CN111327082B - Commutation failure prevention control optimization method and system based on SVM-PSO - Google Patents

Commutation failure prevention control optimization method and system based on SVM-PSO Download PDF

Info

Publication number
CN111327082B
CN111327082B CN202010157359.5A CN202010157359A CN111327082B CN 111327082 B CN111327082 B CN 111327082B CN 202010157359 A CN202010157359 A CN 202010157359A CN 111327082 B CN111327082 B CN 111327082B
Authority
CN
China
Prior art keywords
optimization
reactive
commutation failure
generator
reactive power
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.)
Active
Application number
CN202010157359.5A
Other languages
Chinese (zh)
Other versions
CN111327082A (en
Inventor
李莉
林祺蓉
瞿寒冰
刘博�
李常刚
刘玉田
刘会荣
朱元振
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.)
State Grid Corp of China SGCC
Shandong University
Jinan Power Supply Co of State Grid Shandong Electric Power Co Ltd
Original Assignee
State Grid Corp of China SGCC
Shandong University
Jinan Power Supply Co of State Grid Shandong Electric Power Co Ltd
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 State Grid Corp of China SGCC, Shandong University, Jinan Power Supply Co of State Grid Shandong Electric Power Co Ltd filed Critical State Grid Corp of China SGCC
Priority to CN202010157359.5A priority Critical patent/CN111327082B/en
Publication of CN111327082A publication Critical patent/CN111327082A/en
Application granted granted Critical
Publication of CN111327082B publication Critical patent/CN111327082B/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
    • H02J3/38—Arrangements for parallely feeding a single network by two or more generators, converters or transformers
    • H02J3/46—Controlling of the sharing of output between the generators, converters, or transformers
    • H02J3/50—Controlling the sharing of the out-of-phase component
    • 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
    • H02J3/12—Circuit arrangements for ac mains or ac distribution networks for adjusting voltage in ac networks by changing a characteristic of the network load
    • H02J3/16—Circuit arrangements for ac mains or ac distribution networks for adjusting voltage in ac networks by changing a characteristic of the network load by adjustment of reactive power
    • 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
    • H02J3/36—Arrangements for transfer of electric power between ac networks via a high-tension dc link
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E40/00—Technologies for an efficient electrical power generation, transmission or distribution
    • Y02E40/30—Reactive power compensation
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
    • Y02E60/60—Arrangements for transfer of electric power between AC networks or generators via a high voltage DC link [HVCD]

Landscapes

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

Abstract

The invention discloses a commutation failure prevention control optimization method and a commutation failure prevention control optimization system based on SVM-PSO, which comprise the following steps: taking the minimum reactive control quantity and the number of fault nodes triggering commutation failure as a target function, considering steady-state safety constraint, and establishing a multi-target optimization function for optimizing the reactive power output and reactive compensation capacity of the generator; and solving the multi-objective optimization function by adopting an improved particle swarm algorithm with parameter adaptivity to obtain the optimal reactive power output and reactive power compensation capacity of the generator. The invention has the beneficial effects that: and constructing a support vector machine proxy model to evaluate a commutation result, thereby quickly solving a particle adaptive value.

Description

Commutation failure prevention control optimization method and system based on SVM-PSO
Technical Field
The invention relates to the technical field of direct-current commutation failure prevention and control, in particular to a commutation failure prevention and control optimization method and system based on SVM-PSO.
Background
The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art.
In order to realize cross-region optimal configuration of electric energy, a long-distance and large-capacity high-voltage direct-current transmission technology is widely applied, and direct current falling points inside an alternating-current power grid are dense, so that risks are brought to safe and stable operation of a system. The direct current commutation failure caused by the short-circuit fault of the alternating current system is a common fault of the alternating current and direct current systems, and the large-capacity direct current commutation failure can cause serious faults such as direct current blocking and the like, interrupt power transmission and influence the stability of a receiving-end power grid. Therefore, the risk of phase commutation failure of the system when the alternating current side fails is reduced, and the method has important significance for improving the stability of the system.
The prior art indicates that when a three-phase ground fault occurs on the inversion side of a converter, a phase change failure occurs in a system when the voltage drops below a phase change critical voltage in a transient process. The prior art teaches that increasing the reactive support of the system can avoid the voltage falling below the commutation threshold voltage during transient. In the prior art, under the condition of dynamic reactive compensation with the same capacity, the optimal compensation place of the dynamic reactive is determined based on the dynamic reactive-voltage sensitivity, so that the risk of phase change failure of a system is reduced. In the prior art, aiming at reducing the probability of commutation failure, a dynamic reactive power distribution method is provided based on indexes such as a multi-feed interaction factor, a multi-feed effective short circuit ratio and the like on the assumption that the compensation capacity of each reactive compensation node is given. In the prior art, the reactive compensation capacity of the bus at the inversion side is quantitatively optimized according to the voltage drop amplitude, the voltage supporting capacity of a system is improved, and the phase change failure of a direct current system is avoided. In the prior art, the distribution points and the capacities of a plurality of reactive power compensation devices are optimized by calculating a reactive-commutation failure risk sensitivity index and a commutation failure optimization effect index based on time domain simulation.
The inventor finds that quantitative collaborative optimization is not performed on a plurality of reactive control variables in most of the existing research methods, and global optimization of reactive configuration is difficult to realize; and the optimization effect index is calculated based on time domain simulation, so that the problem of long time consumption exists.
Disclosure of Invention
In view of the above, the invention provides a commutation failure prevention control optimization method and system based on an SVM-PSO, which take the minimization of reactive power control quantity and the number of fault nodes triggering commutation failure as targets, take steady-state safety constraints into consideration, and optimize the reactive power output and reactive power compensation capacity of a generator. Aiming at the characteristics of multiple constraint variables and strong nonlinearity of the problem, an improved particle swarm algorithm with parameter adaptivity is provided, and initial particles are generated by adopting Latin hypercube sampling; when the particle adaptive value is calculated, a proxy model for evaluating commutation failure is constructed based on a support vector machine to replace time domain simulation, and the calculation speed of the particle adaptive value is improved.
In some embodiments, the following technical scheme is adopted:
the commutation failure prevention control optimization method based on the SVM-PSO comprises the following steps:
taking the minimum reactive control quantity and the number of fault nodes triggering commutation failure as a target function, considering steady-state safety constraint, and establishing a multi-target optimization function for optimizing the reactive power output and reactive compensation capacity of the generator;
and solving the multi-target optimization function by adopting an improved particle swarm algorithm with parameter adaptivity to obtain the optimal reactive power output and reactive power compensation capacity of the generator, so that the reactive power control of the generator is realized, and the commutation failure is avoided.
In other embodiments, the following technical solutions are adopted:
an SVM-PSO based commutation failure prevention control optimization system comprises:
the device is used for taking the reactive control quantity and the minimum number of fault nodes triggering commutation failure as target functions, considering steady-state safety constraint and establishing a multi-target optimization function for optimizing the reactive power output and reactive compensation capacity of the generator;
and the device is used for solving the multi-objective optimization function by adopting an improved particle swarm algorithm with parameter adaptivity to obtain the optimal reactive power output and reactive power compensation capacity of the generator.
In other embodiments, the following technical solutions are adopted:
a terminal device comprising a processor and a computer-readable storage medium, the processor being configured to implement instructions; the computer readable storage medium is used for storing a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing the commutation failure prevention control optimization method based on SVM-PSO.
In other embodiments, the following technical solutions are adopted:
a computer-readable storage medium, wherein a plurality of instructions are stored, said instructions being adapted to be loaded by a processor of a terminal device and to perform the above-mentioned SVM-PSO-based commutation failure prevention control optimization method.
Compared with the prior art, the invention has the beneficial effects that:
the invention establishes a reactive power control optimization model for commutation failure prevention, and the optimization model adopts Particle Swarm Optimization (PSO) to solve the optimization model by taking the minimized reactive power control quantity and the number of fault nodes triggering commutation failure as targets.
Because the number of the nodes which trigger the commutation failure is difficult to directly analyze and solve, the invention provides an improved particle swarm algorithm with adaptive parameters to coordinate and optimize a plurality of reactive power control variables so as to reduce the number of the nodes which trigger the commutation failure; in the particle updating process, a Support Vector Machine (SVM) proxy model is constructed to evaluate the commutation result for rapidly calculating the number of nodes triggering commutation failure after optimization, so that the particle adaptive value is rapidly solved. And finally, verifying the effectiveness and the rapidity of the algorithm by taking a certain provincial power grid model as an example.
Drawings
FIG. 1 is a flow chart of a conventional PSO-based optimization;
FIG. 2 is a flowchart of SVM-PSO based optimization according to an embodiment of the present invention;
FIG. 3 is a diagram of a provincial power grid model according to an embodiment of the present invention;
FIGS. 4(a) - (b) are respectively reactive power control quantities before and after optimization based on the PSO algorithm in the embodiment of the present invention;
FIG. 5 shows the accuracy F of the proxy model before and after weight adjustment in an embodiment of the present invention1；
FIGS. 6(a) - (b) are respectively the reactive power control amount before and after the optimization in the embodiment of the present invention;
fig. 7 shows the lowest inverter-side voltage value at the time of a fault according to an embodiment of the present invention.
Detailed Description
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the disclosure. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
The embodiments and features of the embodiments of the present invention may be combined with each other without conflict.
Example one
In one or more embodiments, a commutation failure prevention control optimization method based on SVM-PSO is disclosed, which includes:
taking the minimum reactive control quantity and the number of fault nodes triggering commutation failure as a target function, considering steady-state safety constraint, and establishing a multi-target optimization function for optimizing the reactive power output and reactive compensation capacity of the generator;
and solving the multi-objective optimization function by adopting an improved particle swarm algorithm with parameter adaptivity to obtain the optimal reactive power output and reactive power compensation capacity of the generator.
The above process is explained in detail below.
1. Optimization model
1.1 objective function
The prevention control optimization of commutation failure is to optimize the reactive output and the reactive compensation capacity of the generator on the basis of ensuring the steady-state safety of the system. Transient voltage is improved, and the condition that the lowest voltage value of the inverter side of the converter is lower than commutation critical voltage in the transient process is avoided, so that the risk of commutation failure is reduced. According to the above analysis, the present embodiment takes the minimum reactive control amount and the number of fault nodes triggering commutation failure as the objective function, as shown in the following formula:
Figure BDA0002404552690000051
Figure BDA0002404552690000052
in the formula: f. of1Representing the optimization cost of the reactive control quantity; qGiAnd QCiRespectively representing the reactive power output and the reactive compensation capacity of the optimized generator;
Figure BDA0002404552690000053
and
Figure BDA0002404552690000054
respectively representing the reactive power output and the reactive compensation capacity of the generator before optimization; n is a radical ofGAnd NCRespectively representing a node set with optimally adjustable generator reactive output and reactive compensation capacity; w is aqiAnd wciIs the weight of the respective control measure; f. of2The number of nodes which trigger phase commutation failure when a three-phase grounding short circuit fault occurs is represented; r takes 1 to indicate that the commutation is successful, and takes 0 to indicate that the commutation fails; n is a radical ofFRepresenting a set of nodes that set a three-phase short-to-ground fault.
The optimization problem of this embodiment is a multi-objective optimization problem, and to simplify the problem analysis, different weights are set for two optimization objectives to convert the problem into a single-objective optimization problem, as shown in the following formula:
min F＝s1*f1+s2*f2 (3)
in the formula: s1And s2Respectively, the control cost and the weight of the number of commutation failure nodes. As can be seen from equation (3), the smaller the reactive control cost is, and the smaller the number of nodes triggering commutation failure is, the smaller the objective function F is. Combining the above analysis, the multi-objective optimization problem herein translates into a single-objective optimization problem that minimizes the objective function F.
Constraint conditions
For any optimization scheme, the power flow constraint, the control variable constraint and the state variable constraint are satisfied as shown in the following formula:
Figure BDA0002404552690000061
in the formula: qGi，QCi，QLiThe reactive output of the generator, the reactive compensation capacity and the reactive power of the load; n is a radical ofBIs a set of all bus nodes in the grid; u shapei，UjIs the voltage of nodes i and j; gij，ij，BijIs the conductance, phase angle and susceptance of the line between line nodes i and j; pGiAnd PLiIs the active power consumed by the active output and load of the generator; qGiminAnd QGimaxIs the upper and lower limits of the reactive power output of the generator i; qGiminAnd QGimaxRepresenting the upper limit and the lower limit of the reactive compensation capacity of the node i; u shapeiminAnd UimaxRepresenting the upper and lower voltage limits of node i.
2. PSO-based commutation failure prevention control optimization
2.1 reactive control variable and bus Fault set determination
Before optimization, the reactive power control variable to be optimized needs to be determined. Because the reactive power can not be transmitted in a long distance, the effect of the phase change is little influenced by the generator and the reactive power compensation which are too far away from the direct current falling point. The embodiment selects all possible compensation nodes and generators within a certain range from a direct current falling point according to engineering experience as reactive control variables.
To reduce the computational consumptionIn time, a fault node with an optimization effect needs to be selected for calculation f2. Because the corresponding commutation results are unchanged when the front and rear partial nodes are optimized to have faults, if a bus fault close to a direct-current drop point easily triggers commutation failure, the bus fault far away has weak influence on the commutation results, in this embodiment, when the power grid reactive power control variable reaches the upper limit operating state and the current operating state, the bus nodes in the whole grid are traversed respectively, three-phase metallic ground faults are set, a faulty bus node set N which can send out commutation failure before optimization and can not trigger commutation failure after optimization is selectedF。
2.2 reactive power optimization for commutation failure prevention based on particle swarm optimization
Reactive power optimization of commutation failure prevention control is a nonlinear complex system optimization problem, an objective function cannot be directly solved through an analytical method, and PSO has strong adaptability to a multidimensional optimization problem that the objective function cannot be analyzed. According to the analysis, the PSO algorithm is selected to solve the reactive power optimization problem of commutation failure prevention and control.
The update formula of the particles in the traditional PSO algorithm is as follows:
Figure BDA0002404552690000071
Figure BDA0002404552690000072
in the formula: d represents the dimension of the particle; k represents a particle algebra; w represents an inertial weight; r is1，r2Is a random number between two (0, 1); c. C1，c2Representing individual learning factors and social learning factors.
One particle represents a set of control measures, and each dimension of the particle represents one reactive control variable in the control measure. Before calculating the particle adaptive value, the load flow is solved by combining simulation software, and whether the control measure meets the steady-state constraint is verified. If yes, performing time domain simulation to evaluate the commutation result, and calculating the adaptive value of the particle, otherwise, adapting the particleThe response value is a large value, and particles which do not meet the constraint condition can be automatically removed in the iteration process. Updating the particles according to the formula (5) and the formula (6) in the iterative process, and performing maximum iteration number NmaxAs a condition for stopping iteration of the particle, a control optimization flow based on a PSO algorithm is shown in FIG. 1.
2.3 particle swarm algorithm improvements
The initial particle swarm of the traditional PSO algorithm is randomly generated, and when the search range of the initial particle swarm is small, the algorithm is easy to fall into local optimization. Aiming at the problem of commutation failure control, in order to improve the global search capability of the particle swarm, the particle swarm algorithm is improved by generating the initial position of the particle and updating the particle speed.
2.3.1 Generation of initial particle positions
If the particle positions are generated by completely random sampling, the problem of data focusing is easy to occur, and the particle swarm optimization is trapped in local optimization. Latin Hypercube Sampling (LHS) is hierarchical sampling, and the extracted samples are combinations of different control variables at different levels, so that maximization of solution space information can be obtained by using as few sample points as possible. In order to increase the diversity of the initial particle swarm and improve the global search capability of the algorithm, the initial particle swarm is generated based on the LHS in the embodiment.
2.3.2 particle swarm Algorithm with improved parameter update
Parameters w, c of particle swarm optimization1、c2The optimization capability of the algorithm is greatly influenced. The larger inertia weight is beneficial to the particles to fly to the vicinity of the optimal solution, and the convergence speed of the algorithm is accelerated; the smaller inertia weight enables the particle swarm to be finely searched near the optimal solution, and the optimization searching capability of the algorithm is improved. According to the above, the inertia weight of the adjustment particle in this embodiment decreases with the increase of the number of iterations, and the specific update formula is:
Figure BDA0002404552690000081
in the formula: w is amaxIs the upper inertial weight limit; w is aminIs the lower inertia weight limit; t is a particleThe number of current iterations of the cluster; n is a radical ofmaxIs the maximum number of iterations.
c1Representing the ability of the particle to acquire its own flight experience, c2Representing the ability of the particles to acquire a population flight experience. In the early stage of searching, the particles are beneficial to exploring a new solution space according to the flight experience of the particles; in the later searching stage, the algorithm tends to be convergent, and the particles mainly update positions according to group experiences and search towards the global optimal direction. According to the above analysis, the updating formula of the improved learning factor of the present embodiment is:
Figure BDA0002404552690000082
Figure BDA0002404552690000091
in the formula: c. C1sAnd c1eIs the individual learning factor maximum and minimum; c. C2sAnd c1eIs the maximum value and the minimum value of the social learning factor.
3. Rapid calculation of particle adaptation values based on SVM
3.1 construction of proxy model
The evaluation of commutation failure is essentially a two-class problem, and the evaluation of commutation failure is realized by comparing the lowest value of transient voltage calculated by traditional time domain simulation with commutation critical voltage. The prior art indicates that a support vector machine can rapidly realize the classification of nonlinear problems through a kernel function, therefore, an SVM model is used for evaluating commutation failure, a nonlinear RBF kernel function is introduced to map feature data in a low-dimensional space to a high-dimensional space, and an optimal hyperplane is searched to realize the classification of commutation results. The structure of the RBF is as follows:
Figure BDA0002404552690000092
in the formula: γ is a parameter related to the nuclear radius; x is the number ofiIs the center point; x is an arbitrary point. According to the above analysis, SVMIs a single output classifier, and f is calculated2Therefore, an SVM classifier is constructed for each bus in the fault set to evaluate the commutation result of the converter when the bus has a fault.
3.2 sample Generation
Before training the model, a corresponding sample set is generated. In the embodiment, the commutation result of the converter needs to be evaluated under different reactive power control measures, so that a sample set is constructed by using different control measures and corresponding commutation results, wherein the reactive power control measures are used as input characteristics of the proxy model, and the commutation result is used as output.
In order to reduce the sample size and improve the sampling efficiency, the embodiment selects a set of LHS reactive power generation control measures. The samples generated by the LHS are the combination of all the control factors at different levels, and if the samples are between the upper limit and the lower limit of the reactive power control quantity, all the samples are generated by the LHS only once, so that the whole reactive power output of the system cannot be reflected. Therefore, the embodiment divides the upper limit and the lower limit of the reactive power output of the system into a plurality of different intervals according to expert experience, and generates a set of control measures based on the LHS in the intervals.
After the control measures are collected, the lowest voltage value of the inverter side of the converter is obtained based on time domain simulation when the bus fails, the commutation critical voltage of the converter is 0.8pu, and the lowest voltage is binarized to obtain the commutation result of the converter, wherein the commutation result is shown as the following formula:
Figure BDA0002404552690000101
in the formula: r represents a commutation result, 1 represents commutation success, and 0 represents commutation failure; u. ofminIs the lowest voltage value of the inversion side of the converter.
3.3 handling of sample imbalance problem
There may be a data imbalance problem when classifier samples are generated between different reactive outputs of the system. For example, when the system has less reactive power, the number of samples for which commutation fails may be much greater than the number of samples for which commutation succeeds. When the classification error cost given different classes is the same, the classifier tends to predict the class of samples for which commutation failed to be the class for which commutation succeeded, with a higher error rate for the samples for which commutation failed.
Aiming at the problem that the error rate of the classifier is high when the samples are unbalanced, the punishment weight ratio of various samples is adjusted to be the inverse ratio of the number of various samples, so that the generalization capability of the classifier is improved. The penalty weight for misclassification of the subclass samples is improved, the generalization capability of the algorithm is enhanced, and the classification precision is improved. Meanwhile, in order to reasonably measure the generalization capability of the classifier, harmonic average F of precision and recall is used1The classification accuracy of the model can be estimated more accurately. Only if the classification accuracy of the model is sufficiently high, F1Get the larger value, F1The calculation of (a) is shown as follows:
Figure BDA0002404552690000102
Figure BDA0002404552690000103
Figure BDA0002404552690000104
in the formula: precision is the accuracy of the model and is the ratio of true positive examples in the sample predicted as positive examples; recall is the recall rate of the model, and the proportion of samples predicted as positive samples in the positive samples is; TP is the number of samples of the positive type sample predicted as the positive type sample; FP is the number of the positive samples predicted by the negative samples; TN is the number of the reverse samples predicted as reverse samples; FN is the number of positive samples predicted as negative samples.
3.4 SVM-PSO-based optimization method
In order to obtain a proxy model with higher precision, enough samples need to be generated to ensure the precision of the proxy model. In this embodiment, a small number of samples are generated for the agent model, the samples are divided into a training set and a test set, and after the agent model is trained by using data in the training set, the accuracy of the agent model is verified by using the data in the test set. If the precision is not enough, increasing sampling points, regenerating samples, and verifying the precision of the proxy model again until the requirements are met.
And after a high-precision proxy model is obtained, replacing the evaluation commutation failure based on the time domain simulation with the evaluation commutation failure based on the proxy model. Before calculating the particle adaptive value, checking whether the particles meet the steady-state constraint condition through the moisture-relieving flow, if so, evaluating a commutation result by using an SVM proxy model, otherwise, setting the particle adaptive value as a larger value. The optimization process based on SVM-PSO is shown in FIG. 2.
4. Example analysis
4.1 example grid introduction
In the embodiment, a certain provincial power grid is selected as an example to verify the effectiveness and the optimization efficiency based on the SVM-PSO optimization algorithm. The power grid model is shown in fig. 3, wherein three direct current transmission lines exist, the power grid model is optimized for the commutation failure of one direct current transmission line, and the number 61 is the inversion side bus node of the power grid model. Before optimization, the compensation capacity of all reactive compensation nodes is 0, and reactive output is provided by the generator only. And obtaining 19 reactive control variables according to the method of the 2.1 section, wherein the reactive control variables comprise 7 generator sets, 12 reactive compensation nodes and 12 bus nodes with optimization effects.
In order to generate the agent model sample, 12 classifiers are constructed, the reactive power output range of the generator is specified to be 0.8-1.2 of the current output level, and the reactive power compensation capacity is 0-450 Mvar. 1300 samples with different reactive control measures are obtained by each classifier, 975 samples are used as training agent models, and 325 samples are used as test set verification model generalization capability.
4.2 improved PSO Algorithm validation
In this embodiment, an initial particle swarm is generated by using a PSO algorithm based on fixed parameters and an improved particle swarm algorithm based on adaptive parameters and adopting LHS, and an optimal adaptive value of the initial particle is compared with an optimal adaptive value at the end of iteration. Setting the particle population size N to 35, Nmax＝20，wmax＝0.9，wmin＝0.4，c1s＝2，c1e＝1.49445，c2s＝2，c2e1.49445. In calculating the objective function, the present embodiment specifies the weights w of all reactive power control measuresqiAnd wciAre all 1; since the rated transmission power of the direct-current line is 2000Mw, phase commutation failure will cause large power loss to the receiving-end power grid, in order to reduce the risk of phase commutation failure as much as possible, this embodiment is biased to reduce the number of fault nodes triggering phase commutation failure, and f in formula (3)1Weight s of1Is set to 1, f2Weight s of2Set to 2000. The specific comparison results are shown in table 1.
TABLE 1 optimal adaptation value comparison of conventional PSO and improved PSO
Figure BDA0002404552690000121
As can be seen from table 1, the adaptive value of the initial particle has randomness, and when the PSO algorithm reaches the maximum iteration number, the reactive power control quantity of the improved PSO algorithm is reduced by 381.97Mvar compared with the reactive power control quantity of the conventional PSO algorithm, so that the reactive power control cost is reduced.
In order to analyze the difference between the two methods to obtain the controlled variable, the PSO algorithm based on fixed parameters and the improved particle swarm algorithm based on adaptive parameters need to be compared to obtain the controlled variable, as shown in fig. 4(a) - (b).
As can be seen from fig. 4(a) - (b), the reactive power output of the generator obtained by the two optimization algorithms is relatively close, and the reactive power compensation difference of the node 13 is obvious. The No. 61 node is a bus of the inverter side of the converter, the No. 62, 63 and 64 nodes are bus nodes directly connected with the No. 61 node, and the No. 13 node is connected with the No. 63 node through a line, so that reactive support is provided for the bus of the inverter side of the converter. In the optimization result of improving the PSO, the reactive compensation amount of the nodes 61, 62, 63, 64 is larger, and the compensation amounts of other nodes are all zero. This shows that the optimization result obtained by improving the PSO accurately reflects the principle of reactive local compensation, and the overall optimization control cost is low. However, the reactive control amount of the traditional PSO algorithm for the nodes 61-64 is slightly smaller, and instead, a large amount of reactive compensation is added to the node 13, which is contrary to the reactive local compensation principle, so that the reactive control cost is increased. Therefore, the PSO algorithm is improved aiming at the commutation failure control problem, the convergence and the global optimizing capability of the algorithm can be improved, and the system control cost is reduced.
4.3 Effect of varying penalty weights on proxy model accuracy
When the agent model is trained, sample weights are adjusted, a greater penalty weight is given to the subclass samples, and the obtained agent model precision is compared with the obtained agent model precision without the adjusted weights, wherein the agent model precision is shown in fig. 5. The result shows that after the sample weight is adjusted, the precision of the proxy model with more balanced samples is not obviously improved, but the classification precision of the proxy models with the numbers 37, 54 and 90 with large difference between the numbers of positive and negative samples is greatly improved. As can be seen from FIG. 5, the proxy model establishes a correct nonlinear mapping relationship of the function, which can meet the test requirements of the present document.
4.4 improved SVM-PSO Algorithm optimization calculation
Obtaining the optimal particle when the particle swarm optimization iteration is terminated based on the SVM-PSO algorithm, wherein f corresponding to the particle1Is 2012.65, f2Is 0. For analyzing the optimization efficiency based on the SVM-PSO, the time consumption for calculating the adaptive value of the particle based on time domain simulation software and a proxy model is counted, wherein the time consumption is 34.0448s when the time domain simulation is adopted, and the time consumption is 0.6283s when the proxy model is adopted. The calculation efficiency of the proxy model is about 54 times of that of time domain simulation, and the proxy model can remarkably improve the PSO evolution efficiency.
In order to analyze the difference between the SVM and the time domain simulation, in the improved PSO algorithm, the SVM and the time domain simulation are used to evaluate commutation failure, and the particle adaptation value is calculated, and the detailed optimized scheme is shown in fig. 6(a) - (b).
As can be seen from fig. 6(a) - (b), the reactive power outputs of the generators obtained by the two optimization schemes are not very different, and the reactive control quantities of the bus nodes 62, 63, 64 and 61 are basically the same, but the control quantities of the node 65 are slightly different. The above analysis shows that the SVM-based proxy model of the embodiment can correctly reflect the relationship between the reactive power control measure and the commutation failure result.
In order to verify the effectiveness of the optimization scheme obtained based on the SVM-PSO algorithm, simulation software is used for verifying the optimization scheme. After the optimization scheme is adopted, based on load flow calculation, the maximum voltage of all bus nodes is 1.067pu, the minimum voltage is 0.977pu, and the voltage of the No. 65 node is 1.034pu, so that the steady-state constraint condition of the power grid is met.
In order to verify the accuracy of the SVM-PSO optimization result, the number of fault nodes triggering commutation failure before and after optimization needs to be obtained based on time domain simulation. When the phase change critical voltage is evaluated to fail, the lowest voltage value of the inverter side of the converter at the time of failure needs to be obtained, and the specific voltage is shown in fig. 7.
As can be seen from fig. 7, all bus faults in the fault set before optimization trigger a commutation failure, and the lowest voltage value of all bus faults after optimization is greater than 0.8pu, where when No. 54 bus fault occurs, the lowest voltage of the inverter side of the converter is 0.8022pu, and the number of fault nodes that trigger a commutation failure is 0. The data analysis shows that the SVM-PSO-based optimization method can eliminate the number of fault nodes triggering commutation failure.
Therefore, the time consumption of the optimization algorithm based on the SVM-PSO is far less than that of the PSO algorithm based on time domain simulation, and a reasonable reactive power optimization scheme can be obtained.
By combining the above analysis, the present embodiment establishes a commutation failure prevention control optimization model, which takes the minimum control cost and the number of fault nodes triggering commutation failure as the optimization target, and solves the model by using an SVM-PSO-based optimization method.
In the embodiment, the improved PSO algorithm uses LHS sampling to generate initial particle positions and adaptively adjusts algorithm parameters, so that the optimization capability of the algorithm is improved; greater punishment weight is given to the subclass sample, so that the generalization capability of the agent model is improved;
and evaluating the commutation result based on the proxy model in the particle swarm optimization iteration process, so that the solving time of the optimization model is obviously reduced, and the commutation failure can be accurately evaluated.
The example analysis shows that the optimization algorithm based on the SVM-PSO can quickly obtain an effective reactive power control optimization scheme on the basis of ensuring a certain accuracy.
Example two
In one or more embodiments, a commutation failure prevention control optimization system based on SVM-PSO is disclosed, comprising:
the device is used for taking the reactive control quantity and the minimum number of fault nodes triggering commutation failure as target functions, considering steady-state safety constraint and establishing a multi-target optimization function for optimizing the reactive power output and reactive compensation capacity of the generator;
and the device is used for solving the multi-objective optimization function by adopting an improved particle swarm algorithm with parameter adaptivity to obtain the optimal reactive power output and reactive power compensation capacity of the generator.
The specific implementation manner of the above device is the same as the method disclosed in the first embodiment, and is not described again.
EXAMPLE III
In one or more embodiments, a terminal device is disclosed, which includes a server including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor executes the computer program to implement the commutation failure prevention control optimization method based on SVM-PSO in the first embodiment. For brevity, no further description is provided herein.
It should be understood that in this embodiment, the processor may be a central processing unit CPU, and the processor may also be other general purpose processors, digital signal processors DSP, application specific integrated circuits ASIC, off-the-shelf programmable gate arrays FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and so on. 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 may provide instructions and data to the processor, and a portion of the memory may also include non-volatile random access memory. For example, the memory may also store device type information.
In implementation, the steps of the above method may be performed by integrated logic circuits of hardware in a processor or instructions in the form of software.
The commutation failure prevention control optimization method based on the SVM-PSO in the first embodiment may be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules in the processor. The software modules may be located in ram, flash, rom, prom, or eprom, registers, among other storage media as is well known in the art. The storage medium is located in a memory, and a processor reads information in the memory and completes the steps of the method in combination with hardware of the processor. To avoid repetition, it is not described in detail here.
Those of ordinary skill in the art will appreciate that the various illustrative elements, i.e., algorithm steps, described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
Although the embodiments of the present invention have been described with reference to the accompanying drawings, it is not intended to limit the scope of the present invention, and it should be understood by those skilled in the art that various modifications and variations can be made without inventive efforts by those skilled in the art based on the technical solution of the present invention.

Claims (9)

1. The commutation failure prevention control optimization method based on SVM-PSO is characterized by comprising the following steps:
acquiring reactive power output and reactive compensation capacity data of the generator;
taking the minimum reactive control quantity and the number of fault nodes triggering commutation failure as a target function, considering steady-state safety constraint, and establishing a multi-target optimization function for optimizing the reactive power output and reactive compensation capacity of the generator;
solving the multi-target optimization function by adopting an improved particle swarm algorithm with parameter adaptivity to obtain the optimal reactive power output and reactive compensation capacity of the generator, realizing reactive control on the generator and avoiding commutation failure;
the method comprises the steps of establishing a multi-objective optimization function for optimizing the reactive power output and the reactive compensation capacity of the generator, and taking the minimum number of reactive control quantity and fault nodes triggering commutation failure as the objective function, wherein the objective function is as follows:
Figure FDA0003057344520000011
Figure FDA0003057344520000012
in the formula: f. of1Representing the optimization cost of the reactive control quantity; qGiAnd QCiRespectively representing the reactive power output and the reactive compensation capacity of the optimized generator;
Figure FDA0003057344520000013
and
Figure FDA0003057344520000014
respectively representing the reactive power output and the reactive compensation capacity of the generator before optimization; n is a radical ofGAnd NCRespectively representing a node set with optimally adjustable generator reactive output and reactive compensation capacity; w is aqiAnd wciIs the weight of the respective control measure; f. of2The number of nodes which trigger phase commutation failure when a three-phase grounding short circuit fault occurs is represented; r takes 1 to indicate that the commutation is successful, and takes 0 to indicate that the commutation fails; n is a radical ofFRepresenting a node set for setting a three-phase grounding short-circuit fault;
the optimization adopts multi-objective optimization, different weights are set for two optimization targets, and the two optimization targets are converted into a single-target optimization problem, which is shown as the following formula:
minF＝s1*f1+s2*f2 (3)
in the formula: s1And s2Respectively control cost and commutationThe weight of the number of failed nodes;
for any optimization scheme, the power flow constraint, the control variable constraint and the state variable constraint are satisfied as shown in the following formula:
Figure FDA0003057344520000021
in the formula: qGi，QCi，QLiThe reactive output of the generator, the reactive compensation capacity and the reactive power of the load; n is a radical ofBIs a set of all bus nodes in the grid; u shapei，UjIs the voltage of nodes i and j; gij，ij，BijIs the conductance, phase angle and susceptance of the line between line nodes i and j; pGiAnd PLiIs the active power consumed by the active output and load of the generator; qGiminAnd QGimaxIs the upper and lower limits of the reactive power output of the generator i; qGiminAnd QGimaxRepresenting the upper limit and the lower limit of the reactive compensation capacity of the node i; u shapeiminAnd UimaxRepresenting the upper and lower voltage limits of node i;
introducing a nonlinear RBF kernel function to map the feature data in the low-dimensional space to a high-dimensional space, and searching an optimal hyperplane to realize the classification of commutation results; the structure of the RBF is as follows:
K(x,xi)＝exp(-γ||x-xi||2) (5)
in the formula: γ is a parameter related to the nuclear radius; x is the number ofiIs the center point; x is an arbitrary point.
2. The SVM-PSO-based commutation failure prevention, control and optimization method of claim 1, wherein before solving the multi-objective optimization function by using an improved particle swarm optimization algorithm with parameter adaptivity, the method further comprises:
selecting all possible compensation nodes and generators within a set range from a direct current drop point as reactive control variables;
and respectively traversing the bus nodes of the whole network in the running state and the current running state of the power grid when the reactive power control variable reaches the upper limit, setting three-phase metallic grounding faults, and selecting a fault bus node set which can trigger the commutation failure before optimization and can not trigger the commutation failure after optimization.
3. The SVM-PSO-based commutation failure prevention control optimization method of claim 1, wherein the improved particle swarm algorithm with parameter adaptivity comprises: the inertial weight of the adjustment particle decreases as the number of iterations increases.
4. The SVM-PSO-based commutation failure prevention control optimization method of claim 1, wherein the improved particle swarm algorithm with parameter adaptivity comprises:
for the selection of the learning factor, in the early stage of searching, the particle explores a new solution space according to the flight experience of the particle; in the later stage of searching, the algorithm tends to be convergent, and the particles mainly update positions according to group experiences and search towards the globally optimal direction.
5. The SVM-PSO-based commutation failure prevention control optimization method of claim 1, wherein in the particle update process, a support vector machine proxy model is constructed to evaluate the commutation results, and particle adaptive values are calculated and updated until an iteration termination condition is met.
6. The SVM-PSO-based commutation failure prevention control optimization method of claim 1, wherein a harmonic mean F of model accuracy and model recall is used1Estimating the classification accuracy of the model:
Figure FDA0003057344520000031
wherein precision is the accuracy of the model and is the ratio of true positive examples in the sample of the positive examples predicted; recall is the recall rate of the model, and the proportion of samples predicted as positive samples in the positive samples is;
the method comprises the steps of establishing a multi-objective optimization function for optimizing the reactive power output and the reactive compensation capacity of the generator, and taking the minimum number of reactive control quantity and fault nodes triggering commutation failure as the objective function, wherein the objective function is as follows:
Figure FDA0003057344520000032
Figure FDA0003057344520000033
in the formula: f. of1Representing the optimization cost of the reactive control quantity; qGiAnd QCiRespectively representing the reactive power output and the reactive compensation capacity of the optimized generator;
Figure FDA0003057344520000034
and
Figure FDA0003057344520000035
respectively representing the reactive power output and the reactive compensation capacity of the generator before optimization; n is a radical ofGAnd NCRespectively representing a node set with optimally adjustable generator reactive output and reactive compensation capacity; w is aqiAnd wciIs the weight of the respective control measure; f. of2The number of nodes which trigger phase commutation failure when a three-phase grounding short circuit fault occurs is represented; r takes 1 to indicate that the commutation is successful, and takes 0 to indicate that the commutation fails; n is a radical ofFRepresenting a node set for setting a three-phase grounding short-circuit fault;
the optimization adopts multi-objective optimization, different weights are set for two optimization targets, and the two optimization targets are converted into a single-target optimization problem, which is shown as the following formula:
minF＝s1*f1+s2*f2 (8)
in the formula: s1And s2Respectively controlling the weight of the cost and the number of commutation failure nodes;
for any optimization scheme, the power flow constraint, the control variable constraint and the state variable constraint are satisfied as shown in the following formula:
Figure FDA0003057344520000041
in the formula: qGi，QCi，QLiThe reactive output of the generator, the reactive compensation capacity and the reactive power of the load; n is a radical ofBIs a set of all bus nodes in the grid; u shapei，UjIs the voltage of nodes i and j; gij，ij，BijIs the conductance, phase angle and susceptance of the line between line nodes i and j; pGiAnd PLiIs the active power consumed by the active output and load of the generator; qGiminAnd QGimaxIs the upper and lower limits of the reactive power output of the generator i; qGiminAnd QGimaxRepresenting the upper limit and the lower limit of the reactive compensation capacity of the node i; u shapeiminAnd UimaxRepresenting the upper and lower voltage limits of node i;
introducing a nonlinear RBF kernel function to map the feature data in the low-dimensional space to a high-dimensional space, and searching an optimal hyperplane to realize the classification of commutation results; the structure of the RBF is as follows:
K(x,xi)＝exp(-γ||x-xi||2) (10)
in the formula: γ is a parameter related to the nuclear radius; x is the number ofiIs the center point; x is an arbitrary point.
7. An SVM-PSO-based commutation failure prevention control optimization system is characterized by comprising:
the device is used for taking the reactive control quantity and the minimum number of fault nodes triggering commutation failure as target functions, considering steady-state safety constraint and establishing a multi-target optimization function for optimizing the reactive power output and reactive compensation capacity of the generator;
and the device is used for solving the multi-objective optimization function by adopting an improved particle swarm algorithm with parameter adaptivity to obtain the optimal reactive power output and reactive power compensation capacity of the generator.
8. A terminal device comprising a processor and a computer-readable storage medium, the processor being configured to implement instructions; a computer-readable storage medium storing a plurality of instructions adapted to be loaded by a processor and to perform the SVM-PSO-based commutation failure prevention control optimization method of any one of claims 1 to 6.
9. A computer-readable storage medium having stored thereon a plurality of instructions, wherein the instructions are adapted to be loaded by a processor of a terminal device and to perform the SVM-PSO-based commutation failure prevention control optimization method of any one of claims 1 to 6.
CN202010157359.5A 2020-03-09 2020-03-09 Commutation failure prevention control optimization method and system based on SVM-PSO Active CN111327082B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010157359.5A CN111327082B (en) 2020-03-09 2020-03-09 Commutation failure prevention control optimization method and system based on SVM-PSO

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010157359.5A CN111327082B (en) 2020-03-09 2020-03-09 Commutation failure prevention control optimization method and system based on SVM-PSO

Publications (2)

Publication Number Publication Date
CN111327082A CN111327082A (en) 2020-06-23
CN111327082B true CN111327082B (en) 2021-09-03

Family

ID=71169207

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010157359.5A Active CN111327082B (en) 2020-03-09 2020-03-09 Commutation failure prevention control optimization method and system based on SVM-PSO

Country Status (1)

Country Link
CN (1) CN111327082B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112234641B (en) * 2020-09-21 2022-12-23 清华大学 Direct-current commutation failure blocking and preventing control method
CN114629155A (en) * 2022-03-07 2022-06-14 三峡大学 Model for identifying commutation failure of high-voltage direct-current transmission and prediction method
CN115130246A (en) * 2022-07-05 2022-09-30 北京理工大学 Parametric design method for diameters of intake and exhaust valves of diesel engines with different strengthening degrees

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2017042018A (en) * 2015-08-21 2017-02-23 株式会社東芝 Power converter controller
CN110336298A (en) * 2019-08-17 2019-10-15 广东博慎智库能源科技发展有限公司 A kind of idle planing method of the distribution containing distributed generation resource based on integrated intelligent algorithm

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2017042018A (en) * 2015-08-21 2017-02-23 株式会社東芝 Power converter controller
CN110336298A (en) * 2019-08-17 2019-10-15 广东博慎智库能源科技发展有限公司 A kind of idle planing method of the distribution containing distributed generation resource based on integrated intelligent algorithm

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
含分布式电源的配网无功优化混合算法;王卫平等;《电力系统及其自动化》;20131215;第25卷(第6期);第93-100页 *

Also Published As

Publication number Publication date
CN111327082A (en) 2020-06-23

Similar Documents

Publication Publication Date Title
CN111327082B (en) Commutation failure prevention control optimization method and system based on SVM-PSO
Arya et al. Distributed generation planning using differential evolution accounting voltage stability consideration
Song et al. A novel variable forgetting factor recursive least square algorithm to improve the anti-interference ability of battery model parameters identification
Hou et al. Research on practical power system stability analysis algorithm based on modified SVM
CN110138000B (en) Hybrid direct-current power transmission system control parameter optimization method based on robust multi-objective optimization algorithm
Guo et al. Parameter identification for static var compensator model using sensitivity analysis and improved whale optimization algorithm
CN103530440A (en) Micro-grid harmonic suppression method based on particle swarm optimization algorithm
CN109449947A (en) Isolated island micro-capacitance sensor reactive power/voltage control capability assessment method and its optimization method
Shukla et al. Grey wolf optimization-tuned convolutional neural network for transmission line protection with immunity against symmetrical and asymmetrical power swing
CN115375204A (en) Performance evaluation method for vehicle-mounted intelligent micro-grid
Zhu et al. A noise‐immune model identification method for lithium‐ion battery using two‐swarm cooperative particle swarm optimization algorithm based on adaptive dynamic sliding window
CN117973859A (en) Risk prevention and control method, device, equipment and medium for main distribution network
CN117452287A (en) SO-SVM transformer fault diagnosis method based on dissolved gas sample integration
CN105119285A (en) Wind power storage coordination multi-objective optimization control method based on dynamic weighting
CN114784800B (en) Power distribution network resonance stability analysis method and system
CN116799826A (en) Distributed photovoltaic energy storage system optimal configuration method and related device
CN116826755A (en) Urban power grid partitioning method and system based on adjustable resource analysis and weak link identification
Chen et al. Power system on-line transient stability prediction by margin indices and random forests
Wang et al. Transient stability assessment in bulk power grids using sequential minimal optimization based support vector machine with pinball loss
CN117723983A (en) Energy storage system SOC detection method and device and electronic equipment
Hamid et al. Efficient power scheduling via stability index based tracing technique and blended crossover continuous ant colony optimization
CN114884066A (en) Method and system for determining static voltage stability of power system
Li et al. Parameter Identification of Lithium Battery Model Based on Improved Fireworks Algorithm
CN116231646B (en) PMU optimal configuration method and system based on electric power system weakness and economy
Yao et al. Cycle life prediction of lithium ion battery based on DE-BP neural network

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