CN115713032A - Power grid prevention control method, device, equipment and medium - Google Patents
Power grid prevention control method, device, equipment and medium Download PDFInfo
- Publication number
- CN115713032A CN115713032A CN202211425797.0A CN202211425797A CN115713032A CN 115713032 A CN115713032 A CN 115713032A CN 202211425797 A CN202211425797 A CN 202211425797A CN 115713032 A CN115713032 A CN 115713032A
- Authority
- CN
- China
- Prior art keywords
- power
- power system
- samples
- model
- grid
- 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.)
- Pending
Links
Images
Landscapes
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a power grid prevention control method, a device, equipment and a medium, wherein the method comprises the steps of inputting the current actual value of the active power output of regional power generation into a preset stability evaluation model, and outputting the operation prediction result of a power system by the stability evaluation model; when the operation prediction result of the power system is that the power system is unstable, inputting the current actual value of the active power output of the region power generation into a preset local proxy model, and identifying key characteristics influencing the operation stability of the power system by using the preset local proxy model; generating a power grid safety boundary linear constraint based on the key characteristics; based on the linear constraint of the safety boundary of the power grid, the active power flow of the power system is adjusted by using a direct current optimal power flow method considering the grid loss, so that the power system operates stably. The mechanism modeling of the power system is not needed, the trained model can directly and real-timely perform transient stability assessment and prevention control generation, online application is supported, and the problem of low calculation speed of the traditional method is solved.
Description
Technical Field
The invention belongs to the field of safety analysis and control of power systems, and particularly relates to a power grid prevention and control method, device, equipment and medium.
Background
The interpretation of the complex model can be generally divided into an ante-hoc interpretation and a post-hoc interpretation, wherein the ante-hoc interpretation is realized by training a model with a simple structure and good interpretability or integrating the interpretability into a specific model structure; post-hoc interpretation refers to interpreting a trained machine learning model by developing interpretable techniques. The post-interpretability can be divided into global interpretability and local interpretability according to different interpretation targets and interpretation objects, the global interpretability is used for helping people to understand the overall logic behind the complex model and the internal working mechanism, and the local interpretability is used for helping people to understand the decision process and decision basis of the machine learning model for each input sample.
The instability of a power system unit is one of main reasons for large-scale accidents, and effective transient stability assessment and accident prevention measures are the key points for maintaining the safe and stable operation of the system. In transient stability control of a power system, there are generally two important means, i.e., preventive control and emergency control, and the two methods are complementary in time. The real-time prediction of transient stability after a fault has extremely high requirements on the prediction speed and precision; if the fault can be evaluated before the fault occurs according to the actual working conditions, a large number of irrelevant working conditions can be avoided from being analyzed, the solving difficulty is reduced, and meanwhile, once the current state is judged to be unsafe, a prevention control strategy still has sufficient time to be formulated.
Generally, the preventive control problem requires solving an optimal power flow model (TSCOPF) containing transient stability constraints. The traditional calculation method based on the physical model has the advantages of accurate calculation and high reliability, but the model contains a nonlinear differential algebraic equation, so that the calculation is complex, the calculation time is long, and the requirement of online calculation is difficult to meet. The data driving method has the advantages of potential rule mining and high operation speed, and can avoid complex time domain equation solution when used for transient stability assessment. The model with simple structure is visual, the obtained rule is easy to explain, but the general prediction accuracy is low; the accuracy of a black box model with a complex structure is generally high, but the decision logic of the black box model is not understandable, so that the black box model is difficult to be practically applied to a power system with high safety sensitivity.
Disclosure of Invention
The invention aims to provide a power grid prevention control method, a device, equipment and a medium, which aim to solve the problem of low calculation speed of the traditional method in the background art.
In order to achieve the purpose, the invention adopts the following technical scheme:
in a first aspect, the present invention provides a power grid preventive control method, including the following steps:
acquiring a current actual value of the active power output of the regional power generation;
inputting the current actual value of the active power output of the regional power generation into a preset stability evaluation model, and outputting an operation prediction result of the power system by the stability evaluation model;
when the operation prediction result of the power system is that the power system is unstable, inputting the current actual value of the regional power generation active power output into a preset local proxy model, and identifying key characteristics influencing the operation stability of the power system by using the preset local proxy model;
generating a power grid safety boundary linear constraint based on the key features;
and based on the linear constraint of the safety boundary of the power grid, performing active power flow adjustment on the power system by using a direct current optimal power flow method considering the grid loss so as to ensure that the power system operates stably.
Further, in the step of inputting the current actual value of the active power output of the regional power generation into a preset stability evaluation model, the training mode of the stability evaluation model is as follows:
acquiring an operation mode sample set of the power system;
constructing an XGboost model; the XGboost model is a decision tree forest model consisting of K decision trees;
and determining an objective function for training the decision tree forest model, and performing iterative training in a gradient lifting mode based on the operation mode sample set to finally obtain a trained stability evaluation model.
Further, the step of obtaining the operation mode sample set of the power system specifically includes:
aggregating generators in the region, and performing Latin hypercube sampling on the aggregated generators to obtain a total value sample of the active output of the regional power generation;
randomly sampling generators in the region to obtain an active power output sample of the generator;
adjusting the generator active output samples based on the region generation active output total value samples to enable the sum of the adjusted generator active output samples to be equal to the region generation active output total value samples;
performing diagnosis and adjustment of branch active power flow out-of-limit on the adjusted generator active output sample to obtain an operation mode sample set of the power system;
the operation mode sample set of the power system comprises a destabilization sample and a stabilization sample.
Further, after the step of obtaining the operation mode sample set of the power system, the method further includes the steps of:
when the unstable samples and the stable samples in the operation mode sample set of the power system are unbalanced, the unstable samples or the stable samples are a few samples;
generating k-neighbor samples of the minority class of samples using a SMOTE method;
randomly selecting a plurality of samples from the k neighbor samples, and respectively synthesizing the samples with the minority samples according to a random proportion to obtain new minority samples;
and adding the new few samples into the operation mode sample set of the power system to obtain an operation mode sample set with a balanced number.
Further, in the step of determining an objective function for training the decision tree forest model, the objective function is J:
in the formula: n and K are respectively the number of samples and the number of decision trees; omega is an L2 regularization term of the decision tree;
is a loss function; f. of k Leaf node values for the input features for the kth decision tree pair.
Further, the step of identifying key features affecting the operation stability of the power system by using the preset local agent model specifically includes:
the local agent model is constructed by adopting a classification regression tree; the classification regression tree is a binary tree, and each branch node of the binary tree has two child nodes;
acquiring a local operation mode sample of the power system;
for the characteristics of the local operation mode sample, starting from a root node of a binary tree, advancing to a child node of the binary tree according to the branch conditions of each branch node until a leaf node is finally reached, and determining the category corresponding to the leaf node as the prediction category of the characteristics of the sample;
and determining key characteristics influencing the operation stability of the power grid according to the prediction categories.
Further, in the step of performing active power flow adjustment on the power system by using the direct current optimal power flow method considering the network loss, an objective function of performing active power flow adjustment by using the direct current optimal power flow method is as follows:
in the formula: s G A set of controllable power generation nodes;
P G,i and respectively adjusting the front and rear generated active power output for the controllable power generation node i.
In a second aspect, the present invention provides a grid preventive control apparatus, including:
the acquisition module is used for acquiring the current actual value of the active power output of the regional power generation;
the prediction module is used for inputting the current actual value of the active power output of the region power generation into a preset stability evaluation model, and the stability evaluation model outputs a power system operation prediction result;
the identification module is used for inputting the current actual value of the active power output of the regional power generation into a preset local proxy model when the operation prediction result of the power system is that the power system is unstable in operation, and identifying key characteristics influencing the operation stability of the power system by using the preset local proxy model;
the generating module is used for generating a linear constraint of a safety boundary of the power grid based on the key characteristics;
and the operation adjusting module is used for performing active power flow adjustment on the power system by using a direct current optimal power flow method considering network loss based on the linear constraint of the safety boundary of the power grid so as to ensure that the power system operates stably.
In a third aspect, the present invention provides an electronic device, which includes a processor and a memory, wherein the processor is configured to execute a computer program stored in the memory to implement the power grid prevention control method.
In a fourth aspect, the present invention provides a computer-readable storage medium storing at least one instruction, which when executed by a processor, implements the grid preventive control method described above.
The invention has the following beneficial effects:
1) According to the power grid prevention control method, the current actual value of the active power output of the regional power generation is input into a preset stability evaluation model, and the stability evaluation model outputs the operation prediction result of the power system; when the operation prediction result of the power system is that the power system is unstable in operation, inputting the current actual value of the active power output of the regional power generation into a preset local proxy model, and identifying key characteristics influencing the operation stability of the power system by using the preset local proxy model; generating a power grid safety boundary linear constraint based on the key characteristics; based on the linear constraint of the safety boundary of the power grid, the active power flow of the power system is adjusted by using a direct-current optimal power flow method considering the grid loss, so that the power system operates stably. The mechanism modeling of the power system is not needed, the trained model can directly and real-timely perform transient stability assessment and prevention control generation, online application is supported, and the problem of low calculation speed of the traditional method is solved.
2) The power grid prevention control method provided by the scheme can cope with diversity and uncertainty of operation modes of the power system, self-adaptive extraction of operation characteristics is carried out, and a complex modeling process of a mechanism model is avoided.
3) The method adopts the interpretable local agent model to carry out the traceability analysis of the result, can help the user to master the important basis of the model decision, and avoids the decision risk of the model.
4) In the step of model construction, the SMOTE technology is adopted to balance the samples, and the training effect under the condition of unbalanced machine learning samples is improved.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this application, illustrate embodiments of the invention and, together with the description, serve to explain the invention and not to limit the invention. In the drawings:
FIG. 1 is a schematic diagram of a two-zone interconnect system in an embodiment of the invention;
FIG. 2 is a diagram of a binary classification decision tree and its two-dimensional region partition diagram in an embodiment of the present invention; wherein, (a) is a binary classification decision tree, and (b) is two-dimensional region division;
FIG. 3 is a flowchart illustrating an online grid stability prevention control based on an interpretable proxy model according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of the SMOTE algorithm in an embodiment of the present invention; wherein, (a) is to choose k near neighbor samples, (b) is to synthesize few samples;
FIG. 5 is a histogram of frequency distribution of the result of partitioned sampling and random sampling according to an embodiment of the present invention;
FIG. 6 is a graph of the contribution of features affecting the stability of an IEEE39 node system in an embodiment of the present invention;
FIG. 7 is a power generation force diagram before and after the prevention control adjustment of the IEEE39 node system in the embodiment of the present invention;
FIG. 8 is a simulation graph of the IEEE39 system branch 16-17 before fault prevention control in an embodiment of the present invention; wherein, (a) is a power angle curve of the synchronous motor, and (b) is a voltage amplitude curve of a system node;
FIG. 9 is a simulation graph of IEEE39 system branch 16-17 after fault prevention control in an embodiment of the present invention; wherein, (a) is a power angle curve of the synchronous motor, and (b) is a voltage amplitude curve of a system node;
fig. 10 is a block diagram of a power grid preventive control apparatus according to an embodiment of the present invention;
fig. 11 is a block diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
The present invention will be described in detail below with reference to the embodiments with reference to the attached drawings. It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict.
The following detailed description is exemplary in nature and is intended to provide further explanation of the invention as claimed. Unless otherwise defined, all technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments according to the invention.
Example 1
The scheme provides a power grid prevention control method, which is based on a limit gradient lifting tree algorithm and carries out decision result traceability analysis through an interpretable proxy model. Under the condition of no artificial feature extraction, the method can learn the deeper internal rule of the measured data and the stability margin, is used for fitting the mapping relation between the output of the generator and the stability, has the advantages of strong convergence capacity and good generalization performance, realizes the data-driven acquisition technology of the power generation re-dispatching prevention control optimization strategy aiming at the minimum control cost under the constraint of an expected fault set, helps the power grid operators to find the power grid operation risk in time and provide decision suggestions, and has better popularization value.
As shown in fig. 3, a power grid preventive control method includes the following steps:
s1, constructing a running mode sample set of a power system required by model training
In the scheme, a partition sampling method is used for obtaining the operation mode sample set of the power system. And during sampling, acquiring data such as active power output of the generator required by sampling from original data such as load prediction.
Specifically, the method comprises the following steps:
firstly, generators in a region are considered in a polymerization mode, and Latin hypercube sampling is carried out on the active output of the region power generation to obtain a total value sample of the active output of the region power generation; as shown in fig. 1, which is a schematic diagram of a two-region interconnection system, the regions A1 are considered as an aggregate, and the regions A2 are considered as an aggregate during sampling.
Then, randomly sampling the generators in the region to obtain an active power output sample of the generator; as shown in fig. 1, the generators in regions A1 and A2 are randomly sampled, respectively.
And then, configuring the active output values of all power plants in the region, and adjusting the values of the active output samples of the generators to ensure that the sum of the adjusted active output samples of the generators is equal to the total active output sample of the region generation.
Then, on the basis of obtaining the adjusted generator active power output sample, the branch active power flow out-of-limit diagnosis and adjustment are carried out, and an operation mode sample set with power flow calculation convergence and reasonable node voltage amplitude can be obtained.
The input characteristics of the run mode samples are shown in table 1.
TABLE 1 Steady-State characteristics of run mode sample inputs
And finally, performing transient stability simulation calculation according to a preset fault set, and labeling an output label of the operation mode sample in the operation mode sample set. And if the operation mode sample can be kept stable under any fault of the preset fault set, the output label of the operation mode sample is a stable sample, and otherwise, the operation mode sample is a destabilized sample.
In some embodiments, there is a case where the unstable samples and the stable samples are not balanced, and in order to equalize the numbers of the unstable samples and the stable samples, the SMOTE method is used for oversampling.
For a few classes of sample features x i SMOTE calculates Euclidean distance from the SMOTE to other samples in a minority sample set to obtain k neighbor samples of the SMOTE, and randomly selects a plurality of samples from the k neighbor samples to be respectively matched with x i And synthesizing new minority samples according to a random proportion, so that the number of the minority samples is increased, and adding the new minority samples into the operation mode sample set of the power system to obtain the operation mode sample set of the power system with balanced number.
Specifically, the calculation expression of the k neighbor sample is as follows:
in the formula:
is a minority class sample x i K neighbor samples of (1);
is composed of x i And
synthesizing a new minority sample according to a random proportion; ξ is a random number that follows a uniform distribution over the interval (0, 1).
In the scheme, whether the transient power angle is unstable or not is used as the standard whether the running mode sample is stable or not.
Specifically, the simulation time is 20s, and the criterion of transient power angle instability is as follows:
in the formula: delta delta max 、
The maximum power angle difference between synchronous motors at the simulation end time and the maximum allowable value thereof are respectively adopted in the scheme
Taken at 500.
S2, training stability evaluation model
Constructing an XGboost model for the acquired operation mode sample set of the power system;
the XGboost model constructed by the scheme is a decision tree forest model consisting of K decision trees, and the output of the decision tree forest model is as follows:
in the formula: x is a radical of a fluorine atom i The characteristics of the ith operation mode sample; f. of k (. Cndot.) is the leaf node value for the input feature for the kth decision tree pair.
1) For the {0,1} dichotomy problem of both stable and unstable, first, the output value z of the decision tree forest model is used i Conversion to probability value between 0 and 1 by Logistic function
The expression is as follows:
2) Selecting a logarithmic loss function as a loss function, wherein the expression is as follows:
in the formula: y is i A label for the ith sample; log (-) is a natural log function.
As output value z of a forest model of a decision tree i And converting into a probability value between 0 and 1 through a Logistic function.
3) Determining an objective function of a forest model of a training decision tree as follows:
in the formula: n and K are respectively the number of samples and the number of decision trees; omega is an L2 regularization term of the decision tree; t is the number of leaf nodes of the decision tree; omega j The value of the jth leaf node; γ and λ are regularization coefficients.
Is a loss function; f. of k Leaf node values of the input features for the kth decision tree pair.
4) And performing iterative training based on the extreme gradient lifting tree algorithm.
In the t-th iteration of the gradient lifting process, f is obtained through training t And adding the data into the current decision tree forest, wherein the objective function in the iterative training is as follows:
performing a second-order Taylor expansion on the above equation to obtain:
in the formula: g i First derivative of the loss function of the decision tree, h i The second derivative of the loss function of the decision tree.
Removing the invariant value in equation (9)
Obtaining an approximate expression of the objective function:
the formula is further simplified to obtain:
in the formula: i is j Is the number set of all samples mapped to leaf node j.
In the process of generating the tree structure, one leaf node is changed into a branch node and two corresponding leaf nodes, and all characteristics of the newly generated branch node are tested. For each feature, a linear search is performed on the segmentation points to obtain an optimal segmentation position with the maximum objective function gain, and then a new branch node is formed by using the optimal feature with the maximum objective function gain and the corresponding optimal segmentation position.
The expression for the gain of the objective function is:
in the formula: I.C. A L 、I R Respectively, the number sets of all samples mapped to the newly generated left and right leaf nodes; i is I L 、I R The union of (a).
And training through the steps to obtain a stability evaluation model, identifying the current actual value of the active power output of the region by using the stability evaluation model, and predicting whether the power system is unstable in operation.
S3, constructing a local proxy model
In the scheme, a classification regression tree model is adopted to construct a decision tree agent model.
It should be noted that the CART classification tree used in the present solution is a binary tree, each branch node has two child nodes, and the branch node can be used for a classification problem or a regression problem, and is called a classification tree when used for classification, and is called a regression tree when used for regression.
For the two-classification problem of stabilization and instability, the schematic diagram of the classification decision tree and its corresponding two-dimensional region partition map is shown in fig. 2: (I) The type is represented as unstable, and the region is unsafe (instacure); and (S) indicates that the category is stable and the region is safe (secure).
And acquiring a local operation mode sample of the power system, and for the characteristics of a certain local operation mode sample, proceeding from a root node to a child node of the local operation mode sample according to the branch conditions of each branch node until a leaf node is finally reached, wherein the category corresponding to the leaf node is the prediction category of the sample. In the scheme, the key characteristics influencing the operation stability of the power grid are determined according to the prediction categories.
The CART classification tree is constructed in the following way:
first, based on the Gini index, the best features and best segmentation points are selected that fit the CART classification tree.
Wherein the expression of the Gini index is as follows:
in the formula: d is a sample set; p is a radical of c For the sample probability of class c in D (estimated as the sample frequency), p for each class c The sum is 1; c is the number of classified categories, and C is 2 in the classification problem of the scheme of stability or instability.
Then, the sample set D is segmented according to whether the characteristic a is smaller than the segmentation point A or not to obtain the sample set D 1 And D 2 The kini index expression corresponding to this segmentation is:
in the formula: d and D 1 |、|D 2 Respectively representing sample sets D and D 1 、D 2 The number of samples.
In the scheme, in order to relieve overfitting of a decision tree model and enhance generalization performance of the model, CART cuts off some subtrees from the bottom end of a completely grown decision tree through a cost complexity pruning algorithm, and the pruning loss function expression is as follows:
Cα(T t )＝C(T t )+α|T t | (15)
in the formula: t is t Is a subtree with t as a root node; | T t L is T t The number of leaf nodes; c (-) is the prediction error of the training data, using the measure of the Gini index in the classification tree; alpha is a regularization coefficient. If a subtree with t as root node is pruned, the root node t becomes the only leaf node, and the loss function is:
C α (t)＝C(t)+α (16)
if the value of the loss function does not increase after pruning the subtree with t as the root node, the subtree with t as the root node should be pruned, that is, the pruning condition is C α (t)≤C α (T t ) The following can be obtained:
for a given α value, the optimal subtree corresponding to this α value can be obtained by pruning according to equation (17).
And after the optimization of the steps, obtaining a local proxy model, and identifying key characteristics influencing the operation stability of the power system according to the local operation mode samples to serve as a stability evaluation result.
S4, generating a preventive control strategy based on the stability evaluation result
In the scheme, the objective function of the active power flow optimization adjustment of the transient stability prevention control is set as follows:
in the formula: s G Is a collection of controllable power generation nodes;
P G,i and respectively adjusting the front and rear generated active power output for the controllable power generation node i.
According to the power grid prevention control method provided by the invention, firstly, an operation mode sample of a power system is generated in a sampling manner, and the steady state characteristic quantity of the operation mode sample is obtained through load flow calculation. And performing time domain simulation calculation according to the preset fault set to mark the label of the operation mode sample, wherein if the operation mode sample can be kept stable under any preset fault, the label of the sample is stable, and otherwise, the label is unstable. Training a stability evaluation model according to the sample data of the operation mode, wherein the stability evaluation model is a complex high-performance transient stability evaluation machine learning black box model, such as a decision tree forest model, and a multilayer neural network model, and is used for online transient stability evaluation. And when the result of the online transient stability evaluation is unstable, identifying key characteristics influencing the operation stability of the power system by using a local agent model, and finally generating a preventive control strategy based on the stability evaluation result.
The present solution is further described below with reference to specific application examples
In step 1, because the instability of the power system is not easy to occur, most of the samples generated by sampling are stable samples, the number of the instability samples is small, the categories are unbalanced, the occupation ratio of the instability samples in the total loss function is small, and the classification effect of the trained model on the instability samples is poor.
In the scheme, it is considered that if the proportion of class imbalance exceeds 4, the data set may be an unbalanced data set, the proportion is about 20.
In this scheme, a new Minority class sample is generated by interpolating the Minority class sample through a Minority class oversampling Technique (SMOTE), as shown in fig. 4.
In the aspect of sample generation, the scheme adopts a partition sampling method. As shown in fig. 1, the active power output of each power plant in the region is further configured by random sampling and adjustment of the total power generation amount after the generator aggregation in the region is considered and the latin hypercube sampling is performed on the active power output of the region, and the result is shown in fig. 5.
Compared with a method for directly performing random sampling, the method has the advantages that the total active power output of the areas A1 and A2 obtained by the subarea sampling method and the transmission power of the transmission section between the areas have a wider range, and the distribution in diversity is not exceeded, so that the advantages of the subarea sampling method are displayed.
By adopting the scheme, the stability of the IEEE39, IEEE118 and a certain provincial scale power grid system is evaluated, and various performances are shown in table 2. It can be seen that this scheme can satisfy the stability screening demand under the grid fault.
TABLE 2 evaluation index results of stability evaluation performance on different example systems
By constructing a local proxy model, key features which have important influence on the stability of the power grid are mined, for example, IEEE39 nodes are taken as an example, and as a result, as shown in fig. 6, feature contribution degrees which influence the stability of an IEEE39 node system are obtained, so that key features which influence the stability of the power grid are obtained.
And generating a power grid safety boundary linear constraint based on a local proxy model, and performing active power flow adjustment by using a direct current optimal power flow method considering grid loss, wherein the corresponding power generation output adjustment condition is shown in fig. 7 and table 3. It is observed that the generator output of nodes 33, 34, 35, and 36 is reduced, and the output of the other generators is increased. The output adjustment of each generator is small, and the distance between the sample to be explained and the safety boundary is short.
As can be seen from fig. 7, such adjustment can reduce the transmission power from the lower right to the upper left in the system diagram, which also coincides with the instability of the current operation mode that the transmission power from the lower right to the upper left in the system diagram is too large.
Table 3 results of preventive control active power adjustment of IEEE39 node system to be explained sample 1
Time domain simulation curves of the IEEE39 node system before and after the preventive control under the expected failure of the branches 16 to 17 are shown in fig. 8 and 9.
Time domain simulation verification shows that the power grid can be unstable under the expected faults of the branches 16-17 and can not be unstable under the expected faults of the branches 2-1 or the branches 26-29. The system after the prevention control cannot be unstable under the expected faults of the branch circuits 16-17, 2-1 or 26-29. Therefore, the power grid system recovers the safe state after the prevention and control by adopting the scheme, and can keep stable operation under the expected faults of the branches 16-17.
Example 2
As shown in fig. 10, based on the same inventive concept as the above embodiment, the present invention further provides a power grid prevention and control apparatus, including:
and the obtaining module is used for obtaining the current actual value of the active power output of the regional power generation.
And the prediction module is used for inputting the current actual value of the active power output of the regional power generation into a preset stability evaluation model, and the stability evaluation model outputs the operation prediction result of the power system.
In the prediction module, the training mode of the stability evaluation model is as follows: acquiring an operation mode sample set of the power system; constructing an XGboost model; the XGboost model is a decision tree forest model consisting of K decision trees; and determining an objective function of the forest model of the training decision tree, and performing iterative training by adopting a gradient lifting mode based on the operation mode sample set to finally obtain a trained stability evaluation model.
The method comprises the following steps of obtaining an operation mode sample set of the power system: aggregating the generators in the region, and performing Latin hypercube sampling on the aggregated generators to obtain a total value sample of the active output of the regional power generation; randomly sampling generators in the region to obtain an active output sample of the generator; adjusting the generator active output samples based on the regional generation active output total value samples to enable the sum of the adjusted generator active output samples to be equal to the regional generation active output total value samples; performing branch active power flow out-of-limit diagnosis and adjustment on the adjusted generator active output sample to obtain an operation mode sample set of the power system; the operation mode sample set of the power system comprises a destabilization sample and a stabilization sample. After the step of obtaining the operation mode sample set of the power system, if the unstable samples and the stable samples in the operation mode sample set of the power system are unbalanced and the unstable samples or the stable samples are minority samples, generating k adjacent samples of the minority samples by using a SMOTE method; randomly selecting a plurality of samples from the k neighbor samples, and respectively synthesizing the samples with the minority samples according to a random proportion to obtain new minority samples; and adding the new few samples into the operation mode sample set of the power system to obtain an operation mode sample set with a balanced number.
Determining an objective function of a training decision tree forest model as J:
in the formula: n and K are respectively the number of samples and the number of decision trees; omega is an L2 regularization term of the decision tree;
is a loss function; f. of k Leaf node values of the input features for the kth decision tree pair.
And the identification module is used for inputting the current actual value of the active power output of the regional power generation into a preset local proxy model when the operation prediction result of the power system is that the power system is unstable, and identifying the key characteristics influencing the operation stability of the power system by using the preset local proxy model.
In the identification module, a local proxy model is constructed by adopting a classification regression tree; the classification regression tree is a binary tree, and each branch node of the binary tree has two child nodes; acquiring a local operation mode sample of the power system; for the characteristics of the local operation mode sample, starting from a root node of the binary tree, advancing to child nodes according to the branch conditions of each branch node until a leaf node is finally reached, wherein the category corresponding to the leaf node is the prediction category of the sample characteristics; and determining key characteristics influencing the operation stability of the power grid according to the prediction categories.
And the generating module is used for generating the linear constraint of the safety boundary of the power grid based on the key characteristics.
And the operation adjusting module is used for performing active power flow adjustment on the power system by using a direct current optimal power flow method considering network loss based on the linear constraint of the safety boundary of the power grid so as to ensure that the power system operates stably.
In the operation adjusting module, the objective function of the direct current optimal power flow method for performing active power flow adjustment is as follows:
in the formula: s G Is a collection of controllable power generation nodes;
P G,i and respectively adjusting the front and rear generated active power output for the controllable power generation node i.
Example 3
As shown in fig. 11, the present invention further provides an electronic device 100 for implementing a grid preventive control method; the electronic device 100 comprises a memory 101, at least one processor 102, a computer program 103 stored in the memory 101 and executable on the at least one processor 102, and at least one communication bus 104. The memory 101 may be used for storing a computer program 103, and the processor 102 implements the steps of the power grid preventive control method of embodiment 1 by running or executing the computer program stored in the memory 101 and calling up data stored in the memory 101. The memory 101 may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required by at least one function (such as a sound playing function, an image playing function, etc.), and the like; the storage data area may store data (such as audio data) created according to the use of the electronic apparatus 100, and the like. In addition, the memory 101 may include a non-volatile memory such as a hard disk, a memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), at least one magnetic disk storage device, a Flash memory device, or other non-volatile solid state storage device.
The at least one Processor 102 may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic, discrete hardware components, etc. The processor 102 may be a microprocessor or the processor 102 may be any conventional processor or the like, and the processor 102 is the control center of the electronic device 100 and connects the various parts of the electronic device 100 with various interfaces and lines.
The memory 101 in the electronic device 100 stores a plurality of instructions to implement a grid prevention control method, and the processor 102 may execute the plurality of instructions to implement:
acquiring a current actual value of the active power output of the regional power generation;
inputting the current actual value of the active power output of the regional power generation into a preset stability evaluation model, and outputting a power system operation prediction result by the stability evaluation model;
when the power system operation prediction result is that the power system is unstable in operation, inputting the current actual value of the regional generation active power output into a preset local proxy model, and identifying key characteristics influencing the operation stability of the power system by using the preset local proxy model;
generating a power grid safety boundary linear constraint based on the key features;
and based on the linear constraint of the safety boundary of the power grid, performing active power flow adjustment on the power system by using a direct current optimal power flow method considering the grid loss so as to ensure that the power system operates stably.
Example 4
The integrated modules/units of the electronic device 100 may be stored in a computer-readable storage medium if implemented in the form of software functional units and sold or used as separate products. Based on such understanding, all or part of the flow in the method according to the embodiments of the present invention may also be implemented by a computer program, which may be stored in a computer-readable storage medium and used for instructing relevant hardware, and when the computer program is executed by a processor, the steps of the above-described embodiments of the method may be implemented. Wherein the computer program comprises computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer readable medium may include: any entity or device capable of carrying computer program code, recording medium, U.S. disk, removable hard disk, magnetic disk, optical disk, computer Memory, and Read-Only Memory (ROM).
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
Finally, it should be noted that: although the present invention has been described in detail with reference to the above embodiments, it should be understood by those skilled in the art that: modifications and equivalents may be made to the embodiments of the invention without departing from the spirit and scope of the invention, which is to be covered by the claims.
Claims (10)
1. A power grid preventive control method is characterized by comprising the following steps:
acquiring a current actual value of the active power output of the regional power generation;
inputting the current actual value of the active power output of the regional power generation into a preset stability evaluation model, and outputting an operation prediction result of the power system by the stability evaluation model;
when the power system operation prediction result is that the power system is unstable in operation, inputting the current actual value of the regional generation active power output into a preset local proxy model, and identifying key characteristics influencing the operation stability of the power system by using the preset local proxy model;
generating a power grid safety boundary linear constraint based on the key features;
and based on the power grid safety boundary linear constraint, performing active power flow adjustment on the power system by using a direct current optimal power flow method considering the grid loss so as to ensure that the power system operates stably.
2. The power grid preventive control method according to claim 1, wherein in the step of inputting the current actual value of the regional generated active power output into a preset stability assessment model, the stability assessment model is trained in a manner that:
acquiring an operation mode sample set of the power system;
constructing an XGboost model; the XGboost model is a decision tree forest model consisting of K decision trees;
and determining an objective function for training the decision tree forest model, and performing iterative training in a gradient lifting mode based on the operation mode sample set to finally obtain a trained stability evaluation model.
3. The power grid preventive control method according to claim 2, wherein the step of obtaining the operation mode sample set of the power system specifically includes:
aggregating generators in the region, and performing Latin hypercube sampling on the aggregated generators to obtain a total value sample of the active output of the regional power generation;
randomly sampling generators in the region to obtain an active power output sample of the generator;
adjusting the generator active output samples based on the region power generation active output total value samples to enable the sum of the adjusted generator active output samples to be equal to the region power generation active output total value samples;
diagnosing and adjusting the out-of-limit branch active power flow of the adjusted generator active output sample to obtain an operation mode sample set of the power system;
the operation mode sample set of the power system comprises a destabilizing sample and a stabilizing sample.
4. The grid preventive control method according to claim 3, further comprising, after the step of obtaining the operation mode sample set of the power system, the steps of:
when the unstable samples and the stable samples in the operation mode sample set of the power system are unbalanced, the unstable samples or the stable samples are a few samples;
generating k-neighbor samples of the minority class of samples using a SMOTE method;
randomly selecting a plurality of samples from the k neighbor samples, and respectively synthesizing the samples with the minority samples according to a random proportion to obtain new minority samples;
and adding the new few samples into the operation mode sample set of the power system to obtain an operation mode sample set with a balanced number.
5. The power grid prevention control method according to claim 2, wherein in the step of determining an objective function for training the decision tree forest model, the objective function is J:
in the formula: n and K are respectively the number of samples and the number of decision trees; omega is an L2 regularization term of the decision tree;
is a loss function; f. of k Leaf node values of the input features for the kth decision tree pair.
6. The power grid preventive control method according to claim 1, wherein the step of identifying key features affecting the operational stability of the power system by using a preset local agent model specifically comprises the following steps:
the local proxy model is constructed by adopting a classification regression tree; the classification regression tree is a binary tree, and each branch node of the binary tree has two child nodes;
obtaining a local operation mode sample of the power system;
for the characteristics of the local operation mode sample, starting from a root node of a binary tree, advancing to a child node of the binary tree according to the branch conditions of each branch node until a leaf node is finally reached, and determining the category corresponding to the leaf node as the prediction category of the characteristics of the sample;
and determining key characteristics influencing the operation stability of the power grid according to the prediction categories.
7. The grid preventive control method according to claim 1, wherein in the step of performing active power flow adjustment on the power system using a direct current optimal power flow method taking grid loss into consideration, an objective function of the direct current optimal power flow method for performing active power flow adjustment is as follows:
in the formula: s. the G Is a collection of controllable power generation nodes;
P G,i and adjusting the front and rear generating active power output for the controllable generating node i respectively.
8. A grid preventive control apparatus, comprising:
the acquisition module is used for acquiring the current actual value of the active power output of the regional power generation;
the prediction module is used for inputting the current actual value of the active power output of the region power generation into a preset stability evaluation model, and the stability evaluation model outputs an operation prediction result of the power system;
the identification module is used for inputting the current actual value of the active power output of the regional power generation into a preset local proxy model when the operation prediction result of the power system is that the power system is unstable in operation, and identifying key characteristics influencing the operation stability of the power system by using the preset local proxy model;
the generating module is used for generating a linear constraint of a safety boundary of the power grid based on the key characteristics;
and the operation adjusting module is used for performing active power flow adjustment on the power system by using a direct current optimal power flow method considering network loss based on the linear constraint of the safety boundary of the power grid so as to ensure that the power system operates stably.
9. An electronic device, comprising a processor and a memory, the processor being configured to execute a computer program stored in the memory to implement the grid prevention control method according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores at least one instruction which, when executed by a processor, implements a grid preventive control method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211425797.0A CN115713032A (en) | 2022-11-14 | 2022-11-14 | Power grid prevention control method, device, equipment and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211425797.0A CN115713032A (en) | 2022-11-14 | 2022-11-14 | Power grid prevention control method, device, equipment and medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115713032A true CN115713032A (en) | 2023-02-24 |
Family
ID=85233144
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211425797.0A Pending CN115713032A (en) | 2022-11-14 | 2022-11-14 | Power grid prevention control method, device, equipment and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115713032A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117215205A (en) * | 2023-11-09 | 2023-12-12 | 国网经济技术研究院有限公司 | DC system control parameter analysis method based on decision tree and ISS theory |
-
2022
- 2022-11-14 CN CN202211425797.0A patent/CN115713032A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117215205A (en) * | 2023-11-09 | 2023-12-12 | 国网经济技术研究院有限公司 | DC system control parameter analysis method based on decision tree and ISS theory |
| CN117215205B (en) * | 2023-11-09 | 2024-02-06 | 国网经济技术研究院有限公司 | DC system control parameter analysis method based on decision tree and ISS theory |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Moreno et al. | Wind speed forecasting approach based on singular spectrum analysis and adaptive neuro fuzzy inference system | |
| Jin et al. | An improved ID3 decision tree algorithm | |
| CN108241873B (en) | A kind of intelligent failure diagnosis method towards pumping plant main equipment | |
| CN112086144B (en) | Molecule generation method, device, electronic equipment and storage medium | |
| Oliinyk et al. | The decision tree construction based on a stochastic search for the neuro-fuzzy network synthesis | |
| CN116029221B (en) | Power equipment fault diagnosis method, device, equipment and medium | |
| CN109066651B (en) | Method for calculating limit transmission power of wind power-load scene | |
| CN115713032A (en) | Power grid prevention control method, device, equipment and medium | |
| Zhao et al. | Efficient performance modeling for automated CMOS analog circuit synthesis | |
| CN107590604B (en) | Coherent unit grouping method and system combining S transformation and 2DPCA | |
| CN113904384B (en) | Power grid transient stability coordination control method and system based on gradient elevator | |
| CN113506186B (en) | Method and device for screening disturbed track of power system, electronic equipment and storage medium | |
| CN113127469B (en) | Filling method and system for missing value of three-phase unbalanced data | |
| Kuo et al. | A new symbiotic evolution-based fuzzy-neural approach to fault diagnosis of marine propulsion systems | |
| CN117394313A (en) | Power system transient stability evaluation method, system, chip and equipment | |
| CN117370568A (en) | Power grid main equipment knowledge graph completion method based on pre-training language model | |
| CN117010373A (en) | Recommendation method for category and group to which asset management data of power equipment belong | |
| CN115936926A (en) | SMOTE-GBDT-based unbalanced electricity stealing data classification method and device, computer equipment and storage medium | |
| CN115952493A (en) | Reverse attack method and attack device for black box model and storage medium | |
| CN116070159A (en) | Method and device for realizing fault diagnosis of main power supply system of airplane under unbalanced sample | |
| CN115940253A (en) | Method, device, equipment and medium for judging operation stability of power grid after fault | |
| CN113111588B (en) | NO of gas turbine X Emission concentration prediction method and device | |
| CN115865713A (en) | Importance ordering method, system and terminal for high-order structure in high-order network | |
| CN112698977B (en) | Method, device, equipment and medium for positioning server fault | |
| Sichevska et al. | Information Technology of Forming the Quality of Art and Technical Design of Books. |
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 |