A method for identifying parameters of a power generation load hybrid model

Abstract

The application discloses a parameter identification method for a power generation load mixed model. Firstly, the nature and rough proportion of power generation and load in the power generation load mixed model are identified by using second-level quasi-steady state data of a large number of low-voltage grade measuring points measured by a remote terminal system (RTU). Further, the RTU quasi-steady state data are combined with weather information, and the attributes and rough proportion of different types of power sources such as wind, light and fire in the power generation source are corrected by using evidence theory. Then, the dynamic measurement data of a millisecond-level synchronized phasor measurement device (PMU) with few high-voltage grade measuring points are used to identify the accurate dynamic parameters in the power generation load mixed model by means of a double-error genetic algorithm. The application has high identification precision and solves the problem of the change of load characteristics caused by new energy power generation.

Description

Technical Field

[0001] This invention belongs to the field of power grid operation and control technology, and specifically relates to a method for identifying parameters of a hybrid power generation load model. Background Technology

[0002] With the development of power systems, load identification is an essential step in power system planning, design, operation, and control. In recent years, research on load identification algorithms has been relatively quiet, yet it remains a pressing issue for power systems. Existing technologies have applied load identification models to household appliances, collecting their operating current and using convolutional neural networks to create two-dimensional images. Feature analysis and extraction are then performed on the images, and the resulting data is used for training to determine the load type. Alternatively, deep learning algorithms are integrated with feature extraction, utilizing the advanced feature extraction capabilities of artificial neural networks and backpropagation (BP) neural networks to achieve non-intrusive load identification. Another approach proposes a non-intrusive load identification algorithm based on data mining and machine learning, employing an ensemble learning algorithm for load identification. This algorithm utilizes probability curves and multiple correction methods to achieve load identification. It also involves collecting harmonic characteristics for basic load identification, training a neural network to fit a probability distribution curve, correcting errors, and obtaining the optimal solution.

[0003] The above are the main methods of load identification algorithms in recent years. These methods primarily involve transforming measurable data into other forms and then using intelligent algorithms for identification. However, these methods do not optimize the initial parameters, resulting in low identification accuracy, excessively long identification time, and a limited number of identified load types. Furthermore, with the high penetration of renewable energy generation, the load side often includes a large number of distributed power sources such as renewable energy generation. In this case, traditional load identification models cannot handle the changes in load characteristics brought about by renewable energy generation. Therefore, a mixed generation-load model identification is needed to solve this problem. However, solving the ratio of generation to load in a mixed generation-load model is a complex multi-solution problem, and the solution is extremely complex. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method for identifying parameters in a power generation load hybrid model. First, using second-level quasi-steady-state data from a large number of low-voltage level measuring points obtained by a remote terminal unit (RTU), the nature and approximate proportions of power generation and load in the power generation load hybrid model are identified. Further, the RTU quasi-steady-state data are combined with meteorological information, and evidence theory is used to correct the attributes and approximate proportions of different power sources such as wind, solar, and thermal power. Then, using millisecond-level dynamic measurement data from a small number of high-voltage level measuring points obtained by a phasor measurement unit (PMU), a double-error golden section genetic algorithm is employed to identify the precise dynamic parameters in the power generation load hybrid model. This invention solves the technical problem of load characteristics brought about by new energy power generation.

[0005] The present invention adopts the following technical solution.

[0006] A method for parameter identification in a hybrid power generation load model includes:

[0007] Step 1: Collect measurement data from the remote terminal system, the synchronous phasor measurement device, and meteorological parameters of the power grid location;

[0008] Step 2: Based on the measurement data from the remote terminal system, convert each branch of the power grid into various equivalent models; wherein, the equivalent models include: wind power virtual generator model, photovoltaic virtual generator model, conventional unit model, motor load model, and ZIP load model; the various equivalent models together form a power generation load hybrid model;

[0009] Step 3: Use DS evidence theory to estimate the parameters of various equivalent models; substitute the parameters of various equivalent models into the corresponding equivalent models to calculate the output power value of each type of model; use the power value as the standard for proportional calculation to obtain the sum of the output power of each type of model; compare the output power of each type of model with the sum of the power values ​​to determine the proportion of each type of equivalent model in the power generation load hybrid model.

[0010] Step 4: Based on the genetic algorithm, using the parameters of various equivalent models obtained in Step 3, the measurement data of the synchronous phasor measurement device, and the proportion of the various equivalent models in the power generation load mixed model, the model parameters of each branch are obtained through iterative calculation.

[0011] Step 5: Take the average value of the various equivalent model parameters obtained in Step 4 as the optimal model parameters for the power generation load hybrid model.

[0012] Preferably, in step 1, the measurement data of the remote terminal system includes: power direction, current amplitude, and voltage amplitude;

[0013] The measurement data from the synchronous phasor measurement device includes power, voltage, and current over a period of one week;

[0014] Meteorological parameters include: daily wind speed and sunshine duration.

[0015] The power average value is calculated hourly based on the power measured by the synchronous phasor measurement device. The standard deviation of the power over one week is then calculated using this average value as a benchmark. The standard deviation represents the power stability. The formula for calculating the power standard deviation is as follows:

[0016]

[0017] In the formula, s Here is the standard deviation, and m is the number of hours within a week; P This refers to the actual power. P s This represents the average power per hour over a one-week period.

[0018] Preferably, in step 2, the motor load model adopts the mathematical model of a T-type third-order induction motor as follows:

[0019]

[0020] in, , , ,

[0021] T j The generator rotor inertia time constant; Tm and Te These are the mechanical and electromagnetic torques of the generator, respectively. s For slip; This refers to the induced electromotive force inside the rotor, excluding the stator impedance. This is the current value; A This is the proportionality coefficient related to the square of the resistance torque; B This is the proportionality coefficient between the resistance torque and the rotational speed; C It is a constant that depends on the motor parameters and the initial slip; This represents the small-signal value of the current. r 2 represents the rotor resistance; X 1 represents the stator reactance; X 2 represents the rotor reactance; X m For excitation reactance; The time constant of the stator open-circuit rotor circuit; ; This refers to the no-load torque caused by factors such as friction; ω is the rotor angular velocity.

[0022] In step 2, if the power direction of any branch is positive, the power stability is greater than the set threshold, and the voltage level is 110kV or below, then the branch is equivalent to a wind power virtual generator model. The threshold is set according to different voltage levels. For voltage levels of 110kV or below, it can be set to 1~5, and for voltage levels of 110kV or above, it can be set to 5~10.

[0023] If the power direction of any branch is positive, the power stability is less than the set threshold, and the voltage level is 110kV or below, then the branch is equivalent to a photovoltaic virtual generator model.

[0024] If the power direction of any branch is positive, the power stability is less than the set threshold, and the voltage level is 110kV or above, then the branch is equivalent to a photovoltaic virtual generator model.

[0025] If the power direction of any branch is negative and the power stability is less than a set threshold, then the branch is equivalent to a ZIP load model.

[0026] If the power direction of any branch is negative and the power stability is greater than a set threshold, then the branch is equivalent to a motor load model.

[0027] Preferably, in step 3, the model parameters are estimated using DS evidence theory, a basic probability assignment is established, a normalization constant is calculated based on the basic probability assignment, and the fusion trust of each set of parameters is calculated; the model parameter with the highest fusion trust is substituted into each type of model to obtain the power curve, and the average power is taken to calculate the proportion of each type of equivalent model in the power generation load hybrid model.

[0028] Preferably, in step 4, the golden ratio is used to calculate whether the model parameters need to be mutated; in the nth generation of model parameters, a number r∈(0,1) is randomly generated, if r≤0.382+0.618^(n 1), then the mutation probability is 1 / (N*M), where N is the number of sub-parameters and M is the number of model parameters, that is, to mutate one model parameter.

[0029] In step 4, inputting the model parameters into the power generation load hybrid model yields simulated curves of voltage amplitude, active power, and reactive power parameters. These simulated curves are then compared with the measured curves to obtain voltage amplitude error, active power error, and reactive power error. The error function is constructed as follows:

[0030]

[0031]

[0032]

[0033] In the above formula, is the voltage amplitude error function, is the voltage amplitude error function, is the voltage amplitude error function, s is the number of sampling points of the simulation curve and the measured curve, is the parameter value of each sampling point of the simulation curve, is the parameter value of each sampling point of the measured curve.

[0034] Preferably, the average error is defined as E , E is the average value of the voltage amplitude error, active power error, and reactive power error; use to uniformly represent the voltage amplitude error function, voltage amplitude error function, voltage amplitude error function:

[0035] When < E, output this set of parameters and perform the next round of selection, crossover, and mutation operations;

[0036] When > E, directly perform selection, crossover, and mutation operations on this set of parameters.

[0037] The beneficial effects of the present invention are as follows. Compared with the prior art, the present invention uses a large amount of second-level quasi-steady-state data of low-voltage level measurement points measured by the RTU system, including power direction, power smoothness, current, voltage amplitude, etc., to equivalently classify the power generation load mixture and identify the nature and rough ratio of power generation and load in the power generation load mixture model. At the same time, the present invention adopts the DS evidence theory to correct the attributes and rough ratio of different types of power sources such as wind, light, and fire in the power generation power source to obtain the ratio k; then uses the double-error genetic algorithm to optimize and solve the load model parameters of each branch, obtains the best model parameters of each branch, and calculates the best model parameters of the power generation load mixture model using the best model parameters of each branch. The present invention adopts the golden section genetic algorithm and also adds the error function e to the genetic algorithm, making the evolution speed faster and the result more accurate. The method proposed by the present invention also has the characteristics of wide applicability. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 is a flowchart of a method for identifying parameters of a power generation load mixture model according to the present invention;

[0039] Figure 2 is a method for calculating the proportion of each branch in an embodiment of the present invention;

[0040] Figure 3 is a multi-point crossover method of the genetic algorithm in an embodiment of the present invention;

[0041] Figure 4 is a random mutation method of the genetic algorithm in an embodiment of the present invention;

[0042] Figure 5 This is a comparison chart of the simulated curve and the measured curve in the embodiments of the present invention. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this application are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, other embodiments obtained by those skilled in the art without creative effort are all within the protection scope of this invention.

[0044] A method for parameter identification in a hybrid power generation load model, such as Figure 1 As shown, it includes the following steps:

[0045] Step 1: Collect measurement data from remote terminal systems and synchronous phasor measurement devices corresponding to each branch of the power grid, as well as local meteorological parameters;

[0046] The collected RTU measurement data includes power direction, current, voltage amplitude, etc.; the collected PMU measurement data includes parameters such as power, voltage, and current over a period of one week; meteorological parameters mainly include daily wind speed and sunshine; the power average value is calculated based on the power measured by the PMU in hourly units, and the standard deviation of the power over a period of one week is calculated based on the average value, with the standard deviation value representing the power stability.

[0047] Step 2: Based on the measurement data from the remote terminal system, convert each branch of the power grid into various equivalent models; among which, the equivalent models include: wind power virtual generator model, photovoltaic virtual generator model, conventional unit model, motor load model, and ZIP load model; the various equivalent models together form a power generation load hybrid model;

[0048] Specifically,

[0049] If the power direction is positive, the power stability is greater than 10, and the voltage level is below 110kV, then this branch is equivalent to a wind power virtual generator model.

[0050] If the power direction is positive, the power stability is less than 10, and the voltage level is below 110kV, then this branch is equivalent to a photovoltaic virtual generator model.

[0051] If the power direction is positive, the power stability is less than 10, and the voltage level is above 110kV, then the branch is equivalent to a photovoltaic virtual generator model.

[0052] If the power direction is negative and the power stability is less than 10, then the branch is equivalent to a ZIP load model.

[0053] If the power direction is negative and the power stability is greater than 10, then the branch is equivalent to a motor load model.

[0054] The wind power virtual generator model, the photovoltaic virtual generator model, and the conventional unit model are identified using a third-order practical generator model, and their mathematical models are as follows:

[0055]

[0056] The synchronous reactance of the generator's d-axis; ; For the generator's q-axis transient reactance; The stator open-circuit time constant of the generator rotor d-axis excitation winding; The generator rotor inertia time constant; This represents the generator damping coefficient.

[0057] The electric motor load model adopts the mathematical model of a T-type third-order induction motor as follows:

[0058]

[0059] in, , , ,

[0060] T j The generator rotor inertia time constant; Tm and Te These are the mechanical and electromagnetic torques of the generator, respectively. s For slip; To induce electromotive force; This is the current value; A This is the proportionality coefficient related to the square of the resistance torque; B This is the proportionality coefficient between the resistance torque and the rotational speed; C It is a constant that depends on the motor parameters and the initial slip; r 2 represents the rotor resistance; X 1 represents the stator reactance; X 2 represents the rotor reactance; X m For excitation reactance; The time constant of the stator open-circuit rotor circuit; .

[0061] The mathematical model of a comprehensive static ZIP load consisting of constant impedance, constant current, and constant power is as follows:

[0062]

[0063] in,

[0064]

[0065] The active power of the ZIP load. For ZIP load reactive power, The constant impedance coefficient for active power; The constant current coefficient for active power; The constant impedance coefficient for reactive power; The reactive power constant current coefficient; The constant power coefficient is the active power coefficient. The reactive power constant power coefficient; This represents the active power load of the busbar in steady state. This represents the bus reactive load under steady-state conditions. This is the steady-state voltage of the load bus. This is the real-time voltage of the load bus.

[0066] Step 3: Use DS evidence theory to estimate the parameters of various equivalent models; substitute the parameters of various equivalent models into the corresponding equivalent models to calculate the output power value of each type of model; use the power value as the standard for proportional calculation to obtain the sum of the output power of each type of model; compare the output power of each type of model with the sum of the power values ​​to determine the proportion of each type of equivalent model in the power generation load hybrid model.

[0067] The estimation methods that can be used for model parameters include: (1) analogy algorithm; (2) Delphi / expert estimation; (3) PERT three-point estimation; (4) FP estimation, etc. The estimated model parameters are used to calculate the proportion of each load model in the mixed power generation load model. The present invention will not elaborate on the above methods. The present invention preferably uses DS evidence theory to estimate model parameters and calculate the load model parameters of each branch and the proportion of each load model in the mixed power generation load model.

[0068] The calculation steps of the DS evidence theory are as follows;

[0069] Step 3.1: Establish the hypothesis space, and assign values ​​to all parameters in the parameter library of m models using n different methods, within the range of [ -10, 10 This way, we can obtain n sets of model parameters, and use each set of model parameters as an element of the hypothesis space;

[0070] Step 3.2: Construct the n-dimensional hypothesis space Θ of the evaluation method and model parameters; the evaluation methods include least squares estimation, maximum likelihood estimation, Bayesian estimation, etc.; the hypothesis space is shown in Table 1, with method 1 denoted as X1 and parameter 1 denoted as Y1. Table 1 is also known as the basic probability assignment (BPA).

[0071] Table 1. n-dimensional hypothesis space Θ and probability distribution

[0072]

[0073] In the above formula, P 21 This means that the probability of parameter 2 is calculated using method 1, and other probabilities are calculated similarly;

[0074] Step 3.3: Calculate the normalization constant K;

[0075] The normalization constant is used in the DS evidence theory to reflect conflicting evidence, and its calculation method is as follows:

[0076]

[0077] Step 3.4: Calculate the DS fusion trust for each parameter. The calculation method is as follows:

[0078]

[0079] In the above formula, i The number of parameter sets is [1, m].

[0080] Step 3.5: Select the parameter with the highest fusion trust as the optimal parameter for calculating the load ratio of each branch;

[0081] Step 3.6: Simulate the power curves of each branch load, take the average power, and calculate the proportion k of each branch load in the power generation load mix. The calculation method is the comparison method, which directly compares the power of each branch load with the sum of the power of all branches loads. Substitute the optimal model parameters into the simulation tool to fit the power curves of each branch, obtain the average power of each branch, and calculate the proportion k of each branch load in the power generation load mix.

[0082] Step 4: Based on the genetic algorithm, using the parameters of various equivalent models obtained in Step 3, the measurement data of the synchronous phasor measurement device, and the proportion of the various equivalent models in the power generation load mixed model, the model parameters of each branch are obtained through iterative calculation.

[0083] The PMU measurements are as follows:

[0084] Wind power virtual generator model: three-phase current, three-phase voltage, active power, reactive power, frequency, switching quantities, sequence values, generator power angle, generator internal potential, etc.

[0085] Photovoltaic virtual generator model: three-phase current, three-phase voltage, active power, reactive power, frequency, switching quantities, sequence values, etc.

[0086] ZIP load model: load-side three-phase voltage, three-phase current, active power, reactive power, frequency, load internal resistance, etc.

[0087] Conventional generator unit model: three-phase current at generator terminals, three-phase voltage, active power, reactive power, frequency, switching quantities, sequence values, generator power angle, generator internal electromotive force, speed, etc.

[0088] Motor load model: load-side three-phase voltage, three-phase current, active power, reactive power, frequency, rotor internal resistance, speed, etc.

[0089] The specific steps are as follows:

[0090] Step 4.1: Expand the data of each load model parameter obtained in Step 2. Assume that the model parameters have m sub-parameters, and the sub-parameters contain all the parameters in each model. The expansion method is as follows:

[0091]

[0092] In the above formula, C Here are the sub-parameters; k is the proportion of each branch, with a precision of 0.1. The upper and lower limits of each sub-parameter are expanded by kp, and p is initialized to 1. It can be adjusted according to the ideality of the evolution results. If the evolution results are not ideal, p can be set to 2 to increase the parameter richness. The precision of p is 1. At this time, there are 21kp possible values ​​for the sub-parameters, and a total of 21p^m sets of model parameters. The number of parameters is greatly increased. All possible model parameters are put into the model parameter library for easy access in the following steps.

[0093] Step 4.2: Initialize the genetic generation G=0, and encode the model parameters using binary encoding;

[0094] Step 4.3: Select the model parameters from Step 4.2:

[0095] The basic steps for selecting an operation are as follows:

[0096] Step 4.3.1: Determine the number N of model parameters selected each time;

[0097] Step 4.3.2: Randomly select M groups of model parameters from the model parameter library (each group of model parameters has the same probability of being selected), calculate the fitness value of each sub-parameter in the model parameters, and select the model parameter with the best fitness value to enter the next generation of parameter population;

[0098] Step 4.3.3: Repeat step 4.3.2 multiple times (N times) until the new parameter size reaches the original parameter size.

[0099] Step 4.4: Perform cross operations on the model parameters left over from the selection operation in Step 4.3;

[0100] This invention employs a multi-point intersection method to perform multi-point intersection on model parameters, illustrated by a three-point intersection method, as follows: Figure 3 As shown;

[0101] P1 and P2 are model parameter 1 and model parameter 2, respectively. P1 is crossed with P2 in the sub-parameters a, c, and y to obtain two sets of crossed model parameters. This invention uses a method of randomly selecting from the model parameters to perform the cross operation, and the cross position is also randomly selected from the sub-parameters.

[0102] Step 4.5: Perform mutation operations on the model parameters left over from the crossover operation in Step 4.4;

[0103] This paper uses the real number mutation method to mutate the parameters after the crossover operation in step 4.4.

[0104] The basic steps of mutation operations are as follows:

[0105] Step 4.5.1: Determine whether the model parameters have mutated using a pre-set probability;

[0106] This invention uses the golden ratio to calculate whether model parameters need to be mutated. In the nth generation of model parameters, a number r∈(0,1) is randomly generated. If r≤0.382+0.618^(n) 1), then the mutation probability is 1 / (N*M), (N is the number of sub-parameters, M is the number of model parameters), that is, to allow one model parameter to mutate.

[0107] Step 4.5.2: Randomly select the mutation location for the sub-parameters to be mutated, using the following mutation method: Figure 4 As shown;

[0108] The mutation algorithm in step 4.5.1 is used to calculate whether the model parameters need to be mutated. If mutation is required, a random number between 1 and n is generated, where n is the number of sub-parameters. The generated random number represents the sub-parameter that needs to be mutated. For example... Figure 4 If the generated random number is 3, then the mutated sub-parameter is modified.

[0109] Step 4.6: Based on steps 4.3, 4.4, and 4.5, a set of model parameters can be obtained. Inputting the model parameters into the power generation load hybrid model can obtain the simulation curves of voltage amplitude, active power, and reactive power parameters. Then, compare them with the measured curves to obtain the voltage amplitude error, active power error, and reactive power error. Figure 5 This is a comparison chart of the simulated curve and the measured curve obtained from a certain iteration of evolution.

[0110] Define the average error E , Eis the average value of the voltage amplitude error, active power error, and reactive power error. A new error function is constructed for the present invention as follows:

[0111]

[0112]

[0113]

[0114] In the above formula, is the voltage amplitude error function, is the voltage amplitude error function, is the voltage amplitude error function, s is the number of sampling points of the simulation curve and the measured curve, is the parameter value of each sampling point of the simulation curve, is the parameter value of each sampling point of the measured curve.

[0115] Use to uniformly represent the voltage amplitude error function, voltage amplitude error function, voltage amplitude error function:

[0116] When When < E, output this set of parameters and perform the next round of selection, crossover, and mutation operations.

[0117] When > E, directly perform selection, crossover, and mutation operations on this set of parameters.

[0118] Step 4.7: After steps 4.2 - 4.6 in step 4, the error in the output parameters is the smallest set of parameters as the optimal model parameters.

[0119] Step 5, take the average value of the various equivalent model parameters obtained in step 4 as the optimal model parameters of the power generation load hybrid model.

[0120] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: still can modify the specific implementation manners of the present invention or make equivalent replacements, and any modification or equivalent replacement without departing from the spirit and scope of the present invention shall be covered by the protection scope of the claims of the present invention.

Claims

1. A method for parameter identification in a hybrid power generation load model, characterized in that, include: Step 1: Collect measurement data from the remote terminal system, the synchronous phasor measurement device, and meteorological parameters of the power grid location; Step 2: Based on the measurement data from the remote terminal system, convert each branch of the power grid into various equivalent models; wherein, the equivalent models include: wind power virtual generator model, photovoltaic virtual generator model, conventional unit model, motor load model, and ZIP load model; the various equivalent models together form a power generation load hybrid model; Step 3: Use DS evidence theory to estimate the parameters of various equivalent models; substitute the parameters of various equivalent models into the corresponding equivalent models to calculate the output power value of each type of model; use the power value as the standard for proportional calculation to obtain the sum of the output power of each type of model; compare the output power of each type of model with the sum of the power values ​​to determine the proportion of each type of equivalent model in the power generation load hybrid model. In step 3, the DS evidence theory is used to estimate the model parameters, establish the basic probability assignment, calculate the normalization constant based on the basic probability assignment, and calculate the fusion trust of each group of parameters; the model parameter with the highest fusion trust is substituted into each type of model to obtain the power curve, and the average power is taken to calculate the proportion of each type of equivalent model in the power generation load hybrid model. Step 4: Based on the genetic algorithm, using the parameters of various equivalent models obtained in Step 3, the measurement data of the synchronous phasor measurement device, and the proportion of the various equivalent models in the power generation load mixed model, the model parameters of each branch are obtained through iterative calculation. In step 4, the golden ratio is used to calculate whether the model parameters need to be mutated; in the nth generation of model parameters, a number r∈(0,1) is randomly generated. If r≤0.382+0.618^(n) 1), then the mutation probability is 1 / (N*M), where N is the number of sub-parameters and M is the number of model parameters, allowing one model parameter to mutate; Step 5: Take the average value of the various equivalent model parameters obtained in Step 4 as the optimal model parameters for the power generation load hybrid model.

2. The method for parameter identification of a hybrid power generation load model according to claim 1, characterized in that: In step 1, the measurement data of the remote terminal system includes: power direction, current amplitude, and voltage amplitude; The measurement data from the synchronous phasor measurement device includes power, voltage, and current over a period of one week; Meteorological parameters include: daily wind speed and sunshine duration.

3. The method for parameter identification of a hybrid power generation load model according to claim 2, characterized in that: The power average value is calculated hourly based on the power measured by the synchronous phasor measurement device. The standard deviation of the power over one week is then calculated using this average value as a benchmark. The standard deviation represents the power stability. The formula for calculating the power standard deviation is as follows: In the formula, Standard deviation, The number of hours within a week; 、 This refers to the actual power. This represents the average power per hour over a one-week period.

4. The method for parameter identification of a hybrid power generation load model according to claim 1, characterized in that: In step 2, the motor load model adopts the mathematical model of a T-type third-order induction motor as follows: in, , , , ; The generator rotor inertia time constant; and These are the mechanical and electromagnetic torques of the generator, respectively. For slip; This refers to the induced electromotive force inside the rotor, excluding the stator impedance. This is the current value; This is the proportionality coefficient related to the square of the resistance torque; This is the proportionality coefficient between the resistance torque and the rotational speed; It is a constant that depends on the motor parameters and the initial slip; This represents the small-signal value of the current. Rotor resistance; For stator reactance; For rotor reactance; For excitation reactance; The time constant of the stator open-circuit rotor circuit; For grounding inductance; This refers to the no-load torque caused by friction. ω is the rotor angular velocity.

5. The method for parameter identification of a hybrid power generation load model according to claim 1, characterized in that: In step 2, if the power direction of any branch is positive, the power stability is greater than the set threshold, and the voltage level is 110kV or below, then the branch is equivalent to a wind power virtual generator model; the threshold is set according to different voltage levels, set to 1~5 for voltage levels of 110kV or below, and set to 5~10 for voltage levels above 110kV. If the power direction of any branch is positive, the power stability is less than the set threshold, and the voltage level is 110kV or below, then the branch is equivalent to a photovoltaic virtual generator model. If the power direction of any branch is positive, the power stability is less than the set threshold, and the voltage level is above 110kV, then the branch is equivalent to a photovoltaic virtual generator model. If the power direction of any branch is negative and the power stability is less than a set threshold, then the branch is equivalent to a ZIP load model. If the power direction of any branch is negative and the power stability is greater than a set threshold, then the branch is equivalent to a motor load model.

6. The method for parameter identification of a hybrid power generation load model according to claim 1, characterized in that: In step 4, the model parameters are input into the power generation load hybrid model to obtain the simulated curves of voltage amplitude, active power, and reactive power parameters. These are then compared with the measured curves to obtain the voltage amplitude error, active power error, and reactive power error. The error function is constructed as follows: In the above formula, This is the voltage amplitude error function. This is the voltage amplitude error function. This is the voltage amplitude error function. This represents the number of sampling points for both the simulated and measured curves. For each sampling point of the simulation curve, the parameter value is... These are the parameter values ​​for each sampling point of the measured curve.

7. The method for parameter identification of a hybrid power generation load model according to claim 6, characterized in that: Define the average error E , E Voltage amplitude error Active power error reactive power error The average of the three; using A unified representation of voltage amplitude error function, voltage amplitude error function, and voltage amplitude error function: when < E At that time, the current parameters are output and the next round of selection, crossover, and mutation operations are performed; when > E At that time, selection, crossover, and mutation operations are performed directly on the current parameters.

Images