Distributed Power Supply Wave Blocking Characteristic Parameter Identification Method and System Based on Measured Data

Abstract

The present invention discloses a method and system for identifying the wave-blocking characteristic parameters of distributed power sources based on measured data. The method includes: obtaining n groups of measured data for identifying the parameters to be identified, identifying the measured data according to an improved convolutional neural network to obtain a preliminary result of parameter identification, inputting the preliminary results of parameter identification of the n groups into the single-machine grid-connected model of type B distributed power sources in PSASP software, measuring the electrical quantities of each group of identification results under various working conditions, and screening the minimum value of the weighted average deviation relative error of the n groups of electrical quantities as the final result of parameter identification. It can identify the wave-blocking characteristic parameters of distributed power sources by using multiple groups of measured data, so as to accurately model the low-voltage wave-blocking control of type B distributed power sources.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system analysis, and particularly relates to a method and system for identifying the wave-blocking characteristic parameters of distributed power sources based on measured data. Background Art

[0002] In the stability analysis of power systems, the accurate modeling of distributed photovoltaic power generation systems is a fundamental task. Among them, the inverter, as the core component of the system, the accuracy of its model depends on the accurate acquisition of control parameters. The wave-blocking characteristic parameters of distributed photovoltaic systems, as part of the inverter control parameters, are crucial for ensuring the continuous operation of photovoltaic systems. However, due to manufacturer confidentiality or the dynamic changes in the operating environment, it is difficult to guarantee the accuracy of the wave-blocking characteristic parameters, which affects the accurate analysis of the grid connection characteristics of inverters and photovoltaic power stations. Therefore, conducting research on the identification of model parameters of grid-connected photovoltaic inverters, improving the accuracy and reliability of identification, and constructing a grid-connected photovoltaic inverter model that truly reflects the actual situation have important value for power grid operation planning, accident analysis, and ensuring the safe and stable operation of the power grid after large-scale access of distributed photovoltaic power sources.

[0003] In a detailed distribution network, distributed new energy devices can be mainly divided into the following three types: A distributed new energy (full-power VSC grid-connected new energy with ride-through function); B distributed new energy (full-power VSC grid-connected new energy without ride-through function); doubly-fed wind turbines. Among them, B distributed new energy accounts for a relatively large proportion. Compared with A distributed new energy, B distributed new energy does not consider the operating requirements of centralized new energy during design and manufacturing, does not have low-voltage ride-through control measures, but is equipped with a "wave-blocking (soft disconnection from the grid)" function of protection equipment. The wave-blocking function means that when the voltage is low, in order to protect the commutation equipment, the new energy will stop power output, and then slowly recover to the state before wave-blocking after a period of time. The power recovery process is generally a linear recovery with a fixed slope. In a distributed photovoltaic power generation system, a photovoltaic power generation system connected to the grid through a voltage level of 35 kV or above and a photovoltaic power generation system connected to the public grid through 10 kV is type A distributed photovoltaic; a photovoltaic power generation system connected to the grid through a voltage level of 380 V or below and a photovoltaic power generation system connected to the user side through 10 kV is type B distributed photovoltaic.

[0004] Currently, the accurate method for obtaining the simulation parameters of new energy systems is mainly to identify the model parameters by combining appropriate algorithms with measured data. Common methods include the theoretical analysis method. The theoretical analysis method can relatively completely simulate the internal process of a photovoltaic power generation system. However, with the increase in the complexity of the control system, the solution of non-linear calculus equations becomes difficult, and the existence of some unmeasurable variables will affect the calculation results, resulting in the parameters obtained needing to be adjusted manually many times before they can be applied to photovoltaic control. Summary of the Invention

[0005] The present invention provides a method, system and readable storage medium for identifying the wave-blocking characteristic parameters of distributed power sources based on measured data, which are used to solve the technical problem that the parameters obtained by existing methods need to be manually adjusted multiple times before they can be applied to photovoltaic control.

[0006] In a first aspect, the present invention provides a method for identifying the wave-blocking characteristic parameters of distributed power sources based on measured data, including:

[0007] Obtaining n groups of measured data for identifying the parameters to be identified;

[0008] Identifying the measured data according to an improved convolutional neural network to obtain a preliminary result of parameter identification, wherein the improved convolutional neural network includes an adaptive edge padding full convolution and a valid convolution;

[0009] Inputting the preliminary identification results of n groups of parameters into the single-machine grid-connected model of type B distributed power sources in PSASP software, measuring the electrical quantities of each group of identification results under various working conditions, and screening the minimum value of the weighted average deviation relative error of the n groups of electrical quantities as the final parameter identification result.

[0010] In a second aspect, the present invention provides a system for identifying the wave-blocking characteristic parameters of distributed power sources based on measured data, including:

[0011] An obtaining module configured to obtain n groups of measured data for identifying the parameters to be identified;

[0012] An identifying module configured to identify the measured data according to an improved convolutional neural network to obtain a preliminary result of parameter identification, wherein the improved convolutional neural network includes an adaptive edge padding full convolution and a valid convolution;

[0013] A screening module configured to input the preliminary identification results of n groups of parameters into the single-machine grid-connected model of type B distributed power sources in PSASP software, measure the electrical quantities of each group of identification results under various working conditions, and screen the minimum value of the weighted average deviation relative error of the n groups of electrical quantities as the final parameter identification result.

[0014] In a third aspect, there is provided an electronic device, which includes: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the steps of the method for identifying the wave-blocking characteristic parameters of distributed power sources based on measured data according to any embodiment of the present invention.

[0015] Fourthly, the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the program instructions are executed by a processor, the processor is caused to execute the steps of the method for identifying the wave-blocking characteristic parameters of a distributed power source based on measured data according to any embodiment of the present invention.

[0016] The method and system for identifying the wave-blocking characteristic parameters of a distributed power source based on measured data according to the present application obtain n groups of measured data for identifying the parameters to be identified, identify the measured data according to an improved convolutional neural network to obtain a preliminary result of parameter identification, input the n groups of preliminary results of parameter identification into the single-machine grid-connected model of a Class B distributed power source in PSASP software, measure the electrical quantities of each group of identification results under various working conditions, and screen the minimum value of the weighted average deviation relative error of the n groups of electrical quantities as the final result of parameter identification. It can identify the wave-blocking characteristic parameters of a distributed power source by using multiple groups of measured data, so as to accurately model the low-voltage wave-blocking control of a Class B distributed power source. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0018] Figure 1 It is a flowchart of a method for identifying the wave-blocking characteristic parameters of a distributed power source based on measured data according to an embodiment of the present invention;

[0019] Figure 2 It is a structural block diagram of a system for identifying the wave-blocking characteristic parameters of a distributed power source based on measured data according to an embodiment of the present invention;

[0020] Figure 3 It is a schematic structural diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments of the present invention belong to the scope of protection of the present invention.

[0022] Please refer to Figure 1 , which shows a flowchart of a method for identifying the wave-blocking characteristic parameters of a distributed power source based on measured data according to the present application.

[0023] As Figure 1 shown, the method for identifying the wave-blocking characteristic parameters of distributed power sources based on measured data specifically includes the following steps:

[0024] Step S101, obtain n groups of measured data for identifying the parameters to be identified.

[0025] In this step, select the transient data of the active power response characteristic of the photovoltaic inverter as the observable of the wave-blocking duration and the slope of the active current recovery after wave-blocking, and select the transient data of the reactive power response characteristic of the photovoltaic inverter as the observable of the slope of the reactive current recovery after wave-blocking; set n groups of test conditions, and determine the test point as the grid connection point on the AC side. The n groups of test conditions and test points form n groups of test schemes. Among them, any group of test conditions includes: voltage disturbance amplitude, duration, active power command, and reactive power command.

[0026] Furthermore, set the initial active power and reactive power of the photovoltaic inverter in the test platform, set the voltage disturbance parameters of the grid simulator on the AC side in the test platform according to the voltage disturbance amplitude and duration in any group of test conditions, and provide the corresponding voltage disturbance to the photovoltaic inverter in the test platform so that the test platform can perform transient simulation tests; after the test platform simulation is completed and enters the stable operation state, record the electrical data on the AC side and use it as a group of measured data, and accumulate to obtain n groups of measured data; the measured data includes: active power, reactive power, reactive current, total current, and fundamental voltage of the grid voltage.

[0027] Step S102, identify the measured data according to the improved convolutional neural network to obtain a preliminary result of parameter identification, where the improved convolutional neural network includes adaptive edge padding full convolution and valid convolution.

[0028] In this step, the structure of the improved convolutional neural network includes: input layer: collect data, perform data preprocessing, and convert it into pictures; first convolutional layer: introduce an adaptive edge padding mechanism, and during the convolution process, dynamically adjust the padding value according to the neighborhood characteristics of the edge pixels; pooling layer: perform downsampling on the processed image to obtain the maximum value from the area of the feature map; second convolutional layer, during the convolution process, do not perform zero-padding around the input data, use a convolution kernel of a preset size to perform multiple convolutions, strengthen the local features, and at the same time reduce the size of the output feature map; fully connected layer: adopt softmax full connection, and the obtained activation value is the picture feature extracted by the convolutional neural network; output layer: output the final result calculated by the multi-convolution layer hybrid convolutional neural network algorithm.

[0029] The adaptive edge padding full convolution includes:

[0030] Extract the original input image;

[0031] Use a traditional convolutional layer to perform a preliminary convolution operation on the input image to extract basic edge features;

[0032] Preliminarily extract the edge feature map;

[0033] Use the Fourier transform method to perform frequency analysis on the edge feature map to determine the frequency characteristics of different regions;

[0034] According to the frequency analysis results, apply different filling strategies to different regions. Among them, the filling strategies include: for high-frequency regions, use a preset filling value to retain detailed information, and for low-frequency regions, increase the filling value to smooth the features;

[0035] Output the feature map after adaptive filling.

[0036] Valid convolution includes:

[0037] Extract the original input image;

[0038] Through multiple layers of Valid convolution, gradually extract the high-level features of the image, and finally perform classification through a fully connected layer. Among them, the Valid convolution selects a convolution kernel of a preset size, and the stride is set to 1;

[0039] Output the feature map after Valid convolution.

[0040] It should be noted that analyze the fundamental voltage of the grid voltage in the measured data, and extract the voltage sag depth and voltage sag start time of the disturbance experiment;

[0041] Modify the parameters corresponding to the single-unit grid-connected model of type B distributed power sources according to the extracted voltage sag depth and voltage sag start time of the disturbance experiment. At the same time, set the control parameters of the photovoltaic inverter in the single-unit grid-connected model of type B distributed power sources to build the single-unit grid-connected model of type B distributed power sources required for identification;

[0042] Use the factory value of the photovoltaic inverter in the single-unit grid-connected model of type B distributed power sources as the iterative initial value of the improved convolutional neural network;

[0043] Use the improved convolutional neural network to perform preliminary identification on the measured data to obtain the preliminary results of parameter identification of the wave blocking duration, the active current recovery slope after wave blocking, and the reactive current recovery slope after wave blocking.

[0044] In this embodiment, traditional full convolution adds sufficient zero-padding around the input data to keep the size of the output data the same as that of the input data. However, the edge information may be diluted. Here, an adaptive edge-padding mechanism is introduced to form an adaptive edge-padding full convolution, which dynamically adjusts the padding value according to the neighborhood characteristics of the edge pixels, making the edge information richer and enhancing the expression ability of the edge features. Traditional valid convolution does not add any padding around the input data, and the size of the output data is smaller than that of the input data, reducing the computational amount and gradually shrinking the size of the feature map. However, the features are relatively concentrated. Here, a small-size convolution kernel is used for multiple convolutions, that is, a small-convolution-kernel valid convolution, to further mine the deep information of the local features, enhance the local features, and improve the discriminability of the features. Compared with the single valid convolution or full convolution mode in the conventional convolutional neural network algorithm, the multi-convolution-layer hybrid convolutional neural network algorithm improves the full convolution and valid convolution modes respectively, that is, the adaptive edge-padding full convolution and the small-convolution-kernel valid convolution, improving the accuracy of parameter identification, reducing the computational amount, gradually shrinking the data size, and realizing the improvement of the identification accuracy of the wave-blocking characteristic parameters of distributed power sources.

[0045] Step S103: Input the preliminary identification results of n groups of parameters into the single-machine grid-connected model of Class B distributed power sources in the PSASP software, measure the electrical quantities of each group of identification results under various working conditions, and select the minimum weighted average deviation relative error of the n groups of electrical quantities as the final parameter identification result.

[0046] In this step, fill the preliminary identification results of n groups of parameters into the single-machine grid-connected model of Class B distributed power sources in the PSASP software, run the PSASP simulation software, test the transient simulation results of each group of models under the corresponding working conditions, and record the transient stability values of the active power and reactive power of each group of models under each working condition.

[0047] Calculate the weighted average deviation and the relative error of the weighted average deviation of the active power and reactive power between each group of measured data and its corresponding model, and select the preliminary identification result of the parameters corresponding to the group of measured data with the minimum relative error as the final parameter identification result.

[0048] In summary, the method of this application obtains n groups of measured data for identifying the parameters to be identified, identifies the measured data according to the improved convolutional neural network to obtain the preliminary parameter identification results, inputs the preliminary identification results of n groups of parameters into the single-machine grid-connected model of Class B distributed power sources in the PSASP software, measures the electrical quantities of each group of identification results under various working conditions, and selects the minimum weighted average deviation relative error of the n groups of electrical quantities as the final parameter identification result, which can use multiple groups of measured data to identify the wave-blocking characteristic parameters of distributed power sources, so as to accurately model the low-voltage wave-blocking control of Class B distributed power sources.

[0049] Please refer to Figure 2 , which shows a structural block diagram of a distributed power supply wave-blocking characteristic parameter identification system based on measured data according to the present application.

[0050] As Figure 2 shown, the distributed power supply wave-blocking characteristic parameter identification system 200 includes an acquisition module 210, an identification module 220, and a screening module 230.

[0051] Among them, the acquisition module 210 is configured to acquire n sets of measured data for identifying parameters to be identified;

[0052] The identification module 220 is configured to identify the measured data according to an improved convolutional neural network to obtain a preliminary parameter identification result, where the improved convolutional neural network includes an adaptive edge padding full convolution and a valid convolution;

[0053] The screening module 230 is configured to input the n sets of preliminary parameter identification results into the B-type distributed power supply single-unit grid-connected model in the PSASP software, measure the electrical quantities of each set of identification results under various working conditions, and screen the minimum value of the weighted average deviation relative error of the n sets of electrical quantities as the final parameter identification result.

[0054] It should be understood that Figure 2 the modules described in Figure 1 correspond to the respective steps in the method described with reference to Figure 2 . Therefore, the operations, features, and corresponding technical effects described above for the method also apply to the modules in

[0055] and will not be elaborated here.

[0056] In some other embodiments, the embodiments of the present invention further provide a computer-readable storage medium, on which a computer program is stored. When the program instructions are executed by a processor, the processor is caused to execute the method for identifying the distributed power supply wave-blocking characteristic parameters based on measured data in any of the above method embodiments;

[0057] As an implementation manner, the computer-readable storage medium of the present invention stores computer-executable instructions, and the computer-executable instructions are set as:

[0058] Acquire n sets of measured data for identifying parameters to be identified;

[0059] Input the preliminary identification results of n groups of parameters into the Type-B distributed power single-unit grid-connected model in the PSASP software, measure the electrical quantities of each group of identification results under various working conditions, and select the minimum value of the weighted average deviation relative error of the n groups of electrical quantities as the final parameter identification result.

[0060] The computer-readable storage medium may include a storage program area and a storage data area. Among them, the storage program area can store an operating system and application programs required for at least one function; the storage data area can store data created according to the use of the distributed power wave-blocking characteristic parameter identification system based on measured data, etc. In addition, the computer-readable storage medium may include high-speed random access memory, and may also include memories, such as at least one magnetic disk storage device, a flash memory device, or other non-volatile solid-state storage devices. In some embodiments, the computer-readable storage medium may optionally include a memory remotely provided relative to the processor, and these remote memories may be connected to the distributed power wave-blocking characteristic parameter identification system based on measured data through a network. Examples of the above network include but are not limited to the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0061] Figure 3 It is a schematic structural diagram of the electronic device provided by an embodiment of the present invention, as Figure 3 shown. The device includes: a processor 310 and a memory 320. The electronic device may further include: an input device 330 and an output device 340. The processor 310, the memory 320, the input device 330, and the output device 340 may be connected through a bus or other means, Figure 3 and here, the connection through the bus is taken as an example. The memory 320 is the above-mentioned computer-readable storage medium. The processor 310 executes various functional applications and data processing of the server by running non-volatile software programs, instructions, and modules stored in the memory 320, that is, implements the distributed power wave-blocking characteristic parameter identification method based on measured data in the above method embodiment. The input device 330 can receive input digital or character information, and generate key signal inputs related to the user settings and function controls of the distributed power wave-blocking characteristic parameter identification system based on measured data. The output device 340 may include a display device such as a display screen.

[0062] The above electronic device can execute the method provided by the embodiment of the present invention, and has corresponding functional modules and beneficial effects for executing the method. For technical details not described in detail in this embodiment, reference may be made to the method provided by the embodiment of the present invention.

[0063] As an implementation manner, the above electronic device is applied to a distributed power supply wave-blocking characteristic parameter identification system based on measured data and is used for a client, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to:

[0064] Obtain n groups of measured data for identifying parameters to be identified;

[0065] Identify the measured data according to an improved convolutional neural network to obtain a preliminary parameter identification result, wherein the improved convolutional neural network includes adaptive edge padding full convolution and valid convolution;

[0066] Input the n groups of preliminary parameter identification results into the Type B distributed power supply single-machine grid-connected model in PSASP software, measure the electrical quantities of each group of identification results under various working conditions, and select the minimum weighted average deviation relative error of the n groups of electrical quantities as the final parameter identification result.

[0067] Through the description of the above implementation manners, those skilled in the art can clearly understand that each implementation manner can be realized by means of software plus a necessary general hardware platform, and of course, it can also be realized by hardware. Based on such an understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of each embodiment or some parts of the embodiments.

[0068] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present invention.

Claims

1. A method for identifying the characteristic parameters of distributed power supply based on measured data, characterized in that: include: Obtaining n groups of measured data for identifying parameters to be identified; The measured data is identified according to the improved convolutional neural network to obtain preliminary parameter identification results, wherein the improved convolutional neural network includes adaptive edge filling Full convolution and Valid convolution, and the structure of the improved convolutional neural network includes: Input layer: collect data, preprocess data and convert it into images; The first convolutional layer: introduces an adaptive edge filling mechanism. During the convolution process, the filling value is dynamically adjusted according to the neighborhood characteristics of edge pixels. Pooling layer: downsamples the processed image and obtains the maximum value from the area of ​​the feature map; In the second convolution layer, during the convolution process, no zero padding is performed around the input data. Multiple convolutions are performed using a convolution kernel of a preset size to enhance local features and reduce the size of the output feature map. Fully connected layer: softmax full connection is used, and the activation value obtained is the image feature extracted by the convolutional neural network; Output layer: outputs the final result calculated by the multi-convolutional layer hybrid convolutional neural network algorithm; The adaptive edge filling Full convolution includes: Extract the original input image; Use the traditional convolution layer to perform preliminary convolution operations on the input image to extract basic edge features; Preliminary extraction of edge feature maps; Use Fourier transform method to perform frequency analysis on edge feature map to determine the frequency characteristics of different areas; According to the frequency analysis results, different filling strategies are applied to different regions, wherein the filling strategies include: for high-frequency regions, a preset filling value is used to retain detail information, and for low-frequency regions, the filling value is increased to smooth features; Output the feature map after adaptive padding; The preliminary identification results of n groups of parameters are input into the single-machine grid-connected model of Class B distributed power supply in PSASP software, the electrical quantities of each group of identification results under various working conditions are measured, and the minimum relative error of the weighted average deviation of the n groups of electrical quantities is selected as the final parameter identification result.

2. According to the method for identifying the wave shielding characteristic parameters of a distributed power supply based on measured data in claim 1, it is characterized in that: Before obtaining n groups of measured data for identifying the parameters to be identified, the method further includes: The transient data of the active response characteristics of the photovoltaic inverter are selected as the observed values ​​of the duration of the power outage and the active current recovery slope after the power outage, and the transient data of the reactive response characteristics of the photovoltaic inverter are selected as the observed value of the reactive current recovery slope after the power outage; Set n groups of test conditions, and determine the test point as the AC side grid connection point, and form n groups of test schemes by the n groups of test conditions and the test points, wherein any group of test conditions includes: voltage disturbance amplitude, duration, active power instruction and reactive power instruction.

3. The method for identifying the wave shielding characteristic parameters of a distributed power supply based on measured data according to claim 1 is characterized in that: The step of obtaining n groups of measured data for identifying the parameters to be identified includes: Setting the initial active power and reactive power of the photovoltaic inverter in the test platform, setting the voltage disturbance parameters of the AC side power grid simulator in the test platform according to the voltage disturbance amplitude and duration in any set of test conditions, and providing corresponding voltage disturbance to the photovoltaic inverter in the test platform, so that the test platform can perform transient simulation test; After the simulation of the test platform is completed and enters a stable operating state, the AC side electrical data is recorded and used as a group of measured data, and n groups of measured data are accumulated; the measured data include: active power, reactive power, reactive current, total current and fundamental voltage of the grid voltage.

4. The method for identifying the wave shielding characteristic parameters of a distributed power supply based on measured data according to claim 1 is characterized in that: The Valid convolution includes: Extract the original input image; Through multiple layers of Valid convolution, the high-level features of the image are gradually extracted, and finally classified through the fully connected layer. Among them, the Valid convolution selects a convolution kernel of a preset size and the step size is set to 1; Output the feature map after Valid convolution.

5. The method for identifying the wave shielding characteristic parameters of a distributed power supply based on measured data according to claim 1 is characterized in that: The method of identifying the measured data according to the improved convolutional neural network to obtain preliminary parameter identification results includes: Analyze the fundamental voltage of the power grid voltage in the measured data, and extract the voltage drop depth and voltage drop start time of the disturbance experiment; According to the extracted voltage drop depth and voltage drop start time of the disturbance experiment, the corresponding parameters of the Class B distributed power supply single-machine grid-connected model are modified, and the control parameters of the photovoltaic inverter in the Class B distributed power supply single-machine grid-connected model are set to build the Class B distributed power supply single-machine grid-connected model required for identification; The factory value of the photovoltaic inverter in the Class B distributed power single-machine grid-connected model is used as the iterative initial value of the improved convolutional neural network; The measured data were preliminarily identified using an improved convolutional neural network, and preliminary results of parameter identification were obtained for the duration of wave blocking, the active current recovery slope after wave blocking, and the reactive current recovery slope after wave blocking.

6. The method for identifying the shielding characteristic parameters of a distributed power supply based on measured data according to claim 1 is characterized in that: The method of inputting the preliminary identification results of n groups of parameters into the single-machine grid-connected model of Class B distributed power supply in the PSASP software, measuring the electrical quantity of each group of identification results under various working conditions, and selecting the minimum relative error value of the weighted average deviation of the n groups of electrical quantities as the final parameter identification results includes: Fill the preliminary identification results of n groups of parameters into the single-machine grid-connected model of Class B distributed power supply in PSASP software, run PSASP simulation software, test the transient simulation results of each group of models under corresponding working conditions, and record the transient stability values ​​of active power and reactive power of each group of models under various working conditions; The weighted average deviation of active power and reactive power of each set of measured data and its corresponding model as well as the relative error of the weighted average deviation are calculated, and the preliminary parameter identification result corresponding to the set of measured data with the smallest relative error is selected as the final parameter identification result.

7. A distributed power supply blocking characteristic parameter identification system based on measured data, characterized in that: include: An acquisition module configured to acquire n groups of measured data for identifying parameters to be identified; The identification module is configured to identify the measured data according to the improved convolutional neural network to obtain a preliminary parameter identification result, wherein the improved convolutional neural network includes an adaptive edge filling Full convolution and a Valid convolution, and the structure of the improved convolutional neural network includes: Input layer: collect data, preprocess data and convert it into images; The first convolutional layer: introduces an adaptive edge filling mechanism. During the convolution process, the filling value is dynamically adjusted according to the neighborhood characteristics of edge pixels. Pooling layer: downsamples the processed image and obtains the maximum value from the area of ​​the feature map; In the second convolution layer, during the convolution process, no zero padding is performed around the input data. Multiple convolutions are performed using a convolution kernel of a preset size to enhance local features and reduce the size of the output feature map. Fully connected layer: softmax full connection is used, and the activation value obtained is the image feature extracted by the convolutional neural network; Output layer: outputs the final result calculated by the multi-convolutional layer hybrid convolutional neural network algorithm; The adaptive edge filling Full convolution includes: Extract the original input image; Use the traditional convolution layer to perform preliminary convolution operations on the input image to extract basic edge features; Preliminary extraction of edge feature maps; Use Fourier transform method to perform frequency analysis on edge feature map to determine the frequency characteristics of different areas; According to the frequency analysis results, different filling strategies are applied to different regions, wherein the filling strategies include: for high-frequency regions, a preset filling value is used to retain detail information, and for low-frequency regions, the filling value is increased to smooth features; Output the feature map after adaptive padding; The screening module is configured to input the preliminary identification results of n groups of parameters into the single-machine grid-connected model of Class B distributed power supply in the PSASP software, measure the electrical quantities of each group of identification results under various working conditions, and screen the minimum relative error of the weighted average deviation of the n groups of electrical quantities as the final parameter identification result.

Images