CN117713236A - Flexible resource optimization scheduling method based on multiple time scales - Google Patents

Abstract

The invention discloses a flexible resource optimization scheduling method based on multiple time scales, which comprises the following steps: acquiring flexible resource composition and parameters of a power system to be analyzed, and dividing different time scales according to the flexibility of scheduling of various resources; quantifying the flexible resources under different time scales, and determining corresponding constraint of flexible resource scheduling of the power system according to rated parameters and operation characteristics of various flexible resources; respectively establishing a day-ahead scheduling model taking the minimum total running cost as an objective function, a day-in rolling model taking the minimum running cost as the objective function and a real-time scheduling model taking the minimum adjusting cost as the objective function; solving the three-level model, making a scheduling plan of corresponding flexible resources, and realizing regulation and control of a target power system. According to the invention, the three-level scheduling model is established and solved under the constraint of model operation, and scheduling optimization of flexible resources of the power system is realized by creative design of the three-level scheduling model.

Description

Flexible resource optimization scheduling method based on multiple time scales

Technical Field

The invention relates to the technical field of power system flexible resource optimization scheduling, in particular to a flexible resource optimization scheduling method based on multiple time scales.

Background

In recent years, the development and utilization of renewable energy sources in China are brought into the national energy source development strategy, and with the installation and application of large-scale wind power and photovoltaic, the renewable energy sources in China are greatly developed.

Because of randomness and volatility of wind power and photovoltaic, large-scale flexible resources are needed to ensure safe and stable operation of a power grid, the capacity of absorbing new energy is improved, and the problems of insufficient coordination capacity, low output efficiency, insufficient flexibility and the like among links of a source-grid-charge-storage system are more serious along with the increase of the power generation proportion of the new energy in a power system, so that the optimization solution is needed, and the method has important significance for optimizing, dispatching and managing multiple types of flexible resources, improving the flexibility of the system and realizing safe, stable and reliable operation of the novel power system. Under the background, aiming at the problem of insufficient flexibility of the power system, flexible resources in all aspects of source-network-load-storage are required to be developed, coordinated and optimally scheduled, the output of the coordinated scheduling is enabled to stabilize the fluctuation generated on both sides of the source-load, and meanwhile, the optimal scheduling of the flexible resources is required to be carried out in a time scale because the flexible supply of all kinds of resources has the characteristic of multiple time scales. With the increase of source-network-load-storage various flexible resource capacities of novel power systems in recent years. At present, how to optimally schedule various flexible resource time-sharing scales and establish what scheduling optimization system is a key problem to be solved urgently.

Disclosure of Invention

The embodiment of the invention provides a flexible resource optimization scheduling method based on multiple time scales, which performs scheduling optimization on flexible resource allocation of a power system according to the time scales, so that coordination capacity of each link of source-network-load-storage is improved, and flexibility and stability of the power system are improved.

In a first aspect of the present invention, a method for optimizing and scheduling resources based on multi-time scale flexibility is provided, including:

acquiring flexible resource composition and parameters of a power system to be analyzed, and dividing different time scales according to the flexibility of scheduling of various resources;

quantifying flexible resources under different time scales, and determining corresponding constraint of flexible resource scheduling of the power system according to rated parameters and characteristics of various flexible resources as various resource scheduling constraint conditions;

respectively establishing a day-ahead scheduling model taking the minimum total running cost as an objective function, a day-in rolling model taking the minimum running cost as the objective function and a real-time scheduling model taking the minimum adjusting cost as the objective function based on various resource scheduling constraint conditions to form a three-level model;

solving the three-level model, and according to the output result of the three-level model, making a scheduling plan of the corresponding flexible resource, and realizing the regulation and control of the target power system.

Preferably, the three-level model is solved by the following method:

establishing a gray wolf optimization model, wherein the solution of a day-ahead dispatch model corresponds to a common wolf, the constraint condition and the objective function of the day-ahead dispatch model are used as constraint conditions and objective functions of the common wolf, the solution of a day-ahead scroll model corresponds to a middle-layer wolf, the constraint condition and the objective function of the day-ahead scroll model are used as constraint conditions and objective functions of the middle-layer wolf, the solution of a real-time dispatch model corresponds to a wolf king, and the constraint condition and the objective function of the real-time dispatch model are used as constraint conditions and objective functions of the wolf king; solving the gray wolf optimization model, and taking the solution of the gray wolf optimization model as the solution of the three-level model; the method specifically comprises the following steps:

encoding each scheduling model according to the wolf group level, wherein:

1 wolf Wang Bianma represents a scheduling scheme of a real-time scheduling model;

2, the middle layer wolf code represents a scheduling scheme of an intra-day rolling model;

the 3 common wolf code represents the scheduling scheme of the day-ahead scheduling model.

3. Population initialization

And randomly initializing the codes according to the wolf group level proportion to generate an initial group.

4. Fitness calculation

The total fitness of the wolf group is the sum of the fitness of the wolves of each layer, wherein:

1 wolf Wang Shi is the real-time scheduling cost;

The adaptation of the middle layer wolves is the daily rolling cost;

the 3 common wolf fitness is the day-ahead scheduling cost.

5. Group update

The position of the 1 wolf king is relatively stable, and the middle wolf and the common wolf are subjected to position update;

2, the middle layer wolves move around the wolves to drive the positions of the common wolves to be updated;

and 3, when the positions of the wolves are changed, clustering and moving the whole wolves.

6. Termination condition

The set maximum number of iterations is reached.

7. Solution implementation

Parameter setting (one)

And setting parameters such as group size, iteration number and the like.

Wolf group initialization

And randomly generating wolves of each level according to the set proportion.

(III) evaluation of fitness

And calculating the overall fitness of the current wolf group.

(IV) population update

And updating the positions of the wolves by simulating the cooperative coordination of the wolves hierarchical structure.

(V) termination judgment

And judging whether a termination condition is satisfied.

Analysis of results

And decoding the positions of the wolf, the middle wolf and the common wolf, and analyzing the scheduling results of each model.

8. Characteristics of

1-class wolf cluster movement;

2 attach importance to the guiding status of the radix euphorbiae Fischerianae (real-time layer);

the middle and lower wolves in 3 are greatly affected by the higher wolves.

9. Further optimize

Designing a dynamic adjustment factor to enhance the searching capability;

secondly, the convergence accuracy is improved by adopting a local search technique;

And (III) combining other algorithms to perform mixed operation.

And (5) iteratively obtaining the optimal solution of the three-level scheduling model by simulating the wolf's group level behavior. The strategy fully utilizes the social characteristics of the wolf group, meets the layering requirements of the scheduling optimization problem, and has high calculation efficiency and higher optimization degree of the obtained final result.

In this way, the embodiment of the invention can reduce the cost of operation and adjustment of the flexible resources and improve the consumption level of new energy and the economy of scheduling each resource of the power system by adjusting and controlling the flexible resources of the power system in different time scales.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the description of the embodiments of the present invention will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a flow chart of steps of a flexible resource optimization scheduling method based on multiple time scales according to an embodiment of the present invention;

FIG. 2 is a flexible resource classification diagram of a flexible resource optimization scheduling method based on multiple time scales according to an embodiment of the present invention;

FIG. 3 is a class label diagram of a flexible resource scheduling database of a flexible resource optimization scheduling method based on multiple time scales according to an embodiment of the present invention;

FIG. 4 is a diagram of an operational framework of three-level model multi-time scale scheduling for a multi-time scale based flexible resource optimization scheduling method according to an embodiment of the present invention;

FIG. 5 is a topology diagram of a simulated node system of a flexible resource optimization scheduling method based on multiple time scales according to an embodiment of the present invention;

FIG. 6 is a scheduling diagram of various flexible resources in a scenario of a flexible resource optimization scheduling method based on multiple time scales according to an embodiment of the present invention;

FIG. 7 is a scheduling diagram of various flexible resources in scenario two of a flexible resource optimization scheduling method based on multiple time scales according to an embodiment of the present invention;

fig. 8 is a load diagram of each period of a power system based on a flexible resource optimization scheduling method of multiple time scales according to an embodiment of the present invention.

Detailed Description

The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are some, but not all embodiments of the invention. All other embodiments, which are derived by a person skilled in the art from the embodiments according to the invention without creative efforts, fall within the protection scope of the invention.

The embodiment of the invention discloses a resource optimization scheduling method and a system based on multi-time scale flexibility, as shown in fig. 1, the method comprises the following steps:

step S101: and acquiring the flexible resource composition and parameters of the power system to be analyzed, and dividing different time scales according to the flexibility of scheduling of various resources.

The embodiment of the invention mainly takes flexible resources of different sources at each side of a power system source-network-load-storage to be analyzed as analysis objects, and it is understood that the flexible resources of the power system refer to resources which can increase the flexibility, elasticity and flexibility of the power system and improve the stability of the power system and serve the dynamic supply and demand balance of the power system. It should be appreciated that flexible feeds of various types of resources have the property of multiple time scales, i.e. having different adjustment capabilities at different time scales, the different time scales being divided into categories based on the time properties of the various types of flexible resources, i.e. the adjustability at different time scales.

Specifically, the typical flexible resource composition and parameters thereof for obtaining the power system to be analyzed include the following:

(1) And acquiring relevant parameters of a conventional unit of the power system to be analyzed.

It should be understood that the conventional unit provided by the invention mainly comprises a flexible thermal power unit and an adjustable hydroelectric unit, and belongs to flexible resources on the power supply side.

Specifically, the obtained related parameters of the conventional unit include:

upper limit P of output power of conventional unit _i ^G-max Lower limit P _i ^G-min ；

Climbing rate P during operation of conventional unit _i ^R ；

Obtaining the minimum duration T of the start-stop of each conventional unit _c ；

Acquiring single starting cost of each conventional unitAnd single shut-down cost->

(2) And acquiring relevant parameters of an energy storage unit of the power system to be analyzed.

It should be understood that the energy storage unit provided by the invention mainly comprises an electrochemical energy storage unit and a pumped storage power station, and belongs to flexible resources of an energy storage side.

Obtaining capacity rating E of each electrochemical energy storage unit ^B ；

Acquiring the SOC upper limit S of an electrochemical unit _max And a lower SOC limit S _min Rated state of charge S at end time _n ；

Obtaining the maximum and minimum charge and discharge power P of an electrochemical energy storage unit ^B-max And P ^B-min Charging efficiencyDischarge efficiency->

Obtaining the reservoir maximum water storage capacity V of a pumped storage power station _i ^W-max And minimum water storage volume V _i ^W-min ；

Obtaining maximum and minimum output power P of pumping water of pumped storage power station ^W-max And P ^W-min Conversion efficiency of water quantity and electricity quantity when pumped and discharged by pumped storage power station And->

(3) And acquiring relevant parameters of the load side demand response resource of the power system to be analyzed.

It should be understood that the demand response resources proposed by the present invention mainly include PDR price type demand response load and IDR incentive type demand response load, and belong to load side flexible resources.

Obtaining maximum and minimum call volumes of price type demand response loadIs->

Obtaining maximum increased load amount of excitation type demand response loadMaximum reduced load->

(4) And acquiring relevant parameters of a new energy unit of the electric power system to be analyzed.

It should be understood that the new energy unit provided by the invention mainly refers to a distributed wind generating set and a photovoltaic generating set at the power supply side.

Obtaining upper limit P of output power of new energy unit _i ^N-max Lower limit P _i ^N-min 。

Specifically, according to the quota parameters of flexible resource adjustment of each side of the source network load storage, the flexibility of various resource conditions is calculated, so that different time scales are divided according to the flexibility, and the method comprises the following steps:

it should be understood that, due to different factors affecting the flexibility of source network load storage for various flexible resource adjustment, the time scale division should be performed from each side of resource classification:

specifically, the power source side resources are mainly various conventional units, and the time scale division is mainly influenced by the climbing speed and the starting and stopping time. Under the condition of shorter time scale, the flexibility of power supply side flexible resource adjustment is mainly limited by the climbing speed, and when the time scale is increased, the flexibility of downward adjustment can only be prolonged by closing and stopping the machine set due to the limitation of the upper and lower output limits of the conventional machine set.

Specifically, when the scheduling of the short time scale is participated, the scheduling time calculation formula of the power supply side flexible resource is as follows:

ΔT＝(P _i ^G-max -P _i ^G-min )/P _i ^R

specifically, when the long time scale scheduling is participated, the scheduling time calculation formula of the power supply side flexible resource is as follows:

ΔT＝(P _i ^G-max -P _i ^G-min )/P _i ^R +T _c

specifically, the flexible resources of the energy storage side are mainly various energy storage power stations, and the scheduling flexibility of the energy storage side is mainly influenced by the charge and discharge power. For electrochemical energy storage, the time scale division mainly considers the charge and discharge power and the charge and discharge efficiency, and for pumped storage power stations, the water pumping and discharge power and the water quantity and electricity conversion efficiency are mainly considered.

Specifically, the calculation formula of the scheduling time of the electrochemical energy storage power station is as follows:

specifically, the calculation formula of the scheduling time of the pumped storage power station is as follows:

specifically, the load side flexible resources are mainly various demand response loads, including a PDR price type demand response resource and an IDR incentive type demand response resource. For price type demand response resources, the time scale of the price type demand response resources is related to the pricing mode of electricity price, and in general, the pricing mode of electricity price adopts a day-ahead pricing mode, so that PDR resources need to be determined in day-ahead scheduling, and the price type demand response resources are suitable for long-time scheduling. For the excitation type demand response resource, the length of the power grid dispatching instruction time affecting the time scale division can be generally divided into the following time scales: 1) IDR resources which need to be planned in advance one day; 2) IDR resources with response time of 15min-2 h; 3) IDR resources with response time of 5-15 min; 4) IDR resources that respond in real-time.

The time scale of the adjustment of each flexible resource is obtained by carrying out the processing on each flexible resource participating in the power system, and the obtained time scale is clustered, so that a reasonable scheduling time scale range is obtained.

It should be noted that in order to avoid that the time characteristics of a certain class of flexible resources dominate the clustering process, the adjustment time of the various classes of flexible resources needs to be standardized.

Specifically, the normalization processing of the adjustment time is implemented by adopting a Z-score normalization processing mode through a zscore function in MATLAB.

Specifically, the calculation formula of the Z-score standardized mode is as follows:

where x represents the eigenvalue, μ represents the mean of the feature, and σ represents the standard deviation of the feature.

Because the obtained adjustment time is in linear distribution, the time scales of various flexible resources are clustered by adopting a K-means clustering method after standardization, and different time scales of various flexible resource scheduling are obtained.

Specifically, K-means clustering for various flexible resource adjustment times can be realized through a kmeans function of MATLAB.

Specifically, the K-means clustering comprises the following steps:

1) Randomly selecting k objects from the input flexible resource adjustment time as an initial clustering center;

2) Respectively calculating the distance from each sample point to each cluster center, and distributing the distances to clusters closest to each sample point one by one;

specifically, the distance between the sample point and the clustering center is calculated by using Euclidean distance, and the calculation formula is as follows:

wherein C is _i For the ith cluster center, m is the dimension of the data object. X is x _j And C _ij Is x and C _i Is the j-th attribute value of (c).

3) After all samples are distributed, updating the class center position, wherein the class center is defined as the average value of all objects in the cluster in each dimension;

4) Comparing the obtained clustering center with the clustering center obtained by the previous calculation, if the clustering center changes, turning to a second step, and repeating the steps;

5) And stopping and outputting the clustering result when the class center is not changed any more, and sorting the clustering information.

According to the steps of time scale division and distance, finally dividing the time scale of flexible resource scheduling into short time scales of 0-15min, time scales of 15-60min and long time scales of >1h, wherein each time scale mainly comprises the following flexible resources:

(1) Short time scale of 0-15min

The power supply side comprises a part of flexible thermal power generating units and an adjustable hydropower station, wherein the adjusting capacity influence factors of the thermal power generating units under the short time scale are mainly climbing capacity of the units. The adjustable hydropower station can participate in flexibility adjustment under each time scale, but the unit is mainly influenced by water storage capacity, so that auxiliary services such as peak shaving and the like provided by the hydropower unit are influenced by seasons.

The flexible resource with the rapid adjustment capability on the energy storage side is electrochemical energy storage, and is suitable for adjustment work with high frequency and small fluctuation within 15 min.

The flexible resource with the quick response capability at the load side is an incentive type demand response (IDR), which refers to a preferential policy formulated by a demand response implementation organization to motivate users to respond to a dispatch signal, and mainly comprises direct load control, interruptible load, demand side bidding and emergency demand response. IDR can participate in a broad range of minute to hour responses.

(2) 15-60min medium time scale

Partial thermal power units, such as circulating gas units, cannot participate in scheduling in a short time scale because the climbing rate is relatively slow in comparison with a quick scheduling unit, and are suitable for flexible scheduling in a medium-long time scale.

The compressed air energy storage mode at the energy storage side is suitable for stabilizing power fluctuation with the time scale of tens of minutes due to the slightly slow response speed, and is suitable for flexible scheduling with the medium time scale.

(3) 1h long time scale

On a long time scale, the power supply side can provide flexibility regulation capability and further comprises a coal-fired unit in the thermal power unit, the climbing rate of the coal-fired unit after the flexibility transformation can reach 2% -3% of the limit capacity per minute, and the power supply side can be used for flexibility regulation under the long time scale.

The pumped storage on the energy storage side does not have rapid adjustment capability relative to other energy storage modes, but can participate in providing flexibility adjustment capability on a long time scale.

Price-type demand response on the load side, i.e. PDR, can provide flexibility in adjusting capacity on a long time scale, and price-sensitive power consumers often need more than several hours to react to changes in electricity consumption behavior according to electricity prices. PDR often includes time-of-use power rate responses, real-time power rate responses, and peak power rate responses.

Specifically, different time scales of various flexible resources are divided according to the flexibility adjusting capability as shown in fig. 2.

The basic rated parameters of the relevant units of the electric power system to be analyzed can be obtained through the step, various flexible resources are divided according to a certain time scale, various constraints of scheduling of various flexible resources of the electric power system can be determined according to the parameters obtained through the step, the flexible resources can be classified according to the divided time scale, and a day before day, a day in day and a real-time scheduling model can be respectively established.

Step S102: quantifying the flexible resources under different time scales, and determining corresponding constraint of flexible resource scheduling of the power system according to rated parameters and operation characteristics of various flexible resources;

Specifically, various flexible resources and unit quantization of the power system and determining corresponding scheduling constraints include the following:

(1) And obtaining flexible resource scheduling and optimizing the operation constraint of the conventional unit according to the parameter data of the conventional unit at the power supply side such as the thermal power unit, the adjustable hydropower station and the like.

Specifically, the conventional unit operation constraint comprises a conventional unit output constraint, a conventional unit climbing constraint and a conventional unit start-stop constraint.

Specifically, the conventional unit output constraint is:

wherein,and the output power of the ith conventional unit at the moment t is represented.

It should be understood that the conventional unit climbing constraints include a unit climbing rate constraint and a unit start-stop climbing rate constraint;

specifically, the climbing constraint of the conventional unit is as follows:

wherein,the start-stop state variable of the ith unit at the moment t is 0 or 1, and P _i ^R-U The method is used for representing the starting and stopping climbing speed of the unit, and the calculation formula is as follows:

P _i ^R-U ＝(P _i ^G-max +P _i ^G-min )/3 (5)

it should be appreciated that conventional unit start-stop constraints include unit start-stop time constraints and unit start-stop cost constraints.

Specifically, the conventional unit startup time constraint should be:

wherein,the starting variable of the conventional unit at the moment T is represented, T represents any time period of starting and stopping when the system is operated, and the calculation formula is as follows:

T＝t:min{96,t+T _c -1} (7)

Specifically, the conventional unit downtime constraints should be:

wherein,and the shutdown variable of the conventional unit at the time T is represented.

Specifically, the conventional unit of the system should also consider the cost constraint of starting up and stopping the machine:

wherein,and->The current starting cost and the shutdown cost of the conventional unit t of the system are respectively represented.And->The initial start-up cost and the shutdown cost of the conventional unit of the system are respectively represented.

(2) And according to various parameters of flexible resources of the energy storage side such as electrochemical energy storage, pumped storage and the like, obtaining flexible resource scheduling and optimizing the operation constraint of the energy storage power station.

In particular, the operational constraints of the energy storage power station include electrochemical energy storage power station operational constraints and pumped storage power station operational constraints.

It should be appreciated that the electrochemical energy storage power plant operational constraints include energy storage capacity constraints, energy storage state of charge constraints, and charge-discharge power constraints.

Specifically, the energy storage capacity constraints of the electrochemical energy storage power station are:

in particular, the method comprises the steps of,the energy storage energy of the electrochemical energy storage unit at the ith time t is represented by the following calculation formula:

wherein,the charging power of the ith electrochemical energy storage unit at the t moment is represented; />And the discharge power of the ith electrochemical energy storage unit at the moment t is shown.

Specifically, the energy storage state of charge constraints of the electrochemical energy storage power station are:

S _min ≤S _i,t ≤S _max (15)

Wherein S is _t The calculation formula of the energy storage SOC value at the time t is as follows:

at the same time, the energy storage SOC value S at the time of T end _T Should reach a specified value S _n So that the system can be started and operated normally the next day;

specifically, the state of charge constraint at the end of energy storage should be:

therefore, the real-time energy storage capacity of the energy storage power station should satisfy the constraint:

specifically, the charge-discharge power constraint of the electrochemical energy storage power station is as follows:

wherein,the output power of the ith electrochemical energy storage unit at the t moment is represented by the following calculation formula:

meanwhile, the constraint of the charge and discharge power of the energy storage power station also meets the constraint of the charge and discharge state variable of the energy storage power station:

wherein,and->And the charge and discharge state variables of the electrochemical energy storage unit are respectively represented, and the value is 0 or 1.

Meanwhile, the electrochemical energy storage unit can not simultaneously charge and discharge, and the specific constraints are as follows:

it should be appreciated that the operational constraints of pumped storage plants include reservoir capacity constraints and pumped water power constraints.

Specifically, the reservoir water storage capacity constraint of the pumped storage power station is as follows:

wherein,the water storage capacity of the ith reservoir at the moment t is represented by the following specific calculation formula:

wherein P is _i ^Wc 、P _i ^Wd Respectively representing the pumping and discharging output power of the ith pumped storage power station at the moment t.

Specifically, the water pumping output power constraint of the pumped storage power station is as follows:

wherein,and->And the pumping and discharging start-stop state variables of the ith pumped storage power station at the t moment are respectively represented, and the value is 0 or 1./>And (5) representing the generated power of the ith pumped storage power station at the moment t.

Meanwhile, the pumped storage power station should meet the requirement that the pumping, water charging and discharging processes cannot be carried out at the same time, and the specific constraints are as follows:

(3) And obtaining flexible resource scheduling and optimizing load side demand response constraint according to various parameters of the load side flexible resources such as the incentive type demand response, the price type demand response and the like.

It should be appreciated that the load side demand response constraints include price type demand response constraints, incentive type demand response constraints.

Specifically, the price type demand response constraint is:

wherein P is _PDR,t The amount of modulation of the PDR load at time t is indicated.

Specifically, the incentive type demand response constraint is:

wherein,and->The increase and decrease in IDR load are indicated, respectively. P (P) _IDR,t The amount of the IDR load call at time t is indicated.

(4) And according to the related parameters of the new energy unit integrated by the power system, obtaining the flexible resource scheduling and optimizing the output constraint of the new energy unit.

Wherein,and the predicted power of the ith new energy unit at the moment t is shown. / >And the actual output power of the ith new energy unit at the t moment is shown.

(5) The power system should also meet line flow constraints between nodes when operating.

Specifically, a line power flow P is generated at a time t between a node a and a node b _ab,t The calculation formula is as follows:

P _ab,t ＝γ(A ^G ·P _t ^G +A ^B ·P _t ^B -P _t ^L ) (36)

wherein, gamma represents a power transfer distribution factor, A ^G And A is a ^B And respectively representing the incidence matrixes of the conventional units and the nodes and the incidence matrixes of the energy storage units and the nodes.

Therefore, the line flow constraint among the nodes of the system should be:

(6) According to the output power and the real-time load of each unit of the system, the power load balance constraint of the system can be met at any moment when the system operates.

Wherein P is _t ^L-pre Representing a predicted value of the load on the user side, P _t ^L-F The loss of power of the load at time t is indicated.

(7) According to different calling scenes of various flexible resources, the scheduling of the flexible resources of the power system also needs to consider the adjustment constraint of each scene.

Wherein,is->Respectively representing the power output of a standard scene of a conventional unit and an energy storage unit; />Is->The flexible adjustment capability of the conventional unit and the energy storage unit are respectively shown.

Step S103: based on the relevant parameters of various flexible resources of the obtained power system and the corresponding scheduling constraint conditions, a flexible resource scheduling database is established, and recognition, storage and calling of various flexible resource inputs are realized.

Specifically, the identification of the flexible resource input includes the following:

1) Identifying sources of flexible resources

Specifically, the sources of the flexible resources include a power source side, a load side and an energy storage side, and the identification of the flexible resources on each side is mainly carried out by judging whether the characteristic parameters of the resources on each side exist or not.

Specifically, the characteristic parameters of the flexible resource at the power supply side are mainly the climbing speed and the start-stop variable; the characteristic parameters of the load side are mainly demand response time; the characteristic parameters of the energy storage side are charge and discharge power and charge and discharge efficiency.

2) Identifying time scales for flexible resources

Specifically, the time scale identification mode of the flexible resource is as follows:

according to the characteristic parameters of the flexible resources, the adjustment time of the input flexible resources is calculated according to the calculation formula of the adjustment time of the flexible resources at each side in the step S101.

Clustering the adjustment time of various flexible resources according to the time scale division mode in the step S101, and dividing the corresponding time scale range.

And the input related parameters of the flexible resources are identified, so that the corresponding parameters and constraints are stored in the corresponding storage modules of the database, and the classified scheduling is convenient.

Specifically, the flexible resource scheduling database is stored in a hierarchical storage mode, and hierarchical content comprises the following parts:

it should be understood that the flexible resource scheduling database not only includes relevant parameters of flexible resources on each side of the source load store, but also includes predicted values of the output of the new energy unit and predicted values of the load of the access users of the power system.

Specifically, the predicted values of the output and the load of the new energy unit can be predicted by a future, intra-day and real-time prediction model to obtain the predicted values of the medium-term, short-term and ultra-short term respectively, and the predicted values are summarized into a predicted data set to be recorded in a flexible resource scheduling database.

Specifically, the hierarchical storage of the flexible resource scheduling database is divided into the following four-level labels:

1) Primary label

The primary labels are mainly divided into types of database resources according to the following steps: flexible resources, new energy unit output predicted values and user load predicted values.

2) Two-stage label

The secondary label is mainly divided into a source of flexible resources and mainly comprises a power supply side, an energy storage side and a load side.

3) Three-level label

The three-level label is mainly divided into time scales according to scheduling, and mainly comprises the day before, the day in and the real time.

4) Four-level label

The four-level label is mainly a specific resource type and a predicted value, and is correspondingly stored with basic parameters and corresponding constraints.

Specifically, the classification labels stored in the flexible resource scheduling database are shown in fig. 3.

Specifically, the calling mode of the flexible resource scheduling database comprises the following characteristics:

it should be understood that, according to the power balance mainly scheduled by the power system, the data is mainly called for stabilizing deviation fluctuation of the new energy output predicted value, the user load predicted value and the actual value, and meanwhile, economy of scheduling various flexible resources and efficiency and precision of scheduling are considered, so that scheduling planning needs to be conducted according to requirements under different time scales.

Specifically, the calling function of the database should also satisfy the function of automatically searching labels by the system, and considering the complementary relation between various flexible resources and loads, under the condition of optimal economy, part or all of flexible resources of different types under different time scales are selected to participate in the scheduling so as to satisfy the most reasonable calling mode.

Specifically, the recognition, storage and calling functions of the database can be realized by setting various computer modules respectively.

Specifically, the identification function of the database is realized through a data acquisition module:

it should be appreciated that the data acquisition module should accomplish the collection and identification of data such as flexible resource related parameters, new energy unit output prediction data sets, user load prediction data sets, and the like.

Further dividing, the data acquisition module should include an acquisition subunit and an identification subunit;

specifically, the obtaining subunit is configured to collect the flexible resource basic parameter, the output of the new energy unit, and the user load prediction data set in step S101;

specifically, the identification subunit is configured to identify and classify the collected data, identify the characteristic parameters of various resource data by using the time scale method in step S101, and automatically divide the characteristic parameters into classes.

Specifically, the storage function of the database is realized through a database construction module:

it should be understood that the database construction module should complete the creation of the flexible resource scheduling database and the tag-wise classification and storage of the data collected by the data acquisition module, and update the database in real time according to the input.

Further partitioning, the database construction module should include a build sub-module and a process sub-module.

Specifically, a sub-module is established and used for establishing a flexible resource scheduling database, classifying according to labels and constructing corresponding storage units;

specifically, the processing sub-module is used for processing the data collected by the data acquisition module, mapping the data with the identification sub-unit and processing and warehousing the data;

further partitioning, the build submodule should include a classification subunit and a storage subunit;

specifically, the classification subunit is configured to divide the database storage area according to the four-level label divided in step S103;

specifically, the storage subunit is configured to provide a storage area for parameters and constraints.

Further dividing, the processing sub-module should include a mapping sub-unit, a preprocessing sub-unit, a quality control unit, a binning sub-unit, and an updating sub-unit.

Specifically, the mapping subunit is configured to map the identification subunit of the data acquisition module with the classification subunit of the building sub-module, and correspond the collected data to each tag;

specifically, the preprocessing subunit is used for carrying out standardization processing on the data set;

specifically, the quality control unit is used for performing data quality control on the standardized data set.

Specifically, the warehousing subunit is used for warehousing the standardized data set according to the classified label.

Specifically, the updating subunit is configured to update the stored data in real time, and convert the stored data into historical data as a prediction reference.

Specifically, the calling function of the database is realized through a database calling module:

it should be understood that the invoking module of the database should implement accurate response of the database to the invoking requirement, and can implement scheduling of resources under different labels according to the invoking requirement.

Further partitioning, the database call module should include an identification subunit and a call subunit;

specifically, the identification subunit is used for identifying the type of the calling resource, the time scale and the calling demand;

specifically, the calling subunit is configured to perform class-division and time-scale calling on the identified calling requirement.

Step S104: respectively establishing a day-ahead scheduling model taking the minimum total running cost as an objective function, a day-in rolling model taking the minimum running cost as the objective function and a real-time scheduling model taking the minimum adjusting cost as the objective function based on various resource scheduling constraint conditions;

it should be understood that the time scale of the established day-ahead scheduling model is 1h, the optimization period is 24h, and meanwhile, in order to stabilize the uncertainty and fluctuation of new energy output and load change, the errors under the prediction situation of different loads and new energy output are reduced, and the day-ahead scheduling model is established by a multi-scenario stochastic programming method so as to meet the constraint condition of safe operation of the system.

Specifically, the construction of the day-ahead scheduling model with the minimum total running cost as an objective function comprises the following processes:

(1) And determining cost calculation formulas of various flexible resources when the system is running.

Specifically, the cost of various flexible resources during the operation of the power system comprises: the operation cost of the conventional unit, the operation cost of the new energy unit, the operation cost of the energy storage power station and the user load cost.

Specifically, the operation cost C of the conventional unit ^G The calculation formula is as follows:

wherein p is _s Representing the probability of occurrence of the scene s,represents the electricity generation cost of the ith conventional unit, < ->Representing the power generation capacity of the ith conventional unit at time t in s scene,/>And->And respectively representing the starting cost and the shutdown cost of the ith conventional unit at the moment t under the s scene.

Specifically, the operation cost C of the new energy unit ^N The calculation formula is as follows:

wherein lambda is _i ^N Represents the power generation cost of the ith new energy unit,representing the generated energy of the ith new energy unit at t moment in s scene,/and the like>And->And respectively representing the starting cost and the shutdown cost of the ith new energy unit at the moment t under the s scene. Lambda (lambda) _i ^N-F Indicating the penalty cost of discarding new energy. />Represents the ith new energy unit t in s sceneThe output force of the moment is predicted, And the actual output of the ith new energy unit at the moment t under the s scene is shown.

Specifically, the operating cost C of the energy storage power station ^B The calculation formula is as follows:

wherein,and->Respectively representing the charge electricity price and the discharge electricity price of the ith energy storage power station,/for>And->Respectively representing the charge quantity and the discharge quantity of the ith energy storage power station at the moment t under the scene s, and +.>Representing depreciation cost of the ith energy storage power station, < ->And the start-stop variable at the moment t of the ith energy storage power station in the s scene is represented, and the value is 0-1.

Specifically, user load cost C ^L The calculation formula is as follows:

wherein lambda is ^PDR Representing price type demand response load calling cost, P _PDR,t,s The price type demand response load t moment calling quantity under the s scene is represented; lambda (lambda) ^IDR Representing incentive type demand response load calling cost, P _IDR,t,s And the excitation type demand response load t moment call quantity under the s scene is represented. Lambda (lambda) ^L-F Represents the penalty cost of the loss of power to the load,and the power loss of the load at the moment t under the scene s is shown.

(2) And establishing an objective function with minimum total running cost according to cost calculation formulas of various flexible resources.

Specifically, the objective function expression with the minimum total running cost is:

min{C ^G +C ^N +C ^B +C ^L }

(3) And establishing a flexible resource daily scheduling model taking the minimum total operation cost as an objective function according to various flexible resource scheduling and the constraint of the operation of the power system.

Specifically, the constraint that the flexible resource day-ahead scheduling model should satisfy includes: conventional unit operation constraints (1) - (12), energy storage power station operation constraints (13) - (29), load side demand response constraints (30) - (33), new energy unit output constraints (34) - (35), power system line power flow constraints (36) - (37), power system power balance constraints (38), flexible adjustment constraints for each scene (39) - (40) and various flexible resource scheduling cost calculation formulas (41) - (44).

It should be understood that the time scale of the established daily rolling model is 15min, the optimization period is 4h, and the daily rolling model is usually implemented by feeding back the actually measured power system data in the state of each time period during daily operation to the daily rolling model, and solving the optimal scheduling plan by combining the wind-light output and load prediction data of each time period in 4 h. The intra-day rolling model realizes error correction on a scheduling plan output by a pre-day scheduling model by replacing a predicted value by using an actual measurement value of each time period, so that the scheduling optimization of the intra-day rolling model also takes the minimum system running cost as an objective function, the constraint of various flexible resource scheduling and the constraint of system running are also satisfied, the intra-day rolling model is built by a multi-scene random programming method, and only the calculation formula of the load calling cost and the load calling quantity and the load calling variety are changed.

Specifically, the construction of the day-ahead scheduling model with the minimum system running cost as an objective function comprises the following processes:

(1) And determining cost calculation formulas of various flexible resources when the system is running.

Specifically, the cost of various flexible resources during the operation of the power system comprises: the operation cost of the conventional unit, the operation cost of the new energy unit, the operation cost of the energy storage power station and the user load cost.

The operation cost of the conventional unit, the operation cost of the new energy unit and the operation cost of the energy storage power station are the same as cost calculation formulas (41) - (43) of the day-ahead scheduling model, and the calculation formulas of the user load cost are changed according to the actual measurement value of the load in the day.

Specifically, the calculation formula of the user load cost of the daily scroll model is as follows:

wherein lambda is ^L-OV Indicating the unit cost of the measured load,and the actual measurement load call quantity at the moment t under the scene s is shown.

(2) And establishing an objective function with minimum running cost according to cost calculation formulas of various flexible resources.

Specifically, the objective function expression with the minimum running cost is:

min{C ^G +C ^N +C ^B +C ^L-I }

(3) And establishing a flexible resource daily rolling model with the minimum running cost as an objective function according to various flexible resource scheduling and the constraint of the running of the power system.

It should be understood that, due to the influence of the actual measurement value during the operation of the system in the day, the operation constraint of the new energy unit and the power balance constraint of the system operation already caused by the load side demand response constraint should be changed, and the predicted output and the load call predicted value of the new energy are converted from the long-term predicted value to the short-term predicted value.

Specifically, the load side demand response constraint should also consider the actual call volume constraint:

wherein P is _t ^L-OV The actual amount of load modulation at time t is indicated,and->The upper limit and the lower limit of the actual load adjustment amount are respectively indicated.

Specifically, the power balance constraint of the system operation should be:

specifically, constraints that the flexible resource intra-day rolling model should satisfy include: conventional unit operation constraints (1) - (12), energy storage power station operation constraints (13) - (29), load side demand response constraints (30) - (33) (46), new energy unit output constraints (34) - (35), power system line flow constraints (36) - (37), power system power balance constraints (47), flexible adjustment constraints for each scene (39) - (40) and various flexible resource scheduling cost calculation formulas (41) - (43) (45).

It should be understood that the time scale of the established real-time scheduling model is 5min, and the real-time scheduling model corrects the deviation of the output force of the daily rolling model according to the latest ultra-short-term predicted value based on the output result of the daily rolling model, so that the real-time scheduling model is more refined scheduling optimization. The multi-scene stochastic programming method is not suitable for the requirements of high precision and robustness, so that the rotation standby cost and constraint are introduced into the real-time scheduling model, and the opportunity constraint method is introduced, so that the rotation standby capacity constraint condition is not smaller than a certain confidence level, and meanwhile, the uncertainty of the ultra-short-term predicted value is considered, and the uncertainty of the predicted value in the opportunity constraint is processed by introducing the fuzzy parameter.

Specifically, the establishment of the real-time scheduling model with minimum adjustment cost as an objective function comprises the following processes:

(1) And determining cost calculation formulas of various flexible resources when the system is running.

Specifically, the cost of various flexible resources during the operation of the power system comprises: the operation cost of the conventional unit, the operation cost of the new energy unit, the operation cost of the energy storage power station, the user load cost and the rotation standby cost.

The operation cost of the conventional unit, the operation cost of the new energy unit, the operation cost of the energy storage power station and the user load cost are the same as the cost calculation formulas (41) - (43) (45) of the daily rolling model, and a rotary standby cost calculation formula is added in the cost calculation formula.

Specifically, the rotational standby cost calculation formula is:

wherein lambda is ^M-G Representing the unit rotation standby cost of the conventional unit,and->Respectively represents the positive rotation and the negative rotation of the conventional unit for standby, lambda ^M-B Representing the unit rotation standby cost of the energy storage unit, +.>And->Indicating the positive and negative rotation of the energy storage unit for standby.

(2) And establishing an objective function with minimum running cost according to cost calculation formulas of various flexible resources.

Specifically, the objective function expression with the minimum running cost is:

min{C ^G +C ^N +C ^B +C ^L-I +C ^M }

(3) And establishing a flexible resource real-time scheduling model taking the minimum adjustment cost as an objective function according to various flexible resource scheduling and the constraint of the operation of the power system.

It should be appreciated that due to the refinement and robustness requirements of the real-time scheduling model, the rotational reserve capacity constraint is added to the real-time scheduling model and expressed in the form of an opportunity constraint, the opportunity constraint is processed through fuzzy parameters, so that uncertainty is eliminated, and the predicted output of the new energy and the predicted value of the load scheduling amount are converted from a short-term predicted value to an ultra-short-term predicted value.

Specifically, the rotational reserve capacity constraint is:

where Pr { } represents the probability that the constraint is true,and->Fuzzy parameters respectively representing new energy and load ultra-short term predicted values in a real-time scheduling period, wherein alpha is the confidence level of establishment of reserve capacity constraint, and P _t ^L The total call quantity of the real-time adjustable load before, in the day at the time t is represented.

Specifically, fuzzy parameters of new energy and load ultra-short-term predicted values in real-time scheduling periodAnd->Can be represented by a membership parameter quaternion:

wherein,membership parameter of ultra-short term predicted value for new energy unit output and load call (omega) ₁ ,ω ₂ ,ω ₃ ,ω ₄ ) And the coefficient is a proportionality coefficient and is related to the historical data of the output and the load of the new energy unit.

Meanwhile, in order to facilitate solving of a real-time scheduling model containing fuzzy opportunity constraints, the constraints are subjected to clear equivalent processing by adopting trigonometric function fuzzy parameters.

Specifically, the clear equivalent of the rotational spare capacity fuzzy opportunity constraint is:

specifically, the constraint that the flexible resource real-time scheduling model should satisfy includes: conventional unit operation constraints (1) - (12), energy storage power station operation constraints (13) - (29), load side demand response constraints (30) - (33) (46), new energy unit output constraints (34) - (35), power system line flow constraints (36) - (37), power system power balance constraints (47), scene flexible adjustment constraints (39) - (40), rotational reserve capacity constraints (53) - (54) and various flexible resource scheduling cost calculation formulas (41) - (43) (45) (48).

Step S105: and predicting the user load by using an intelligent algorithm of machine learning, and dynamically monitoring the actual user load, so that the user load is dynamically classified into three-level scheduling models under different time scales, and accurate scheduling of flexible resources is realized.

Specifically, the intelligent algorithm of machine learning is utilized to predict the load of the user, which comprises the following contents:

(1) And selecting load data of a typical day according to the load historical data of the power system, and constructing a prediction data set.

Specifically, the power system load history data is the history data of the past access load of the power system collected to the flexible resource database and the data of the influence factors causing the fluctuation of the power load, such as weather, temperature, season, peak and valley of the power consumption of the user, and the like. From these historical data, a dataset of power system load predictions is established.

(2) And constructing a power load prediction model, and training through an intelligent algorithm to improve the prediction precision.

Specifically, the embodiment of the invention realizes the prediction of the power load through a long-short-term memory neural network (LSTM) model, and the LSTM neural network can learn and memorize the modes and the characteristics in the historical data through the memory cells, and predicts the future power load according to the modes and the characteristics, thereby having higher prediction precision and stability.

Specifically, based on the collected power load prediction data set, the prediction set, the training set and the verification set are reasonably divided, and the LSTM model is trained, and the machine learning process of the LSTM model is realized through two gating cyclic input gates and a forgetting gate. The input gate controls the degree of data input at the current moment, and the forget gate controls the degree of retention of historical information. Through these two gating cycles, the LSTM can select important information for processing and forget unimportant information. In addition, LSTM introduces output gates to control the output of the prediction results.

Specifically, the predicting the power load using the LSTM model includes:

first, the historical power load data needs to be sorted and cleaned to eliminate outliers and missing values. Then, an LSTM is utilized to construct a prediction model, and the prediction precision is improved by adjusting the network structure and parameters. And finally, evaluating the prediction result to ensure the reliability and stability of the model.

Specifically, the mathematical expression corresponding to the LSTM internal structure is:

f _t ＝σ(W _f ·[h _t-1 ,x _t ]+b _f )

i _t ＝σ(i _f ·[h _t-1 ,x _t ]+b _i )

g _t ＝tanh(W _c ·[h _t-1 ,x _t ]+b _c )

o _t ＝σ(W _o ·[h _t-1 ,x _t ]+b ₀ )

wherein f _t An output representing a forget gate; sigma represents a sigmoid activation function, h _t-1 、x _t In the input sigma function, the output is between 0 and 1, 0 means that the content is completely discarded, and 1 means that the content is reserved; i.e _t Is the output of the input gate, controls the deletion or retention of the candidate memory cell; o (o) _t Is an output door; w represents the weight parameter corresponding to each gate; b represents the corresponding bias of each gate.

The LSTM model can be used for carrying out data prediction of higher progress on the power load, so that predicted data of the user load is obtained.

(3) And generating a user load prediction graph according to the predicted data of the user load, and realizing prediction of the user load under different time scales.

Specifically, the dynamic monitoring of the actual user load includes the following:

(1) Classifying the power users, giving the users corresponding power labels, and monitoring the power consumption conditions of the users according to the classification;

specifically, the classification of the power users can be carried out by giving user labels according to user files established when the users access the network, and establishing corresponding user load dynamic monitoring models for various obtained label users, so that the power users have the monitoring functions of classifying resource types, voltage grades, industry types and electricity utilization types in the dynamic monitoring of actual user loads.

(2) Describing and depicting the load characteristics of the user in real time through a monitoring model;

specifically, the real-time characteristics of the user load are characterized by a differential equation model, and the general form is as follows:

wherein P is _k And Q _k Respectively representing active power and reactive power of user at time k, U and I respectively representing voltage and current of user equipment, n _P 、n _Q 、n _U 、n _I The number of occurrences of active power, reactive power, voltage, and current are represented, respectively.

The dynamic power data of the user equipment can be calculated in real time through the formula.

(3) And dynamically generating a user load characteristic diagram according to the real-time characteristic of the user load, and realizing the dynamic monitoring of the actual user load.

And converting the power data of the user equipment obtained according to the calculation formula into dynamic power load data of the power user, dynamically monitoring and analyzing the data, dynamically generating an actual user load characteristic trend chart in real time, and realizing dynamic monitoring of the actual user load.

Specifically, the dynamic classification of the user load is realized by predicting the deviation change of the load and the actual load, the deviation value between the predicted user load curve and the actual user characteristic trend is calculated by comparing the predicted user load curve and the actual user characteristic trend, and the deviation of the user load value is processed in a corresponding daily front, daily inner and real-time scheduling model, so that the dynamic monitoring and the dynamic classification of the user load are realized, and the scheduling of flexible resources is more accurate.

Step S106: solving the three-level model, and according to the output result of the three-level model, making a scheduling plan of the corresponding flexible resource, and realizing the regulation and control of the target power system.

Specifically, for solving three-stage models of day-ahead scheduling, day-ahead rolling and real-time scheduling, the linear optimization solver CPLEX is utilized to solve the problem, but according to different scheduling time scales of flexible resources, input and output parameters of each stage of models are different, so that a scheduling plan of the flexible resources under different time scales can be obtained by solving each stage of models.

Specifically, the solving process of the day-ahead scheduling model is as follows:

1) According to the obtained relevant rated parameters of various units of the power system, determining the input parameters of a scheduling model before the flexible resource day;

Specifically, the input parameters of the day-ahead scheduling model are: conventional unit related parameters, energy storage power station related parameters, load side demand response resource related parameters, new energy unit related parameters and power grid network related parameters.

Specifically, the relevant parameters of the conventional unit are as follows: the number of the conventional units, the upper and lower limits of the output power of the conventional units, the up-down climbing rate and the start-stop climbing rate of the conventional units, the starting cost, the closing cost and the minimum duration of the start-stop of the conventional units, the reference scene output power of the conventional units and the flexible adjustment capability of the conventional units.

Specifically, the relevant parameters of the energy storage power station are as follows: the energy storage power station comprises the quantity of energy storage power stations, the charge and discharge power and the charge and discharge efficiency of an electrochemical energy storage unit, the maximum and minimum capacity of the electrochemical energy storage unit, the capacity at the initial moment and the capacity at the end moment in the day, the charge state of the electrochemical energy storage unit, the maximum and minimum water storage capacity of a water storage power station reservoir, the maximum and minimum output power of water pumping and discharging of the water pumping and storage power station, the water quantity and electric quantity conversion efficiency of the water pumping and storage power station during water pumping and discharging, the output power of a reference scene of the energy storage unit and the flexible adjustment capacity of the energy storage unit.

Specifically, the load side demand response resource related parameters are: maximum and minimum call volume of price type demand response load and maximum and minimum call volume of incentive type demand response load.

Specifically, the relevant parameters of the new energy unit are as follows: the number of new energy units, the maximum and minimum output power of the new energy units, and the output power of the new energy units is predicted for a long time in the new energy units.

Specifically, the grid network related parameters include: the number of network nodes, the number of network lines, the network line power transfer distribution factor and the association matrix among the nodes.

2) And solving the flexible resource day-ahead scheduling model through a mathematical optimization toolbox yalminip and CPLEX optimization solver to obtain a model output result.

Specifically, the output result of the day-ahead scheduling model is: the starting and stopping time of the conventional unit, the starting and stopping state variable of the conventional unit, the output power of each unit of the conventional unit, the pumping and discharging time of the pumped storage power station, the pumping and discharging state variable of the pumped storage power station, the output power of the pumped storage power station, the calling quantity of the PDR load and the calling quantity of the IDR load.

3) And (3) according to the output result, a start-stop plan of a conventional unit, a charge-discharge plan of a pumped storage unit, a call plan of a PDR load and a call plan of an IDR load are compiled.

Specifically, the solving process of the intra-day rolling model is as follows:

1) Determining the input parameters of the flexible resource intra-day rolling model according to the input parameters and the output result of the pre-day scheduling model;

Specifically, the input parameters of the daily rolling model include conventional unit related parameters, energy storage power station related parameters, load side demand response resource related parameters, new energy unit related parameters, power grid network related parameters and output results of a daily scheduling model, and it should be noted that in the daily rolling model, predicted values of output and load of the new energy unit are changed from medium-long-term predicted values to short-term predicted values.

2) And solving the flexible resource daily rolling model through a mathematical optimization toolbox yalminip and CPLEX optimization solver to obtain a model output result.

Specifically, the output result of the intra-day rolling model is: the method comprises the steps of charging and discharging time of an electrochemical energy storage power station, output power of the electrochemical energy storage power station, charging and discharging state variables of the electrochemical energy storage power station, actual output power of a new energy unit and adjustment amount of IDR load.

3) And (5) compiling an output plan of the new energy unit, a charging and discharging plan of the electrochemical energy storage power station and an IDR load calling plan according to the output result.

Specifically, the solving process of the real-time scheduling model is as follows:

1) Determining the input parameters of the flexible resource real-time scheduling model according to the input parameters and output results of the day-ahead scheduling model and the day-in rolling model and the related parameters of the rotation reserve quantity and the fuzzy opportunity constraint;

Specifically, the input parameters of the real-time scheduling model include: the method comprises the following steps of conventional unit related parameters, energy storage power station related parameters, load side demand response resource related parameters, new energy unit related parameters, power grid network related parameters, output results of a day-ahead scheduling model, output results of a day-ahead rolling model, positive and negative rotation reserve of the conventional unit, positive and negative rotation reserve of the energy storage unit and confidence level of establishment of rotation reserve capacity constraint, and fuzzy parameters of new energy output and load prediction, wherein the problem that in a real-time scheduling model, the predicted values of the new energy unit output and load are changed from short-term predicted values to ultra-short-term predicted values is solved.

2) And solving the flexible resource real-time scheduling model through a mathematical optimization toolbox yalminip and CPLEX optimization solver to obtain a model output result.

Specifically, the output result of the real-time scheduling model is: the rotation reserve capacity of the conventional unit, the rotation reserve capacity of the energy storage unit and the adjustment amount of various flexible resources.

3) And (5) compiling a power-out plan, a rotating standby plan and a call plan of DR load of all units according to the output results.

Specifically, a running framework diagram of the three-level model multi-time scale scheduling is shown in fig. 4.

A second embodiment of solving the three-level model is given below, where the variables used in this embodiment are independent of the previous description:

the concrete implementation mode for solving the three-level scheduling model based on the gray wolf optimization algorithm is as follows:

1. model building

The problem to be solved comprises three layers of a day-ahead scheduling model, a day-in rolling model and a real-time scheduling model, wherein the three layers have a hierarchical relationship. The following symbols are defined:

x ₁ : scheduling decision variables before the day; x is x ₂ : rolling the decision variable in the day; x is x ₃ : scheduling decision variables in real time; f (f) ₁ (x ₁ ): a day-ahead scheduling model objective function representing total running cost; f (f) ₂ (x ₂ ): an intra-day rolling model objective function representing running cost; f (f) ₃ (x ₃ ): real-time scheduling model objective function representing adjustment costs

2. Coding design

Each possible solution in the solution space is represented using real number encoding, and a mapping relation is constructed:

X＝{x ₁ ，x ₂ ，x ₃ }

wherein X represents a scheduling scheme.

3. Algorithm flow

The wolf optimization algorithm realizes optimization solution by simulating the survival habit of the wolf group. The flow is as follows:

1. population initialization

N wolves were randomly generated as the initial population: x is X ₁ ，X ₂ ，…，X _N 。

2. Fitness calculation

Defining the overall fitness of the wolf group:

F(X)＝ω ₁ f ₁ (x ₁ )+ω ₂ f ₂ (x ₂ )+ω ₃ f ₃ (x ₃ )

wherein omega ₁ ，ω ₂ ，ω ₃ Is a weight coefficient.

The fitness of each wolf was calculated by equation.

3. Ticket wolf determination

Three wolves with the best adaptability are selected as leading wolves: x is X _l1 ，X _l2 ，X _l3 。

4. Surrounding hunting object

The remaining wolves are gathered toward the lead wolf position.

5. Location update

X _k+1 ＝X _k +B·Y _k

Wherein B is a scaling factor, Y _k Indicating the current search direction.

6. Termination judgment

And (5) if the termination condition is met, exiting the iteration, otherwise, returning to the step (2).

4. Implementation flow

1. Parameter setting

Determining algorithm parameters: n, maximum number of iterations, etc.

2. Encoding and initialization

The initial population is randomly generated by encoding according to the second section method.

3. Evaluation of fitness

And calculating the current wolf group fitness.

4. Iterative search

The flow of the third section is repeated until the stop condition is satisfied.

5. Decoding and analysis

Decoding the optimal solution, and analyzing the obtained three model scheduling results.

5. Complexity analysis

Time complexity: o (N), spatial complexity: o (N).

6. Further optimize

Introducing adaptive parameter adjustment

Combining other intelligent optimization algorithms

Adding random disturbance to avoid sinking into local optimum

The wolf crowd characteristic in the algorithm accords with the layering characteristic of the three-level scheduling model, and the layering is clear. The information exchange behavior of the simulated wolf clusters can promote cooperative coordination among different layers. According to the characteristics of the three-level model, the mapping relation between the wolf group level codes and the hierarchical model can be designed, so that the corresponding wolf king is actually scheduled, and the hierarchical decision mechanism of the scheduling system is approximately simulated. The cooperative effect among different scheduling levels can be enhanced by setting related strategies of wolf cluster movement, so that the scheduling scheme has clear levels and meets the actual demands. In conclusion, the gray wolf optimization algorithm can be well combined with the three-level scheduling model, the layered thought of three-level scheduling is reflected to a certain extent, and the overall optimized and layered clear scheduling scheme can be searched through the self-organizing and self-learning characteristics of the gray wolf optimization algorithm.

By the method, the flexible resources on each side of the power system source network charge storage can be subjected to multi-time scale scheduling, and the optimal scheduling of the flexible resources is realized from the viewpoint of minimum cost.

The technical scheme of the embodiment of the invention is further described below by using a specific application example.

Taking a regional node system as an example, as shown in fig. 5, the node system has 30 nodes in total, 41 branches in total, 6 conventional generator sets, 1 new energy set and 1 electrochemical energy storage power station, and the rest nodes comprise various load side users. According to the related requirements and the technical scheme of the invention, the related parameters of various units of the power system are obtained, a corresponding scheduling model is constructed, and scheduling optimization is performed on flexible resources of the power system.

According to the related data of the power system, the technical scheme according to the embodiment of the invention calculates the call quantity of various flexible resources of the system and the load value of each period of the power system under the condition of meeting the minimum cost, as shown in fig. 6, 7 and 8.

In summary, the embodiment of the invention constructs a reasonable multi-time scale resource calling plan by considering the calling cost of various flexible resources participating in the scheduling under the multi-time scale based on the relevant parameters of each unit of each node of the power system to be analyzed, thereby realizing more efficient and economical scheduling of various flexible resources of the power system.

The foregoing is merely illustrative of the present invention, and the present invention is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present invention. Therefore, the protection scope of the invention is subject to the protection scope of the claims.

Claims (8)

1. The flexible resource optimization scheduling method based on the multiple time scales is characterized by comprising the following steps of:

acquiring flexible resource composition and parameters of a power system to be analyzed, and dividing different time scales according to the flexibility of scheduling of various resources;

quantifying flexible resources under different time scales, and determining corresponding constraint of flexible resource scheduling of the power system according to rated parameters and characteristics of various flexible resources as various resource scheduling constraint conditions;

respectively establishing a day-ahead scheduling model taking the minimum total running cost as an objective function, a day-in rolling model taking the minimum running cost as the objective function and a real-time scheduling model taking the minimum adjusting cost as the objective function based on various resource scheduling constraint conditions to form a three-level model;

solving the three-level model, and according to the output result of the three-level model, making a scheduling plan of the corresponding flexible resource, and realizing the regulation and control of the target power system.

2. The method for optimizing and scheduling flexible resources based on multiple time scales according to claim 1, wherein the step of obtaining the flexible resource composition and parameters of the power system to be analyzed and dividing different time scales according to the flexibility of scheduling each flexible resource comprises the steps of:

step 1, acquiring relevant parameters of a conventional unit of a power system to be analyzed:

obtaining the upper limit P of the output power of a conventional unit _i ^G-max Lower limit P _i ^G-min ；

Obtaining the climbing rate P of a conventional unit during operation _i ^R ；

Obtaining the minimum duration T of the start-stop of each conventional unit _c ；

Acquiring single start cost C of each conventional unit _i ^U-S And a single shut-down cost C _i ^D-S ；

Step 2, acquiring relevant parameters of an energy storage unit of the power system to be analyzed:

obtaining capacity rating E of each electrochemical energy storage unit _B ；

Acquiring the SOC upper limit S of an electrochemical unit _max And a lower SOC limit S _min Rated state of charge S at end time _n ；

Obtaining the maximum and minimum charge and discharge power P of an electrochemical energy storage unit ^B-max And P ^B-min Charging efficiencyDischarge efficiency

Obtaining the reservoir maximum water storage capacity V of a pumped storage power station _i ^W-max And minimum water storage volume V _i ^W-min ；

Obtaining maximum and minimum output power P of pumping water of pumped storage power station ^W-max And P ^W-min Conversion efficiency of water quantity and electricity quantity when pumped and discharged by pumped storage power station And->

Step 3, acquiring relevant parameters of load side demand response resources of the power system to be analyzed:

obtaining maximum and minimum call volumes of price type demand response loadIs->

Acquisition ofMaximum increased load capacity of excitation type demand response loadMaximum reduced load->

Step 4, acquiring relevant parameters of new energy units of the power system to be analyzed:

obtaining upper limit P of output power of new energy unit _i ^N-max Lower limit P _i ^N-min 。

3. The multi-time scale-based flexible resource optimization scheduling method according to claim 1, wherein the step of quantifying flexible resources under different time scales and determining corresponding constraints of power system flexible resource scheduling according to rated parameters and characteristics of various flexible resources comprises:

obtaining flexible resource scheduling optimization conventional unit operation constraint according to each item of parameter data of a power supply side conventional unit, wherein the power supply side comprises a thermal power unit and an adjustable hydropower station;

according to each parameter of flexible resources of an energy storage side, obtaining flexible resource scheduling and optimizing operation constraint of the energy storage power station, wherein the energy storage side comprises electrochemical energy storage and pumped storage;

obtaining flexible resource scheduling optimization load side demand response constraint according to various parameters of the load side flexible resource, including incentive type demand response and price type demand response;

Obtaining flexible resource scheduling and optimizing new energy unit output constraint according to related parameters of the new energy unit incorporated by the power system;

when the power system operates, line flow constraint among all nodes is met;

according to the output power and the real-time load of each unit of the system, the power load balance constraint of the system should be satisfied at any moment when the system operates;

according to different calling scenes of various flexible resources, scheduling the flexible resources of the power system needs to consider the adjustment constraint of each scene.

4. The flexible resource optimization scheduling method based on multiple time scales according to claim 1, wherein the step of establishing a day-ahead scheduling model with minimum total running cost as an objective function based on various resource scheduling constraint conditions comprises:

determining a cost calculation formula of various flexible resources when the system operates;

according to cost calculation formulas of various flexible resources, an objective function with minimum total running cost is established;

according to various flexible resource scheduling and constraint of power system operation, a flexible resource day-ahead scheduling model taking the minimum total operation cost as an objective function is established;

the cost calculation formula of each flexible resource comprises the following steps:

Running cost C of conventional unit ^G The calculation formula is as follows:

running cost C of new energy unit ^N The calculation formula is as follows:

operating cost C of energy storage power station ^{Occupying the area of} The calculation formula is as follows:

user load cost C ^L The calculation formula is as follows:

the calculation formula of the objective function of the day-ahead scheduling model is as follows:

min{C ^G +C ^N +C ^B +C ^L }

wherein C is ^G Representing the running cost of a conventional unit of the energy storage system, C ^N Representing the running cost of the new energy unit, C ^B Representing the running cost of an energy storage power station of an energy storage system, C ^L Representing the cost of the user load.

5. The flexible resource optimization scheduling method based on multiple time scales according to claim 1, wherein the step of establishing a daily rolling model with minimum running cost as an objective function based on various resource scheduling constraint conditions comprises the following steps:

determining a cost calculation formula of various flexible resources when the system operates;

according to cost calculation formulas of various flexible resources, an objective function with minimum total running cost is established;

according to various flexible resource scheduling and constraint of power system operation, establishing a flexible resource daily rolling model with the minimum total operation cost as an objective function;

the cost calculation formula of each flexible resource comprises the following steps:

Running cost C of conventional unit ^G A calculation formula;

running cost C of new energy unit ^N A calculation formula;

operating cost C of energy storage power station ^B A calculation formula;

user load cost C ^L-I The calculation formula is as follows:

the calculation formula of the intra-day rolling model objective function is as follows:

min{C ^G +C ^N +C ^B +C ^L-I }

wherein C is ^G Representing the running cost of a conventional unit of the energy storage system, C ^N Representing the running cost of the new energy unit, C ^B Representing the running cost of an energy storage power station of an energy storage system, C ^L-I Representing the cost of the user load.

6. The flexible resource optimization scheduling method based on multiple time scales according to claim 1, wherein the step of establishing a real-time scheduling model with minimum adjustment cost as an objective function based on various resource scheduling constraint conditions comprises the following steps:

determining a cost calculation formula of various flexible resources when the system operates;

according to cost calculation formulas of various flexible resources, an objective function with minimum total running cost is established;

according to various flexible resource scheduling and the constraint of the operation of the power system, a flexible resource real-time scheduling model taking the minimum adjustment cost as an objective function is established;

the cost calculation type content of the various flexible resources comprises the following components:

running cost C of conventional unit ^G A calculation formula;

running cost C of new energy unit ^N A calculation formula;

operating cost C of energy storage power station ^{Occupying the area of} A calculation formula;

user load cost C ^L-I A calculation formula;

cost of rotation reserve C ^M The calculation formula is as follows:

the calculation formula of the real-time scheduling model objective function is as follows:

min{C ^G +C ^N +C ^{occupying the area of} +C ^L-I +C ^M }

Wherein C is ^G Representing the running cost of a conventional unit of the energy storage system, C ^N Representing the running cost of the new energy unit, C ^B Representing the running cost of an energy storage power station of an energy storage system, C ^L-I Representing user load cost, C ^M Indicating the cost of spinning reserve.

7. The flexible resource optimization scheduling method based on multiple time scales as claimed in claim 1, wherein the step of solving the three-level model comprises:

the solving step of three-level models of day-ahead scheduling, day-ahead rolling and real-time scheduling by using a mathematical optimization toolbox yalminip and a linear optimization solver CPLEX comprises the following steps:

solving a scheduling model before the day of the flexible resource;

solving a flexible resource daily rolling model;

solving a flexible resource real-time scheduling model;

the solving process of the flexible resource day-ahead scheduling model comprises the following steps:

according to the obtained relevant rated parameters of various units of the power system, determining the input parameters of a scheduling model before the flexible resource day;

Solving a flexible resource day-ahead scheduling model through a mathematical optimization toolbox yalminip and CPLEX optimization solver to obtain a model output result;

according to the output result, a start-stop plan of a conventional unit, a charge-discharge plan of a pumped storage unit, a PDR load calling plan and an IDR load calling plan are compiled;

the solving process for the flexible resource intra-day rolling model comprises the following steps:

determining the input parameters of the flexible resource intra-day rolling model according to the input parameters and the output result of the pre-day scheduling model;

solving a flexible resource daily rolling model through a mathematical optimization toolbox yalminip and CPLEX optimization solver to obtain a model output result;

according to the output result, a new energy unit output plan, an electrochemical energy storage power station charge-discharge plan and an IDR load calling plan are compiled;

the solving process of the flexible resource real-time scheduling model comprises the following steps:

determining the input parameters of the flexible resource real-time scheduling model according to the input parameters and output results of the day-ahead scheduling model and the day-in rolling model and the related parameters of the rotation reserve quantity and the fuzzy opportunity constraint;

and solving the flexible resource real-time scheduling model through a mathematical optimization toolbox yalminip and CPLEX optimization solver to obtain a three-level model output result.

8. The flexible resource optimization scheduling method based on multiple time scales according to claim 1, wherein the step of solving the three-level model specifically comprises:

establishing a gray wolf optimization model, wherein the solution of a day-ahead dispatch model corresponds to a common wolf, the constraint condition and the objective function of the day-ahead dispatch model are used as constraint conditions and objective functions of the common wolf, the solution of a day-ahead scroll model corresponds to a middle-layer wolf, the constraint condition and the objective function of the day-ahead scroll model are used as constraint conditions and objective functions of the middle-layer wolf, the solution of a real-time dispatch model corresponds to a wolf king, and the constraint condition and the objective function of the real-time dispatch model are used as constraint conditions and objective functions of the wolf king; solving the gray wolf optimization model, and taking the solution of the gray wolf optimization model as the solution of the three-level model.