Wind power plant sweep frequency impedance analysis method and device based on differential neural network, electronic equipment and storage medium

Abstract

The invention discloses a wind power plant sweep frequency impedance analysis method and device based on a differential neural network, electronic equipment and a storage medium. The method comprises the steps of obtaining a wind power plant simulation model, a wind speed of a wind power plant at a to-be-analyzed moment, a bus voltage of the wind power plant at the to-be-analyzed moment and a plurality of to-be-input disturbance voltages; adding each disturbance voltage and the bus voltage to obtain a plurality of superposed voltages; converting each superimposed voltage to obtain a plurality of superimposed direct current side voltages; combining each superposed direct current side voltage with the wind speed to generate a plurality of groups of input data; inputting each group of input data into the wind power plant simulation model in sequence to generate each disturbance fan side current; calculating and generating an impedance value corresponding to each disturbance voltage according to each disturbance fan side current and the corresponding disturbance voltage; and according to the impedance value corresponding to each disturbance voltage to be input, wind power plant sweep frequency impedance analysis is carried out. By implementing the method, the efficiency of frequency sweep impedance analysis of the wind power plant can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of wind farm simulation calculation, and in particular to a wind farm swept frequency impedance analysis method, device, electronic equipment and storage medium based on differential neural network. Background Art

[0002] With the rapid development of new energy technologies, wind power generation, as an important component of clean energy, continues to increase its share in the global power system. In order to optimize the design and operation performance of wind farms, simulation technology is widely used in the dynamic characteristics analysis and system optimization of wind farms. In related technologies, the swept frequency impedance analysis of large-scale wind farms usually needs to be completed by building a wind farm simulation model. These simulation models can effectively simulate the dynamic behavior of wind farms under different working conditions, and provide important support for evaluating the interaction characteristics between wind farms and power grids. Research on the construction of mathematical models and simulation methods for doubly fed and direct-drive wind turbines has achieved many results, and laid a theoretical foundation for the implementation of simulation models in engineering applications.

[0003] However, existing technologies still face significant problems when performing swept frequency impedance analysis of wind farms. For example, traditional simulation methods usually rely on a large amount of hardware parameters and operating data, and the modeling process is complex and time-consuming, which not only increases the difficulty of analysis, but also limits the flexibility and adaptability of simulation methods. At the same time, the lack of a lightweight impedance analysis method makes it difficult to quickly and efficiently complete the impedance assessment of large-scale wind farms. These problems mainly stem from the high dependence of traditional modeling methods on physical parameters and the complexity of the operating characteristics of wind farms themselves, which makes it difficult for existing technologies to meet the needs of rapid analysis and optimization. Summary of the invention

[0004] The embodiment of the present invention provides a method, device, electronic device and storage medium for wind farm frequency sweep impedance analysis based on differential neural network. The efficiency of wind farm frequency sweep impedance analysis can be improved by implementing the present invention.

[0005] An embodiment of the present invention provides a wind farm swept frequency impedance analysis method based on a differential neural network, comprising:

[0006] Obtaining a wind farm simulation model, the wind speed of the wind farm at the time to be analyzed, the bus voltage of the wind farm at the time to be analyzed, and several disturbance voltages to be input; wherein the wind farm simulation model is constructed based on a differential neural network;

[0007] Add each disturbance voltage to be input to the bus voltage of the wind farm at the time to be analyzed to obtain a number of superimposed voltages; transform each superimposed voltage to obtain a number of superimposed DC side voltages; combine each superimposed DC side voltage with the wind speed of the wind farm at the time to be analyzed to generate a number of groups of input data;

[0008] Inputting each group of input data into the wind farm simulation model in turn, so that the differential neural network in the wind farm simulation model generates a corresponding disturbed wind turbine side current according to the current input data;

[0009] According to each disturbed wind turbine side current and the corresponding disturbed voltage to be input, an impedance value corresponding to each disturbed voltage to be input is calculated and generated;

[0010] According to the impedance value corresponding to each disturbance voltage to be input, the wind farm swept frequency impedance analysis is performed.

[0011] Furthermore, the training of the differential neural network includes:

[0012] Obtain a conventional operation training set and a high-frequency disturbance training set for model training; wherein each sample in the conventional operation training set includes a first DC side current label corresponding to the DC side voltage without superimposed disturbance, wind speed, and the DC side voltage without superimposed disturbance; each sample in the high-frequency disturbance training set includes a second DC side current label corresponding to the DC side voltage with superimposed high-frequency disturbance, wind speed, and the DC side voltage with superimposed high-frequency disturbance; the DC side voltage without superimposed disturbance is the DC side voltage corresponding to the normal operation of the wind farm under the conventional bus voltage condition; the DC side voltage with superimposed high-frequency disturbance is the DC side voltage corresponding to the operation of the wind farm after applying high-frequency disturbance under the conventional bus voltage condition;

[0013] Randomly divide the regular operation training set into a plurality of batches of first training samples according to a preset number;

[0014] The first training samples of each batch are sequentially input into the differential neural network, and the differential neural network is trained until a preset number of training times is reached; wherein, when the differential neural network receives each batch of the first training samples, it outputs the current value of the wind power side corresponding to the first training sample; according to the current value of the wind power side and the corresponding first DC side current label, the loss function value is calculated by the loss function; and the differential neural network is updated according to the loss function value by using the optimizer;

[0015] Randomly divide the high-frequency perturbation training set into a plurality of batches of second training samples according to a preset number;

[0016] The second training samples of each batch are sequentially input into the differential neural network, and the differential neural network is trained until a preset number of training times is reached; wherein, when the differential neural network receives each batch of second training samples, it outputs the current value of the wind power side corresponding to the second training sample; according to the current value of the wind power side and the corresponding second DC side current label, the loss function value is calculated by the loss function; and the differential neural network is updated according to the loss function value by using the optimizer.

[0017] Furthermore, before sequentially inputting each group of input data into the wind farm simulation model so that the differential neural network in the wind farm simulation model generates the corresponding disturbed wind turbine side current according to the current input data, the method further includes:

[0018] Obtain the bus current of the wind farm at the time to be analyzed;

[0019] The bus current of the wind farm at the time to be analyzed and the bus voltage of the wind farm at the time to be analyzed are used as the initial simulation environment of the wind farm simulation model;

[0020] Transform the bus voltage of the wind farm at the time to be analyzed to obtain the DC side initialization voltage;

[0021] Repeat the simulation environment update operation until the absolute value of the difference between the bus initialization current and the bus current of the wind farm at the time to be analyzed does not exceed a preset threshold;

[0022] The simulation environment updating operation includes:

[0023] In the current simulation environment, the DC side initialization voltage and the wind speed of the wind farm at the time to be analyzed are input into the wind farm simulation model, so that the differential neural network in the wind farm simulation model generates the current DC side current according to the initialization voltage and the wind speed of the wind farm at the time to be analyzed; wherein the initial simulation environment is the initial simulation environment;

[0024] Perform inverse transformation on the current DC side current to obtain the current bus current;

[0025] When the absolute difference between the current bus current and the bus current of the wind farm at the time to be analyzed is greater than a preset threshold, the current bus current and the bus voltage of the wind farm at the time to be analyzed are used as the updated initial simulation environment.

[0026] Further, the calculation and generation of the impedance value corresponding to each disturbance voltage to be input according to each disturbance wind turbine side current and the corresponding disturbance voltage to be input includes:

[0027] The quotient between the current disturbance voltage and the wind turbine side current after the disturbance is calculated to generate an impedance value corresponding to the current disturbance voltage.

[0028] Furthermore, a regular run training set is generated in the following way:

[0029] Obtain several combinations of DC side voltage and wind speed without superimposed disturbance as a set of typical data;

[0030] Input each set of typical data into a preset wind farm physical model for simulation, and use the DC side current obtained by simulating the wind farm physical model as the first DC side label of the current typical data;

[0031] Combining each group of typical data with the corresponding first DC side label to generate a number of first samples;

[0032] Generate a regular operation training set according to the first sample;

[0033] The wind farm physical model is generated in the following way:

[0034] Obtain the equipment parameters of the wind farm, the operating environment data of the wind farm, and the operation control strategy of the wind farm, and build a physical model of the wind farm based on CloudPSS SimStudio software.

[0035] Furthermore, a high-frequency perturbation training set is generated in the following way:

[0036] Obtain several combinations of DC side voltage and wind speed with superimposed high-frequency disturbances as a set of typical data;

[0037] Input each set of typical data into a preset wind farm physical model, and use the DC side current of the wind farm physical model as the second DC side label of the current typical data;

[0038] Combining each group of typical data with the corresponding second DC side label to generate a number of second samples;

[0039] A high-frequency perturbation training set is generated according to the second sample.

[0040] Based on the above method embodiment, the present invention provides a corresponding device embodiment.

[0041] An embodiment of the present invention provides a wind farm swept frequency impedance analysis device based on differential neural network, comprising: a data acquisition module, an input data generation module, a disturbance wind turbine side current generation module, an impedance value calculation module and a swept frequency impedance analysis module;

[0042] The data acquisition module is used to acquire a wind farm simulation model, a wind speed at a time to be analyzed in the wind farm, a bus voltage at a time to be analyzed in the wind farm, and a number of disturbance voltages to be input; wherein the wind farm simulation model is constructed based on a differential neural network;

[0043] The input data generation module is used to add each disturbance voltage to be input to the bus voltage of the wind farm at the time to be analyzed to obtain a number of superimposed voltages; transform each superimposed voltage to obtain a number of superimposed DC side voltages; combine each superimposed DC side voltage with the wind speed of the wind farm at the time to be analyzed to generate a number of groups of input data;

[0044] The disturbance wind turbine side current generation module is used to input each group of input data into the wind farm simulation model in sequence, so that the differential neural network in the wind farm simulation model generates the corresponding disturbance wind turbine side current according to the current input data;

[0045] The impedance value calculation module is used to calculate and generate the impedance value corresponding to each disturbance voltage to be input according to each disturbance fan side current and the corresponding disturbance voltage to be input;

[0046] The swept frequency impedance analysis module is used to perform a swept frequency impedance analysis of the wind farm according to the impedance value corresponding to each disturbance voltage to be input.

[0047] Furthermore, the wind farm swept frequency impedance analysis device based on differential neural network further includes: a simulation model steady-state operation module;

[0048] The simulation model steady-state operation module is used to obtain the bus current of the wind farm at the time to be analyzed;

[0049] The bus current of the wind farm at the time to be analyzed and the bus voltage of the wind farm at the time to be analyzed are used as the initial simulation environment of the wind farm simulation model;

[0050] Transform the bus voltage of the wind farm at the time to be analyzed to obtain the DC side initialization voltage;

[0051] Repeat the simulation environment update operation until the absolute value of the difference between the bus initialization current and the bus current of the wind farm at the time to be analyzed does not exceed a preset threshold;

[0052] The simulation environment updating operation includes:

[0053] In the current simulation environment, the DC side initialization voltage and the wind speed of the wind farm at the time to be analyzed are input into the wind farm simulation model, so that the differential neural network in the wind farm simulation model generates the current DC side current according to the initialization voltage and the wind speed of the wind farm at the time to be analyzed; wherein the initial simulation environment is the initial simulation environment;

[0054] Perform inverse transformation on the current DC side current to obtain the current bus current;

[0055] When the absolute difference between the current bus current and the bus current of the wind farm at the time to be analyzed is greater than a preset threshold, the current bus current and the bus voltage of the wind farm at the time to be analyzed are used as the updated initial simulation environment.

[0056] Based on the above method item embodiments, the present invention provides corresponding electronic device item embodiments.

[0057] An embodiment of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the wind farm swept frequency impedance analysis method based on a differential neural network as described in any one of the above method embodiments can be implemented.

[0058] Based on the above method item embodiments, the present invention provides a corresponding storage medium item embodiment.

[0059] An embodiment of the present invention provides a storage medium having a computer program stored thereon. When the computer program is executed by a processor, the wind farm swept frequency impedance analysis method based on a differential neural network as described in any one of the above method embodiments can be implemented.

[0060] Compared with the prior art, the present invention has the following beneficial effects:

[0061] The embodiment of the present invention provides a method, device, electronic device and storage medium for wind farm sweep impedance analysis based on differential neural network. The method adds each disturbance voltage to be input and the bus voltage at the time to be analyzed to obtain a number of superimposed voltages; transforms each superimposed voltage to generate a number of superimposed DC side voltages; combines these superimposed DC side voltages with the wind speed at the time to be analyzed to form a number of input data; sequentially injects the input data into the wind farm simulation model so that the differential neural network in the model generates the corresponding disturbance wind turbine side current; calculates the impedance value corresponding to each disturbance voltage according to the disturbance wind turbine side current and the corresponding disturbance voltage to be input; and uses these impedance values ​​to complete the sweep impedance analysis of the wind farm.

[0062] The present invention establishes a simulation model of a wind farm through a differential neural network, and fits the dynamic characteristics of the wind farm through the differential neural network, thereby solving the problem that a large amount of hardware parameters and operating data are required when performing swept frequency impedance analysis using a wind farm simulation model in the prior art, and improving the efficiency of swept frequency impedance analysis of the wind farm. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 It is a flow chart of a wind farm swept frequency impedance analysis method based on differential neural network provided in one embodiment of the present invention.

[0064] Figure 2 It is a structural schematic diagram of a wind farm swept frequency impedance analysis device based on differential neural network provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0065] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0066] like Figure 1 As shown, an embodiment of the present invention provides a wind farm swept frequency impedance analysis method based on a differential neural network, which at least includes the following steps:

[0067] Step S1, obtaining a wind farm simulation model, a wind speed of the wind farm at a time to be analyzed, a bus voltage of the wind farm at a time to be analyzed, and a plurality of disturbance voltages to be input.

[0068] Specifically, the wind farm simulation model is built based on a differential neural network; the blank python component template provided by CloudPSS SimStudio can be used to complete the construction of the swept frequency component containing the differential neural network. In this process, the differential neural network is integrated into the swept frequency component, and its powerful nonlinear modeling capabilities are used to quickly simulate the behavior of complex systems. After the construction of the swept frequency component is completed, it can be added to the wind farm simulation model to enable the model to handle multi-frequency disturbance responses. Through the above steps, the construction and data preparation of the wind farm simulation model can be completed efficiently, laying a solid foundation for the subsequent swept frequency impedance analysis.

[0069] Step S2, adding each disturbance voltage to be input to the bus voltage of the wind farm at the time to be analyzed to obtain a number of superimposed voltages; transforming each superimposed voltage to obtain a number of superimposed DC side voltages; combining each superimposed DC side voltage with the wind speed of the wind farm at the time to be analyzed to generate a number of groups of input data.

[0070] Specifically, each disturbance voltage to be input is added to the bus voltage of the wind farm at the time to be analyzed to generate several superimposed voltages. These superimposed voltages are used to simulate the operating state of the wind farm under different disturbance conditions, covering a variety of possible dynamic behaviors. Next, each superimposed voltage is π Transform to generate the corresponding superimposed DC side voltage. π Transformation is an effective voltage transformation method that can map the voltage parameters on the AC side to the DC side, thereby more directly reflecting the dynamic characteristics inside the wind farm. This step greatly improves data processing efficiency and calculation stability by converting complex AC voltage signals into a relatively simplified DC side expression.

[0071] In completion πAfter the transformation, each superimposed DC side voltage is combined with the wind speed of the wind farm at the time to be analyzed to generate several sets of input data. These input data contain both electrical characteristics (voltage) and environmental characteristics (wind speed), which can fully characterize the operating status of the wind farm. This combination method enables subsequent simulation analysis to more accurately capture the nonlinear dynamic response of the wind farm and provide high-quality input information for impedance analysis.

[0072] Step S3: input each group of input data into the wind farm simulation model in sequence, so that the differential neural network in the wind farm simulation model generates the corresponding disturbed wind turbine side current according to the current input data.

[0073] It should be noted here that each set of input data is input into the wind farm simulation model in turn, so that the simulation model can perform dynamic response calculations according to the input conditions. In the wind farm simulation model, the differential neural network is used as the core calculation module, making full use of its powerful nonlinear mapping ability and dynamic characteristic modeling ability to generate the corresponding disturbance wind turbine side current according to the current input data.

[0074] Specifically, the superimposed DC side voltage and wind speed information contained in the input data jointly determine the operating state of the wind farm under specific disturbance conditions. The differential neural network can efficiently generate the disturbed wind turbine side current that reflects the response characteristics of the wind farm by deeply analyzing these input features and combining its embedded multi-layer dynamic mapping structure. This calculation process can fully capture the nonlinear dynamic behavior inside the wind farm, including the electrical and mechanical coupling characteristics and the stability changes after the disturbance.

[0075] In a preferred embodiment, the training of the differential neural network includes:

[0076] Obtain a conventional operation training set and a high-frequency disturbance training set for model training; wherein each sample in the conventional operation training set includes a first DC side current label corresponding to the DC side voltage without superimposed disturbance, wind speed, and the DC side voltage without superimposed disturbance; each sample in the high-frequency disturbance training set includes a second DC side current label corresponding to the DC side voltage with superimposed high-frequency disturbance, wind speed, and the DC side voltage with superimposed high-frequency disturbance; the DC side voltage without superimposed disturbance is the DC side voltage corresponding to the normal operation of the wind farm under the conventional bus voltage condition; the DC side voltage with superimposed high-frequency disturbance is the DC side voltage corresponding to the operation of the wind farm after applying high-frequency disturbance under the conventional bus voltage condition;

[0077] Randomly divide the regular operation training set into a plurality of batches of first training samples according to a preset number;

[0078] The first training samples of each batch are sequentially input into the differential neural network, and the differential neural network is trained until a preset number of training times is reached; wherein, when the differential neural network receives each batch of the first training samples, it outputs the current value of the wind power side corresponding to the first training sample; according to the current value of the wind power side and the corresponding first DC side current label, the loss function value is calculated by the loss function; and the differential neural network is updated according to the loss function value by using the optimizer;

[0079] Randomly divide the high-frequency perturbation training set into a plurality of batches of second training samples according to a preset number;

[0080] The second training samples of each batch are sequentially input into the differential neural network, and the differential neural network is trained until a preset number of training times is reached; wherein, when the differential neural network receives each batch of second training samples, it outputs the current value of the wind power side corresponding to the second training sample; according to the current value of the wind power side and the corresponding second DC side current label, the loss function value is calculated by the loss function; and the differential neural network is updated according to the loss function value by using the optimizer.

[0081] In a preferred embodiment, the routine operation training set is generated by:

[0082] Obtain several combinations of DC side voltage and wind speed without superimposed disturbance as a set of typical data;

[0083] It should be noted here that the DC side voltage without superimposed disturbance passes through the bus voltage without superimposed disturbance. π Transformation is obtained; the bus voltage without superimposed disturbances is generated by random numbers, but the generation range is strictly limited to the bus voltage range that can ensure the long-term stable operation of the wind farm. Specifically, the setting of this range needs to comprehensively consider the steady-state characteristics and dynamic characteristics of the wind farm operation to ensure that the generated bus voltage value will neither exceed the rated operating capacity of the wind farm equipment nor meet the requirements of the power system for the stability of the wind farm. Optionally, since the wind turbine does not operate under constant working conditions when the actual wind farm is connected to the grid, but the grid-connected voltage will fluctuate slightly during operation, according to my country's national standards, the allowable fluctuation is ±6% of the line voltage. Therefore, the bus voltage without superimposed disturbances can also fluctuate by ±6% on the basis of the above-mentioned bus voltage range that can ensure the long-term stable operation of the wind farm.

[0084] In actual implementation, an upper and lower voltage range can be set according to the wind farm design specifications and operating parameters. The bus voltage value generated by the random number will be strictly limited to this range to avoid abnormal operation of the internal equipment of the wind farm or instability of the grid interaction characteristics due to excessively high or low voltage. This method not only retains the flexibility of random generation, but also ensures the reality and rationality of the simulation conditions, providing a reliable basis for the input preparation of the subsequent simulation model.

[0085] In addition, wind farms generally operate at a wind speed of 8 m / s. In order to better simulate the range of wind speed changes in actual wind farms, the wind speed range is taken as 5.5-7.9 m / s as low wind speed, 8-10.7 m / s as medium wind speed, and 10.8-13.8 m / s as high wind speed. Random numbers are also used to generate the wind speed in each gear.

[0086] Input each set of typical data into a preset wind farm physical model for simulation, and use the DC side current obtained by simulating the wind farm physical model as the first DC side label of the current typical data;

[0087] Combining each group of typical data with the corresponding first DC side label to generate a number of first samples;

[0088] Generate a regular operation training set according to the first sample;

[0089] The wind farm physical model is generated in the following way:

[0090] Obtain the equipment parameters of the wind farm, the operating environment data of the wind farm, and the operation control strategy of the wind farm, and build a physical model of the wind farm based on CloudPSS SimStudio software.

[0091] Optionally, obtain the equipment parameters, operating environment data and operation control strategy of the wind farm, and build a physical model of the wind farm based on CloudPSS SimStudio software. Among them, the equipment parameters of the wind farm include the type of wind turbine, power level, rated voltage, converter configuration, step-up transformer parameters, etc., which are used to accurately describe the hardware characteristics and electrical topology of the wind farm; the operating environment data include wind speed distribution, temperature, air pressure, etc., which reflect the external conditions of the area where the wind farm is located and are important factors affecting the power generation performance and dynamic characteristics of the wind farm; the operation control strategy covers the maximum power point tracking (MPPT) of the generator, reactive power compensation, power factor adjustment and other control mechanisms, which directly determine the dynamic response of the wind farm and the interactive characteristics with the power grid.

[0092] In the CloudPSS SimStudio software, by importing the above information and combining it with the modeling tools provided by the software, the construction of the physical model of the wind farm can be quickly completed. Specifically, users can use the software's built-in component library and simulation modules to assemble the models of wind turbines, transformers, lines and control systems into a complete wind farm simulation system. In addition, the programmable components provided by the software (such as Python component templates) can further enhance the flexibility of the model. Users can write scripts to customize modeling and parameter adjustments for special operating conditions or design requirements of wind farms. The constructed physical model can be used as the basis for wind farm dynamic characteristics analysis, swept frequency impedance testing and optimization design.

[0093] In a preferred embodiment, the high-frequency perturbation training set is generated in the following manner:

[0094] Obtain several combinations of DC side voltage and wind speed with superimposed high-frequency disturbances as a set of typical data;

[0095] It should be explained here that the DC side voltage with superimposed high-frequency disturbance is added with disturbance voltage based on the above-mentioned bus voltage range that can ensure the long-term stable operation of the wind farm. In the present invention, a fixed frequency is selected each time the disturbance voltage is injected, and the frequency can be 100Hz, 600Hz, 1100Hz, 1600Hz or 2100Hz, and the amplitude of the disturbance voltage is controlled between 1% and 5% of the line voltage. In this way, the frequency characteristics of each disturbance signal are ensured to be clear, and at the same time, the stable operation of the wind farm will not be damaged. In the batch simulation process, by randomly selecting the amplitude and frequency of the disturbance voltage, the diversity and reliability of the simulation results are further enhanced, providing comprehensive data support for the swept frequency impedance analysis of the wind farm. In addition, the way of generating the wind speed here is similar to the wind speed of the above-mentioned conventional operation training set, and will not be repeated.

[0096] Input each set of typical data into a preset wind farm physical model, and use the DC side current of the wind farm physical model as the second DC side label of the current typical data;

[0097] Combining each group of typical data with the corresponding second DC side label to generate a number of second samples;

[0098] A high-frequency perturbation training set is generated according to the second sample.

[0099] In an optional embodiment, before sequentially inputting each group of input data into the wind farm simulation model so that the differential neural network in the wind farm simulation model generates the corresponding disturbed wind turbine side current according to the current input data, the method further includes:

[0100] Obtain the bus current of the wind farm at the time to be analyzed;

[0101] The bus current of the wind farm at the time to be analyzed and the bus voltage of the wind farm at the time to be analyzed are used as the initial simulation environment of the wind farm simulation model;

[0102] Transform the bus voltage of the wind farm at the time to be analyzed to obtain the DC side initialization voltage;

[0103] Repeat the simulation environment update operation until the absolute value of the difference between the bus initialization current and the bus current of the wind farm at the time to be analyzed does not exceed a preset threshold;

[0104] The simulation environment updating operation includes:

[0105] In the current simulation environment, the DC side initialization voltage and the wind speed of the wind farm at the time to be analyzed are input into the wind farm simulation model, so that the differential neural network in the wind farm simulation model generates the current DC side current according to the initialization voltage and the wind speed of the wind farm at the time to be analyzed; wherein the initial simulation environment is the initial simulation environment;

[0106] Perform inverse transformation on the current DC side current to obtain the current bus current;

[0107] When the absolute difference between the current bus current and the bus current of the wind farm at the time to be analyzed is greater than a preset threshold, the current bus current and the bus voltage of the wind farm at the time to be analyzed are used as the updated initial simulation environment.

[0108] Step S4: Calculate and generate an impedance value corresponding to each disturbance voltage to be input according to each disturbance fan-side current and the corresponding disturbance voltage to be input.

[0109] In a preferred embodiment, the step of calculating and generating an impedance value corresponding to each disturbance voltage to be input according to each disturbance wind turbine side current and the corresponding disturbance voltage to be input includes:

[0110] The quotient between the current disturbance voltage and the wind turbine side current after the disturbance is calculated to generate an impedance value corresponding to the current disturbance voltage.

[0111] Specifically, the quotient between the current disturbance voltage and the wind turbine side current after the disturbance is calculated to generate the impedance value corresponding to the current disturbance voltage. It is particularly important to emphasize that if the disturbance voltage and the wind turbine side current after the disturbance are both expressed in complex form, then the calculation of the impedance value is the division operation of the complex voltage and the complex current. In the complex division, the real and imaginary parts of the impedance value reflect the active and reactive impedance characteristics of the wind farm respectively, so as to more accurately describe the dynamic response characteristics of the wind farm under different frequency disturbances. This complex operation method ensures the comprehensive capture of the impedance characteristics of the wind farm and provides accurate basic data support for the subsequent swept frequency impedance analysis.

[0112] Step S5: performing a wind farm swept frequency impedance analysis according to the impedance values ​​corresponding to the disturbance voltages to be input.

[0113] It should be noted here that the swept frequency impedance analysis of the wind farm is completed according to the impedance value corresponding to each disturbance voltage to be input. Specifically, by calculating and summarizing the disturbance voltage input at different frequencies and its corresponding impedance value, a frequency-impedance characteristic curve of the wind farm is constructed. This characteristic curve can intuitively reflect the dynamic response behavior of the wind farm under different frequency disturbances, and reveal its key characteristics of stability and interaction with the power grid. In practical applications, this analysis method can be used to evaluate the resonance risk of wind farms within a specific frequency range, and provide a reliable basis for the design optimization of wind farms, the formulation of operation control strategies, and the evaluation of power grid interconnection schemes. This technical means based on swept frequency impedance analysis not only improves the accuracy and efficiency of the analysis, but also significantly reduces the complexity and time cost of traditional analysis methods.

[0114] Based on the above method embodiment, the present invention provides a corresponding device embodiment.

[0115] like Figure 2 As shown, an embodiment of the present invention provides a wind farm sweep frequency impedance analysis device based on differential neural network, comprising: a data acquisition module, an input data generation module, a disturbance wind turbine side current generation module, an impedance value calculation module and a sweep frequency impedance analysis module;

[0116] The data acquisition module is used to acquire a wind farm simulation model, a wind speed at a time to be analyzed in the wind farm, a bus voltage at a time to be analyzed in the wind farm, and a number of disturbance voltages to be input; wherein the wind farm simulation model is constructed based on a differential neural network;

[0117] The input data generation module is used to add each disturbance voltage to be input to the bus voltage of the wind farm at the time to be analyzed to obtain a number of superimposed voltages; transform each superimposed voltage to obtain a number of superimposed DC side voltages; combine each superimposed DC side voltage with the wind speed of the wind farm at the time to be analyzed to generate a number of groups of input data;

[0118] The disturbance wind turbine side current generation module is used to input each group of input data into the wind farm simulation model in sequence, so that the differential neural network in the wind farm simulation model generates the corresponding disturbance wind turbine side current according to the current input data;

[0119] The impedance value calculation module is used to calculate and generate the impedance value corresponding to each disturbance voltage to be input according to each disturbance fan side current and the corresponding disturbance voltage to be input;

[0120] The swept frequency impedance analysis module is used to perform a swept frequency impedance analysis of the wind farm according to the impedance value corresponding to each disturbance voltage to be input.

[0121] In a preferred embodiment, the wind farm swept frequency impedance analysis device based on differential neural network further includes: a simulation model steady-state operation module;

[0122] The simulation model steady-state operation module is used to obtain the bus current of the wind farm at the time to be analyzed;

[0123] The bus current of the wind farm at the time to be analyzed and the bus voltage of the wind farm at the time to be analyzed are used as the initial simulation environment of the wind farm simulation model;

[0124] Transform the bus voltage of the wind farm at the time to be analyzed to obtain the DC side initialization voltage;

[0125] Repeat the simulation environment update operation until the absolute value of the difference between the bus initialization current and the bus current of the wind farm at the time to be analyzed does not exceed a preset threshold;

[0126] The simulation environment updating operation includes:

[0127] In the current simulation environment, the DC side initialization voltage and the wind speed of the wind farm at the time to be analyzed are input into the wind farm simulation model, so that the differential neural network in the wind farm simulation model generates the current DC side current according to the initialization voltage and the wind speed of the wind farm at the time to be analyzed; wherein the initial simulation environment is the initial simulation environment;

[0128] Perform inverse transformation on the current DC side current to obtain the current bus current;

[0129] When the absolute difference between the current bus current and the bus current of the wind farm at the time to be analyzed is greater than a preset threshold, the current bus current and the bus voltage of the wind farm at the time to be analyzed are used as the updated initial simulation environment.

[0130] It should be noted that the embodiment of the device described above corresponds to the above-mentioned embodiment of the present invention, and it can implement any of the above-mentioned wind farm swept frequency impedance analysis methods based on differential neural networks of the present invention. In addition, the embodiment of the above-mentioned device is merely schematic, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the accompanying drawings of the embodiment of the device provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which can be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art can understand and implement it without paying any creative labor.

[0131] Based on the above method embodiment of the present invention, a corresponding electronic device embodiment is provided.

[0132] An embodiment of the present invention provides an electronic device, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the wind farm swept frequency impedance analysis method based on differential neural network described in any one of the present invention is implemented, or, when the processor executes the computer program, the functions of each module in the above-mentioned device embodiments are implemented.

[0133] Exemplarily, the computer program may be divided into one or more modules, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules may be a series of computer program instruction segments capable of implementing specific functions, which are used to describe the execution process of the computer program in the terminal device.

[0134] The terminal device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0135] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the terminal device, and uses various interfaces and lines to connect various parts of the entire terminal device.

[0136] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the terminal device by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required for a function, etc.; the data storage area can store data created according to the use of the mobile phone, etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0137] Based on the above method embodiment, the present invention provides a storage medium embodiment;

[0138] Another embodiment of the present invention provides a storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute any one of the above-mentioned wind farm swept frequency impedance analysis methods based on differential neural networks of the present invention.

[0139] The storage medium is a computer-readable storage medium, and the computer program includes computer program code, which may be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0140] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example" or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine different embodiments or examples described in this specification and the features of different embodiments or examples, unless they are contradictory.

[0141] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A wind farm swept frequency impedance analysis method based on differential neural network, characterized in that: include: Obtaining a wind farm simulation model, the wind speed of the wind farm at the time to be analyzed, the bus voltage of the wind farm at the time to be analyzed, and several disturbance voltages to be input; wherein the wind farm simulation model is constructed based on a differential neural network; Add each disturbance voltage to be input to the bus voltage of the wind farm at the time to be analyzed to obtain a number of superimposed voltages; transform each superimposed voltage to obtain a number of superimposed DC side voltages; combine each superimposed DC side voltage with the wind speed of the wind farm at the time to be analyzed to generate a number of groups of input data; Inputting each group of input data into the wind farm simulation model in turn, so that the differential neural network in the wind farm simulation model generates a corresponding disturbed wind turbine side current according to the current input data; According to each disturbed wind turbine side current and the corresponding disturbed voltage to be input, an impedance value corresponding to each disturbed voltage to be input is calculated and generated; According to the impedance value corresponding to each disturbance voltage to be input, the wind farm swept frequency impedance analysis is performed.

2. The wind farm swept frequency impedance analysis method based on differential neural network according to claim 1, characterized in that: The training of the differential neural network includes: Obtain a conventional operation training set and a high-frequency disturbance training set for model training; wherein each sample in the conventional operation training set includes a first DC side current label corresponding to the DC side voltage without superimposed disturbance, wind speed, and the DC side voltage without superimposed disturbance; each sample in the high-frequency disturbance training set includes a second DC side current label corresponding to the DC side voltage with superimposed high-frequency disturbance, wind speed, and the DC side voltage with superimposed high-frequency disturbance; the DC side voltage without superimposed disturbance is the DC side voltage corresponding to the normal operation of the wind farm under the conventional bus voltage condition; the DC side voltage with superimposed high-frequency disturbance is the DC side voltage corresponding to the operation of the wind farm after applying high-frequency disturbance under the conventional bus voltage condition; Randomly divide the regular operation training set into a plurality of batches of first training samples according to a preset number; The first training samples of each batch are sequentially input into the differential neural network, and the differential neural network is trained until a preset number of training times is reached; wherein, when the differential neural network receives each batch of the first training samples, it outputs the current value of the wind power side corresponding to the first training sample; according to the current value of the wind power side and the corresponding first DC side current label, the loss function value is calculated by the loss function; and the differential neural network is updated according to the loss function value by using the optimizer; Randomly divide the high-frequency perturbation training set into a plurality of batches of second training samples according to a preset number; The second training samples of each batch are sequentially input into the differential neural network, and the differential neural network is trained until a preset number of training times is reached; wherein, when the differential neural network receives each batch of second training samples, it outputs the current value of the wind power side corresponding to the second training sample; according to the current value of the wind power side and the corresponding second DC side current label, the loss function value is calculated by the loss function; and the differential neural network is updated according to the loss function value by using the optimizer.

3. The wind farm swept frequency impedance analysis method based on differential neural network according to claim 2, characterized in that: Before sequentially inputting each group of input data into the wind farm simulation model so that the differential neural network in the wind farm simulation model generates the corresponding disturbed wind turbine side current according to the current input data, the method further includes: Obtain the bus current of the wind farm at the time to be analyzed; The bus current of the wind farm at the time to be analyzed and the bus voltage of the wind farm at the time to be analyzed are used as the initial simulation environment of the wind farm simulation model; Transform the bus voltage of the wind farm at the time to be analyzed to obtain the DC side initialization voltage; Repeat the simulation environment update operation until the absolute value of the difference between the bus initialization current and the bus current of the wind farm at the time to be analyzed does not exceed a preset threshold; The simulation environment updating operation includes: In the current simulation environment, the DC side initialization voltage and the wind speed of the wind farm at the time to be analyzed are input into the wind farm simulation model, so that the differential neural network in the wind farm simulation model generates the current DC side current according to the initialization voltage and the wind speed of the wind farm at the time to be analyzed; wherein the initial simulation environment is the initial simulation environment; Perform inverse transformation on the current DC side current to obtain the current bus current; When the absolute difference between the current bus current and the bus current of the wind farm at the time to be analyzed is greater than a preset threshold, the current bus current and the bus voltage of the wind farm at the time to be analyzed are used as the updated initial simulation environment.

4. The wind farm swept frequency impedance analysis method based on differential neural network according to claim 3, characterized in that: The step of calculating and generating an impedance value corresponding to each disturbance voltage to be input according to each disturbance fan side current and the corresponding disturbance voltage to be input includes: The quotient between the current disturbance voltage and the wind turbine side current after the disturbance is calculated to generate an impedance value corresponding to the current disturbance voltage.

5. The wind farm swept frequency impedance analysis method based on differential neural network according to claim 4, characterized in that: Generate a regular run training set by: Obtain several combinations of DC side voltage and wind speed without superimposed disturbance as a set of typical data; Input each set of typical data into a preset wind farm physical model for simulation, and use the DC side current obtained by simulating the wind farm physical model as the first DC side label of the current typical data; Combining each group of typical data with the corresponding first DC side label to generate a number of first samples; Generate a regular operation training set according to the first sample; The wind farm physical model is generated in the following way: Obtain the equipment parameters of the wind farm, the operating environment data of the wind farm, and the operation control strategy of the wind farm, and build a physical model of the wind farm based on CloudPSS SimStudio software.

6. The wind farm swept frequency impedance analysis method based on differential neural network according to claim 5, characterized in that: The high-frequency perturbation training set is generated by: Obtain several combinations of DC side voltage and wind speed with superimposed high-frequency disturbances as a set of typical data; Input each set of typical data into a preset wind farm physical model, and use the DC side current of the wind farm physical model as the second DC side label of the current typical data; Combining each group of typical data with the corresponding second DC side label to generate a number of second samples; A high-frequency perturbation training set is generated according to the second sample.

7. A wind farm sweep frequency impedance analysis device based on differential neural network, characterized in that: include: Data acquisition module, input data generation module, disturbance fan side current generation module, impedance value calculation module and swept frequency impedance analysis module; The data acquisition module is used to acquire a wind farm simulation model, a wind speed at a time to be analyzed in the wind farm, a bus voltage at a time to be analyzed in the wind farm, and a number of disturbance voltages to be input; wherein the wind farm simulation model is constructed based on a differential neural network; The input data generation module is used to add each disturbance voltage to be input to the bus voltage of the wind farm at the time to be analyzed to obtain a number of superimposed voltages; transform each superimposed voltage to obtain a number of superimposed DC side voltages; combine each superimposed DC side voltage with the wind speed of the wind farm at the time to be analyzed to generate a number of groups of input data; The disturbance wind turbine side current generation module is used to input each group of input data into the wind farm simulation model in sequence, so that the differential neural network in the wind farm simulation model generates the corresponding disturbance wind turbine side current according to the current input data; The impedance value calculation module is used to calculate and generate the impedance value corresponding to each disturbance voltage to be input according to each disturbance fan side current and the corresponding disturbance voltage to be input; The swept frequency impedance analysis module is used to perform a swept frequency impedance analysis of the wind farm according to the impedance value corresponding to each disturbance voltage to be input.

8. The wind farm swept frequency impedance analysis device based on differential neural network according to claim 7, characterized in that: Also includes: Simulation model steady-state operation module; The simulation model steady-state operation module is used to obtain the bus current of the wind farm at the time to be analyzed; The bus current of the wind farm at the time to be analyzed and the bus voltage of the wind farm at the time to be analyzed are used as the initial simulation environment of the wind farm simulation model; Transform the bus voltage of the wind farm at the time to be analyzed to obtain the DC side initialization voltage; Repeat the simulation environment update operation until the absolute value of the difference between the bus initialization current and the bus current of the wind farm at the time to be analyzed does not exceed a preset threshold; The simulation environment updating operation includes: In the current simulation environment, the DC side initialization voltage and the wind speed of the wind farm at the time to be analyzed are input into the wind farm simulation model, so that the differential neural network in the wind farm simulation model generates the current DC side current according to the initialization voltage and the wind speed of the wind farm at the time to be analyzed; wherein the initial simulation environment is the initial simulation environment; Perform inverse transformation on the current DC side current to obtain the current bus current; When the absolute difference between the current bus current and the bus current of the wind farm at the time to be analyzed is greater than a preset threshold, the current bus current and the bus voltage of the wind farm at the time to be analyzed are used as the updated initial simulation environment.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the wind farm swept frequency impedance analysis method based on differential neural network described in any one of claims 1 to 6 can be implemented.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it can implement the wind farm swept frequency impedance analysis method based on differential neural network as described in any one of claims 1 to 6.

Images