A permanent magnet fan control method based on BP neural network improved active disturbance rejection

Abstract

This invention relates to a permanent magnet wind turbine control method based on an improved active disturbance rejection system (ADRS) using a BP neural network, belonging to the field of new energy power generation grid-connected control technology. The method includes the following steps: S1: Establishing a combined aerodynamic-mechanical-electromagnetic model of the permanent magnet synchronous wind turbine, back-to-back converter, DC capacitor, grid-side filter, and power grid; S2: Establishing an inertial analogy relationship between the synchronous generator rotor energy and the DC capacitor energy storage; S3: Constructing an inertial synchronous control channel for the grid-side DC capacitor; S4: Establishing a small-signal model of the power fluctuations on the turbine side and the voltage fluctuations on the DC capacitor; S5: Designing an improved linearly extended state observer for the turbine side speed loop; S6: Constructing an ADRS control law with online parameter tuning using a BP neural network; S7: Executing coordinated grid-connected control.

Description

Technical Field

[0001] This invention belongs to the field of grid-connected control technology for new energy power generation, and relates to a control method for permanent magnet wind turbines based on improved self-disturbance rejection based on BP neural network. Background Technology

[0002] With the continuous growth of installed capacity of new energy power generation, the power system is gradually showing a trend of simultaneous increase in the proportion of new energy power sources and the proportion of power electronic interface equipment. Wind turbines are connected to the grid through back-to-back converters, and their synchronization method, power regulation method, and response characteristics to grid disturbances directly affect the stable operation of the grid-connected system.

[0003] Existing grid-connected wind power converters typically use phase-locked loops (PLLs) to obtain the grid voltage phase and frequency, and control the grid-connected current accordingly. This control method has a fast response speed and facilitates maximum power point tracking, but the converter mainly functions as a controlled current source, lacking the inherent rotational inertia, damping, and active frequency support capabilities of synchronous generators. Furthermore, when grid strength decreases or the penetration rate of new energy sources increases, the coupling between the PLL and the grid impedance may reduce the system stability margin.

[0004] Grid-based control simulates the electromagnetic relationships and rotor motion characteristics of a synchronous generator in converter control, enabling the converter to establish voltage phase and actively support the power grid without relying on phase-locked loops. Common grid-based methods include power synchronization control, droop control, and virtual synchronization control. These methods typically use output active power as the synchronization variable and are more suitable for devices such as energy storage converters where the output power can be directly adjusted.

[0005] For wind turbines employing permanent magnet synchronous generators and full-power back-to-back converters, the generator-side converter controls generator speed and active power, while the grid-side converter controls DC bus voltage and reactive power. DC capacitor inertial synchronization control leverages the mathematical similarity between DC capacitor energy storage and synchronous generator rotor kinetic energy, analogizing the DC capacitor voltage to the synchronous generator speed. The voltage deviation is used to generate the grid-side modulation voltage phase angle, thus achieving autonomous synchronization without a phase-locked loop.

[0006] Under ideal power balance conditions, the DC capacitor voltage can vary with the grid frequency and establish a quantitative mapping relationship. However, the DC capacitor is also located in the energy exchange link between the generator-side power and the grid-side power. When wind speed, reference speed, or actual speed fluctuates, the change in generator-side output power will cause the DC capacitor to charge and discharge, resulting in voltage fluctuations that are not caused by changes in grid frequency. This leads to frequency perception deviation and weakens the stability of inertial synchronization control.

[0007] Permanent magnet synchronous generators (PMSGs) exhibit nonlinear and strongly coupled characteristics. Traditional generator-side speed loops often employ fixed-parameter proportional-integral (PI) control, which is prone to overshoot or oscillation when the reference speed changes abruptly, the actual speed fluctuates, or the model parameters change. This makes it difficult to simultaneously achieve fast tracking and power fluctuation suppression. Traditional active disturbance rejection (ADRC) can utilize extended state observers to estimate and compensate for model uncertainties and external disturbances; however, its observers and feedback parameters typically rely on manual tuning, limiting its adaptability to continuously changing wind speeds and engine speeds. Backpropagation (BP) neural networks, by updating network weights using error backpropagation (BP) and gradient descent, can correct control parameters online based on operating errors, providing ADRCs with parameter adjustment capabilities that adapt to changing operating conditions.

[0008] Therefore, it is necessary to construct a wind turbine control method based on BP neural network to improve active disturbance rejection. Without changing the basic logic of controlling active power on the turbine side and controlling DC capacitor voltage on the grid side, the active disturbance rejection parameters of the speed loop should be adjusted online to suppress the transmission of wind speed and actual speed fluctuations to the DC side, and maintain the quantitative sensing capability of DC capacitor voltage to grid frequency. Summary of the Invention

[0009] In view of this, the purpose of this invention is to provide a permanent magnet fan control method based on BP neural network-based improved self-disturbance rejection.

[0010] To achieve the above objectives, the present invention provides the following technical solution: A method for controlling a permanent magnet fan with improved active disturbance rejection based on a BP neural network includes the following steps: S1: Establish a combined aerodynamic-mechanical-electromagnetic model of permanent magnet synchronous wind turbine, back-to-back converter, DC capacitor, grid-side filter and power grid; S2: Establish an inertial analogy between the rotor energy of a synchronous generator and the energy storage of a DC capacitor; S3: Construct a grid-side DC capacitor inertial synchronization control channel; S4: Establish a small-signal model of power fluctuations on the generator side to DC capacitor voltage fluctuations; S5: Design of an improved linearly expanded state observer for the machine-side speed loop; S6: Construct the active disturbance rejection control law for online parameter tuning of a BP neural network; S7: Perform coordinated grid connection control.

[0011] Furthermore, in step S1, the generator-side converter adopts rotor field-oriented control and maximum power point tracking, while the grid-side converter adopts DC capacitor inertial synchronous control, and collects wind speed, reference speed, actual speed, DC capacitor voltage, generator-side power, grid-side power, and grid frequency.

[0012] Furthermore, in step S2, the per-unit relationship between the rotor energy storage and back electromotive force amplitude of the synchronous generator is as follows:

[0013] in, E w Energy storage for the rotor of a synchronous generator; J * This is the per-unit value of the moment of inertia; oh m This is the per-unit value of the mechanical angular velocity; m This is the per-unit value of the back electromotive force amplitude; ψ m * This is the per-unit value of the permanent magnet flux linkage; oh g This is the per-unit value of the electric angular velocity, and in this analogy... oh g = oh m ; The per-unit relationship between DC capacitor energy storage and modulation voltage amplitude is as follows:

[0014] in, E C Energy is stored in a DC capacitor; C * This is the per-unit value of the capacitor; U dc This is the DC capacitor voltage value; m This represents the per-unit value of the grid-side modulation voltage amplitude. The modulation ratio is the per-unit value; dc This is the per-unit value of the DC capacitor voltage; U t This is the per-unit value of the output voltage amplitude.

[0015] Furthermore, in step S3, the deviation between the actual value and the reference value of the DC capacitor voltage is superimposed to the rated angular frequency through a synchronous proportional circuit, and then integrated to generate the phase angle of the grid-side modulation voltage.

[0016] in, i The phase angle of the grid-side modulation voltage; oh b The reference angular frequency; s For Laplace variables; oh 0 represents the per-unit value of the rated angular frequency; K zThis is the proportional coefficient for the synchronization process; U dc This is the actual value of the DC capacitor voltage; U dc0 This is the reference value for the DC capacitor voltage; U dcref This is the reference value for the DC capacitor voltage.

[0017] Furthermore, in step S4, ignoring the back-to-back converter losses, the operating point linearization is performed on the equations for grid-side output power, power angle, and DC capacitor voltage. With the grid frequency and grid voltage amplitude remaining constant, the dynamic relationship between the change in wind power and the change in DC capacitor voltage is obtained:

[0018] in, D dc This represents the per-unit change in DC capacitor voltage. D w This represents the per-unit variation of wind power. s For Laplace variables; H C The inertial time constant of the DC capacitor; U t0 This represents the steady-state modulation voltage amplitude. g0 This represents the per-unit value of the steady-state grid voltage amplitude. dc0 This is the per-unit value of the steady-state DC capacitor voltage; d 0 represents the steady-state work angle; g This is the per-unit value of the equivalent reactance on the grid side; oh 0 represents the steady-state electrical angular frequency; this relationship indicates that the power fluctuations on the machine side caused by wind speed and rotational speed will directly generate additional disturbances in the DC capacitor voltage; in order to suppress this propagation path from the source of the disturbance, an improved active disturbance rejection control model is introduced into the machine side rotational speed loop.

[0019] Furthermore, in step S5, the d-axis current of the permanent magnet synchronous motor is set to zero, and the mechanical torque change, friction, model parameter perturbation, and external disturbance are unified into the total system disturbance. The state observation value and the total disturbance observation value are updated using the speed state and its observation error:

[0020] in, e 1 represents the error in observing the rotational speed. z 1 represents the observed value of the rotational speed. 1 indicates the rotational speed; z 2 represents the total observed disturbance of the system;a 1 and a 2 represents the observer gain; b 0 represents the control gain; i q * for q Shaft current reference value; 1. z 2 and e The upper point in 1 represents the derivative of the corresponding variable with respect to time; the observer uses the error derivative term to correct the total disturbance estimate.

[0021] Furthermore, in step S6, an error feedback control quantity is generated based on the deviation between the reference speed and the observed speed, and feedforward compensation is performed using the total disturbance observation value:

[0022] in, u 0 represents the error feedback control value; k p For the proportional parameters of the active disturbance rejection controller; oh m * For reference mechanical speed; z 1 represents the observed value of the rotational speed. i q * for q Shaft current reference value; z 2 represents the total observed disturbance of the system; b 0 represents the control gain.

[0023] A neural network evaluation index is constructed using the squared error between the reference speed and the actual speed.

[0024] in, E As evaluation indicators; oh m * For reference mechanical speed; oh m The actual mechanical rotation speed is used; the neural network adopts a structure of four input nodes, five hidden layer nodes, and one output node, and updates the connection weights from the input layer to the hidden layer and from the hidden layer to the output layer through backpropagation (BP).

[0025]

[0026] in, Δw kj ( n ) is the firstn Hidden layer nodes during the next iteration j To the output layer node k The connection weight correction amount; D w ji ( n ) is the input layer node i To hidden layer nodes j The connection weight correction amount; or For learning rate; α The inertia coefficient; E ( n ) is the first n Evaluation metrics for each iteration; w kj ( n )and w ji ( n ) represents the connection weights for this iteration; Δw kj ( n -1) and Δw ji ( n -1) represents the previous weight adjustment amount; the neural network output is used to adjust the scaling parameters online. k p This allows the controller to adapt to fluctuations in wind speed and actual rotational speed.

[0027] Furthermore, in step S7, the reference value of the q-axis current on the machine side is sent into the inner current loop to generate voltage commands for the d-axis and q-axis of the rotating coordinate system. These commands are then driven by the machine-side converter through Parker inverse transformation and space vector pulse width modulation. The grid-side converter generates a synchronous phase angle based on the DC capacitor voltage and generates a modulation voltage amplitude based on the reactive power deviation and stabilization circuit. The reference speed, actual speed, and DC capacitor voltage are collected cyclically, and the neural network weights and active disturbance rejection parameters are updated online to suppress additional fluctuations in the DC capacitor voltage caused by machine-side power fluctuations.

[0028] The beneficial effects of this invention are as follows: This invention uses the machine-side speed tracking error as the evaluation criterion, combining online weight learning of a BP neural network with an improved extended state observer and a total disturbance compensation control law; simultaneously, it retains the frequency-voltage quantitative mapping of DC capacitor inertial synchronization control, and reduces the interference of wind speed and speed fluctuations on grid-side frequency sensing by adaptively adjusting the speed loop on the disturbance source side. Specifically, this invention: 1. Improve dynamic speed tracking performance The BP neural network adjusts the active disturbance rejection control parameters online according to the operating status, which can shorten the tracking process, reduce overshoot, and reduce engine-side speed fluctuations when the reference speed changes.

[0029] 2. Enhance the DC capacitor voltage immunity to interference. The improved extended state observer provides unified estimation and compensation for model uncertainties, mechanical torque variations, and external disturbances, which can significantly suppress DC capacitor voltage oscillations caused by random fluctuations in wind speed and actual rotational speed.

[0030] 3. Improve the accuracy of power grid frequency sensing The method retains the quantitative mapping relationship between DC capacitor voltage and grid frequency, while reducing the interference of generator-side power changes on this mapping relationship, thereby improving the frequency sensing accuracy of phase-locked loop-free inertial synchronization control.

[0031] 4. Maintain existing control logic and improve grid connection stability. The method does not change the basic structure of the permanent magnet synchronous wind turbine generator set, which controls the active power on the generator side and the DC capacitor voltage on the grid side. It can be combined with the existing back-to-back converter control system and improve the stability of inertial synchronous grid connection under time-varying wind speed conditions.

[0032] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0033] To make the objectives, technical solutions, and advantages of the present invention clearer, the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, wherein: Figure 1 This is a schematic diagram illustrating the similarity between a grid-side converter and a synchronous generator. Figure 2 This is a block diagram of the inertial synchronization control for the grid-side converter. Figure 3 Diagram of the improved active disturbance rejection machine side control structure based on BP neural network; Figure 4 This is the block diagram of the outer loop active disturbance rejection control for rotational speed; Figure 5 Improve the active disturbance rejection control block diagram for BP neural network; Figure 6 The comparison shows the speed tracking performance under wind speed changes, where (a) is the traditional proportional-integral control and (b) is the method of the present invention. Figure 7 The DC capacitor voltage under wind speed fluctuation when the grid frequency decreases by 0.1Hz is given, where (a) is the traditional proportional-integral control and (b) is the method of the present invention. Figure 8The DC capacitor voltage is given when the grid frequency decreases by 0.1 Hz, where (a) is the conventional proportional-integral control and (b) is the method of the present invention. Detailed Implementation

[0034] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0035] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Therefore, the drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0036] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the invention. However, it will be apparent to those skilled in the art that embodiments of the invention may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the invention.

[0037] Example 1: This invention provides an inertial synchronization anti-interference control method for permanent magnet synchronous wind turbines based on improved active disturbance rejection using a BP neural network, comprising the following steps: S1: Establish a combined aerodynamic-mechanical-electromagnetic model of the permanent magnet synchronous wind turbine, back-to-back converter, DC capacitor, grid-side filter, and power grid. The turbine-side converter adopts rotor field-oriented control and maximum power point tracking, while the grid-side converter adopts DC capacitor inertial synchronous control. Wind speed, reference speed, actual speed, DC capacitor voltage, turbine-side power, grid-side power, and grid frequency are collected.

[0038] S2: As Figure 1 As shown, an inertial analogy relationship is established between the rotor energy of a synchronous generator and the energy stored in a DC capacitor. The per-unit relationship between the rotor energy storage and the back electromotive force amplitude of a synchronous generator is as follows:

[0039] in, Ew Energy storage for the rotor of a synchronous generator; J * This is the per-unit value of the moment of inertia; oh m This is the per-unit value of the mechanical angular velocity; m This is the per-unit value of the back electromotive force amplitude; ψ m * This is the per-unit value of the permanent magnet flux linkage; oh g This is the per-unit value of the electric angular velocity, and in this analogy... oh g = oh m .

[0040] The per-unit relationship between DC capacitor energy storage and modulation voltage amplitude is as follows:

[0041] in, E C Energy is stored in a DC capacitor; C * This is the per-unit value of the capacitor; U dc This is the DC capacitor voltage value; m This represents the per-unit value of the grid-side modulation voltage amplitude. The modulation ratio is the per-unit value; dc This is the per-unit value of the DC capacitor voltage; U t This is the per-unit value of the output voltage amplitude.

[0042] S3: As Figure 2 As shown, a grid-side DC capacitor inertial synchronization control channel is constructed. The deviation between the actual and reference values ​​of the DC capacitor voltage is superimposed to the rated angular frequency via a synchronization proportional element, and then integrated to generate the grid-side modulation voltage phase angle.

[0043] in, i The phase angle of the grid-side modulation voltage; oh b The reference angular frequency; s For Laplace variables; oh 0 represents the per-unit value of the rated angular frequency; K z This is the proportional coefficient for the synchronization process; U dcThis is the actual value of the DC capacitor voltage; U dc0 This is the reference value for the DC capacitor voltage; U dcref This is the reference value for the DC capacitor voltage.

[0044] S4: Establish a small-signal model of the power fluctuations on the turbine side and the DC capacitor voltage fluctuations. Ignoring back-to-back converter losses, linearize the operating point equations for grid-side output power, power angle, and DC capacitor voltage. With the grid frequency and grid voltage amplitude remaining constant, obtain the dynamic relationship between the wind power change and the DC capacitor voltage change:

[0045] in, D dc This represents the per-unit change in DC capacitor voltage. D w This represents the per-unit variation of wind power. s For Laplace variables; H C The inertial time constant of the DC capacitor; U t0 This represents the steady-state modulation voltage amplitude. g0 This represents the per-unit value of the steady-state grid voltage amplitude. dc0 This is the per-unit value of the steady-state DC capacitor voltage; d 0 represents the steady-state work angle; g This is the per-unit value of the equivalent reactance on the grid side; oh 0 represents the steady-state electrical angular frequency. This relationship indicates that power fluctuations on the turbine side caused by wind speed and engine speed directly generate additional disturbances in the DC capacitor voltage. To suppress this propagation path from the source of the disturbance, an improved active disturbance rejection control is introduced into the turbine side speed loop. The overall turbine side structure is as follows: Figure 3 As shown.

[0046] S5: Design an improved linearly extended state observer for the machine-side speed loop. Set the d-axis current of the permanent magnet synchronous motor to zero, unify mechanical torque variations, friction, model parameter perturbations, and external disturbances into a total system disturbance, and update the state observations and total disturbance observations using the speed state and its observation error:

[0047] in, e 1 represents the error in observing the rotational speed. z 1 represents the observed value of the rotational speed. 1 indicates the rotational speed; z2 represents the total observed disturbance of the system; a 1 and a 2 represents the observer gain; b 0 represents the control gain; i q * for q Shaft current reference value; 1. z 2 and e In Figure 1, the upper point represents the derivative of the corresponding variable with respect to time. The observer uses the error derivative term to correct the total disturbance estimate, and its closed-loop structure is as follows: Figure 4 As shown.

[0048] S6: Construct an active disturbance rejection control law with online parameter tuning using a BP neural network. The error feedback control quantity is generated based on the deviation between the reference speed and the observed speed, and feedforward compensation is performed using the total disturbance observation value.

[0049] in, u 0 represents the error feedback control value; k p For the proportional parameters of the active disturbance rejection controller; oh m * For reference mechanical speed; z 1 represents the observed value of the rotational speed. i q * for q Shaft current reference value; z 2 represents the total observed disturbance of the system; b 0 represents the control gain.

[0050] A neural network evaluation index is constructed using the squared error between the reference speed and the actual speed.

[0051] in, E As evaluation indicators; oh m * For reference mechanical speed; oh m This represents the actual mechanical rotational speed. The neural network employs a structure with four input nodes, five hidden layer nodes, and one output node. It updates the connection weights between the input layer and the hidden layers, and between the hidden layers and the output layer, using backpropagation (BP) to detect errors.

[0052]

[0053] in, Δw kj ( n ) is the first n Hidden layer nodes during the next iteration j To the output layer node k The connection weight correction amount; D w ji ( n ) is the input layer node i To hidden layer nodes j The connection weight correction amount; or For learning rate; α The inertia coefficient; E ( n ) is the first n Evaluation metrics for each iteration; w kj ( n )and w ji ( n ) represents the connection weights for this iteration; Δw kj ( n -1) and Δw ji ( n -1) represents the previous weight adjustment amount. The neural network output is used to adjust the scaling parameters online. k p This allows the controller to adapt to fluctuations in wind speed and actual rotational speed. The complete control structure is as follows: Figure 5 As shown.

[0054] S7: Execute coordinated grid-connected control. The machine-side q-axis current reference value is sent to the inner current loop to generate d-axis and q-axis voltage commands in the rotating coordinate system. These commands are then driven by Parker inverse transform and space vector pulse width modulation to drive the machine-side converter. The grid-side converter generates a synchronous phase angle based on the DC capacitor voltage and generates a modulation voltage amplitude based on reactive power deviation and stabilization. The reference speed, actual speed, and DC capacitor voltage are cyclically collected, and the neural network weights and active disturbance rejection parameters are updated online to suppress additional fluctuations in DC capacitor voltage caused by machine-side power fluctuations.

[0055] Example 2: This embodiment establishes an inertial synchronization grid-connected model of a permanent magnet direct-drive wind power system on the MATLAB / Simulink simulation platform. The active power of the wind turbine is 6.25MW, the rated DC capacitor voltage is 2400V, the DC capacitor is 0.02F, the rated grid frequency is 50Hz, the filter inductor is 3mH, the filter capacitor is 0.00035F, the filter time constant is 0.1s, and the proportional coefficient of the synchronization link is π / 24. The turbine-side speed loop adopts both traditional proportional-integral control and the BP neural network-improved active disturbance rejection control of this invention.

[0056] The simulated wind speed increased from 11.104 m / s to 12 m / s in steps between 1.5 and 2 seconds, and from 11 m / s to 10 m / s in steps between 3 and 4 seconds. Figure 6 As can be seen in (a) and (b), the actual rotational speed of the method of the present invention can follow the reference rotational speed more quickly, and the overshoot of the step process is significantly reduced.

[0057] After 2 seconds of simulation, the power grid frequency decreases by 0.1 Hz, corresponding to a decrease of 0.002 pu per unit of frequency. At the same time, a random fluctuation with a standard deviation of 0.1 is superimposed on the rated wind speed. Figure 7 As shown in (a) and (b), both control methods can quantitatively sense frequency changes in DC capacitor voltage, but the capacitor voltage exhibits significant oscillations under traditional proportional-integral control, while the method of this invention only produces slight fluctuations.

[0058] After 2 seconds of simulation, the grid frequency also decreased by 0.1 Hz, and a random fluctuation with a standard deviation of 0.1 was superimposed on the actual rotational speed. Figure 8 Figures (a) and (b) show that the method of the present invention can significantly reduce the amplitude of DC capacitor voltage fluctuations and improve the stability of inertial synchronization grid connection while maintaining the ability to quantitatively sense frequency changes.

[0059] Example 3: An electronic device, including a memory and a processor; The memory is used to store computer programs; The processor is configured to implement the method described in Embodiment 1 when executing the computer program.

[0060] Example 4: A computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in Embodiment 1.

[0061] Example 5: A computer program product includes a computer program that, when executed by a processor, implements the method described in Example 1.

[0062] In the above embodiments, the reference to "this embodiment" in the specification indicates that a specific feature, structure, or characteristic described in connection with the embodiment is included in at least some embodiments, but not necessarily all embodiments. Multiple appearances of "this embodiment" do not necessarily refer to the same embodiment.

[0063] In the above embodiments, although the invention has been described in conjunction with specific embodiments thereof, many substitutions, modifications, and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory structures (e.g., dynamic RAM (DRAM)) may be used with the embodiments discussed. The embodiments of the invention are intended to cover all such substitutions, modifications, and variations falling within the broad scope of the appended claims.

[0064] As will be understood by those skilled in the art, the computer-readable storage medium described in this embodiment allows for the implementation of all or part of the steps in the above method embodiments by computer program-related hardware. The aforementioned computer program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0065] The electronic terminal provided in this embodiment includes a processor, a memory, a transceiver, and a communication interface. The memory and the communication interface are connected to the processor and the transceiver and complete communication between them. The memory is used to store computer programs, the communication interface is used to perform communication, and the processor and the transceiver are used to run the computer programs, so that the electronic terminal performs the steps of the above method.

[0066] In this embodiment, the memory may include random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage device.

[0067] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0068] This invention can be used in a wide range of general-purpose or special-purpose computing system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices, etc.

[0069] This invention can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This invention can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0070] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A control method for permanent magnet fans based on improved active disturbance rejection using a BP neural network, characterized in that: Includes the following steps: S1: Establish a combined aerodynamic-mechanical-electromagnetic model of permanent magnet synchronous wind turbine, back-to-back converter, DC capacitor, grid-side filter and power grid; S2: Establish an inertial analogy between the rotor energy of a synchronous generator and the energy storage of a DC capacitor; S3: Construct a grid-side DC capacitor inertial synchronization control channel; S4: Establish a small-signal model of power fluctuations on the generator side to DC capacitor voltage fluctuations; S5: Design of an improved linearly expanded state observer for the machine-side speed loop; S6: Construct the active disturbance rejection control law for online parameter tuning of a BP neural network; S7: Perform coordinated grid connection control.

2. The permanent magnet fan control method based on BP neural network with improved active disturbance rejection as described in claim 1, characterized in that: In step S1, the generator-side converter adopts rotor field-oriented control and maximum power point tracking, while the grid-side converter adopts DC capacitor inertial synchronous control. The wind speed, reference speed, actual speed, DC capacitor voltage, generator-side power, grid-side power, and grid frequency are collected.

3. The permanent magnet fan control method based on BP neural network with improved active disturbance rejection as described in claim 1, characterized in that: In step S2, the per-unit relationship between the rotor energy storage and back electromotive force amplitude of the synchronous generator is as follows: in, E w Energy storage for the rotor of a synchronous generator; J * This is the per-unit value of the moment of inertia; ω m This is the per-unit value of the mechanical angular velocity; m This is the per-unit value of the back electromotive force amplitude; ψ m * This is the per-unit value of the permanent magnet flux linkage; ω g This is the per-unit value of the electric angular velocity, and in this analogy... ω g = ω m ; The per-unit relationship between DC capacitor energy storage and modulation voltage amplitude is as follows: in, E C Energy is stored in a DC capacitor; C * This is the per-unit value of the capacitor; U dc This is the DC capacitor voltage value; m This represents the per-unit value of the grid-side modulation voltage amplitude. The modulation ratio is the per-unit value; dc This is the per-unit value of the DC capacitor voltage; U t This is the per-unit value of the output voltage amplitude.

4. The permanent magnet fan control method based on BP neural network with improved active disturbance rejection as described in claim 1, characterized in that: In step S3, the deviation between the actual value and the reference value of the DC capacitor voltage is superimposed to the rated angular frequency through a synchronous proportional circuit, and then integrated to generate the phase angle of the grid-side modulation voltage. in, θ The phase angle of the grid-side modulation voltage; ω b The reference angular frequency; s For Laplace variables; ω 0 represents the per-unit value of the rated angular frequency; K z This is the proportional coefficient for the synchronization process; U dc This is the actual value of the DC capacitor voltage; U dc0 This is the reference value for the DC capacitor voltage; U dcref This is the reference value for the DC capacitor voltage.

5. The permanent magnet fan control method based on BP neural network with improved active disturbance rejection as described in claim 1, characterized in that: In step S4, ignoring the back-to-back converter losses, the operating point linearization is performed on the equations for grid-side output power, power angle, and DC capacitor voltage. With the grid frequency and grid voltage amplitude remaining constant, the dynamic relationship between the change in wind power and the change in DC capacitor voltage is obtained: in, Δ dc This represents the per-unit change in DC capacitor voltage. Δ w This represents the per-unit variation of wind power. s For Laplace variables; H C The inertial time constant of the DC capacitor; U t0 This represents the steady-state modulation voltage amplitude. g0 This represents the per-unit value of the steady-state grid voltage amplitude. dc0 This is the per-unit value of the steady-state DC capacitor voltage; δ 0 represents the steady-state work angle; g This is the per-unit value of the equivalent reactance on the grid side; ω 0 represents the steady-state electrical angular frequency; this relationship indicates that the power fluctuations on the machine side caused by wind speed and rotational speed will directly generate additional disturbances in the DC capacitor voltage; in order to suppress this propagation path from the source of the disturbance, an improved active disturbance rejection control model is introduced into the machine side rotational speed loop.

6. The permanent magnet fan control method based on BP neural network with improved active disturbance rejection as described in claim 1, characterized in that: In step S5, the d-axis current of the permanent magnet synchronous motor is set to zero. The mechanical torque variation, friction, model parameter perturbation, and external disturbance are unified into the total system disturbance. The state observations and total disturbance observations are updated using the speed state and its observation error. in, e 1 represents the error in observing the rotational speed. z 1 represents the observed value of the rotational speed. 1 indicates the rotational speed; z 2 represents the total observed disturbance of the system; a 1 and a 2 represents the observer gain; b 0 represents the control gain; i q * for q Shaft current reference value; 1. z 2 and e The upper point in 1 represents the derivative of the corresponding variable with respect to time; the observer uses the error derivative term to correct the total disturbance estimate.

7. The permanent magnet fan control method based on BP neural network with improved active disturbance rejection as described in claim 1, characterized in that: In step S6, an error feedback control quantity is generated based on the deviation between the reference speed and the observed speed, and feedforward compensation is performed using the total disturbance observation value: in, u 0 represents the error feedback control value; k p For the proportional parameters of the active disturbance rejection controller; ω m * For reference mechanical speed; z 1 represents the observed value of the rotational speed. i q * for q Shaft current reference value; z 2 represents the total observed disturbance of the system; b 0 represents the control gain. A neural network evaluation index is constructed using the squared error between the reference speed and the actual speed. in, E As evaluation indicators; ω m * For reference mechanical speed; ω m The actual mechanical rotation speed is used; the neural network adopts a structure of four input nodes, five hidden layer nodes, and one output node, and updates the connection weights from the input layer to the hidden layer and from the hidden layer to the output layer through backpropagation (BP). in, Δw kj ( n ) is the first n Hidden layer nodes during the next iteration j To the output layer node k The connection weight correction amount; Δw ji ( n ) is the input layer node i To hidden layer nodes j The connection weight correction amount; η For learning rate; α The inertia coefficient; E ( n ) is the first n Evaluation metrics for each iteration; w kj ( n )and w ji ( n ) represents the connection weights for this iteration; Δw kj ( n -1) and Δw ji ( n -1) represents the previous weight adjustment amount; the neural network output is used to adjust the scaling parameters online. k p This allows the controller to adapt to fluctuations in wind speed and actual rotational speed.

8. The permanent magnet fan control method based on BP neural network with improved active disturbance rejection as described in claim 1, characterized in that: In step S7, the reference value of the q-axis current on the machine side is sent into the inner current loop to generate voltage commands for the d-axis and q-axis of the rotating coordinate system. These commands are then driven by the machine-side converter through Park inverse transformation and space vector pulse width modulation. The grid-side converter generates a synchronous phase angle based on the DC capacitor voltage and generates a modulation voltage amplitude based on the reactive power deviation and stabilization circuit. The reference speed, actual speed, and DC capacitor voltage are collected cyclically, and the neural network weights and active disturbance rejection parameters are updated online to suppress additional fluctuations in DC capacitor voltage caused by machine-side power fluctuations.

Images