135] In the formula, is the enlargement factor; is a positive coefficient and the product of the enlargement factor ;

[0136] For historical scenarios, through the above inclusion relationship, a sequence formed by enlargement factors is obtained, and the maximum value in the sequence is the enlargement factor in the scaling of the convex hull. The final data-driven uncertain set is represented by the rhombus convex hull. In addition, since the scaling factor in the original formula is determined by itself, the larger the area contained by the uncertain set, the more conservative the decision, so the scaling factor can be modified according to a certain confidence (i.e. the percentage of data contained by the rhombus convex hull).

[0137] Further, based on the seasonal, periodical distribution characteristics of the new energy and load prediction output data and the matching of the day-ahead prediction value to the corresponding sub-database, the rhombus convex hull uncertain set is generated based on a certain confidence using the above method, and the standby demand curve set representing the extreme scenarios is generated by extracting the set vertices. The envelope line of all extreme scenarios under the corresponding confidence is the standby demand capacity of the system at each time, thereby generating the corresponding system operation standby constraint:

[0138] ; ;

[0139] In the formula, and for The system adjusts and lowers its reserve capacity during specific time periods. and These represent both increasing and decreasing the reserve capacity for each type of reserve resource. and These are the coefficient matrices representing the proportion of increased and decreased reserve capacity for each type of reserve resource.

[0140] Furthermore, the scheduling and operation models for various types of backup resources include synchronous generator operation models, wind farm operation models, photovoltaic power plant operation models, ultra-high voltage direct current transmission line operation models, regional AC tie line operation models, energy storage device operation models, and demand-side response load operation models.

[0141] (1) Synchronous unit operation model

[0142] Traditional synchronous generator units possess excellent adjustability and controllability, effectively mitigating the volatility and uncertainty of renewable energy output. They are a crucial source of reserve capacity under current technological conditions. The operational reserve capacity of synchronous generator units is closely related to their output and ramp-up rate; the mathematical model for providing backup auxiliary services can be expressed as:

[0143] ;

[0144] ;

[0145] ;

[0146] In the formula, and thermal power units The minimum and maximum technical output values; For thermal power units exist Active power output during a given time period; and thermal power units Maximum uphill and maximum downhill gradients; and thermal power units exist The period includes both upward and downward adjustments to reserve capacity.

[0147] (2) Wind farm operation model

[0148] When the active power of the wind farm exceeds 20% of the total rated output, the unit can participate in the upward adjustment of the reserve auxiliary service through overspeed control or pitch angle; when the active power in the system is excessive, it can also provide downward reserve capacity for the system by reducing its active output.

[0149] ;

[0150] ;

[0151] ;

[0152] where, is the day-ahead power prediction value of the wind farm in the time period; is the dispatching plan value of the wind farm in the time period; and are the maximum active power output (rated installed capacity) and the minimum active power output (minimum grid-connected power) of the wind farm respectively; and are the steady-state upward reserve and downward reserve capacity of the wind farm in the time period; is the wind curtailment of the wind farm in the time period.

[0153] (3) Photovoltaic power station operation model

[0154] Similar to wind energy resources, photovoltaic power generation also exhibits intermittent and unstable characteristics. Secondly, photovoltaic arrays are fixed devices with zero rotational inertia, and their power fluctuations will have a more significant impact on the power grid (rotating components of wind turbines can buffer wind resource fluctuations to some extent). The reserve model of photovoltaic power station operation can be expressed as:

[0155] ;

[0156] ;

[0157] ;

[0158] where, is the day-ahead power prediction value of the photovoltaic power station in the time period; is the dispatching plan value of the photovoltaic power station in the time period; and respectively the maximum active power output (rated installed capacity) and the minimum active power output (minimum grid-connected power) of the photovoltaic power plant; and respectively the maximum active power output (rated installed capacity) and the minimum active power output (minimum grid-connected power) of the photovoltaic power plant; and respectively the maximum active power output (rated installed capacity) and the minimum active power output (minimum grid-connected power) of the photovoltaic power plant; the steady-state up-regulation and down-regulation reserve capacity of the photovoltaic power plant in the time period the steady-state up-regulation and down-regulation reserve capacity of the photovoltaic power plant in the time period the light curtailment amount of the photovoltaic power plant in the time period the light curtailment amount of the photovoltaic power plant in the time period

[0159] (4) Operation model of UHVDC transmission line

[0160] The construction of the operation reserve mode of UHVDC should follow the following principles: it not only needs to exert certain cross-regional regulation capacity, but also needs to maintain the power quality of the receiving side (when the power grid is the transmission end) or the transmission side (when the power grid is the receiving end). Under this principle, in addition to the conventional constraints, the power adjustment times of UHVDC, the maximum adjustment capacity per cycle, and the power transaction deviation during dispatching should also be limited, and the operation reserve model of UHVDC can be expressed as:

[0161] ;

[0162] ;

[0163] ;

[0164] ;

[0165] ;

[0166] In the formula, is the power transmission up-regulation reserve capacity of UHVDC in the time period is the power transmission down-regulation reserve capacity of UHVDC in the time period is the maximum outgoing power of UHVDC; is the minimum outgoing power of UHVDC; is the actual outgoing power of UHVDC; is the maximum upward ramping amount of UHVDC per unit dispatching period; is the maximum downward ramping amount of UHVDC per unit dispatching period, and UHVDC can be operated in short-time overload; is a binary variable used to represent whether the DC line is in the time period compared with the time period is a binary variable used to represent whether the DC line is in the time period compared with the time period 1 if the transmission power is adjusted, 0 if not; is the time interval, here 1h. is the maximum number of adjustment times of the UHVDC transmission power in the dispatch period; is the DC line power deviation limit of the transmission power from the initial active power schedule value; is the UHVDC transmission power in the power transmission schedule value of the time interval; is the DC line power deviation limit of the transmission power from the initial active power schedule value; is the time interval, here 1h.

[0167] (5) Regional AC tie-line operation model

[0168] Similar to the UHVDC transmission, in order to ensure the power quality in a large area, it is necessary to limit the adjustment times of the regional AC tie-line (HVAC), the maximum adjustment power in each time interval, and the power deviation during the dispatch period. Under the regional reserve sharing mechanism, the operation reserve model is as follows:

[0169]

[0170]

[0171]

[0172]

[0173]

[0174]

[0175]

[0176] wherein, is the upward reserve amount of the regional AC tie-line in the time interval; is the downward reserve amount of the regional AC tie-line in the time interval; is the power transmission value of the tie-line in the time interval; is the maximum transmission power of the tie-line, which is limited by the rated transmission capacity of the tie-line and the transmission limit capacity of the sending / receiving end power grid (thermal stability, transient stability, etc.); is the rated transmission capacity of the regional AC tie-line; is the stability transmission limit capacity of the regional AC tie-line;​​​​​​​ is the minimum transmission power of the tie-line; is the maximum up-ramp of the regional AC tie-line; is the maximum down-ramp of the regional AC tie-line; is a binary variable, used to represent whether the transmission power of the regional AC tie-line is adjusted in the time interval compared to the time interval , 1 represents adjustment, 0 represents no adjustment; is a time number set; is the upper limit of the number of transmission power adjustment of the regional AC tie-line in the scheduling period; is the upper limit of the power deviation of the transmission power of the regional AC tie-line from the initial active arrangement value; is the power transmission planning value of the regional AC tie-line in the time interval ; is a time interval, which is taken as 1h here; is the upper limit of the power deviation of the transmission power of the regional AC tie-line from the initial active arrangement value; is the set of all tie-lines in the section ; is the maximum transmission power of the section .

[0177] (6) Energy storage device operation model

[0178] Energy storage devices have advantages such as energy space-time transmission and bidirectional power regulation, and are a key technology for solving the stable operation of intermittent renewable energy power. In general, electrochemical energy storage and pumped storage are the most widely used energy storage technologies in current engineering applications. Taking electrochemistry as an example, its operation standby model can be represented as:

[0179] ;

[0180] ;

[0181] ;

[0182] ;

[0183] ;

[0184] In the formula, is the charging power value of the energy storage device in the time interval ; is the discharging power value of the energy storage device in the time interval ; is the rated charging power of the energy storage device . the rated discharge power of the energy storage device ; and is a binary variable, indicating the state of the energy storage device in the time period, 1 represents the charging state, and 0 represents the non-charging state; the remaining power of the energy storage device in the time period; the charging efficiency of the energy storage device ; the discharge efficiency of the energy storage device ; is a time interval, which is taken as 1h here; the minimum value of the power of the energy storage device ; the maximum value of the power of the energy storage device ; the upward reserve capacity of the energy storage device in the time period; the downward reserve capacity of the energy storage device in the time period.

[0185] (7) Demand-side response load operation model

[0186] On the load side, demand-side response is also an important way to improve the flexibility of system operation. Demand-side response refers to a power consumption behavior in which users actively adjust their power consumption patterns according to price signals or incentive mechanisms. By starting orderly power consumption in the day-ahead dispatching plan, demand-side response loads can provide equivalent upward reserve capacity for the system during peak load periods. The operation reserve model can be characterized as:

[0187] ;

[0188] ;

[0189] ;

[0190] ;

[0191] In the formula, is the actual value of the load in the time period; is the day-ahead predicted value of the load in the time period; is the actual value of the load in The ordered electricity consumption response scale of the time period; The load In The equivalent up-regulation reserve of the time period; And The load The maximum response and minimum response proportion of the ordered electricity consumption; The load set participating in the ordered electricity consumption in the system.

[0192] In addition to the system operation reserve constraints, the resources such as conventional synchronous units, regional tie lines, etc. need to be regulated in the day-ahead scheduling stage to meet the real-time balance constraints and line capacity constraints of the power generation side and the electricity consumption side.

[0193] ;

[0194] ;

[0195] In the formula, The synchronous unit number set; The wind farm number set; The photovoltaic power station number set; The energy storage device number set; The node load number set; The UHVDC transmission number set; The regional AC tie line number set; The The active power output of the synchronous unit at the node (bus node) in the time period; The active power output of the wind farm at the node in the time period; The active power output of the photovoltaic power station at the node in the time period; The discharging active power of the energy storage at the node in the time period; The charging active power of the energy storage at the node in the time period; The active power output of the UHVDC transmission at the node in the time period; The active power output of the regional AC tie line at the node in the time period; The active power output of the load at the node in the time period; The active power output of the load at the node in the time period; The active power output of the load at the node in the time period; The active power output of the load at the node in the time period; The active power output of the load at the node is the node power transfer matrix, indicating the node injected unit active power, line active power flow change amount; is the rated transmission capacity of line .

[0196] Further, the objective function of the multi-type backup resource collaborative optimization model is:

[0197] ;

[0198] In the formula, , and respectively fuel cost, start-up cost and shutdown cost of synchronous unit in time period; is the new energy curtailment penalty coefficient; is the load shedding penalty coefficient of node ; is the new energy power station curtailment penalty coefficient; is the maximum output of new energy power station in time period; and respectively total upward reserve cost and total downward reserve cost of system in time period, the value is the sum of reserve capacity of each flexible resource subject multiplied by reserve cost; is the probability of different limit scenarios; is the load shedding amount of node in time period under scenario ; is the curtailment power of new energy station in time period under scenario .

[0199] Further, the two-stage system reserve optimization model of the present application is prepared in advance through unit commitment and reserve scheme. In the day-ahead stage, various limit reserve scenarios are fully considered, and various flexible reserve resources are called to balance the real-time power of the system, so that the system operation reserve constraints are met under the premise that the new energy curtailment and load shedding are as small as possible, thereby realizing the effective trade-off between system operation reliability and economy. The model of the present application is optimized and solved by Benders decomposition method, and the algorithm flow is as shown in Figure 2 Fig. 1, more specifically, the day-ahead pre-dispatch master problem contains the aforementioned operating constraints; the day-ahead rescheduling sub-problem further adds the system operating constraints based on limit reserve scenarios.

[0200] The multi-type reserve resource collaborative optimization model considering the accurate description of source load uncertainty can be solved by calling an existing mature commercial solver, and the reserve demand capacity of the system at each time can be obtained, and a corresponding multi-type flexible resource reserve configuration scheme can be obtained.

[0201] In Figure 3 , the system operation reserve allocation scene (Case2) of the multi-type reserve resource collaborative optimization model considering the accurate description of source load uncertainty provided by the application is taken as an example, and the system operation reserve allocation scene (Case1) of the conventional reserve resource collaborative optimization model based on the traditional deterministic reserve configuration principle (fixed proportion maximum load) is compared, as shown in Figure 4 , Figure 5 It can be seen from the above that the conventional reserve resource allocation method based on the traditional deterministic reserve configuration principle cannot respond to the dynamic changes of the reserve demand, the redundant reserve configuration in some periods leads to the increase of the synchronous unit operation cost, and the insufficient reserve configuration in some periods leads to the occurrence of large-scale new energy curtailment; the reserve resource allocation method of the application effectively responds to the dynamic changes of the reserve demand, and greatly reduces the wind and light curtailment; in addition, it can be seen from Table 1 that the application reduces the total system cost by 7.09%, reduces the synchronous unit operation cost by 3.32%, and reduces the synchronous unit start-up capacity by 11.42%, which indicates that compared with the conventional reserve capacity configuration method based on the traditional deterministic reserve configuration principle, the application avoids the redundant start-up capacity of the synchronous unit through the accurate quantitative evaluation of the system reserve demand, and effectively improves the dispatching and operation economy of the system.

[0202] Table 1 System simulation running results under different scenes

[0203]

[0204] Referring to Figure 6 , the application further provides a multi-type reserve resource collaborative optimization device considering source load uncertainty, which is used to realize the multi-type reserve resource collaborative optimization method considering source load uncertainty, and the device comprises:

[0205] A reserve constraint acquisition module is configured to analyze the prediction error distribution characteristics of the new energy and the load, generate a plurality of new energy and load prediction error sub-databases based on data clustering, generate a rhombus convex hull uncertainty set based on the sub-databases, extract the top points of the set to generate a reserve demand curve set of the limit scene, and generate system operation reserve constraints according to the reserve demand capacity of the system at each time.

[0206] A model construction module is configured to construct a multi-type backup resource collaborative optimization model and a multi-type backup resource scheduling operation model, the multi-type backup resource collaborative optimization model taking the minimum total cost as an objective and taking system operation backup constraints as constraint conditions, the total cost including a day-ahead stage cost and a day-ahead stage cost, the day-ahead stage cost including a generation cost, a new energy curtailment cost and a backup cost, and the day-ahead stage cost including new energy curtailment and load shedding penalty under each backup limit scenario.

[0207] A model solution module is configured to solve the multi-type backup resource collaborative optimization model to obtain backup demand capacity of the system at each time and a corresponding multi-type backup resource configuration scheme.

[0208] Referring to Figure 7 The application further provides a multi-type backup resource collaborative optimization device considering source-load uncertainty, including a memory and a processor.

[0209] The memory is configured to store computer program codes and transmit the computer program codes to the processor.

[0210] The processor is configured to execute the method for multi-type backup resource collaborative optimization considering source-load uncertainty according to instructions in the computer program codes.

[0211] The application further provides a computer readable storage medium, which stores computer programs, and the computer programs are executed by the processor to realize the method for multi-type backup resource collaborative optimization considering source-load uncertainty.

[0212] Generally, computer instructions used to realize the method of the application can be carried by any combination of one or more computer readable storage media. The non-transitory computer readable storage medium can include any computer readable medium except a signal in the process of transmission.

[0213] The computer readable storage medium can be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination of the above. More specific examples (non-exhaustive list) of the computer readable storage medium include an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EKROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, the computer readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device or component.

[0214] Computer program code for carrying out operations of the present application can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages, and specifically Python language and platform frameworks based on TensorFlow, PyTorch, etc. suitable for neural network computing. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0215] The above apparatus and non-transitory computer readable storage medium can refer to the specific description of a multi-type backup resource collaborative optimization method considering source load uncertainty and its beneficial effects, which will not be described here.

[0216] Although the embodiments of the present application have been shown and described above, it should be understood by those skilled in the art that the above embodiments are exemplary and cannot be interpreted as a limitation of the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A method for multi-type reserve resource collaborative optimization considering source load uncertainty, characterized in that, include: The distribution characteristics of prediction errors for new energy sources and loads are analyzed, and multiple sub-databases of prediction errors for new energy sources and loads are generated based on data clustering. Based on the sub-database, a rhomboid convex hull uncertainty set is generated. The vertices of the set are extracted to generate a set of backup demand curves for extreme scenarios. Based on the backup demand capacity of the system at each time point, backup constraints for system operation are generated. A collaborative optimization model for multiple types of reserve resources and a scheduling and operation model for multiple types of reserve resources are constructed. The collaborative optimization model for multiple types of reserve resources aims to minimize the total cost and uses system operation reserve constraints as constraints. The total cost includes day-ahead stage cost and intraday stage cost. The day-ahead stage cost includes generation cost, renewable energy curtailment cost and reserve cost. The intraday stage cost includes renewable energy curtailment and load shedding penalties under various reserve extreme scenarios. Solve the multi-type backup resource collaborative optimization model to obtain the backup capacity requirement of the system at each time point, as well as the corresponding configuration scheme of the multi-type backup resources.

2. The method of claim 1, wherein, The method for generating the new energy and load forecasting error sub-database is as follows: The new energy forecast error data are arranged in order of magnitude based on the day-ahead forecast power output of new energy sources, and selected sequentially. A sub-database of new energy prediction errors is composed of several samples; among them... This represents the sample size for the current daytime forecast of new energy sources. This represents the number of sample groups for the day-ahead forecast of new energy sources. The K-means clustering algorithm is used to obtain monthly groups of load forecast errors before the day of the day. Based on the load forecast values ​​before the day of the day, K-means clustering is performed on the monthly groups to obtain a sub-database of load forecast errors.

3. The method of claim 1, wherein, The method for obtaining the uncertain set of the rhombic convex hull includes: The collected historical scenes are denoted as wherein, is the number of collected historical scenes; Solve for a high-dimensional ellipsoid to encompass all historical scenes, and then solve for the parameters of the high-dimensional ellipsoid to address the following optimization problem: ； wherein is a constant, representing the volume of the unit sphere in dimension is a positive definite matrix; is a translation parameter; The convex set surrounded by the original high-dimensional ellipsoid of The convex set surrounded by the original high-dimensional ellipsoid​ ； ； wherein is a diagonal matrix; and is a diagonal matrix eigenvalues of is a transformation matrix; The original high-dimensional ellipsoid After rotation and translation, the mathematical expression of the high-dimensional ellipsoid is as follows: ； wherein is the high dimensional ellipsoid equation after rotation and translation; is a random variable is the coordinate value in the rotated coordinate; is an n*n matrix; High-dimensional ellipsoid The vertex coordinates are: ； In the formula, is a high-dimensional ellipsoid coordinate values of the first vertex; is the number of vertices, ; The original high-dimensional ellipsoid can be obtained by using the inverse transformation of the above equation. The vertex equation is derived from the original high-dimensional ellipsoid. The equation of the rhombus convex hull formed by the vertices is shown below: ； In the formula, is a rhombus convex hull equation surrounded by high-dimensional ellipsoid vertices; is an element of the transformation matrix is an element of the transformation matrix Introducing a magnification factor to enlarge and reduce the rhomboid convex hull, the corrected uncertain set is: ； ； wherein, is a rhombus hull uncertainty set; is a limit scenario obtained after hull scaling; is a scaling factor in hull scaling.

4. The method of claim 3, wherein, The method for calculating the magnification factor in the convex hull scaling is as follows: Collecting historical scenarios Rotating and translating to new historical scenario collection ; According to the convex optimization theory, if the new historical scenario can be covered by a rhombic convex hull with the high-dimensional ellipsoid vertex, the following formula is established: ； ； wherein is a positive coefficient; for Given several historical scenarios, we establish the following optimization problem to determine the inclusion relationship between the historical scenarios and the convex hull: ； ； wherein is the first amplification factor; is a positive coefficient of the amplification factor ; The magnification factor is obtained by the above inclusion relationship The sequence is formed by the magnification factors, and the maximum value in the sequence is the magnification factor in the convex hull scaling .

5. The method of claim 1, wherein, The system's operational standby constraint is: ； ； wherein and are up and down reserve capacities of the period system; and are up and down reserve capacities of each type of reserve resource, respectively; and are the proportionality coefficient matrices of up and down reserve capacities of each type of reserve resource, respectively.

6. The method of claim 1, wherein, The objective function of the multi-type backup resource collaborative optimization model is: ； In the formula, , and are synchronous units The fuel cost, start-up cost and shutdown cost of the unit in the time period ; is the new energy curtailment penalty coefficient; is the load shedding penalty coefficient of node ; is the curtailment penalty coefficient of the new energy power plant ; is the maximum output of the new energy power plant in the time period ; and are the total upward reserve cost and the total downward reserve cost of the system in the time period ; is the probability of different limit scenarios; is the load shedding amount of node in the time period under scenario ; is the curtailment power of the new energy power station in the time period under scenario .

7. The method of claim 1, wherein, The scheduling and operation models for the various types of backup resources include synchronous generator operation models, wind farm operation models, photovoltaic power plant operation models, ultra-high voltage direct current transmission line operation models, regional AC tie line operation models, energy storage device operation models, and demand-side response load operation models.

8. A multi-type reserve resource collaborative optimization device considering source load uncertainty, characterized in that, The apparatus is used to implement the method according to any one of claims 1-7, the apparatus comprising: The backup constraint acquisition module is used to analyze the distribution characteristics of prediction errors of new energy sources and loads, generate multiple new energy and load prediction error sub-databases based on data clustering, generate a rhomboid convex hull uncertainty set based on the sub-databases, extract the vertices of the set to generate a set of backup demand curves for extreme scenarios, and generate system operation backup constraints based on the backup demand capacity of the system at each time point. The model building module is used to build a collaborative optimization model for multiple types of reserve resources and a scheduling and operation model for multiple types of reserve resources. The collaborative optimization model for multiple types of reserve resources aims to minimize the total cost and uses system operation reserve constraints as constraints. The total cost includes day-ahead stage cost and intraday stage cost. The day-ahead stage cost includes generation cost, renewable energy curtailment cost and reserve cost. The intraday stage cost includes renewable energy curtailment and load shedding penalties under various reserve extreme scenarios. A model solving module is configured to solve the multi-type reserve resource collaborative optimization model to obtain a reserve demand capacity of the system at each time and a corresponding configuration scheme of the multi-type reserve resource. 9.A multi-type reserve resource collaborative optimization device considering source load uncertainty, comprising: a memory and a processor; the memory is configured to store computer program codes and transmit the computer program codes to the processor; the processor is configured to execute the method according to any one of claims 1 to 7 according to instructions in the computer program codes.

10. A computer-readable storage medium, characterized in that, The computer program is stored on the computer readable storage medium, and the computer program is executed by the processor to implement the method according to any one of claims 1 to 7.

Images