Method and system for correcting a speed regulation model of a doubly-fed electric machine system

Abstract

The application discloses a double-fed motor system speed regulation model correction method and system, and belongs to the technical field of power equipment. The application obtains multi-source operation data of a double-fed motor, wherein the multi-source operation data comprises electromagnetic dimension data, hydraulic dimension data and environmental dimension data; determines an excitation inductance prediction value according to the multi-source operation data and a preset deep learning model; determines an excitation inductance correction value according to the excitation inductance prediction value and a preset check mechanism; and updates speed regulation model parameters of the double-fed motor according to the excitation inductance correction value, thereby achieving the beneficial effect of improving the stability of power generation equipment.

Description

Technical Field

[0001] This invention relates to the field of power equipment technology, and in particular to a method and system for correcting the speed regulation model of a doubly fed motor system. Background Technology

[0002] With the development of technology, doubly-fed motors have been used in a variety of different scenarios. Currently, doubly-fed motors are also gradually being used in variable-speed pumped storage power stations for variable-speed pumped storage, power generation, and phase regulation. Common doubly-fed motors generally adopt vector control technology, which specifically equates the complex motor to a DC motor, making active and reactive power outputs independent. However, in the scenario of variable-speed pumped storage, the extreme operating conditions of the unit lead to the saturation of the iron core inside the motor, resulting in strong coupling between active and reactive power control, which in turn leads to the failure of vector control and affects the stability of the power generation equipment. Summary of the Invention

[0003] The present invention aims to solve at least one of the problems existing in the prior art, and provides a method and system for correcting the speed regulation model of a doubly fed motor system.

[0004] One aspect of the present invention provides a method for correcting the speed regulation model of a doubly-fed induction generator system, comprising: Acquire multi-source operating data of the doubly fed motor, including electromagnetic dimension data, hydraulic dimension data, and environmental dimension data; The predicted value of the excitation inductance is determined based on the multi-source operating data and the preset deep learning model. The magnetizing inductance correction value is determined based on the predicted value of the magnetizing inductance and the preset verification mechanism. The speed control model parameters of the doubly fed motor are updated based on the excitation inductance correction value.

[0005] Optionally, determining the predicted value of the magnetizing inductance based on the multi-source operating data and the preset deep learning model includes: The multi-source operational data is fused to obtain a fused feature vector; The fused feature vector is input into a preset deep learning model to obtain the initial magnetizing inductance prediction value; The predicted excitation inductance is obtained by performing time-series smoothing based on the initial predicted excitation inductance value and the excitation inductance value at the previous moment.

[0006] Optionally, before inputting the fused feature vector into a preset deep learning model to obtain the initial magnetizing inductance prediction value, the method further includes: Obtain a historical multi-source operational dataset, which includes multiple sets of historical electromagnetic dimension data, historical hydraulic dimension data, and historical environmental dimension data; A training sample set is constructed based on the historical multi-source operation dataset and the corresponding historical excitation inductance calibration values; The initial long short-term memory network model is trained based on the training sample set to obtain the preset deep learning model.

[0007] Optionally, training the initial long short-term memory network model based on the training sample set to obtain the preset deep learning model includes: Construct input time series features based on the historical multi-source runtime dataset; Construct an output tag based on the historical excitation inductance calibration value; The initial long short-term memory network model is trained based on the input time series features and the output labels to obtain the preset deep learning model.

[0008] Optionally, determining the magnetizing inductance correction value based on the predicted magnetizing inductance value and a preset verification mechanism includes: The theoretical stator voltage is calculated based on the predicted value of the excitation inductance and the motor steady-state equation. Calculate the voltage residual based on the theoretical stator voltage and the measured stator voltage; The physical verification result is determined based on the voltage residual and the preset threshold.

[0009] Optionally, determining the magnetizing inductance correction value based on the predicted magnetizing inductance value and the preset verification mechanism further includes: The observed magnetizing inductance value is calculated based on the magnetic flux observer and the multi-source operating data. The observation deviation is calculated based on the observed excitation inductance value and the predicted excitation inductance value. The verification results of the mechanism model are determined based on the observed deviations.

[0010] Optionally, acquiring the multi-source operating data of the doubly-fed motor includes: The stator three-phase voltage, stator three-phase current, rotor three-phase voltage, rotor three-phase current, active power, and reactive power are collected as the electromagnetic dimension data. The water head, guide vane opening, and flow rate of the upper and lower reservoirs are collected as the hydraulic dimension data. The winding temperature, ambient humidity, and running time are collected as the environmental dimension data. The multi-source operational data is determined based on the electromagnetic dimension data, the hydraulic dimension data, and the environmental dimension data.

[0011] Another aspect of the present invention provides a speed regulation model correction system for a doubly fed motor system, comprising: The acquisition module is used to acquire multi-source operating data of the doubly fed motor, including electromagnetic dimension data, hydraulic dimension data and environmental dimension data. The prediction module is used to determine the predicted value of the excitation inductance based on the multi-source operating data and the preset deep learning model. The verification module is used to determine the correction value of the magnetizing inductance based on the predicted value of the magnetizing inductance and the preset verification mechanism; The correction module is used to update the speed regulation model parameters of the doubly fed motor according to the correction value of the excitation inductance.

[0012] In another aspect, the present invention provides a doubly-fed motor system speed regulation model correction device, the doubly-fed motor system speed regulation model correction device comprising: a memory, a processor, and a doubly-fed motor system speed regulation model correction program stored in the memory and executable on the processor, the doubly-fed motor system speed regulation model correction program being configured to implement the steps of the doubly-fed motor system speed regulation model correction method described above.

[0013] In another aspect, the present invention provides a storage medium storing a speed control model correction program for a doubly fed motor system, wherein when the doubly fed motor system speed control model correction program is executed by a processor, the program implements the steps of the speed control model correction method for a doubly fed motor system as described above.

[0014] The speed control model correction method and system for a doubly fed motor system of the present invention obtains multi-source operating data of the doubly fed motor, determines the predicted value of the excitation inductance based on the multi-source operating data and a preset deep learning model, determines the corrected value of the excitation inductance based on the predicted value of the excitation inductance and a preset verification mechanism, and updates the speed control model parameters of the doubly fed motor based on the corrected value of the excitation inductance, thereby improving the stability of the doubly fed motor operation. Attached Figure Description

[0015] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.

[0016] Figure 1 This is a schematic diagram of the structure of the speed regulation model correction device for the doubly fed motor system in the hardware operating environment involved in the embodiments of the present invention; Figure 2 This is a flowchart illustrating the first embodiment of the speed regulation model correction method for a doubly fed motor system according to the present invention. Figure 3 This is a flowchart illustrating the second embodiment of the speed regulation model correction method for a doubly fed motor system according to the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the various embodiments of the present invention will be described in detail below with reference to the accompanying drawings. However, those skilled in the art will understand that many technical details are presented in the embodiments of the present invention to facilitate a better understanding of the invention. However, the technical solutions claimed in the present invention can be implemented even without these technical details and various variations and modifications based on the following embodiments. The division of the following embodiments is for ease of description and should not constitute any limitation on the specific implementation of the present invention. The various embodiments can be combined with and referenced by each other without contradiction.

[0018] Reference Figure 1 , Figure 1 This is a schematic diagram of the structure of the speed regulation model correction device for the doubly fed motor system in the hardware operating environment involved in the embodiments of the present invention.

[0019] like Figure 1 As shown, the speed regulation model correction device for the doubly-fed motor system may include: a processor 1001, a communication bus 1002, an interactive device 1003, a network interface 1004, and a memory 1005. The processor 1001 may be a central processing unit (CPU). The communication bus 1002 is used to enable communication between these components. The interactive device 1003 may include a display screen or an input unit such as a keyboard. Optionally, the interactive device 1003 may also be connected to the communication bus via standard wired or wireless interfaces. The network interface 1004 may optionally include standard wired or wireless interfaces (such as a Wi-Fi interface). The memory 1005 may be a high-speed random access memory (RAM) or a stable non-volatile memory (NVM), such as a disk drive. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0020] Those skilled in the art will understand that Figure 1 The structure shown does not constitute a limitation on the speed control model correction device for a doubly fed motor system, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0021] The memory 1005, which serves as a storage medium, may include an operating system, a data storage module, a network communication module, a user interface module, and a speed control model correction program for a doubly-fed motor system.

[0022] exist Figure 1 In the doubly fed motor system speed regulation model correction device shown, the network interface 1004 is mainly used for data communication with other devices; the interactive device 1003 is mainly used for data interaction with the user; the processor 1001 and memory 1005 in the doubly fed motor system speed regulation model correction device of the present invention can be set in the doubly fed motor system speed regulation model correction device. The doubly fed motor system speed regulation model correction device calls the doubly fed motor system speed regulation model correction program stored in the memory 1005 through the processor 1001 and executes the doubly fed motor system speed regulation model correction method provided in the embodiment of the present invention.

[0023] This invention provides a method for correcting the speed regulation model of a doubly-fed motor system, referring to... Figure 2 , Figure 2 This is a flowchart illustrating the first embodiment of a speed regulation model correction method for a doubly fed motor system according to the present invention.

[0024] In this embodiment, the speed regulation model correction method for the doubly fed motor system includes steps S1 to S4.

[0025] Step S1: Obtain multi-source operating data of the doubly fed motor, including electromagnetic dimension data, hydraulic dimension data and environmental dimension data.

[0026] In this embodiment, hydraulic and environmental dimension data are collected by sensors, and the electromagnetic dimension data is obtained by deriving the control data of the doubly-fed motor control system. It should be noted that the electromagnetic dimension data here can refer to electromagnetically related data, such as current data and voltage data. Optionally, electromagnetic field data is generally obtained through calculation. This is because directly measuring the magnetic field requires significant cost.

[0027] Step S2: Determine the predicted value of the excitation inductance based on the multi-source operating data and the preset deep learning model.

[0028] It should be noted that the predicted excitation inductance value here refers to the value representing the magnetization capability of the motor's main magnetic circuit at the current moment, calculated in real time using a deep learning model and multi-source data. In the equivalent circuit model of a doubly-fed induction generator (DFIG), the excitation inductance is located on the coupling branch of the stator and rotor. The predicted excitation inductance value is used as a variable to characterize the core saturation level in real time. The deeper the core saturation, the smaller the predicted excitation inductance value. This is a negatively correlated and non-linear relationship. Optionally, the multi-source operating data is normalized, and noise or other abnormal data is removed through data cleaning. Generally, a one-dimensional vector is obtained, which is then input into the preset deep learning model, and the data output by the preset deep learning model is used as the predicted excitation inductance value.

[0029] Step S3: Determine the magnetizing inductance correction value based on the predicted value of the magnetizing inductance and the preset verification mechanism.

[0030] In this embodiment, the obtained predicted value of the excitation inductance is input into a preset verification mechanism. This mechanism integrates physical constraint rules and statistical confidence intervals to evaluate the reliability of the predicted value and calculates the confidence interval. It is mainly used to determine whether the predicted value falls within a reasonable range determined based on the theory of motor magnetization curve and to eliminate abnormal predictions caused by sensor failure or data drift.

[0031] Step S4: Update the speed regulation model parameters of the doubly fed motor according to the excitation inductance correction value.

[0032] Specifically, the verified excitation inductance correction value is injected in real time into the flux linkage observer and current decoupling controller of the doubly-fed induction generator speed control model, replacing or integrating the original offline calibration parameters. Optionally, the observation gain of rotor and stator flux linkage is corrected by updating the mutual inductance coefficient in the flux linkage calculation equation, thereby adjusting the decoupling compensation term and the feedback gain of the speed estimator in the vector control, so that the control model always matches the actual magnetic circuit saturation state of the motor.

[0033] In this embodiment, by acquiring multi-source operating data of the doubly-fed motor, determining the predicted value of the excitation inductance based on the multi-source operating data and a preset deep learning model, determining the corrected value of the excitation inductance based on the predicted value of the excitation inductance and a preset verification mechanism, and updating the speed regulation model parameters of the doubly-fed motor based on the corrected value of the excitation inductance, the stability of the doubly-fed motor operation process can be improved.

[0034] Furthermore, based on the first embodiment, a second embodiment of the speed regulation model correction method for doubly-fed motor systems of the present invention is proposed. In this embodiment, referring to... Figure 3 The step of determining the predicted value of the excitation inductance based on the multi-source operating data and the preset deep learning model includes steps S21 to S23.

[0035] Step S21: Perform feature fusion on the multi-source running data to obtain a fused feature vector.

[0036] Specifically, multimodal feature fusion is performed on the electromagnetic, hydraulic, and environmental dimension data obtained in step S1. First, the original data of each dimension are normalized and denoised to eliminate dimensional differences and sensor noise. Then, an attention weighting mechanism is used to map different types of data to a unified latent space to construct a fused feature vector.

[0037] Step S22: Input the fused feature vector into a preset deep learning model to obtain the initial magnetizing inductance prediction value.

[0038] Specifically, the fused feature vector generated in step S21 is input into a preset deep learning model; the model performs forward inference calculations and outputs the predicted value of the initial excitation inductance at the current moment, which reflects the instantaneous magnetization capability of the motor's main magnetic circuit under the current multi-source operating conditions. Optionally, the preset deep learning model here adopts a neural network architecture with a long short-term memory network or an attention mechanism.

[0039] Step S23: Perform time-series smoothing processing based on the initial magnetizing inductance prediction value and the previous magnetizing inductance value to obtain the magnetizing inductance prediction value.

[0040] In this embodiment, specifically, the predicted excitation inductance value is calculated by averaging the initial predicted excitation inductance value and the excitation inductance value at the previous moment. In other embodiments, the predicted excitation inductance value can be calculated using a weighted average. Optionally, Kalman filtering can be used for timing smoothing. Kalman filtering is a common algorithm and will not be discussed in detail here.

[0041] In this embodiment, by fusing the features of the multi-source operating data to obtain a fused feature vector, the fused feature vector is input into a preset deep learning model to obtain an initial magnetizing inductance prediction value. Then, based on the initial magnetizing inductance prediction value and the magnetizing inductance value at the previous moment, a time-series smoothing process is performed to obtain the magnetizing inductance prediction value, thereby improving the accuracy of the magnetizing inductance prediction value.

[0042] Furthermore, based on the first and second embodiments, a third embodiment of the speed regulation model correction method for doubly-fed motor systems of the present invention is proposed. In this embodiment, before inputting the fused feature vector into a preset deep learning model to obtain the initial excitation inductance prediction value, the method further includes: Obtain a historical multi-source operational dataset, which includes multiple sets of historical electromagnetic dimension data, historical hydraulic dimension data, and historical environmental dimension data; A training sample set is constructed based on the historical multi-source operation dataset and the corresponding historical excitation inductance calibration values; The initial long short-term memory network model is trained based on the training sample set to obtain the preset deep learning model.

[0043] In this embodiment, historical data covering different seasons, loads, and operating conditions are collected from the long-term operation archives of the doubly-fed induction generator (DFIG). This includes electromagnetic dimensions such as stator and rotor voltage and current, hydraulic dimensions such as cooling medium flow rate and pressure, and environmental dimensions such as temperature and humidity. These multi-source data are continuously recorded by a sensor network and pre-processed for noise reduction to form a historical database containing rich variations in operating conditions. Since historical excitation inductance is not directly detectable, it is generally obtained through offline simulation or during experiments. Each sample contains multi-dimensional operating states within a specific time and historical window, along with their corresponding standard excitation inductance values, forming a structured training sample set. The historical multi-source operating dataset is time-aligned and paired with the historical excitation inductance ground truth values ​​obtained through offline experiments or finite element simulation calibration, forming supervised samples with a one-to-one correspondence between input features and output labels. This sample set establishes a quantitative relationship between multi-source physical quantities and the saturation degree of the motor's magnetic circuit, providing a labeled data foundation for the Long Short-Term Memory (LSTM) network to learn time-dependent features.

[0044] Furthermore, the step of training the initial long short-term memory network model based on the training sample set to obtain the preset deep learning model includes: Construct input time series features based on the historical multi-source runtime dataset; Construct an output tag based on the historical excitation inductance calibration value; The initial long short-term memory network model is trained based on the input time series features and the output labels to obtain the preset deep learning model.

[0045] In this embodiment, the historical multi-source operational dataset is segmented using a sliding window, and electromagnetic, hydraulic, and environmental data from the current moment and several previous moments are selected to construct time-series features. Through feature normalization and temporal alignment, a fixed-length multi-dimensional input vector is formed. In this embodiment, the input time-series features are fed into the input, forget, and output gates of a Long Short-Term Memory (LSTM) network. The gating mechanism selectively memorizes long-term dependencies and filters short-term noise, progressively extracting temporal patterns and coupling features from the multi-source data in the hidden layers. Optionally, stochastic gradient descent can be used to update network parameters, and an early stopping strategy can be employed to prevent overfitting until the model reaches a preset accuracy on the validation set. The resulting preset deep learning model possesses accurate temporal prediction capabilities and can directly execute step S22 for inference, achieving real-time estimation of the magnetizing inductance.

[0046] In this embodiment, by constructing input time series features based on the historical multi-source running dataset, constructing output labels based on the historical excitation inductance calibration values, and training the initial long short-term memory network model based on the input time series features and the output labels, the preset deep learning model is obtained, thereby improving the accuracy of the preset deep learning model.

[0047] Furthermore, based on any of the above embodiments, a fourth embodiment of the speed regulation model correction method for doubly-fed motor systems of the present invention is proposed. In this embodiment, determining the correction value of the excitation inductance based on the predicted value of the excitation inductance and the preset verification mechanism includes: The theoretical stator voltage is calculated based on the predicted value of the excitation inductance and the motor steady-state equation. Calculate the voltage residual based on the theoretical stator voltage and the measured stator voltage; The physical verification result is determined based on the voltage residual and the preset threshold.

[0048] In this embodiment, the magnitude of the voltage residual is compared with a preset physical threshold. When the residual is less than the threshold, the physical verification is deemed successful, and the predicted value of the excitation inductance is directly output as the correction value. When the residual exceeds the threshold but is less than the rejection threshold, a weighted correction strategy is triggered, using the residual gradient to fine-tune the predicted value. When the residual exceeds the rejection threshold, the predicted value is deemed physically unreliable, and the smoothed value or observed value from the previous time step is used instead. This mechanism ensures that the final output corrected value of the excitation inductance fully utilizes the ability of deep learning to capture nonlinear saturation while strictly satisfying the physical consistency constraints of the motor steady-state equation, thus avoiding the erroneous guidance of abnormal predictions on the speed regulation model.

[0049] Furthermore, determining the magnetizing inductance correction value based on the predicted magnetizing inductance value and the preset verification mechanism further includes: The observed magnetizing inductance value is calculated based on the magnetic flux observer and the multi-source operating data. The observation deviation is calculated based on the observed excitation inductance value and the predicted excitation inductance value. The verification results of the mechanism model are determined based on the observed deviations.

[0050] In this embodiment, optionally, the mechanism model verification result is determined based on the amplitude and distribution characteristics of the observation deviation: when the deviation is less than the mechanism confidence threshold, the data-driven prediction is considered reliable, and the predicted value is used as the correction value; when the deviation exceeds the threshold, the weight of the predicted value is reduced, the proportion of the observed excitation inductance value in the correction result is increased, or the online model update mechanism is triggered; when the deviation is continuously abnormal, it is determined that the deep learning model may fail, and the model is completely switched to the output of the observer-based mechanism model.

[0051] Furthermore, based on any of the above embodiments, a fifth embodiment of the speed regulation model correction method for a doubly-fed motor system of the present invention is proposed. In this embodiment, obtaining multi-source operating data of the doubly-fed motor includes: The stator three-phase voltage, stator three-phase current, rotor three-phase voltage, rotor three-phase current, active power, and reactive power are collected as the electromagnetic dimension data. The water head, guide vane opening, and flow rate of the upper and lower reservoirs are collected as the hydraulic dimension data. The winding temperature, ambient humidity, and running time are collected as the environmental dimension data. The multi-source operational data is determined based on the electromagnetic dimension data, the hydraulic dimension data, and the environmental dimension data.

[0052] In this embodiment, it should be noted that humidity often has an indirect impact on core saturation, primarily leading to changes in heat dissipation conditions, which may result in nonlinear effects. Optionally, voltage and current transformers installed on the stator and rotor sides are used to synchronously collect stator three-phase voltage, stator three-phase current, rotor three-phase voltage, and rotor three-phase current at a high-frequency sampling rate. Furthermore, in some embodiments, hydraulic dimension data such as water head, guide vane opening, and flow rate through the turbine are acquired from the turbine governor and reservoir monitoring terminal; the turbine output mechanical power is estimated by multiplying the water head and flow rate, and the dynamic variation of the load torque is analyzed in conjunction with the guide vane opening.

[0053] Furthermore, this invention also proposes a speed regulation model correction system for a doubly fed motor system, comprising an acquisition module, a prediction module, a verification module, and a correction module.

[0054] The acquisition module is used to acquire multi-source operating data of the doubly fed motor, including electromagnetic dimension data, hydraulic dimension data and environmental dimension data.

[0055] The prediction module is used to determine the predicted value of the excitation inductance based on the multi-source operating data and the preset deep learning model.

[0056] The verification module is used to determine the correction value of the magnetizing inductance based on the predicted value of the magnetizing inductance and the preset verification mechanism.

[0057] The correction module is used to update the speed regulation model parameters of the doubly fed motor according to the correction value of the excitation inductance.

[0058] Furthermore, this invention also proposes a doubly-fed motor system speed regulation model correction device, which includes: a memory, a processor, and a doubly-fed motor system speed regulation model correction program stored in the memory and executable on the processor. The doubly-fed motor system speed regulation model correction program is configured to implement the steps of the doubly-fed motor system speed regulation model correction method described in any of the above embodiments.

[0059] Furthermore, this embodiment of the invention also proposes a storage medium storing a speed regulation model correction program for a doubly fed motor system. When the doubly fed motor system speed regulation model correction program is executed by a processor, it implements the steps of the doubly fed motor system speed regulation model correction method described in any of the above embodiments.

[0060] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0061] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0062] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0063] Those skilled in the art will understand that the above embodiments are specific implementations of the present invention, and in practical applications, various changes can be made in form and detail without departing from the spirit and scope of the present invention.

Claims

1. A method for correcting the speed regulation model of a doubly-fed induction generator system, characterized in that, include: Acquire multi-source operating data of the doubly fed motor, including electromagnetic dimension data, hydraulic dimension data, and environmental dimension data; The predicted value of the excitation inductance is determined based on the multi-source operating data and the preset deep learning model. The magnetizing inductance correction value is determined based on the predicted value of the magnetizing inductance and the preset verification mechanism. The speed control model parameters of the doubly fed motor are updated based on the excitation inductance correction value.

2. The speed regulation model correction method for a doubly-fed motor system according to claim 1, characterized in that, The step of determining the predicted value of the excitation inductor based on the multi-source operating data and the preset deep learning model includes: The multi-source operational data is fused to obtain a fused feature vector; The fused feature vector is input into a preset deep learning model to obtain the initial magnetizing inductance prediction value; The predicted excitation inductance is obtained by performing time-series smoothing based on the initial predicted excitation inductance value and the excitation inductance value at the previous moment.

3. The speed regulation model correction method for a doubly-fed motor system according to claim 2, characterized in that, Before inputting the fused feature vector into a preset deep learning model to obtain the initial magnetizing inductance prediction value, the method further includes: Obtain a historical multi-source operational dataset, which includes multiple sets of historical electromagnetic dimension data, historical hydraulic dimension data, and historical environmental dimension data; A training sample set is constructed based on the historical multi-source operation dataset and the corresponding historical excitation inductance calibration values; The initial long short-term memory network model is trained based on the training sample set to obtain the preset deep learning model.

4. The speed regulation model correction method for a doubly-fed motor system according to claim 3, characterized in that, The step of training the initial long short-term memory network model based on the training sample set to obtain the preset deep learning model includes: Construct input time series features based on the historical multi-source runtime dataset; Construct an output tag based on the historical excitation inductance calibration value; The initial long short-term memory network model is trained based on the input time series features and the output labels to obtain the preset deep learning model.

5. The speed regulation model correction method for a doubly-fed motor system according to claim 1, characterized in that, The step of determining the magnetizing inductance correction value based on the predicted value of the magnetizing inductance and the preset verification mechanism includes: The theoretical stator voltage is calculated based on the predicted value of the excitation inductance and the motor steady-state equation. Calculate the voltage residual based on the theoretical stator voltage and the measured stator voltage; The physical verification result is determined based on the voltage residual and the preset threshold.

6. The speed regulation model correction method for a doubly-fed motor system according to claim 5, characterized in that, The step of determining the magnetizing inductance correction value based on the predicted value of the magnetizing inductance and the preset verification mechanism further includes: The observed magnetizing inductance value is calculated based on the magnetic flux observer and the multi-source operating data. The observation deviation is calculated based on the observed excitation inductance value and the predicted excitation inductance value. The verification results of the mechanism model are determined based on the observed deviations.

7. The method for correcting the speed regulation model of a doubly-fed motor system according to any one of claims 1 to 6, characterized in that, The acquisition of multi-source operating data of the doubly-fed motor includes: The stator three-phase voltage, stator three-phase current, rotor three-phase voltage, rotor three-phase current, active power, and reactive power are collected as the electromagnetic dimension data. The water head, guide vane opening, and flow rate of the upper and lower reservoirs are collected as the hydraulic dimension data. The winding temperature, ambient humidity, and running time are collected as the environmental dimension data. The multi-source operational data is determined based on the electromagnetic dimension data, the hydraulic dimension data, and the environmental dimension data.

8. A speed regulation model correction system for a doubly-fed induction generator system, characterized in that, include: The acquisition module is used to acquire multi-source operating data of the doubly fed motor, including electromagnetic dimension data, hydraulic dimension data and environmental dimension data. The prediction module is used to determine the predicted value of the magnetizing inductance based on the multi-source operating data and the preset deep learning model. The verification module is used to determine the correction value of the magnetizing inductance based on the predicted value of the magnetizing inductance and the preset verification mechanism. The correction module is used to update the speed regulation model parameters of the doubly fed motor according to the correction value of the excitation inductance.

9. A speed regulation model correction device for a doubly-fed induction generator system, characterized in that, The doubly fed motor system speed control model correction device includes: a memory, a processor, and a doubly fed motor system speed control model correction program stored in the memory and executable on the processor, wherein the doubly fed motor system speed control model correction program is configured to implement the steps of the doubly fed motor system speed control model correction method according to any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium stores a speed control model correction program for a doubly fed motor system. When the doubly fed motor system speed control model correction program is executed by the processor, it implements the steps of the speed control model correction method for a doubly fed motor system according to any one of claims 1 to 7.

Images