A Method, Device and Storage Medium for Optimizing Grid Voltage Stability Control Parameters

Abstract

A method for optimizing power grid voltage stability control parameters, belonging to the technical field of power system control, solves the problem of how to improve the voltage stability ability of a new power system in a complex environment. The present invention uses Gaussian kernel function probability density estimation to fit new energy random variables to obtain a probability density function, and obtains a random scenario characterizing the randomness of new energy. Based on the density clustering algorithm, the random scenarios are classified with the specific parameter values of the random variables as features, and typical scenarios of the random variables are generated. Obtain typical scenarios under the coupling of new energy randomness and faults. Based on the typical scenarios, a system control parameter optimization model is established. The snake optimization algorithm is used to solve the system control parameter optimization model. The present invention can obtain voltage stable operation control parameters that make the system have stronger robustness, so as to improve the voltage stability ability of the new power system in a complex environment.

Description

Technical Field

[0001] The present invention belongs to the technical field of novel power system control, and relates to a method, a device and a storage medium for optimizing power grid voltage stability control parameters based on unsupervised learning. Background Art

[0002] The new power system has obvious "double high" characteristics of a high proportion of new energy and a high proportion of power electronics. The randomness of new energy has brought unprecedented challenges to the stability of system voltage. Especially when encountering extreme weather, power system failures occur frequently, further exacerbating the complexity and severity of stability problems. Therefore, in the context of the new power system, fully considering the randomness of new energy and optimizing system control parameters under fault conditions is of great practical significance for ensuring the stable operation of the power system.

[0003] At present, the characterization methods for the randomness of renewable energy output are mainly divided into probability model methods and machine learning methods. Among them, probability statistics methods mainly include: probability distribution models (such as normal distribution, Weibull distribution), time series analysis (such as autoregressive model (AR), autoregressive integrated moving average model (ARIMA)), correlation analysis (such as autocorrelation function (ACF)), random process models (such as Markov chain (Markov Chain), stochastic differential equation (SDE)), non-parametric statistical methods (such as kernel density estimation (KDE)). The machine learning methods used to characterize the randomness of renewable energy output mainly include: Variational Autoencoder (VAE), Long Short-Term Memory (LSTM), Generative Adversarial Network (GAN), Random Forest (RF), Support Vector Machine (SVM), and Convolutional Neural Network (CNN) and Graph Neural Network (GNN) based on deep learning.

[0004] However, in terms of system control parameter optimization, existing technical research is limited to certain deterministic scenarios. There are also a small number of works that study random scenario generation methods, but these studies do not consider the system control parameter optimization problem under random scenarios and fault conditions.

[0005] In the existing technology, control strategy optimization models are usually only formulated for certain deterministic scenarios in random scenarios, and the impact of renewable energy randomness on system voltage stability is not fully considered. In addition, in the study of power system control parameter optimization, the existing technology often ignores the combined impact of renewable energy randomness and fault coupling scenarios.

[0006] 1) When characterizing the randomness of new energy, existing technologies fail to effectively classify and analyze a large number of scenarios caused by randomness. If only a few scenarios are randomly selected for control parameter optimization, it is difficult to ensure that the optimization results perform well in all scenarios; if all scenarios are considered, the computational complexity will be significantly increased, and may even make the calculation impossible to complete, making it difficult to achieve the optimization requirements in practical applications.

[0007] 2) Existing control strategies lack consideration of fault scenarios. When the system encounters disturbances caused by faults, it is difficult to achieve effective fault ride-through capability, resulting in poor robustness of the system in the face of faults and failure to meet voltage stability requirements. Summary of the invention

[0008] The technical solution of the present invention is used to solve the problem of how to improve the voltage stability capability of a new power system in a complex environment.

[0009] The present invention solves the above technical problems through the following technical solutions:

[0010] The present invention provides a method for optimizing power grid voltage stability control parameters, comprising:

[0011] S1 uses Gaussian kernel function probability density estimation to fit the new energy random variables to obtain the probability density function, and generates the specific value of the random variable through the probability density function, so as to obtain the random scene that characterizes the randomness of the new energy;

[0012] S2 classifies random scenarios based on the density clustering algorithm with the specific parameter values ​​of random variables as features to generate random typical scenarios;

[0013] S3 obtains typical scenarios under the coupling of new energy randomness and faults;

[0014] S4 Establish system control parameter optimization model based on typical scenarios;

[0015] S5 uses the snake optimization algorithm to solve the system control parameter optimization model.

[0016] Furthermore, the method of fitting the random variable to obtain the probability density function using the Gaussian kernel function probability density estimation is as follows: for each estimation point, the weights of all sample points are calculated, and the weights are accumulated and multiplied by the normalization factor to obtain the final probability density function estimate.

[0017] Furthermore, based on the density clustering algorithm, the random scenes are classified with the specific parameter values ​​of the random variables as the features, and the method of generating random typical scenes is as follows:

[0018] Set neighborhood radius , then set Minimum number of points , a certain point of The points whose neighborhood satisfies formula (3) hour, and Then they are grouped into one category;

[0019] (3)

[0020] In the formula, Yes and Point The Euclidean distance of express of The number of neighbors, is the sample space;

[0021] like ,but As the core point, if lie in of In the neighborhood, and is the core point, then from It is density-reachable. We propagate under this density-reachable condition, cluster the samples, and thus cluster the scenes generated by random variables. Finally, we get The cluster center is A random typical scene.

[0022] Furthermore, the method to obtain typical scenarios under the randomness of new energy and fault coupling is: The random typical scenarios are traversed for N-1 faults, and the stability index under the fault scenario is extracted as the feature. Then, the random typical scenarios are clustered using the density clustering algorithm to obtain the common features of the typical scenarios considering the randomness of new energy and fault coupling. indivual.

[0023] Furthermore, a method for establishing a system control parameter optimization model based on typical scenarios is as follows: through the mathematical relationship model of new energy power generation and the system model, a mathematical relationship model between new energy random variables and system control parameters and system voltage stability indicators is obtained, thereby obtaining a system control parameter optimization model under typical scenarios.

[0024] Furthermore, the new energy random variables include: random variables of wind power generation and random variables of photovoltaic power generation. The random variables of wind power generation are wind speed, and the random variables of photovoltaic power generation are temperature and light intensity.

[0025] Furthermore, the formula of the system control parameter optimization model is as follows:

[0026] (8)

[0027] In the formula, It represents the deviation variable in the optimization target, which is used to measure the degree of deviation of the system operation status; represents the number of typical scenarios considering the randomness of new energy and fault coupling, Represents the set of all nodes in the system. Representation Node The active power output of the generator, Representation Node Active power demand of the load; Representation Node The reactive power output of the generator, Representation Node Reactive power demand of the load; Indicates line The apparent power, Indicates line The apparent power limit of Represents the conductance of the network admittance matrix, reflecting the line Active power transfer capability, Represents the admittance matrix, reflecting the line Reactive power transfer capability; and Respectively represent nodes and nodes The voltage amplitude, Representation Node and nodes The phase angle difference, is the objective function, For Node The active power is minimum at For Node The maximum reactive power at For Node The minimum reactive power at For Node The maximum reactive power at , They are node space and branch space respectively.

[0028] Furthermore, the method of solving the system control parameter optimization model using the snake optimization algorithm is as follows:

[0029] The population size and maximum number of iterations of the initial control optimization parameters are and , randomly initialize the position of each snake in the population;

[0030] The population Divided into male , female indivual;

[0031] Defines the evaluation parameter temperature for the iteration and food quantity ;

[0032] When there is no food, the food search phase is carried out and the snake position is updated;

[0033] When food is available, there is a food-foraging phase and a fighting-mating phase;

[0034] Update the snake's worst position.

[0035] The present invention also provides a device, including a memory and a processor, wherein the memory is used to store a program that supports the processor to execute the above-mentioned grid voltage stability control parameter optimization method, and the processor is configured to execute the program stored in the memory.

[0036] The present invention also provides a storage medium on which a computer program is stored. When the computer program is run by a processor, the steps of the above-mentioned method for optimizing the parameters of power grid voltage stability control are executed.

[0037] The advantages of the present invention are:

[0038] The present invention adopts Gaussian kernel function probability density estimation to fit new energy random variables to obtain probability density functions, generates specific values ​​of random variables through probability density functions, and thus obtains random scenarios that characterize the randomness of new energy; classifies random scenarios based on density clustering algorithms with specific parameter values ​​of random variables as features, generates typical scenarios of random variables, namely, random typical scenarios; obtains typical scenarios under the randomness and fault coupling of new energy; establishes a system control parameter optimization model based on typical scenarios; solves the system control parameter optimization model using a snake optimization algorithm; the present invention takes the randomness of new energy output and fault coupling as features, obtains typical scenarios that take into account the randomness and faults of new energy through a clustering method, and then constructs a control parameter optimization model based on the typical scenarios, and optimizes the system control parameters using an intelligent optimization algorithm. The present invention can obtain voltage stability operation control parameters that make the system more robust, so as to enhance the voltage stability capability of the new power system in a complex environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 is a flow chart of a method for optimizing a grid voltage stability control parameter of the present invention;

[0040] Figure 2 is a structural diagram of a direct-drive wind power generation system model of the present invention;

[0041] Figure 3 is a model structure diagram of the photovoltaic power generation system of the present invention;

[0042] Figure 4 is a system structure diagram of a standard example of the present invention;

[0043] FIG5 (a) is a schematic diagram of the clustering result of the present invention;

[0044] FIG5( b ) is a schematic diagram of a typical scene obtained by clustering the simulation results of the present invention;

[0045] Figure 6 It is a schematic diagram of control parameters obtained by the control parameter optimization simulation model constructed through typical scenarios of the present invention. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in combination with the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and specific embodiments:

[0048] Embodiment 1

[0049] like Figure 1 As shown, a method for optimizing power grid voltage stability control parameters based on unsupervised learning of the present invention comprises the following steps:

[0050] Step 1: Use Gaussian kernel function probability density estimation to fit the new energy random variable to obtain the probability density function, and generate the specific value of the random variable through the probability density function to obtain the random scene that characterizes the randomness of the new energy.

[0051] The random variables include random variables of wind power generation and random variables of photovoltaic power generation. The random variables of wind power generation are wind speed, and the random variables of photovoltaic power generation are temperature and light intensity.

[0052] The method of using Gaussian kernel function probability density estimation to fit random variables to obtain probability density function is as follows:

[0053] For each estimated point , calculate the weights of all sample points , and then multiply it by the normalization factor , thus obtaining the final probability density function estimate , the specific formula is as follows:

[0054] (1)

[0055] (2)

[0056] In the formula, and denote the mean and standard deviation of the Gaussian kernel function, respectively. and represent the number of samples and kernel width respectively, Represents the i-th estimation point, according to the empirical formula: .

[0057] Step 2: Classify random scenarios based on the density clustering algorithm using the specific parameter values ​​of random variables as features to generate typical scenarios of random variables, namely, random typical scenarios.

[0058] Set neighborhood radius , used to determine the neighborhood range around a point, and then set Minimum number of points , used to indicate The minimum number of points required to form a sufficiently dense area within the neighborhood.

[0059] Some point of The points whose neighborhood satisfies formula (3) hour, and Then they are grouped into one category.

[0060] (3)

[0061] In the formula, Yes and Point The Euclidean distance of express of The number of neighbors, is the sample space.

[0062] like ,but As the core point, if lie in of In the neighborhood, and is the core point, then from The density clustering algorithm propagates under this density-reachable condition, clusters the samples, and thus clusters the scenes generated by random variables, and finally obtains The cluster center is A random typical scene.

[0063] Step 3: Obtain typical scenarios under the randomness and fault coupling of new energy sources.

[0064] right The random typical scenarios are traversed for N-1 faults, and the stability index under the fault scenario is extracted as the feature. Then, the random typical scenarios are clustered using the density clustering algorithm to obtain the common features of the typical scenarios considering the randomness of new energy and fault coupling. indivual.

[0065] Step 4: Establish a system control parameter optimization model based on typical scenarios.

[0066] The wind turbine grid-connected system mainly includes new energy units, converters, transformers, transmission lines, loads and power grids. The new energy units include direct-drive wind power generation and photovoltaic power generation systems. Taking the direct-drive wind power generation system as an example, the direct-drive wind power generation system model is as follows: Figure 2 As shown in Figure 2, the photovoltaic power generation system model is as follows: Figure 3 shown.

[0067] Wind power is converted into mechanical power through wind turbines, and the relationship is:

[0068] (4)

[0069] In the formula, is the mechanical power, is the air density, is the swept area of ​​the wind turbine, is the wind energy utilization coefficient (ratio to blade tip speed and pitch angle Related), is the wind speed.

[0070] The relationship between mechanical power conversion to grid-side inverter DC voltage is:

[0071] (5)

[0072] In the formula, It is the conversion rate of mechanical energy into electrical energy.

[0073] In Sinusoidal Pulse Width Modulation (SPWM), the relationship between AC output voltage and DC voltage is:

[0074] (6)

[0075] In the formula, is the modulation ratio, which is determined by the modulation signal and the carrier signal.

[0076] (7)

[0077] In the formula, At standard light intensity and reference temperature The DC voltage under can be determined by the voltage at the maximum power point, is the temperature coefficient, which can be determined by the open circuit voltage and the temperature change difference.

[0078] Through the above-mentioned mathematical relationship model of renewable energy power generation and system model, we can obtain the mathematical relationship model between renewable energy random variables and system control parameters and system voltage stability index, and thus obtain the control parameter optimization model under typical scenarios. The control parameter optimization model is shown in formula (8):

[0079] (8)

[0080] In the formula, It represents the deviation variable in the optimization target, which is used to measure the degree of deviation of the system operation status; represents the number of typical scenarios considering the randomness of new energy and fault coupling, Represents the set of all nodes in the system. Representation Node The active power output of the generator, Representation Node Active power demand of the load; Representation Node The reactive power output of the generator, Representation Node Reactive power demand of the load; Indicates line The apparent power, Indicates line The apparent power limit of Represents the conductance of the network admittance matrix, reflecting the line Active power transfer capability, Represents the admittance matrix, reflecting the line Reactive power transfer capability; and Respectively represent nodes and nodes The voltage amplitude, Representation Node and nodes The phase angle difference, is the objective function, For Node The active power is minimum at For Node The maximum reactive power at For Node The minimum reactive power at For Node The maximum reactive power at , They are node space and branch space respectively.

[0081] Step 5: Use the snake optimization algorithm to solve the system control parameter optimization model

[0082] The Snake optimization algorithm (SOA) is used to solve the problem. The population size and maximum number of iterations of the initial control optimization parameters are and , the position of each snake in the population is randomly initialized as:

[0083] (9)

[0084] In the formula, For the The position of the snake, yes A random number in the range, and are the upper and lower bounds of the snake's position respectively.

[0085] The population Divided into male , female , namely:

[0086] (10)

[0087] (11)

[0088] Defines the evaluation parameter temperature for the iteration and food quantity As shown in formulas (12) and (13):

[0089] (12)

[0090] (13)

[0091] In the formula, is the current iteration number, is a constant, and its empirical value can be 0.5.

[0092] The first stage is the food search stage when there is no food. ( To set the parameter, we can take 0.25 according to experience), and update the snake position by the following formula:

[0093] (14)

[0094] (15)

[0095] In the formula, , are the positions of male and female snakes, , are the random positions of male and female snakes, , They are , The fitness value of , They are , The fitness value of is a constant, and its empirical value is 1.

[0096] The second stage, when food is available, is the food searching stage and the fighting and mating stage.

[0097] when and When , the position update formula is:

[0098] (16)

[0099] In the formula, is the position of the individual snake, The optimal position for individual snakes, is a constant, and its empirical value is 2. , And random number Entering combat mode, the positions of male and female snakes are updated as shown below:

[0100] (17)

[0101] (18)

[0102] In the formula, and The best positions for male and female snakes, and are the fitness values ​​of the best positions for female and male snakes, respectively.

[0103] when , And random number When the male and female snakes enter the mating mode, the positions of the male and female snakes are updated as shown below:

[0104] (19)

[0105] (20)

[0106] In the third stage, the worst position of the snake is updated as shown below:

[0107] (twenty one)

[0108] (twenty two)

[0109] In the formula, and This is the best position for male and female snakes.

[0110] The optimal control parameters of the system can be obtained through the three stages of the snake optimization algorithm.

[0111] Case Analysis

[0112] A comparative analysis of the voltage stabilization effect is performed on a standard example using the control parameters before and after optimization by the method of the present invention. In this example, only the illumination intensity and temperature of photovoltaic power generation are considered as random variables for analysis. The structure of the standard example is as follows: Figure 4 As shown. Based on the light intensity (300-800 W / m²) and temperature (0-40℃) range in the Sichuan Basin, 1440 groups of random variable samples of photovoltaic power generation are randomly generated by Python, and these samples are clustered using the method of the present invention to obtain the clustering results shown in Figure 5 (a). Subsequently, the cluster centers in Figure 5 (a) are used as random typical scenarios for batch simulation, and their stability indicators are calculated. The simulation results are clustered again, and finally the typical scenario shown in Figure 5 (b) is obtained. Figure 6 The control parameters obtained by the control parameter simulation model constructed through typical scenarios and the voltage stability simulation results under the random variable disturbance of photovoltaic power generation are shown. As can be seen from the figure, the control parameters optimized by the present invention show stronger robustness in dealing with random variable disturbances.

[0113] Embodiment 2

[0114] A device includes a memory and a processor, wherein the memory is used to store a program that supports the processor to execute the grid voltage stability control parameter optimization method in embodiment 1, and the processor is configured to execute the program stored in the memory.

[0115] Embodiment 3

[0116] A storage medium stores a computer program, which executes the steps of the method for optimizing power grid voltage stability control parameters in embodiment 1 when the computer program is executed by a processor.

[0117] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing power grid voltage stability control parameters, characterized in that: include: S1 uses Gaussian kernel function probability density estimation to fit the new energy random variables to obtain the probability density function, and generates the specific value of the random variable through the probability density function, so as to obtain the random scene that characterizes the randomness of the new energy; S2 classifies random scenarios based on the density clustering algorithm with the specific parameter values ​​of random variables as features to generate random typical scenarios; S3 obtains typical scenarios under the coupling of new energy randomness and faults; S4 establishes a system control parameter optimization model based on typical scenarios as follows: In the formula, It represents the deviation variable in the optimization target, which is used to measure the degree of deviation of the system operation status; represents the number of typical scenarios considering the randomness of new energy and fault coupling, Represents the set of all nodes in the system. Representation Node The active power output of the generator, Representation Node Active power demand of the load; Representation Node The reactive power output of the generator, Representation Node Reactive power demand of the load; Indicates line The apparent power, Indicates line The apparent power limit of Represents the conductance of the network admittance matrix, reflecting the line Active power transfer capability, Represents the admittance matrix, reflecting the line Reactive power transfer capability; and Respectively represent nodes and nodes The voltage amplitude, Representation Node and nodes The phase angle difference, is the objective function, For Node The active power is minimum at For Node The maximum reactive power at For Node The minimum reactive power at For Node The maximum reactive power at , They are node space and branch space respectively; S5 uses the snake optimization algorithm to solve the system control parameter optimization model.

2. The method for optimizing grid voltage stability control parameters according to claim 1, characterized in that: The method of fitting the random variable using the Gaussian kernel function probability density estimation to obtain the probability density function is as follows: for each estimation point, the weights of all sample points are calculated, and the weights are accumulated and multiplied by the normalization factor to obtain the final probability density function estimate.

3. The method for optimizing grid voltage stability control parameters according to claim 1, characterized in that: Based on the density clustering algorithm, the random scenes are classified with the specific parameter values ​​of random variables as the characteristics. The method of generating random typical scenes is as follows: Set neighborhood radius , then set Minimum number of points , a certain point of The points whose neighborhood satisfies formula (3) hour, and Then they are grouped into one category; （3） In the formula, Yes and Point The Euclidean distance of express of The number of neighbors, is the sample space; like ,but As the core point, if lie in of In the neighborhood, and is the core point, then from It is density-reachable. We propagate under this density-reachable condition, cluster the samples, and thus cluster the scenes generated by random variables. Finally, we get The cluster center is A random typical scene.

4. The method for optimizing grid voltage stability control parameters according to claim 1, characterized in that: The method to obtain typical scenarios under the randomness of new energy and fault coupling is: The random typical scenarios are traversed for N-1 faults, and the stability index under the fault scenario is extracted as the feature. Then, the random typical scenarios are clustered using the density clustering algorithm to obtain the common features of the typical scenarios considering the randomness of new energy and fault coupling. indivual.

5. The method for optimizing grid voltage stability control parameters according to claim 1, characterized in that: The method for establishing a system control parameter optimization model based on typical scenarios is as follows: through the mathematical relationship model of new energy power generation and the system model, the mathematical relationship model between new energy random variables and system control parameters and system voltage stability indicators is obtained, thereby obtaining the system control parameter optimization model under typical scenarios.

6. The method for optimizing grid voltage stability control parameters according to claim 5, characterized in that: New energy random variables include: random variables of wind power generation and random variables of photovoltaic power generation. The random variables of wind power generation are wind speed, and the random variables of photovoltaic power generation are temperature and light intensity.

7. The method for optimizing grid voltage stability control parameters according to claim 1, characterized in that: The method of solving the system control parameter optimization model using the snake optimization algorithm is as follows: The population size and maximum number of iterations of the initial control optimization parameters are and , randomly initialize the position of each snake in the population; The population Divided into male , female indivual; Defines the evaluation parameter temperature for the iteration and food quantity ; When there is no food, the food search phase is carried out and the snake position is updated; When food is available, there is a food-foraging phase and a fighting-mating phase; Update the snake's worst position.

8. A device comprising a memory and a processor, characterized in that: The memory is used to store a program that supports the processor to execute the grid voltage stability control parameter optimization method according to any one of claims 1 to 7, and the processor is configured to execute the program stored in the memory.

9. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for optimizing the grid voltage stability control parameters as described in any one of claims 1 to 7 are executed.

Images