Dynamic discrete equivalent model identification method of photovoltaic power generation system under different scenarios

Abstract

The application discloses a dynamic discrete equivalent model identification method of a photovoltaic power generation system under different scenes and belongs to the technical field of power system researches; specifically comprises the following steps: S1) establishing a dynamic discrete equivalent model of the photovoltaic power generation system, S2) analyzing and determining a dynamic discrete equivalent model parameter identification method of the photovoltaic power generation system, and S3) comparing and verifying the description ability of the parameter identification method under different scenes; the application establishes the dynamic discrete equivalent model of the photovoltaic power generation system based on the electromechanical transient characteristics of the grid-connected photovoltaic power generation system, adopts the IEEE14 node system to research the generalization effect of the model by using the least square method and the bat algorithm for parameter identification under different scenes, and determines the applicable parameter identification method under different scenes.

Description

Technical Field

[0001] The method for identifying dynamic discrete equivalent models of photovoltaic power generation systems under different scenarios, as described in this invention, belongs to the field of power system research technology. Background Technology

[0002] With the continuous increase in grid-connected photovoltaic (PV) capacity, the operating characteristics of the power grid have changed significantly. The contradiction of "high-proportion inverter clusters and weak grid systems" has become prominent, and the transient stability of grid-connected PV power generation systems seriously affects the safe and stable operation of the entire power system. Establishing a model structure and parameters that conform to actual operating conditions, and selecting appropriate identification methods for the model and parameters, can not only accurately describe the dynamic characteristics of grid-connected PV power generation systems, but also provide a theoretical basis for rapid and accurate transient simulation and fault protection of transmission and distribution networks.

[0003] In the dynamic modeling of grid-connected photovoltaic (PV) power generation systems, current research focuses primarily on detailed modeling of various components of the PV system. For example: 1. Based on the forward and reverse effects and dynamic effects of solar cells, a Bishop model based on the forward and reverse characteristics and dynamic properties of solar cells has been established; 2. Considering the deviation between the output of the PV array model and the actual system output under static shading, a PV array model based on dynamic shading has been established; 3. Addressing the harmonic characteristics of inverters, an inverter model based on the Norton equivalent circuit has been established. Regarding dynamic model parameter identification methods, numerous intelligent algorithms have been introduced based on traditional parameter identification methods. For example: 1. Combining the advantages and disadvantages of genetic algorithms and particle swarm optimization algorithms, a genetic particle swarm algorithm has been proposed for parameter identification of PV grid-connected inverters; 2. Based on chaotic mapping theory and the seagull optimization algorithm, a chaotic seagull optimization algorithm has been proposed for parameter identification of PV-involved load models; 3. Based on fish school behavior and frog jumping characteristics, a hybrid algorithm combining artificial fish schools and frog jumping has been proposed for multi-scenario identification of PV arrays; 4. Considering habitat selection strategies and adaptive compensation mechanisms, a novel bat algorithm has been proposed for parameter identification of ship motion models.

[0004] In summary, the dynamic modeling of grid-connected photovoltaic power generation systems lacks research on the characteristics of external faults and has not formed an accurate and unified equivalent model. The model structure and parameters determine the accuracy of the simulation results. The least squares method is the most commonly used parameter identification method, but under different fault scenarios, its generalization effect on the model should be considered. By introducing intelligent algorithms for comparison, the appropriate parameter identification method for different fault scenarios can be determined. Summary of the Invention

[0005] This invention overcomes the shortcomings of existing technologies and provides a method for identifying dynamic discrete equivalent models of photovoltaic power generation systems under different scenarios, and determines the parameter identification methods applicable to different scenarios.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for identifying dynamic discrete equivalent models of photovoltaic power generation systems under different scenarios, comprising the following steps:

[0007] S1) Establish a dynamic discrete equivalent model of the photovoltaic power generation system;

[0008] S2) Analyze and determine the parameter identification method for the dynamic discrete equivalent model of the photovoltaic power generation system;

[0009] S3) Comparative verification of the descriptive capabilities of parameter identification methods in different scenarios.

[0010] Beneficial Effects: This invention primarily studies the electromechanical transient characteristics of photovoltaic (PV) power generation systems, specifically the relationship between the AC power of the PV inverter and the power and voltage at the grid connection point after voltage drops of varying depths occur when a grid fault occurs after PV grid connection. A dynamic discrete equivalent model of the PV power generation system is derived through corresponding formulas, accurately describing the dynamic characteristics after PV grid connection. Simulation analysis is also conducted on the parameter identification methods applicable to the dynamic discrete equivalent model of the PV power generation system under different scenarios, yielding the following conclusions:

[0011] (1) Under single-scenario faults with low photovoltaic penetration and high voltage drop depth, the least squares method has better generalization ability than the bat algorithm and the identified parameters are closer to the actual situation. In the parameter identification of dynamic models of power systems with photovoltaic participation in power distribution, the least squares method is more suitable for single-scenario faults with high voltage drop depth.

[0012] (2) Under multiple fault scenarios with different photovoltaic penetration rates and different voltage drop depths, the Bat Algorithm has better generalization ability than the least squares method and identifies parameters that are closer to the actual situation. In the identification of parameters of dynamic models of power systems with photovoltaic participation in power distribution, the Bat Algorithm is more suitable for multiple fault scenarios with low voltage drop depths.

[0013] (3) Whether it is a single scenario or multiple scenarios, the least squares method and the bat algorithm are used to identify the dynamic discrete equivalent model of the photovoltaic power generation system. Although there are certain differences in the generalization ability of the model, the fitting residuals of the model are relatively small, which shows the rationality and effectiveness of the parameter identification method of the dynamic discrete equivalent model of the photovoltaic power generation system based on the least squares method and the bat algorithm. Attached Figure Description

[0014] The present invention will now be described in further detail with reference to the accompanying drawings;

[0015] Figure 1 This is a model of a grid-connected photovoltaic power generation system in this invention;

[0016] Figure 2 This is a single-phase equivalent circuit model of the photovoltaic power generation system in this invention;

[0017] Figure 3 This invention aims to improve the adaptability of the bat algorithm to parameter identification in the dynamic discrete equivalent model of photovoltaic power generation systems.

[0018] Figure 4 This is a schematic diagram of the photovoltaic power generation system being integrated into the IEEE 14-node system in a verification embodiment of the present invention;

[0019] Figure 5 This is a verification embodiment of the present invention for least squares parameter fitting in a single scenario;

[0020] Figure 6 This is a verification embodiment of the present invention for fitting bat algorithm parameters in a single scenario;

[0021] Figure 7 This is a verification embodiment of the present invention for fitting the real part parameters of the least squares method in multiple scenarios;

[0022] Figure 8 This is a verification embodiment of the present invention for fitting the real part parameters of the bat algorithm in multiple scenarios;

[0023] Figure 9 This is a comparison of the real part parameter fitting effects of two algorithms in multiple scenarios in the verification embodiment of the present invention;

[0024] Figure 10 This is a comparison of the fitting effects of the imaginary part parameters of the two algorithms in multiple scenarios in the verification embodiment of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments; based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] This invention adopts an isolated two-stage grid-connected photovoltaic power generation system model, such as... Figure 1 As shown, the system includes a photovoltaic array, a DC / DC boost converter, a three-phase DC / AC inverter, an LC filter, an isolation transformer, and a power grid. The photovoltaic array converts solar energy into low-voltage direct current (DC) through the photoelectric effect, which is then fed into the DC / DC converter. The boost circuit controls the generation of stable high-voltage DC, while the MPPT (Maximum Power Point Tracking) controller ensures maximum output power from the photovoltaic array. The DC power is then converted into AC power (AC) in phase and frequency with the grid by the grid-connected inverter control module, thus feeding the DC power into the grid while meeting grid connection requirements.

[0027] The photovoltaic array modeling adopts an engineering practical model capable of simulating any type of photovoltaic module. It considers the transient process of the voltage regulator capacitor Cs, assuming that the light intensity and temperature of the photovoltaic array remain constant over a short period. The MPPT uses the perturbation-observation method. The DC / DC boost converter section ignores switching actions and uses a dynamic mathematical model described by second-order differential equations. The inverter employs a dual closed-loop control structure with an outer voltage loop and an inner current loop for PWM pulse width modulation. This invention mainly studies the electromechanical transient characteristics of photovoltaic power generation systems, specifically the relationship between the AC side of the inverter and the power and voltage at the grid connection point when a grid fault occurs (the bus voltage drops to varying depths). Therefore, the models of the photovoltaic array and boost converter section are not described in detail. The power loss of the inverter is ignored, and the AC side of the inverter is considered as the grid interface, equivalent to a voltage source.

[0028] A method for identifying dynamic discrete equivalent models of photovoltaic power generation systems under different scenarios includes the following steps:

[0029] S1) Establish a dynamic discrete equivalent model of the photovoltaic power generation system;

[0030] S2) Analyze and determine the parameter identification method for the dynamic discrete equivalent model of the photovoltaic power generation system;

[0031] S3) Comparative verification of the descriptive capabilities of parameter identification methods in different scenarios.

[0032] Step S1) establishes a dynamic discrete equivalent model of the photovoltaic power generation system, the specific content of which is as follows:

[0033] S11) Based on the transient third-order mechanism model

[0034] according to Figure 2 The single-phase equivalent circuit model of the photovoltaic power generation system shown is used to obtain the dynamic differential equation of the photovoltaic power generation system:

[0035] (1);

[0036] In the formula: I L This is the grid-connected current; U i This refers to the AC side voltage of the inverter. U g This is the grid-connected voltage; R , L These are the equivalent resistance and inductance from the inverter to the grid connection point, respectively. i pv Output current for the photovoltaic array; i dc , u dc Input current and voltage to the DC side of the inverter; CFor the DC-side capacitor of the photovoltaic power generation system;

[0037] By performing the Park transform on equation (1), we obtain the transient third-order mechanism model:

[0038] (2);

[0039] In the formula: I L.d , I L.q and U g.d , U g.q These represent the d-axis and q-axis components of the grid-connected current and voltage of the photovoltaic power generation system, respectively. U i.d , U i.q These are the d-axis and q-axis components of the inverter's AC side voltage, respectively. S d , S q These are the inverter modulation parameters; ω The grid-connected voltage angular frequency;

[0040] After determining the initial conditions, the transient third-order mechanism model solves equation (1) to obtain the d-axis and q-axis components of the grid-connected current. The Park transformation is then applied to obtain the third-order equivalent descriptive model, as shown in equation (3).

[0041] (3);

[0042] S12) Establish a dynamic discrete equivalent model

[0043] The transient third-order mechanism model requires determining the initial conditions of the model, which is not convenient for the dynamic characteristic analysis of photovoltaic power generation system. In this invention, the Laplace transform of equation (2) is performed and linearized to obtain the incremental form relationship of the d-axis and q-axis components of the current at the grid connection of the photovoltaic power generation system, as shown in equation (4):

[0044] (4);

[0045] In the formula: , , , , , , , , , , , , , , ;

[0046] By transforming the incremental form (4), we obtain the load model in the frequency domain, as shown in equation (5):

[0047] (5);

[0048] In the formula:

[0049] ,

[0050] ,

[0051] ,

[0052] ;

[0053] By performing a coordinate transformation on equation (5), the relationship between the real and imaginary parts of the grid-connected voltage and current is obtained, as shown in equation (6):

[0054] (6);

[0055] In the formula:

[0056] , , , ;

[0057] The transfer functions of the real and imaginary parts of the current relative to the grid voltage are then obtained, as shown in equation (7):

[0058] (7);

[0059] In the formula:

[0060] , , , , , ;

[0061] Based on the bilinear transformation method, equation (7) is transformed into a difference equation model, as shown in equation (8):

[0062] (8);

[0063] In the formula: k This refers to the system sampling time; θ αi ( i =1~9) are the real part coefficients of the model. θ βi ( i=1~9) are the imaginary part coefficients of the model, and their expressions are shown in equations (A1)-(A18):

[0064] (A1);

[0065] (A2);

[0066] (A3);

[0067] (A4);

[0068] (A5);

[0069] (A6);

[0070] (A7);

[0071] (A8);

[0072] (A9);

[0073] (A10);

[0074] (A11);

[0075] (A12);

[0076] (A13);

[0077] (A14);

[0078] (A15);

[0079] (A16);

[0080] (A17);

[0081] (A18);

[0082] Step S2) involves analyzing and determining the parameter identification method for the dynamic discrete equivalent model of the photovoltaic power generation system. The specific steps are as follows:

[0083] S21) Basic Principles of Analytical Parameter Identification

[0084] Parameter identification consists of three modules: input and output data, mathematical model, and identification criteria. The mathematical model is optimized based on the system's input and output data to find a set of optimal parameters that meet the identification criteria, so that the model's output results are close to the actual system's output results.

[0085] Based on the mathematical model of the difference equation concerning the real and imaginary parts of the current established in step S1), and given the voltage, active power, and reactive power at the grid connection point, identify the real part coefficients in the model. θ αi ( i =1~9) and imaginary part coefficient θ βi ( i =1~9), using the root mean square error (RMSE) as the identification criterion, as shown in equation (9), the output error between the dynamic discrete equivalent model and the actual system is quantified:

[0086] (9);

[0087] In the formula: N This represents the number of system sampling points. E , Ê The root mean square error (RMSE) of the real and imaginary currents of the photovoltaic power generation system. I rk , I jk For the first in the actual system k Real and imaginary currents at each sampling point; Î rk , Î jk In the dynamic discrete equivalent model, the first k Real and imaginary currents at each sampling point;

[0088] S22) Least Squares Parameter Identification

[0089] (1) Principle of least squares method

[0090] The least squares method is a mathematical model that best fits the actual system in the sense of minimizing variance. It is the most basic method for model parameter identification and is widely used in time domain and frequency domain identification. It can be calculated offline or online recursively.

[0091] Considering the discreteness of actual sampling, an nth-order difference equation is used to represent the discrete model, as shown in equation (10):

[0092] (10);

[0093] In the formula: u ( k ),y ( k () refers to system inputs and outputs; e ( k ) represents the system fitting error; n+N represents the length of the measurement data; and n represents the system order.

[0094] The difference equation above can be rewritten in matrix form as shown in equation (11):

[0095] (11);

[0096] In the formula:

[0097] , , , .

[0098] The system variance J is shown in equation (12):

[0099] (12);

[0100] Differentiating J with respect to θ and setting it to zero yields the optimal parameter estimate, as shown in equation (13):

[0101] (13);

[0102] (2) Parameter identification process

[0103] The voltage, active power, and reactive power data at the grid connection point are obtained to identify the model parameters. The real current, imaginary current, and bus voltage dynamic data at times k, k+1, k+2, k+3, and k+4 are recorded. The parameter estimates are corrected with the goal of minimizing the variance. The corresponding nonlinear model equations are analyzed by selecting m sets of observation data, as shown in equation (14). The variance of the system is obtained, as shown in equation (15). A unique set of optimal parameter estimates is determined, as shown in equation (16).

[0104] (14);

[0105] (15);

[0106] (16);

[0107] In the formula:

[0108] x 1= ΔI r ( k +3), x 2= ΔI r (k +2)， x 3= ΔI r ( k +1)， x 4= ΔI r ( k )， x 5= ΔU ( k +4)， x 6= ΔU ( k +3)， x 7= ΔU ( k +2)， x 8= ΔU ( k +1)， x 9= ΔU ( k )， x 10 = ΔI j ( k +3)， x 11 = ΔI j ( k +2)， x 12 = ΔI j ( k +1)， x 13 = ΔI j ( k )， x 14 = ΔU ( k +4)， x 15 = ΔU ( k +3)， x 16 = ΔU ( k +2)， x 17 = ΔU ( k +1)， x 18 = ΔU ( k )； y 1= ΔI r ( k +4)， y2= ΔI j ( k +4);

[0109] S23) Bat Algorithm Parameter Identification

[0110] (1) Principle of Bat Algorithm

[0111] With the development of intelligent algorithms, many intelligent algorithms have been adopted and adapted to be new methods suitable for power system load parameter identification. The bat algorithm (BA) is a representative example. It has the characteristics of simple model structure, few target parameters, strong versatility and easy implementation. It mainly simulates the behavior of bats using echolocation to hunt. It can realize both local search and global search and has good convergence characteristics.

[0112] The process of using ultrasound to locate prey is described by the following formulas, with the pulse frequency updated according to formulas (17) to (19). f i ,speed v i and location x i Pulse loudness during bat search A i and frequency R i The pulse loudness and frequency are continuously updated according to equations (20) to (21). The optimal solution is found through local search, and multiple flights are performed to generate multiple new solutions. Finally, a global search is performed to obtain the globally optimal solution.

[0113] (17);

[0114] (18);

[0115] (19);

[0116] In the formula: β It can be any value within [0,1]. This is the optimal value for a global search. f min , f max These are the minimum and maximum values ​​of the pulse frequency;

[0117] (20);

[0118] (twenty one);

[0119] In the formula: α The loudness attenuation coefficient, 0 <α <1; γ To increase the frequency coefficient, γ> 0;

[0120] (2) Parameter identification process

[0121] The adaptability of the Bat algorithm to parameter identification in dynamic discrete equivalent models of photovoltaic power generation systems, such as... Figure 3 As shown, the bat algorithm parameter identification is to find the parameter values ​​corresponding to the minimum RMSE of the objective function based on the same input and output of the given model and the actual system. The specific steps are: to obtain the input and output data required for parameter identification through simulation experiments; to determine the parameters to be identified and the objective function to be optimized in the established model; to continuously optimize the parameters in the model using the bat algorithm to find the model parameters corresponding to the minimum RMSE of the objective function; and to output the final result.

[0122] The specific content of step S3) comparing and verifying the descriptive capabilities of parameter identification methods under different scenarios is as follows:

[0123] A power transmission and distribution system model with a centralized photovoltaic array was built using MATLAB / Simulink. The transmission system, distribution system, and their parameters were set up. The model was validated in single and multiple scenarios. The least squares method and the bat algorithm were used to identify the dynamic discrete equivalent model of the photovoltaic power generation system. The generalization ability of the parameters identified by different methods to the real-world situation was compared, and the appropriate parameter identification method for the corresponding scenario was determined. The conclusions are as follows:

[0124] (1) In the parameter identification of dynamic model of power system with photovoltaic participation in power distribution, the least squares method is used for single-scenario faults under high voltage drop depth;

[0125] (2) In the parameter identification of dynamic model of power system with photovoltaic participation in power distribution, the bat algorithm is used for multi-scenario faults under low voltage drop depth.

[0126] The following specific examples demonstrate the comparative verification of the parameter identification method's descriptive capabilities in different scenarios.

[0127] To study parameter identification methods applicable to dynamic discrete equivalent models of photovoltaic power generation systems under different scenarios, a power transmission and distribution system model containing a centralized photovoltaic array was built based on MATLAB / Simulink, such as... Figure 4As shown. The transmission system is an IEEE 14-bus system with the following parameters: reference voltage 23kV, system frequency 50Hz, and a series RL impedance with an inductance of 0.618H and a resistance of 0.4Ω. The distribution system is a photovoltaic (PV) power generation system with the following parameters: open-circuit voltage 44.5V, short-circuit current 8.2A, optimal operating voltage 35.5V, optimal operating current 7.51A, 300 PV arrays in series, 200 in parallel, and a module conversion efficiency of 16.9%; an LC filter with an inductance of 0.17mH and a capacitance of 0.0018F; and a 63kW capacity, 35 / 230kV dual-winding step-up transformer. The PV power generation system is connected to the IEEE 14-bus system through grid connection node 7.

[0128] The simulation time was set to 1.5s, the simulation step size to 0.001s, and the sampling frequency to 1000Hz. A short-circuit grounding fault occurred at 0.5s. The simulation was studied in single-scenario (specific photovoltaic penetration rate, specific voltage drop depth) and multi-scenario (different photovoltaic penetration rates, different voltage drop depths) scenarios. The least squares method and the bat algorithm were used to identify the dynamic discrete equivalent model of the photovoltaic power generation system. The generalization ability of the parameters identified by different methods to the actual situation was compared to determine the appropriate parameter identification method for each scenario.

[0129] Simulation verification in a single scenario

[0130] When a three-phase short-circuit ground fault occurs in the power transmission and distribution network, with a photovoltaic penetration rate of 20% and a voltage drop depth of 50%, the least squares method and the bat algorithm are used to identify the dynamic discrete equivalent model of the photovoltaic power generation system, respectively. The generalization effect of the model is as follows: Figure 5 and Figure 6 As shown, where, Ir , I j These are measured values. Ir - U , Ij - U The values ​​are the model fit values, and the model fit residuals are shown in Table 1.

[0131] Table 1 Model Fitting Residuals in a Single Scenario

[0132] Table 1 Residual of model fitting in single scene

[0133]

[0134] Depend on Figure 5 and Figure 6It can be seen that, in a single scenario with a photovoltaic penetration rate of 20% and a voltage drop depth of 50%, the dynamic response of the dynamic discrete equivalent model of the photovoltaic grid-connected power generation system basically matches the measured curve of the actual output of the transmission and distribution network system, demonstrating the rationality and effectiveness of the parameter identification method for the dynamic discrete equivalent model of the photovoltaic power generation system based on the least squares method and the bat algorithm. However, the fitting effect of the least squares method is better than that of the bat algorithm. Table 1 also shows that, in a single scenario, the fitting residual of the model by the bat algorithm is approximately twice that of the least squares method. Therefore, in specific scenarios with low photovoltaic penetration and high voltage drop depth, the parameters identified by the least squares method are closer to the actual situation than those identified by the bat algorithm. Thus, in the parameter identification of the dynamic model of power systems with photovoltaic participation in distribution, the least squares method is more suitable for single-scenario faults with high voltage drop depths.

[0135] Simulation verification in multiple scenarios

[0136] When a three-phase short-circuit ground fault occurs in the power transmission and distribution network, the photovoltaic penetration rate is increased from 10% to 80% (with a spacing of 10%), and the voltage drop depth is increased from 5% to 50% (with a spacing of 10%). The least squares method and the bat algorithm are used to identify the dynamic discrete equivalent model of the photovoltaic power generation system, respectively, resulting in a set of general identification parameters. These parameters are then fitted to various fault scenarios under different photovoltaic penetration rates and voltage drop depths. Taking the real part of the dynamic discrete equivalent model of the photovoltaic power generation system as an example, the generalization effect of the model's real part is as follows: Figure 7 and Figure 8 As shown, the vertical axis represents photovoltaic penetration rate, and the horizontal axis represents voltage drop depth. To further analyze the generalization ability and applicability of the least squares method and the bat algorithm for the model, the ability of the two methods to describe the actual situation for different penetration rate ranges is compared, such as... Figure 9 and Figure 10 As shown in the figure, the vertical axis represents the residual rate, and the horizontal axis represents the photovoltaic penetration rate.

[0137] Depend on Figure 7 and Figure 8 It can be seen that when the photovoltaic penetration rate is within the range of 20% to 30% and the voltage drop depth is within the range of 15% to 35%, the generalization effect of the least squares method on the model is far inferior to that of the bat algorithm. Even with a very low photovoltaic penetration rate, the generalization effect still has some error when the voltage drop depth reaches a certain value. Meanwhile, with a photovoltaic penetration rate of 10% to 80%, when the voltage drop depth exceeds 40%, the multi-scenario fitting of the bat algorithm shows relatively good generalization performance for the model. Figure 9 and Figure 10It can be seen that when the photovoltaic penetration rate is within the range of 75% and 65%, respectively, the Bat Algorithm outperforms the least squares method in generalizing the real and imaginary parts of the dynamic discrete equivalent model of the photovoltaic power generation system. However, its generalization ability to the imaginary part of the model is more sensitive than its generalization ability to the real part. Therefore, under various scenarios with different photovoltaic penetration rates and voltage drop depths, the parameters identified by the Bat Algorithm are closer to the actual situation than those identified by the least squares method. Thus, in the parameter identification of dynamic models of power systems with photovoltaic participation in distribution, the Bat Algorithm is more suitable for multi-scenario faults under low voltage drop depths.

[0138] This invention primarily studies the electromechanical transient characteristics of photovoltaic (PV) power generation systems, specifically the relationship between the AC measurement of the PV inverter and the power and voltage at the grid connection point after voltage drops of varying depths occur when a grid fault occurs after PV grid connection. A dynamic discrete equivalent model of the PV power generation system is derived through corresponding formulas, accurately describing the dynamic characteristics after PV grid connection. Simulation analysis is also conducted on the parameter identification methods applicable to the dynamic discrete equivalent model of the PV power generation system under different scenarios, yielding the following conclusions:

[0139] (1) Under single-scenario faults with low photovoltaic penetration and high voltage drop depth, the least squares method has better generalization ability than the bat algorithm and the identified parameters are closer to the actual situation. In the parameter identification of dynamic models of power systems with photovoltaic participation in power distribution, the least squares method is more suitable for single-scenario faults with high voltage drop depth.

[0140] (2) Under multiple fault scenarios with different photovoltaic penetration rates and different voltage drop depths, the Bat Algorithm has better generalization ability than the least squares method and identifies parameters that are closer to the actual situation. In the identification of parameters of dynamic models of power systems with photovoltaic participation in power distribution, the Bat Algorithm is more suitable for multiple fault scenarios with low voltage drop depths.

[0141] (3) Whether it is a single scenario or multiple scenarios, the least squares method and the bat algorithm are used to identify the dynamic discrete equivalent model of the photovoltaic power generation system. Although there are certain differences in the generalization ability of the model, the fitting residuals of the model are relatively small, which shows the rationality and effectiveness of the parameter identification method of the dynamic discrete equivalent model of the photovoltaic power generation system based on the least squares method and the bat algorithm.

[0142] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying dynamic discrete equivalent models of photovoltaic power generation systems under different scenarios, characterized in that... Includes the following steps: S1) Establish a dynamic discrete equivalent model of the photovoltaic power generation system. Specifically, based on the single-phase equivalent circuit model of the photovoltaic power generation system, obtain the dynamic differential equation of the photovoltaic power generation system. Perform Park transformation on the dynamic differential equation of the photovoltaic power generation system to obtain a transient third-order mechanism model. Perform Laplace transformation on the transient third-order mechanism model and linearize it to obtain the incremental form relationship of the d-axis and q-axis components of the current at the grid connection of the photovoltaic power generation system. Transform the incremental form relationship to obtain the load model in the frequency domain. Perform coordinate transformation on the load model to obtain the relationship between the real and imaginary parts of the grid voltage and current. Then obtain the transfer function of the real and imaginary parts of the current relative to the grid voltage. Based on the bilinear transformation method, the transfer function of the real and imaginary parts of the current relative to the grid voltage is transformed into a difference equation model. S2) Analyze and determine the parameter identification method for the dynamic discrete equivalent model of the photovoltaic power generation system. The parameter identification method includes least squares parameter identification and bat algorithm parameter identification. S3) Comparative verification of the descriptive capabilities of parameter identification methods under different scenarios. Specifically, a power transmission and distribution system model containing a centralized photovoltaic array was built based on MATLAB / Simulink. The transmission system, distribution system, and their parameters were set. The least squares method and the bat algorithm were used to identify the dynamic discrete equivalent model of the photovoltaic power generation system under single and multiple scenarios. The generalization ability of the parameters identified by different identification methods to the actual situation was compared, and the appropriate parameter identification method for the corresponding scenario was determined. The conclusion is as follows: (1) In the parameter identification of dynamic model of power system with photovoltaic participation in power distribution, the least squares method is used for single-scenario faults under high voltage drop depth; (2) In the parameter identification of dynamic model of power system with photovoltaic participation in power distribution, the bat algorithm is used for multi-scenario faults under low voltage drop depth.

2. The method for identifying dynamic discrete equivalent models of photovoltaic power generation systems under different scenarios according to claim 1, characterized in that, Step S1) establishes a dynamic discrete equivalent model of the photovoltaic power generation system, the specific content of which is as follows: S11) Based on the transient third-order mechanism model Based on the single-phase equivalent circuit model of the photovoltaic power generation system, the dynamic differential equation of the photovoltaic power generation system is obtained: （1）； In the formula: I L This is the grid-connected current; U i This refers to the AC side voltage of the inverter. U g This is the grid-connected voltage; R , L These are the equivalent resistance and inductance from the inverter to the grid connection point, respectively. i pv Output current for the photovoltaic array; i dc , u dc Input current and voltage to the DC side of the inverter; C For the DC-side capacitor of the photovoltaic power generation system; By performing the Park transform on equation (1), we obtain the transient third-order mechanism model: （2）； In the formula: I L.d , I L.q and U g.d , U g.q These represent the d-axis and q-axis components of the grid-connected current and voltage of the photovoltaic power generation system, respectively. U i.d , U i.q These are the d-axis and q-axis components of the inverter's AC side voltage, respectively. S d , S q These are the inverter modulation parameters; ω The grid-connected voltage angular frequency; S12) Establish a dynamic discrete equivalent model By performing a Laplace transform and linearizing equation (2), the incremental relationship of the d-axis and q-axis components of the current at the grid connection point of the photovoltaic power generation system is obtained, as shown in equation (4): （4）； In the formula: , , , , , , , , , , , , , , ; By transforming the incremental form (4), we obtain the load model in the frequency domain, as shown in equation (5): （5）； In the formula: ， ， ， ； By performing a coordinate transformation on equation (5), the relationship between the real and imaginary parts of the grid-connected voltage and current is obtained, as shown in equation (6): （6）； In the formula: ， ， ， ； The transfer functions of the real and imaginary parts of the current relative to the grid voltage are then obtained, as shown in equation (7): （7）； In the formula: ， ， ， ， ， ； Based on the bilinear transformation method, equation (7) is transformed into a difference equation model, as shown in equation (8): （8）； In the formula: k This refers to the system sampling time; θ αi ( i =1~9) are the real part coefficients of the model. θ βi ( i =1~9) are the imaginary part coefficients of the model, and their expressions are shown in equations (A1)-(A18): （A1）； （A2）； （A3）； （A4）； （A5）； （A6）； （A7）； （A8）； （A9）； （A10）； （A11）； （A12）； （A13）； （A14）； （A15）； （A16）； （A17）； （A18）。 3. The method for identifying dynamic discrete equivalent models of photovoltaic power generation systems under different scenarios according to claim 2, characterized in that, Step S2) involves analyzing and determining the parameter identification method for the dynamic discrete equivalent model of the photovoltaic power generation system. The specific steps are as follows: S21) Basic Principles of Analytical Parameter Identification Parameter identification consists of three modules: input and output data, mathematical model, and identification criteria. The mathematical model is optimized based on the system's input and output data to find a set of optimal parameters that meet the identification criteria, so that the model's output results are close to the actual system's output results. Based on the mathematical model of the difference equation concerning the real and imaginary parts of the current established in step S1), and given the voltage, active power, and reactive power at the grid connection point, identify the real part coefficients in the model. θ αi ( i =1~9) and imaginary part coefficient θ βi ( i =1~9), using the root mean square error as the identification criterion, as shown in equation (9), the output error between the dynamic discrete equivalent model and the actual system is quantified: （9）； In the formula: N This represents the number of system sampling points. E , Ê The root mean square error of the real and imaginary currents of the grid-connected photovoltaic power generation system; I rk , I jk For the first in the actual system k Real and imaginary currents at each sampling point; Î rk , Î jk In the dynamic discrete equivalent model, the first k Real and imaginary currents at each sampling point; S22) Least Squares Parameter Identification (1) Principle of least squares method Considering the discreteness of actual sampling, an nth-order difference equation is used to represent the discrete model, as shown in equation (10): （10）； In the formula: u ( k ), y ( k () refers to system inputs and outputs; e ( k ) represents the system fitting error; n+N represents the length of the measurement data; and n represents the system order. The difference equation above can be rewritten in matrix form as shown in equation (11): （11）； In the formula: ， ， ， ； The system variance J is shown in equation (12): （12）； Differentiating J with respect to θ and setting it to zero yields the optimal parameter estimate, as shown in equation (13): （13）； (2) Parameter identification process The voltage, active power, and reactive power data at the grid connection point are obtained to identify the model parameters. The real current, imaginary current, and bus voltage dynamic data at times k, k+1, k+2, k+3, and k+4 are recorded. The parameter estimates are corrected with the goal of minimizing the variance. The corresponding nonlinear model equations are analyzed by selecting m sets of observation data, as shown in equation (14). The variance of the system is obtained, as shown in equation (15). A unique set of optimal parameter estimates is determined, as shown in equation (16). （14）； （15）； （16）； In the formula: x 1= ΔI r ( k +3)， x 2= ΔI r ( k +2)， x 3= ΔI r ( k +1)， x 4= ΔI r ( k )， x 5= ΔU ( k +4)， x 6= ΔU ( k +3)， x 7= ΔU ( k +2)， x 8= ΔU ( k +1)， x 9= ΔU ( k )， x 10 = ΔI j ( k +3)， x 11 = ΔI j ( k +2)， x 12 = ΔI j ( k +1)， x 13 = ΔI j ( k )， x 14 = ΔU ( k +4)， x 15 = ΔU ( k +3)， x 16 = ΔU ( k +2)， x 17 = ΔU ( k +1)， x 18 = ΔU ( k )； y 1= ΔI r ( k +4)， y 2= ΔI j ( k +4)； S23) Bat Algorithm Parameter Identification (1) Principle of Bat Algorithm The process of using ultrasound to locate prey is described by the following formulas, with the pulse frequency updated according to formulas (17) to (19). f i ,speed v i and location x i Pulse loudness during bat search A i and frequency R i The pulse loudness and frequency are continuously updated according to equations (20) to (21). The optimal solution is found through local search, and multiple flights are performed to generate multiple new solutions. Finally, a global search is performed to obtain the globally optimal solution. （17）； （18）； （19）； In the formula: β It can be any value within [0,1]. This is the optimal value for a global search. f min , f max These are the minimum and maximum values ​​of the pulse frequency; （20）； （21）； In the formula: α The loudness attenuation coefficient, 0 < α <1; γ To increase the frequency coefficient, γ> 0; (2) Parameter identification process Bat algorithm parameter identification involves finding the parameter values ​​that minimize the RMSE of the objective function, given a model and the actual system with the same inputs and outputs. The specific steps are as follows: obtain the input and output data required for parameter identification through simulation experiments; determine the parameters to be identified and the objective function to be optimized in the established model; continuously optimize the parameters in the model using the bat algorithm to find the model parameters that minimize the RMSE of the objective function; and output the final result.

Images