CN112821420A - XGboost-based prediction method and system for dynamic damping factor and multidimensional frequency index in ASFR model - Google Patents

XGboost-based prediction method and system for dynamic damping factor and multidimensional frequency index in ASFR model Download PDF

Info

Publication number
CN112821420A
CN112821420A CN202110103617.6A CN202110103617A CN112821420A CN 112821420 A CN112821420 A CN 112821420A CN 202110103617 A CN202110103617 A CN 202110103617A CN 112821420 A CN112821420 A CN 112821420A
Authority
CN
China
Prior art keywords
damping factor
dynamic damping
model
asfr
frequency
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202110103617.6A
Other languages
Chinese (zh)
Other versions
CN112821420B (en
Inventor
文云峰
黄明增
赵荣臻
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hunan University
Original Assignee
Hunan University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Hunan University filed Critical Hunan University
Priority to CN202110103617.6A priority Critical patent/CN112821420B/en
Publication of CN112821420A publication Critical patent/CN112821420A/en
Application granted granted Critical
Publication of CN112821420B publication Critical patent/CN112821420B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H02J3/24—Arrangements for preventing or reducing oscillations of power in networks
    • H02J3/241—The oscillation concerning frequency
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
    • H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]

Landscapes

  • Engineering & Computer Science (AREA)
  • Power Engineering (AREA)
  • Supply And Distribution Of Alternating Current (AREA)

Abstract

The invention discloses a method and a system for predicting dynamic damping factors and multidimensional frequency indexes in an ASFR model based on XGboost, wherein the multidimensional frequency index prediction method comprises the following steps: constructing an aggregation multi-machine system frequency response model (ASFR); acquiring a dynamic damping factor data set; training an XGboost model by using a dynamic damping factor data set, and determining the optimal hyperparameter of the XGboost model through Bayesian optimization to obtain a dynamic damping factor prediction model; and inputting system operation data at the fault clearing moment into the dynamic damping factor prediction model to obtain a dynamic damping factor, and transmitting the obtained dynamic damping factor to the ASFR model to predict extreme frequency and/or quasi-steady-state frequency. According to the method, the XGboost model is used for realizing real-time correction of the dynamic damping factor in the ASFR model, so that the ASFR model can accurately predict the multidimensional frequency index after a disturbance accident, and a basis is provided for subsequent frequency stability evaluation and control decision of the power system.

Description

XGboost-based prediction method and system for dynamic damping factor and multidimensional frequency index in ASFR model
Technical Field
The invention belongs to the technical field of power systems, and particularly relates to a method and a system for predicting dynamic damping factors and multidimensional frequency indexes in an ASFR model based on XGboost.
Background
The frequency measures the active output and the load balance degree, and is an important index for reflecting the electric energy quality and the safe and stable operation of the system. If the system frequency can not be maintained in the allowable range of the power grid after a disturbance accident, devices such as low-frequency load shedding devices, high-frequency load shedding devices and the like are triggered to act, so that the system is subjected to the risk of large-range load shedding and load shedding, and even frequency collapse can be caused in serious cases, and huge social and economic losses are caused.
A System Frequency Response (SFR) model is one of the main methods of the current frequency stability analysis, and a whole-network generator/load model is equivalent to a single-machine centralized load model on the basis of neglecting a high-order nonlinear link and an amplitude limiting link. However, current research often analyzes that the same type of units are aggregated into one unit, and an actual power system is composed of a plurality of different types of units, such as a hydroelectric unit and a thermal power unit, which are not completely matched. In order to construct a frequency response model suitable for an actual power grid, a multiple unit system frequency response model (ASFR) for aggregating multiple units of different types in the power grid into a single unit of the same type with high precision still needs to be researched.
Meanwhile, in order to accurately predict the frequency index after the disturbance accident, the parameters of the ASFR model need to be accurate. However, as for the damping coefficient, since the influence factors thereof are complicated and various, and the damping coefficient changes with the change of the frequency deviation, it is difficult to accurately set the damping coefficient in the ASFR model. For simplicity, only the effect of the load frequency characteristic on the damping factor is usually considered and set to a constant value based on manual experience, e.g., D1-2 pu. However, in the power system, the damping of the power system includes other elements, such as damping due to the rotational speed characteristics of the turbine generator, damping due to the excitation system, and damping due to the governor system, in addition to the load damping factor. The influence of the running state of the power system, the excitation system and the speed regulation system on the dynamic damping factor is ignored, and the fixed damping factor is adopted, so that a considerable error occurs when the ASFR model is used for calculating the frequency index of the disturbed power system, and the method is particularly applied to a practical large power grid.
Therefore, constructing an accurate ASFR model and correcting the dynamic damping factor of the ASFR model in real time is a critical factor for improving the calculation accuracy and efficiency of the ASFR model, and is also urgent to be researched.
Disclosure of Invention
The invention aims to provide a method and a system for predicting dynamic damping factors and multidimensional frequency indexes in an ASFR model based on XGboost. .
On one hand, the invention provides a dynamic damping factor prediction method in an ASFR model based on XGboost, which comprises the following steps:
step 1: calculating equivalent parameters of the multi-computer system, and further constructing an aggregate multi-computer system frequency response model ASFR;
step 2: acquiring a dynamic damping factor data set based on a simulation result of a mass of expected fault scenes, wherein the dynamic damping factor data set comprises system operation data and dynamic damping factor data;
and step 3: performing iterative training on the XGboost model by using the dynamic damping factor data set to establish a nonlinear mapping relation between system operation data and a dynamic damping factor so as to obtain a dynamic damping factor prediction model;
the input characteristic variable of the dynamic damping factor prediction model is system operation data, and the output characteristic variable is a dynamic damping factor;
and 4, step 4: and (3) acquiring system operation data at the moment of eliminating the fault, inputting the system operation data into the dynamic damping factor prediction model determined in the step (3), and predicting the dynamic damping factor corresponding to the fault.
According to the method, the XGboost is used for establishing the dynamic damping factor prediction model, so that the dynamic damping factor in the ASFR model can be corrected in real time, and a foundation is laid for the ASFR model to accurately predict the multidimensional frequency index after the disturbance accident. Wherein, the system operation data is as follows: the system comprises a generator electromagnetic power/reactive power, a bus voltage amplitude/voltage phase angle, a system active load/reactive load, a line active power, a reserve capacity and a power deficit value after disturbance.
In a second aspect, the invention provides a multidimensional frequency index prediction method fusing ASFR and XGBoost, comprising the following steps:
s1: acquiring an ASFR model constructed by the prediction method of the dynamic damping factor in the ASFR model based on the XGboost and a dynamic damping factor prediction model based on the XGboost;
s2: acquiring system operation data at the moment of eliminating the fault, inputting the system operation data into the dynamic damping factor prediction model determined in the step S1 to obtain a dynamic damping factor corresponding to the fault, and inputting the dynamic damping factor into the ASFR model constructed in the step S1;
s3: and predicting extreme value frequency and/or quasi-steady-state frequency after disturbance fault by using the ASFR model.
According to the invention, the XGboost is utilized to establish a dynamic damping factor prediction model, so that the real-time correction of the dynamic damping factor in the ASFR model is realized, and finally, the ASFR model can accurately predict the multidimensional frequency index after a disturbance accident.
Optionally, the ASFR model is constructed by the following process:
calculating equivalent parameters of a multi-machine system and a unit by a weighted dynamic equivalent method;
respectively equating a plurality of units of each type in the multi-unit system to a single unit of the same type based on the equivalent parameters; recombining into the aggregate multi-machine system frequency response model (ASFR);
optionally, if the types of the units in the multi-unit system include a thermal power unit and a hydroelectric power unit, equating the multiple thermal power units and the multiple hydroelectric power units to one thermal power unit and one hydroelectric power unit respectively, and then constructing an ASFR model, wherein the corresponding equivalent parameters include: the equivalent inertia time constant, the equivalent difference adjustment coefficient, the unit standardization gain, the equivalent parameter of the hydroelectric generating set prime motor-speed regulator and the equivalent parameter of the thermal generating set prime motor-speed regulator;
wherein the equivalent inertia time constant H of the multi-machine systemsysIs calculated as follows:
Figure BDA0002916518140000031
in the formula, Hi,sysIs the equivalent inertia time constant of the ith unit, HiRepresenting the inertia time constant of the ith unit; n is the total number of the units; siThe rated power of the ith unit is represented; ssysRepresents the rated power of the system;
the equivalent adjustment coefficient R of the multi-machine system is calculated as follows:
Figure BDA0002916518140000032
in the formula, RiIs the equivalent difference adjustment coefficient, K, of the ith unitmiGain, k, for the ith unitiEquivalent gain of the ith unit;
the unit normalized gain is calculated as follows:
Figure BDA0002916518140000033
in the formula: k'iThe standard gain of the unit i is obtained;
the equivalent parameters of the prime motor-speed regulator of the thermal power generating unit are calculated as follows:
Figure BDA0002916518140000034
in formula (II), X'1,jIs an equivalent parameter corresponding to the jth parameter in the thermal power generating unit,
Figure BDA0002916518140000035
representing the jth parameter, n, of the ith thermal power generating unit1The number of units of the thermal power generating unit, n3The number of parameters of the thermal power generating unit is set;
the equivalent parameters of the hydroelectric generating set prime motor-speed regulator are calculated as follows:
Figure BDA0002916518140000036
in formula (II), X'2,jIs an equivalent parameter corresponding to the jth parameter in the hydroelectric generating set,
Figure BDA0002916518140000037
representing the jth parameter, n, of the ith hydroelectric generating set2The number n of the units of the hydroelectric generating set4The number of parameters of the hydroelectric generating set is shown.
In the prior art, a plurality of thermal power generating units are often aggregated into one thermal power generating unit, and an actual power system consists of a plurality of different types of units, such as a hydroelectric generating unit and a thermal power generating unit, which are not completely matched. In order to construct a frequency response model suitable for an actual power grid, a plurality of different types of machine sets of the power grid are respectively aggregated into a single machine set of the same type with high precision, and then an ASFR model is constructed.
Optionally, the dynamic damping factor prediction model is as follows:
Figure BDA0002916518140000041
in the formula:
Figure BDA0002916518140000042
for the dynamic damping factor predicted value, f, corresponding to the ith samplek(xi) Represents the prediction value, x, of the dynamic damping factor of the kth tree to the ith sampleiAnd K is the number of the established regression trees for the system operation data corresponding to the ith sample.
Optionally, a bayesian optimization method is adopted in the iterative training process of the XGBoost model to determine the optimal hyperparameter of the XGBoost model, and the XGBoost model under the determined optimal hyperparameter is trained again to obtain the dynamic damping factor prediction model.
Optionally, the dynamic damping factor in the dynamic damping factor data set is obtained as follows:
simulating a mass of expected fault scenes by a time domain simulation method, and executing the following processes for each expected fault scene:
adjusting a dynamic damping factor of the ASFR model under a predicted fault scene to enable extreme value frequency and/or quasi-steady-state frequency obtained by the ASFR model to be consistent with extreme value frequency and/or quasi-steady-state frequency obtained by time domain simulation; and further determining a dynamic damping factor corresponding to the extreme value frequency and/or the quasi-steady-state frequency in the expected fault scene.
In a third aspect, the present invention provides a prediction system, comprising: the system comprises a polymerization multi-machine system frequency response model building module, a dynamic damping factor data set building module, a dynamic damping factor prediction model building module and a dynamic damping factor prediction module;
the system comprises a frequency response model building module, a frequency response model calculation module and a frequency response model calculation module, wherein the frequency response model building module of the multi-machine system is used for calculating equivalent parameters of the multi-machine system so as to build an ASFR (amplitude-dependent frequency response) model of the multi-machine system;
the dynamic damping factor data set construction module is used for acquiring a dynamic damping factor data set based on a simulation result of a mass of expected fault scenes, and the dynamic damping factor data set comprises system operation data and dynamic damping factor data;
the dynamic damping factor prediction model construction module is used for performing iterative training on the XGboost model by using the dynamic damping factor data set to establish a nonlinear mapping relation between system operation data and a dynamic damping factor so as to obtain a dynamic damping factor prediction model;
the input characteristic variable of the dynamic damping factor prediction model is system operation data, and the output characteristic variable is a dynamic damping factor;
and the dynamic damping factor prediction module is used for acquiring system operation data at the moment of fault elimination, inputting the system operation data to the dynamic damping factor prediction model and predicting the dynamic damping factor corresponding to the fault.
Optionally, the prediction system further comprises: the frequency index prediction module is used for inputting the dynamic damping factor into an ASFR model and then predicting extreme value frequency and/or quasi-steady-state frequency after disturbance fault by using the ASFR model.
When the dynamic damping factor data set construction module acquires the dynamic damping factor data set, the acquired dynamic damping factor is a damping factor corresponding to extreme value frequency and/or quasi-steady-state frequency in a fault scene.
In a fourth aspect, the present invention also provides a system comprising a processor and a memory, the memory storing a computer program, the processor invoking the computer program to implement the steps of the dynamic damping factor prediction method or the multi-dimensional frequency index prediction method.
In a fifth aspect, the present invention provides a storage medium storing a computer program, which is called by a processor to execute the steps of the dynamic damping factor prediction method or the multi-dimensional frequency index prediction method.
Advantageous effects
1. According to the method, the mapping relation between the input characteristic variable and the output dynamic damping factor is established through the XGboost model, so that a dynamic damping factor prediction model is obtained and is used for correcting the dynamic damping factor of the ASFR model in real time, and the phenomenon that the frequency index prediction precision is sacrificed due to the fact that the ASFR model adopts a fixed dynamic damping factor is avoided; under the condition of the same test sample, compared with an ASFR model and an XGboost model which adopt fixed damping, the method disclosed by the invention has higher prediction precision.
2. The invention considers that the shallow neural network is difficult to accurately predict the dynamic damping factor; deep learning has strong complex function representation capability, but needs a large number of samples for iterative learning, and has the defect of long training time; the XGboost is an integrated learning algorithm, an efficient integrated learning mode of serially integrating a plurality of regression trees is adopted, requirements on offline training sample size and feature data types are low, generalization capability is strong, multithreading parallel computing can be automatically adopted, and operation speed is extremely high. Therefore, the dynamic damping factor prediction model is constructed based on XGboost.
3. In a further preferred scheme of the invention, the invention also provides a technical means for automatically adjusting and optimizing the hyperparameters in the XGboost model. The dynamic damping factor prediction is realized by considering that the XGboost is based on setting hyper-parameters such as the number of trees, the maximum depth of the trees, the regularization coefficient, the learning rate and the like. The regularization coefficient measures the degree of over-fitting of the inhibition model, and the number, maximum depth and learning rate of the trees influence the training speed and prediction accuracy of the model. In order to avoid the inefficiency and the non-optimality of manual parameter adjustment, the accuracy of dynamic damping factor prediction is improved, and a Bayesian optimization algorithm is introduced to realize automatic optimization of the hyperparameters in the XGboost model.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is an aggregate multi-chassis system frequency response model (ASFR).
Fig. 2 is a prediction framework of dynamic damping factors and multidimensional frequency indexes in an ASFR model based on XGBoost.
Fig. 3 is a diagram of an improved new england 10 machine 39 node system.
FIG. 4 shows the dynamic damping factors corresponding to the extreme frequency and the quasi-steady-state frequency under different disturbance faults.
FIG. 5 is a diagram illustrating an error distribution of extreme frequency prediction according to the method of the present invention.
FIG. 6 is a diagram illustrating the error distribution of the method for predicting the quasi-steady-state frequency according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without inventive effort based on the embodiments of the present invention, are within the scope of the present invention.
Example 1:
the invention provides a multidimensional frequency index prediction method fusing ASFR and XGboost, which comprises the following steps:
step 1: and calculating equivalent parameters of the multi-computer system, and further constructing an aggregate multi-computer system frequency response model (ASFR).
As shown in fig. 1, in this embodiment, a plurality of thermal power generating units and a plurality of hydroelectric power generating units in a multi-unit system are respectively equivalent to one thermal power generating unit and one hydroelectric power generating unit. Namely, equivalent replacement is performed based on equivalent parameters, and the equivalent parameters related in this embodiment specifically include: the equivalent inertia time constant and the equivalent difference adjusting coefficient of the multi-machine system, the unit standardization gain, the equivalent parameter of the hydroelectric generating set prime motor-speed regulator and the equivalent parameter of the thermal generating set prime motor-speed regulator. Other possible embodiments are not limited to the above equivalent parameters and to the above equivalent parameter combinations.
The process of constructing the frequency response model of the multi-machine aggregation system in the embodiment is as follows:
s11: calculating equivalent parameters of the multi-computer system by a weighted dynamic equivalent method; s12: respectively equating multiple units under each type in the system to a single unit of the same type based on equivalent parameters; and then combined into the aggregate multi-chassis system frequency response model (ASFR).
Wherein the equivalent inertia time constant H of the multi-machine systemsys：
Calculating the equivalent inertia time constant of each unit in the multi-unit system according to the following formula; then summing to obtain the equivalent inertia time constant H of the multi-machine systemsys；
Figure BDA0002916518140000061
Figure BDA0002916518140000062
In the formula, Hi，sysIs the equivalent inertia time constant of the ith unit, HiRepresenting the inertia time constant of the ith unit; n is the total number of the units; si、SjRespectively representing rated power of the ith unit and the jth unit; ssysIndicating the power rating of the multi-machine system.
Furthermore, by using the equivalent inertia time constant, the inertia center frequency f of the multi-machine system can be solvedCOI；
Figure BDA0002916518140000071
In the formula (f)iThe frequency of generator node i.
Equivalent difference adjustment coefficient of the multi-machine system:
the total gain of the speed regulators of a plurality of turbines being the gain K of each groupmiThe sum (K)mi＝Si/Ssys) Then, the equivalent difference coefficient R of the whole multi-machine system can be expressed as:
Figure BDA0002916518140000072
in the formula: riIs the equivalent adjustment coefficient, kappa, of the ith unitiAnd the equivalent gain of the ith unit.
For the thermal power generating unit, the equivalent difference adjustment coefficient R of the thermal power generating unit1Can be expressed as:
Figure BDA0002916518140000073
for a hydroelectric generating set, the equivalent difference adjustment coefficient R of the hydroelectric generating set2Can be expressed as:
Figure BDA0002916518140000074
wherein n is1The number of units of the thermal power generating unit, n2The number of the hydroelectric generating sets.
Unit standardized gain
Normalized gain for each unit:
Figure BDA0002916518140000075
in the formula: k'iIs the normalized gain for unit i.
Due to the greater κ'iThe value means that the ith generator has larger rated power and more sensitive frequency reduction, and the unit has larger influence on the equivalent parameter X of the ASFR model. Therefore, the ASFR model equivalent parameter X is represented by a weighted average of the parameters of each generator.
For a thermal power generating unit, the equivalent parameters of the thermal power generating unit are expressed as follows:
Figure BDA0002916518140000076
in formula (II), X'1,jIs an equivalent parameter corresponding to the jth parameter in the thermal power generating unit,
Figure BDA0002916518140000077
representing the jth parameter, n, of the ith thermal power generating unit3The number of parameters of the thermal power generating unit is set;
wherein, the parameter of thermal power unit includes: time constant T of servo3Time constant of steam volume TCHReheater time constant TRHHigh pressure turbine stage power ratio FHP。
For hydroelectric generating set, equivalent parameter X'2,jCan be expressed as:
Figure BDA0002916518140000081
in formula (II), X'2,jIs an equivalent parameter corresponding to the jth parameter in the hydroelectric generating set,
Figure BDA0002916518140000082
j (th) representing ith hydroelectric generating setParameter, n4The number of parameters of the hydroelectric generating set is shown.
The parameters of the hydroelectric generating set comprise: speed regulator response time TGPilot valve time constant TpSoft feedback time constant TdTime constant of water hammer effect TwCoefficient of soft feedback link Dd。
Method for constructing aggregation model of thermal power generating unit by using equivalent parameters and transfer function h of speed regulator of aggregation model1Can be expressed as:
Figure BDA0002916518140000083
method for constructing aggregation model of hydroelectric generating set by using equivalent parameters and speed regulator transfer function h of aggregation model2Can be expressed as:
Figure BDA0002916518140000084
where s is the complex frequency.
As shown in fig. 1, which is a schematic diagram of a frequency response model of a multiple unit system, it can be known from the diagram that the equivalent thermal power generating units and hydroelectric power generating units are aggregated to obtain a multiple unit system frequency response model (ASFR).
Equivalent parameters of a speed regulator and a prime motor in an aggregated multi-machine system frequency response model (ASFR) can be obtained through the formula, however, for a damping coefficient, due to the fact that influence factors are complex and various, accurate setting of a dynamic damping factor in the ASFR model is difficult. In order to realize the correction of the dynamic damping factor in the frequency response model (ASFR) of the polymerization multi-machine system, the dynamic damping factor prediction model based on the XGboost is constructed. Then, it is first necessary to acquire a dynamic damping factor data set.
step 2: and acquiring a dynamic damping factor data set based on a simulation result of a mass expected fault scene, wherein the dynamic damping factor data set comprises system operation data and dynamic damping factor data.
In this embodiment, in order to obtain more reliable extremal frequency and quasi-steady-state frequency, the dynamic damping factor corresponding to the two frequency indexes is selected as the standard.
The concrete mode is as follows:
simulating mass expected fault scenes by a time domain simulation method, and acquiring system operation data under each disturbance fault; in addition, dynamic damping factors of the ASFR model under the predicted fault scene are adjusted by adopting a bisection method, so that the extreme value frequency and the quasi-steady-state frequency obtained by the ASFR model are consistent with the extreme value frequency and/or the quasi-steady-state frequency of time domain simulation, and further, the dynamic damping factors D corresponding to the extreme value frequency and the quasi-steady-state frequency under the fault scene are determined1And D2. Further obtaining system operation data and dynamic damping factor D of a fault scene1And D2And participating in XGboost model training as a sample. In other possible embodiments, the standard for obtaining the dynamic damping factor may be set according to actual requirements, and the present invention is not limited to this specifically.
In this embodiment, the preferred system operation data includes: the system comprises a generator electromagnetic power/reactive power, a bus voltage amplitude/voltage phase angle, a system active load/reactive load, a line active power, a reserve capacity and a power deficit value after disturbance. Other possible embodiments are not limited to the combinations described above nor to the characteristic quantities described above.
step 3: and performing iterative training on the XGboost model by using the dynamic damping factor data set to establish a nonlinear mapping relation between system operation data and a dynamic damping factor so as to obtain a dynamic damping factor prediction model.
The dynamic damping factor dataset is represented as: d { (x)i,Di) N samples are contained in the data set D, the characteristic quantity of each sample is m, and the input characteristic variable and the real value of the corresponding dynamic damping factor are x respectivelyi、Di. Assuming that the XGBoost model has K regression trees, the XGBoost-based dynamic damping factor prediction model can be expressed as:
Figure BDA0002916518140000091
in the formula: f. ofkRepresenting the kth regression tree. From the above, the predicted value is the sum of the predicted values of the dynamic damping factors of the K regression trees.
The target function of the XGboost is set as follows:
Figure BDA0002916518140000092
in the formula: l () is the function of the error,
Figure BDA0002916518140000093
for measuring dynamic damping factor predicted value
Figure BDA0002916518140000094
With the true value DiThe difference between the two represents the degree of model fitting data;
Figure BDA0002916518140000095
the predicted value of the dynamic damping factor of the ith sample is obtained; diIs the true value of the dynamic damping factor of the ith sample. To prevent overfitting, a regularization term Ω (f) is definedk) By controlling the complexity of the model, the regularization term Ω (f) corresponding to the kth regression treek) The expression of (a) is:
Figure BDA0002916518140000096
in the formula: gamma is the punishment coefficient of the leaf node; epsilon is a regular term coefficient; w 'and w' represent the number of leaves and the weight of leaves of the kth tree, respectively;
in the invention, a forward step algorithm is utilized to train the target function in the training process, and the target function can be expressed as the following predicted value in the t iteration:
Figure BDA0002916518140000097
in the formula: omega (f)t) The regularization term of the regression tree at the t iteration is a regularization term which should satisfy the sum of the regularization terms of all the regression trees at the t iteration:
Figure BDA0002916518140000101
ft(xi) Inputting the characteristic variable x for the ith sample by the regression tree at the t iterationiShould satisfy the sum of the dynamic damping factor predicted values of all the regression trees at the t-th iteration:
Figure BDA0002916518140000102
using a second order Taylor expansion, the function is simplified and the constant term is removed to yield the following equation:
Figure BDA0002916518140000103
Figure BDA0002916518140000104
Figure BDA0002916518140000105
in the formula: giAnd hiFirst and second derivatives of the ith sample loss function, respectively, and then the objective function can be written as:
Figure BDA0002916518140000106
in this embodiment, based on the objective function, a prediction model for predicting the dynamic damping factor may be obtained by performing model training on the basis of the hyper-parameter determination. In order to improve the accuracy of the prediction model, the invention preferably utilizes a Bayesian optimization method to carry out the optimization of the hyperparameter in the iteration process to obtain the optimal hyperparameter of the XGboost.
Bayesian Optimization (BO) is a very effective global Optimization algorithm with the goal of finding a globally optimal solution to the search space as follows:
Figure BDA0002916518140000107
wherein λ is*For the optimized super-parameter vector, lambda is the XGboost model super-parameter vector, chi represents a decision space, and f represents an objective function.
The bayesian optimization framework mainly comprises two core parts, namely a non-parametric Gaussian Process (GP) and an Acquisition Function (AF).
GP is used to build unknown objective function models, which can model a finite set of hyper-parametric vector samples as a multivariate gaussian distribution. In GP, the combination of finite hyperparametric vector samples can be represented as:
f(λ)～GP(m(λ),b(λ,λ′))
in the formula: m (-) represents a mean function; b (λ, λ') is a covariance function defined as:
Figure BDA0002916518140000108
in the formula: and theta is a parameter of parameter adjustment step length.
The new data point is denoted as λtThe corresponding function target value is marked as DtUsing GP characteristics, will
Figure BDA0002916518140000111
And DtUnion, expressed as:
Figure BDA0002916518140000112
in the formula: b ═ B (λ)1,λt),b(λ2,λt)...b(λt0,λt0)]；
Figure BDA0002916518140000113
To measure noise.
When adding a new point λtAfter that, the new prediction distribution is expressed as:
Figure BDA0002916518140000114
the predicted mean and covariance are as follows:
Figure BDA0002916518140000115
Figure BDA0002916518140000116
in Bayesian optimization, the active strategy for selecting the next evaluation point is the acquisition function from the observed data set D1:tThe obtained posterior distribution structure is constructed, and the next evaluation point lambda is selected through the guidance of maximization of the posterior distribution structuret+1：
λt+1＝maxαt(λ；D1:t)
Because the Bayesian optimization algorithm is the existing algorithm and the content of the algorithm is not optimized, the optimization mechanism process is not elaborated in detail. It should be understood that the optimal hyperparameter vector of the XGboost model is determined by using a Bayesian optimization algorithm, and then the XGboost model under the optimal hyperparameter is further trained to obtain a dynamic damping factor prediction model.
step 4: and acquiring system operation data at the moment of eliminating the fault, inputting the system operation data into the dynamic damping factor prediction model determined by step3, and predicting the dynamic damping factor system operation data corresponding to the fault.
Step 5: and predicting extreme value frequency and/or quasi-steady-state frequency after disturbance fault by using the ASFR model.
In the embodiment, the XGBoost is used to establish a dynamic damping factor prediction model, so as to realize real-time correction of the dynamic damping factor in the ASFR model, so that the ASFR model can accurately predict the multidimensional frequency index after the disturbance accident.
Example 2:
the purpose of this embodiment is to obtain a real-time dynamic damping factor, and therefore, this embodiment provides a method for predicting a damping factor in an ASFR model based on XGBoost, which includes the following steps:
step 1: calculating equivalent parameters of the multi-machine system, and further constructing an aggregate multi-machine system frequency response model (ASFR);
step 2: acquiring a dynamic damping factor data set based on a simulation result of a mass of expected fault scenes, wherein the dynamic damping factor data set comprises system operation data and dynamic damping factor data;
and step 3: performing iterative training on the XGboost model by using the dynamic damping factor data set to establish a nonlinear mapping relation between system operation data and a dynamic damping factor so as to obtain a dynamic damping factor prediction model;
the input characteristic variable of the dynamic damping factor prediction model is system operation data, and the output characteristic variable is a dynamic damping factor;
and 4, step 4: and (3) acquiring system operation data at the moment of eliminating the fault, inputting the system operation data into the dynamic damping factor prediction model determined in the step (3), and predicting the dynamic damping factor corresponding to the fault.
The implementation process of each step is consistent with the corresponding step in embodiment 1, and the difference between this embodiment and embodiment 1 is only that frequency prediction is not performed after the dynamic damping factor is obtained in this embodiment. The dynamic damping factor is obtained, and the dynamic damping factor can be applied to system inertia level evaluation, demand reserve evaluation and the like besides related application of frequency prediction.
In other feasible manners, the present invention further provides a prediction system based on the damping factor prediction method or the multidimensional frequency index prediction method, including: the system comprises a polymerization multi-machine system frequency response model building module, a dynamic damping factor data set building module, a dynamic damping factor prediction model building module and a dynamic damping factor prediction module;
the frequency response model building module of the multi-machine system is used for calculating equivalent parameters of the multi-machine system so as to build an ASFR (advanced standard frequency response) model of the multi-machine system;
the dynamic damping factor data set construction module is used for acquiring a dynamic damping factor data set based on a simulation result of a mass of expected fault scenes, and the dynamic damping factor data set comprises system operation data and dynamic damping factor data;
the dynamic damping factor prediction model construction module is used for performing iterative training on the XGboost model by using the dynamic damping factor data set to establish a nonlinear mapping relation between system operation data and a dynamic damping factor so as to obtain a dynamic damping factor prediction model;
and the dynamic damping factor prediction module is used for acquiring system operation data at the moment of fault elimination, inputting the system operation data to the dynamic damping factor prediction model and predicting the dynamic damping factor corresponding to the fault.
In some possible implementations, the prediction system further includes: and the frequency index prediction module is further used for predicting extreme value frequency and/or quasi-steady-state frequency after disturbance fault by using the ASFR model after the dynamic damping factor is input into the ASFR model, wherein when the dynamic damping factor data set construction module acquires the dynamic damping factor data set, the acquired dynamic damping factor is a damping factor corresponding to the extreme value frequency and/or the quasi-steady-state frequency in a fault scene.
For the specific implementation process of each module, please refer to the above method content, which is not described herein again. It should be understood that, the specific implementation process of the above unit module refers to the method content, and the present invention is not described herein in detail, and the division of the above functional module unit is only a division of a logic function, and there may be another division manner in the actual implementation, for example, multiple units or components may be combined or may be integrated into another system, or some features may be omitted, or may not be executed. Meanwhile, the integrated unit can be realized in a hardware form, and can also be realized in a software functional unit form.
In some possible manners, the present invention further provides a prediction system, including a processor and a memory, where the memory stores a computer program, and the processor calls the computer program to implement a prediction method of a dynamic damping factor in the ASFR model based on XGBoost or a step of the multidimensional frequency index prediction method that merges ASFR and XGBoost.
In some possible manners, the present invention provides a storage medium storing a computer program, where the computer program is called by a processor to execute the step of the prediction method for the dynamic damping factor in the ASFR model based on the XGBoost or the multidimensional frequency index prediction method fusing the ASFR and the XGBoost.
The specific implementation process of each step refers to the explanation of the foregoing method.
It should be understood that in the embodiments of the present invention, the Processor may be a Central Processing Unit (CPU), and the Processor may also be other general purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) or other Programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The memory may include both read-only memory and random access memory, and provides instructions and data to the processor. The portion of memory may also include non-volatile random access memory. For example, the memory may also store device type information.
The storage medium is a computer storage medium, which may be an internal storage unit of the controller according to any of the foregoing embodiments, for example, a hard disk or a memory of the controller. The storage medium may also be an external storage device of the controller, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), and the like provided on the controller. Further, the storage medium may also include both an internal storage unit of the controller and an external storage device. The storage medium is used to store the computer program and other programs and data required by the controller. The storage medium may also be used to temporarily store data that has been output or is to be output.
Based on such understanding, the technical solution of the present invention essentially or partially contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
To verify the feasibility and effectiveness of the present invention, an example analysis was performed on the improved new england 10 machine 39 node system. Wherein, the computer program is compiled and finished on the computer by using python and MATLAB, and the computer is configured as follows: CPU Intel Core i5-4200 and memory 8 GB.
Application example 1: the improved new england 10 machine 39 node system is taken as a test system, as shown in fig. 3, wherein the units 33, 34, 35 and 36 are changed into hydroelectric generating units. And performing off-line simulation by using a PSD-BPA platform and an MATLAB platform to generate a dynamic damping factor data set. The load level was adjusted from 75% to 120% with 1% intervals. In order to guarantee the trend convergence, the active output and the reactive output of the generator are correspondingly changed when the load level changes. The fault considers the tripping fault of a single unit and the tripping fault of any two units. Simulating a frequency response process under each active disturbance through PSD-BPA to obtain system operation data, and simultaneously adjusting a dynamic damping factor in the ASFR model of the fault scene by adopting a bisection method to enable the ASFR model to obtain an extreme value frequency and a quasi-steady state frequency andthe PSD-BPA results are consistent, so that the dynamic damping factor D corresponding to the extreme value frequency and the quasi-steady state frequency under each disturbance fault is obtained1And D2. Based on system operating data and dynamic damping factor data (D)1、D2) Constructing a dynamic damping factor data set; finally, 2382 samples are obtained in total, and 1882 samples are used as training samples, and 500 samples are used as testing samples.
FIG. 4 shows that the most suitable dynamic damping factor D corresponding to the extreme frequency and the quasi-steady-state frequency of the test sample is obtained by the bisection method1And D2. As can be seen from FIG. 4, under different disturbance faults, the dynamic damping factors corresponding to the extreme frequency and the quasi-steady-state frequency are different, such as D corresponding to the extreme frequency1In the interval [2-8 ]]And (4) changing. In addition, quasi-steady-state frequency corresponds to D2Overall ratio D1Is large. The dynamic change characteristics of the dynamic damping factors are visually displayed, and the extreme value frequency and/or quasi-steady-state frequency after a disturbance accident are difficult to accurately predict by adopting the fixed dynamic damping factors in the ASFR model.
In order to verify the effectiveness of the method provided by the invention, an XGboost and ASFR model of a fixed dynamic damping factor are used as a comparison algorithm, and the testing and training are carried out on the same data, and the results are shown in tables 1 and 2.
TABLE 1 comparison of extreme frequency prediction accuracy for different methods
Figure BDA0002916518140000141
TABLE 2 comparison of quasi-steady-state frequency prediction accuracy by different methods
Figure BDA0002916518140000142
As can be seen from tables 1 and 2, the ASFR model using the fixed dynamic damping factor has a large prediction error, absolute values of errors corresponding to the predicted extreme frequency and the quasi-steady-state frequency are 1.27Hz and 0.23Hz, and RMSEs corresponding to the predicted extreme frequency and the quasi-steady-state frequency are 0.281Hz and 0.057Hz, respectively, which are difficult to meet the high-precision requirement of online application. The frequency stability information hidden in the high dimensionality is excavated through the XGboost model, the prediction precision is improved to a certain extent, but the XGboost model is poor in interpretability because the training process is a black box. The dynamic damping factor in the ASFR model is corrected through the XGboost model, the strong causal relationship among electrical information is reserved, meanwhile, the prediction precision of the ASFR model is greatly improved, and RMSEs corresponding to the predicted extreme frequency and the quasi-steady-state frequency are 74% and 12% of the XGboost respectively.
Fig. 5 and fig. 6 show the error distributions of the predicted extreme frequency and the quasi-steady-state frequency, respectively, according to the method of the present invention. When the extreme frequency is predicted, the number of samples with the maximum error smaller than 0.05 accounts for 90.6%, the number of samples with the predicted quasi-steady-state frequency error smaller than 0.01 accounts for 94.4%, and the prediction precision is excellent. In addition, although the error is larger than 0.1 when the extreme frequency is predicted, the proportion of the sample is only 1.6%, which is acceptable for the actual power grid.
It should be emphasized that the examples described herein are illustrative and not restrictive, and thus the invention is not to be limited to the examples described herein, but rather to other embodiments that may be devised by those skilled in the art based on the teachings herein, and that various modifications, alterations, and substitutions are possible without departing from the spirit and scope of the present invention.

Claims (10)

1. A dynamic damping factor prediction method in an ASFR model based on XGboost is characterized in that: the method comprises the following steps:
step 1: calculating equivalent parameters of the multi-computer system, and further constructing an aggregate multi-computer system frequency response model ASFR;
step 2: acquiring a dynamic damping factor data set based on a simulation result of a mass of expected fault scenes, wherein the dynamic damping factor data set comprises system operation data and dynamic damping factor data;
and step 3: performing iterative training on the XGboost model by using the dynamic damping factor data set to establish a nonlinear mapping relation between system operation data and a dynamic damping factor so as to obtain a dynamic damping factor prediction model;
the input characteristic variable of the dynamic damping factor prediction model is system operation data, and the output characteristic variable is a dynamic damping factor;
and 4, step 4: and (3) acquiring system operation data at the moment of eliminating the fault, inputting the system operation data into the dynamic damping factor prediction model determined in the step (3), and predicting the dynamic damping factor corresponding to the fault.
2. A multidimensional frequency index prediction method fusing ASFR and XGboost is characterized by comprising the following steps: the method comprises the following steps:
s1: acquiring an aggregate multi-machine system frequency response model ASFR and an XGboost-based dynamic damping factor prediction model which are constructed according to claim 1;
s2: acquiring system operation data at the moment of eliminating the fault, inputting the system operation data into the dynamic damping factor prediction model determined in the step S1 to obtain a dynamic damping factor corresponding to the fault, and inputting the dynamic damping factor into the ASFR model constructed in the step S1;
s3: and predicting extreme value frequency and/or quasi-steady-state frequency after disturbance fault by using the ASFR model.
3. The method of claim 2, wherein: the construction process of the ASFR model comprises the following steps:
calculating equivalent parameters of a multi-machine system and a unit by a weighted dynamic equivalent method;
respectively equating a plurality of units of each type in the multi-unit system to a single unit of the same type based on the equivalent parameters; and recombining into the ASFR model.
4. The method of claim 3, wherein: if the types of the units in the multi-unit system comprise a thermal power unit and a hydroelectric power unit, equating the multiple thermal power units and the multiple hydroelectric power units into one thermal power unit and one hydroelectric power unit respectively, and then constructing an ASFR model, wherein the corresponding equivalent parameters comprise: the equivalent inertia time constant, the equivalent difference adjustment coefficient, the unit standardization gain, the equivalent parameter of the hydroelectric generating set prime motor-speed regulator and the equivalent parameter of the thermal generating set prime motor-speed regulator;
wherein the equivalent inertia time constant H of the multi-machine systemsysIs calculated as follows:
Figure FDA0002916518130000011
in the formula, Hi,sysIs the equivalent inertia time constant of the ith unit, HiRepresenting the inertia time constant of the ith unit; n is the total number of the units; siThe rated power of the ith unit is represented; ssysIndicating the rated power of the multi-machine system;
the equivalent adjustment coefficient R of the multi-machine system is calculated as follows:
Figure FDA0002916518130000021
in the formula, RiIs the equivalent difference adjustment coefficient, K, of the ith unitmiGain for the ith unit, κiEquivalent gain of the ith unit;
the unit normalized gain is calculated as follows:
Figure FDA0002916518130000022
in the formula: k'iThe standard gain of the unit i is obtained;
the equivalent parameters of the prime motor-speed regulator of the thermal power generating unit are calculated as follows:
Figure FDA0002916518130000023
in formula (II), X'1，jIs an equivalent parameter corresponding to the jth parameter in the thermal power generating unit,
Figure FDA0002916518130000024
representing the jth parameter, n, of the ith thermal power generating unit1The number of units of the thermal power generating unit, n3The number of parameters of the thermal power generating unit is set;
the equivalent parameters of the hydroelectric generating set prime motor-speed regulator are calculated as follows:
Figure FDA0002916518130000025
in formula (II), X'2,jIs an equivalent parameter corresponding to the jth parameter in the hydroelectric generating set,
Figure FDA0002916518130000026
representing the jth parameter, n, of the ith hydroelectric generating set2The number n of the units of the hydroelectric generating set4The number of parameters of the hydroelectric generating set is shown.
5. The method of claim 2, wherein: the dynamic damping factor prediction model is as follows:
Figure FDA0002916518130000027
in the formula:
Figure FDA0002916518130000028
for the dynamic damping factor predicted value, f, corresponding to the ith samplek(xi) Represents the prediction value, x, of the dynamic damping factor of the kth tree to the ith sampleiAnd K is the number of the established regression trees for the system operation data corresponding to the ith sample.
6. The method of claim 2, wherein: and determining the optimal hyperparameter of the XGboost model by adopting a Bayesian optimization method in the iterative training process of the XGboost model, and training the XGboost model determined under the optimal hyperparameter again to obtain the dynamic damping factor prediction model.
7. The method of claim 2, wherein: the dynamic damping factor in the dynamic damping factor data set is obtained as follows:
simulating a mass of expected fault scenes by a time domain simulation method, and executing the following processes for each expected fault scene:
adjusting a dynamic damping factor of the ASFR model under a predicted fault scene to enable extreme value frequency and/or quasi-steady-state frequency obtained by the ASFR model to be consistent with extreme value frequency and/or quasi-steady-state frequency obtained by time domain simulation; and further determining a dynamic damping factor corresponding to the extreme value frequency and/or the quasi-steady-state frequency in the expected fault scene.
8. A predictive system, characterized by: the method comprises the following steps: the system comprises a polymerization multi-machine system frequency response model building module, a dynamic damping factor data set building module, a dynamic damping factor prediction model building module and a dynamic damping factor prediction module;
the system comprises a frequency response model building module, a frequency response model calculation module and a frequency response model calculation module, wherein the frequency response model building module of the multi-machine system is used for calculating equivalent parameters of the multi-machine system so as to build an ASFR (amplitude-dependent frequency response) model of the multi-machine system;
the dynamic damping factor data set construction module is used for acquiring a dynamic damping factor data set based on a simulation result of a mass of expected fault scenes, and the dynamic damping factor data set comprises system operation data and dynamic damping factor data;
the dynamic damping factor prediction model construction module is used for performing iterative training on the XGboost model by using the dynamic damping factor data set to establish a nonlinear mapping relation between system operation data and a dynamic damping factor so as to obtain a dynamic damping factor prediction model;
the input characteristic variable of the dynamic damping factor prediction model is system operation data, and the output characteristic variable is a dynamic damping factor;
and the dynamic damping factor prediction module is used for acquiring system operation data at the moment of fault elimination, inputting the system operation data to the dynamic damping factor prediction model and predicting the dynamic damping factor corresponding to the fault.
9. The prediction system of claim 8, wherein: the frequency index prediction module is used for inputting the dynamic damping factor into an ASFR model and then predicting extreme value frequency and/or quasi-steady-state frequency after disturbance fault by using the ASFR model;
when the dynamic damping factor data set construction module acquires the dynamic damping factor data set, the acquired dynamic damping factor is a damping factor corresponding to extreme value frequency and/or quasi-steady-state frequency in a fault scene.
10. A predictive system, characterized by: comprising a processor and a memory, the memory storing a computer program that the processor calls to perform: the method for predicting dynamic damping factors in an ASFR model based on XGBoost as claimed in claim 1 or the method for predicting multidimensional frequency index fusing ASFR and XGBoost as claimed in claim 2.
CN202110103617.6A 2021-01-26 2021-01-26 XGboost-based prediction method and system for dynamic damping factor and multidimensional frequency index in ASFR model Active CN112821420B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110103617.6A CN112821420B (en) 2021-01-26 2021-01-26 XGboost-based prediction method and system for dynamic damping factor and multidimensional frequency index in ASFR model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110103617.6A CN112821420B (en) 2021-01-26 2021-01-26 XGboost-based prediction method and system for dynamic damping factor and multidimensional frequency index in ASFR model

Publications (2)

Publication Number Publication Date
CN112821420A true CN112821420A (en) 2021-05-18
CN112821420B CN112821420B (en) 2022-06-14

Family

ID=75859319

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110103617.6A Active CN112821420B (en) 2021-01-26 2021-01-26 XGboost-based prediction method and system for dynamic damping factor and multidimensional frequency index in ASFR model

Country Status (1)

Country Link
CN (1) CN112821420B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114139809A (en) * 2021-12-03 2022-03-04 国网河南省电力公司经济技术研究院 Dynamic frequency response curve prediction method and system based on XGboost power system
WO2023280316A1 (en) * 2021-07-09 2023-01-12 杨光 Data analysis method based on improved xgboost class method, and pricing method and related device
CN116961033A (en) * 2023-09-18 2023-10-27 昆明理工大学 Wind and water through direct current delivery system frequency characteristic analysis method considering generalized load

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
QINGXIN SHI: "Analytical Method to Aggregate Multi-Machine SFR Model With Applications in Power System Dynamic Studies", 《IEEE TRANSACTIONS ON POWER SYSTEMS》 *
赵荣臻: "电力系统预想故障下动态频率响应特性指标智能化评估方法研究", 《万方》 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2023280316A1 (en) * 2021-07-09 2023-01-12 杨光 Data analysis method based on improved xgboost class method, and pricing method and related device
CN114139809A (en) * 2021-12-03 2022-03-04 国网河南省电力公司经济技术研究院 Dynamic frequency response curve prediction method and system based on XGboost power system
CN116961033A (en) * 2023-09-18 2023-10-27 昆明理工大学 Wind and water through direct current delivery system frequency characteristic analysis method considering generalized load
CN116961033B (en) * 2023-09-18 2023-11-28 昆明理工大学 Wind and water through direct current delivery system frequency characteristic analysis method considering generalized load

Also Published As

Publication number Publication date
CN112821420B (en) 2022-06-14

Similar Documents

Publication Publication Date Title
CN112821420B (en) XGboost-based prediction method and system for dynamic damping factor and multidimensional frequency index in ASFR model
CN111224404B (en) Power flow rapid control method for electric power system with controllable phase shifter
US20210133376A1 (en) Systems and methods of parameter calibration for dynamic models of electric power systems
CN103825269B (en) Rapid probabilistic load flow calculation method considering static power frequency characteristics of electric power system
CN109638815B (en) Method for determining safety and stability prevention control strategy of medium-and-long-term voltage of power system
CN112310980A (en) Safety and stability evaluation method and system for direct-current blocking frequency of alternating-current and direct-current series-parallel power grid
CN115392141B (en) Self-adaptive current transformer error evaluation method
CN105305439A (en) Probability dynamic power flow computing method and system in view of input variable correlation
CN109698505B (en) Regulation and control quantitative mapping calculation method for large power grid static voltage stability online prevention and control
CN115940148A (en) Minimum inertia requirement evaluation method and device, electronic equipment and storage medium
Hamodat et al. Automated Generation Control of Multiple‐Area Electrical System with an Availability‐Based Tariff Pricing Scheme Regulated by Whale Optimized Fuzzy PID Controller
Hu et al. Frequency prediction model combining ISFR model and LSTM network
Cai et al. Data-driven predictive based load frequency robust control of power system with renewables
CN116865343B (en) Model-free self-adaptive control method, device and medium for distributed photovoltaic power distribution network
Razmi et al. Neural network based on a genetic algorithm for power system loading margin estimation
CN117318088A (en) Frequency safety margin calculation method and system for new energy power system
CN116845886A (en) Multi-port autonomous photovoltaic system network construction control method based on model prediction
CN115549093B (en) Method and system for online modeling and oscillation analysis of new energy power system
Yang et al. Simplified Prediction Model of Frequency Nadir for Power Systems Penetrated With Renewable Energy
CN115907192A (en) Method and device for generating wind power fluctuation interval prediction model and electronic equipment
CN114139809A (en) Dynamic frequency response curve prediction method and system based on XGboost power system
CN113721461A (en) New energy unit parameter identification method and system based on multiple test scenes
CN106296461B (en) Power grid parameter identification method based on particle swarm optimization algorithm and local state estimation
Sarkar et al. Fractional Order PID-PSS Design using Hybrid Deep Learning Approach for Damping Power System Oscillations
Li et al. Equivalent Virtual Inertia Estimation of Wind Turbines Based on GS-LSTM

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant