Multivariable control method for permanent magnetic doubly-fed generator set

Abstract

The invention discloses a permanent magnetic doubly-fed generator set multivariable control method for a high-temperature adiabatic compressed air energy storage system, relates to the technical field of power generation technology coordination control, and aims to solve the problems of high input and output variable coupling degree and large system internal disturbance in the energy release control process of the high-temperature adiabatic compressed air energy storage system. According to the technical scheme, the method is characterized in that variable-speed and constant-frequency operation in the energy release process is achieved through the characteristics of a permanent magnetic doubly-fed generator, and a larger turbine rotating speed adjustment feasible region is provided for sliding pressure operation in the energy release process; and meanwhile, decoupling among the rotor winding current frequency in the generator, the turbine valve opening, the output power and the output frequency is realized through a BP neural network inverse model, and the flexibility and the stability of the system are improved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of coordinated control of power generation technology, and particularly relates to a permanent magnetized double-fed generator set multivariable control method for a high-temperature adiabatic compressed air energy storage system. BACKGROUND

[0002] With the substantial increase of renewable energy installed capacity in China, large-scale wind and solar power consumption has become a major test of power grid stability. Building large-scale energy storage power stations is one of the main tasks of new power system construction, among which compressed air energy storage is a key direction of new energy storage development due to its large capacity, short construction period, high efficiency, environmental friendliness and other characteristics. The high-temperature adiabatic compressed air energy storage system stores air in a high-temperature and high-pressure state through a compressor during the energy storage stage, and stores heat through a heat storage system. The high-pressure air is injected into a salt cave or other gas storage device for storage. During the energy release stage, the heat storage system releases heat to the high-pressure air, which enters the turbine to do work and drive the generator set to generate electricity. The entire process does not produce any carbon emissions and can provide rotational inertia to assist power grid frequency modulation, which is a large-scale energy storage form that has attracted much attention.

[0003] During the energy release stage of the high-temperature adiabatic compressed air energy storage system, the air in the gas storage device continuously releases work. Since the gas storage device is a constant volume container, the outlet air pressure inevitably changes continuously with the operation process. Most compressed air energy storage units currently use throttling gas distribution, which adjusts the air pressure entering the turbine through a throttle valve. In the throttling process, there is a huge pressure energy loss, which reduces the system efficiency. At the same time, some units also use sliding pressure gas distribution, which can avoid throttling loss. However, the turbine's work capacity and speed have a strong coupling, and traditional generators rely on constant speed to maintain alternating current output frequency to meet the requirements of the power grid, resulting in a great dependence of the system on mechanical speed regulating devices.

[0004] Unlike common generators, the permanent magnetized double-fed generator includes a stator and inner and outer rotors. The outer rotor is a permanent magnet rotor, and permanent magnets are attached to the inner and outer rotors, dividing the entire generator into two inner and outer motors formed by the inner rotor and the permanent magnet outer rotor, and the permanent magnet outer rotor and the stator. The stator is directly connected to the power grid, and the inner rotor is connected to the power grid through a power converter. The air gap magnetic field provided by the permanent magnet outer rotor can decouple the inner rotor speed and the output alternating current frequency, widening the feasible domain of the turbine speed.

[0005] The application provides a permanent magnetized double-fed generator set multivariable control method for a high-temperature adiabatic compressed air energy storage system, which aims to realize decoupling between the inner rotor winding current frequency of the generator, the turbine valve opening degree and the output power and output frequency through a BP neural network inverse model, improve the decoupling degree of the system and the permanent magnetized double-fed generator set, and improve the flexibility and stability of the system. Summary of the Invention

[0006] This application provides a multivariable control method for a permanent magnet doubly fed generator set in a high-temperature adiabatic compressed air energy storage system. The technical objective is to decouple the rotor winding current frequency, turbine valve opening, output power, and output frequency in the generator by using a BP neural network inverse model, thereby enabling independent adjustment of output power and output frequency and solving the problem of strong coupling between the feasible domain requirement of speed change and output power during energy release regulation.

[0007] The above-mentioned technical objective of this application is achieved through the following technical solution:

[0008] A multivariable control method for a permanent magnet doubly-fed generator set, used in a high-temperature adiabatic compressed air energy storage system, includes the following steps:

[0009] S1: Train the inverse model of the BP neural network as a decoupler to achieve decoupling of input and output variables;

[0010] S2: Based on the system characteristics, build a control logic framework and use the feedback output of the BP neural network inverse model decoupler to achieve multivariate regulation of multiple control variables.

[0011] Furthermore, the high-temperature adiabatic compressed air energy storage system includes a compressor, an air storage device, an air turbine, and a permanent magnet doubly fed generator connected in sequence. The permanent magnet doubly fed generator includes a stator and two rotors, an inner rotor and an outer rotor. The outer rotor is a permanent magnet rotor with permanent magnets attached to both its inner and outer surfaces. The entire generator is divided into two motors, an inner motor and an outer permanent magnet rotor, and an outer permanent magnet rotor and a stator, respectively. The stator is directly connected to the power grid, and the inner rotor is connected to the power grid through a power converter.

[0012] Further, step S1 specifically includes:

[0013] S11: Initialize the structural parameters of the BP neural network inverse model, including selecting the appropriate number of hidden layers and the number of hidden layer nodes;

[0014] S12: Construct training and testing sets for the inverse BP neural network model using simulation data and actual field data. Train and validate the inverse neural network model. Calculate the network output during forward propagation and update the weights and biases of each node using gradient descent during backpropagation. Use the root mean square error and the square of the correlation coefficient as the objective function of the genetic algorithm to optimize the parameters of the inverse BP neural network model. Repeat the above steps until the root mean square error of the training and testing sets is <1% and the square of the correlation coefficient is >0.99.

[0015] Furthermore, in step S2, a control logic framework is built based on the system characteristics, specifically including:

[0016] S21: Select the rotor winding current frequency ω of the permanent magnet doubly fed generator. ire The current i of the inner rotor under the dq synchronous shaft system gd and i gq The valve opening μ of the air turbine is taken as the input variable, and the active power P, reactive power Q and output frequency f of the generator are selected as the output variables.

[0017] S22: Determine the coupling relationship between input and output variables based on system characteristics.

[0018] S23: Utilize the feedback output of the BP neural network inverse model decoupler to achieve multivariate regulation of multiple control variables.

[0019] Furthermore, the coupling relationship between the input and output variables specifically includes:

[0020] The valve opening degree μ affects the airflow into the turbine, which in turn affects the internal rotor speed n. i The current i of the inner rotor in the dq synchronous shaft system gd and i gq It affects the generator's active power P, reactive power Q, and also the internal rotor speed n. i The internal rotor speed n i With the frequency ω of the inner rotor winding current ire Both input and output quantities affect the generator output frequency f. Analysis shows that there is a strong coupling relationship between the input and output quantities.

[0021] Furthermore, the multivariate regulation includes:

[0022] By using a BP neural network inverse model decoupler, the deviation values ​​of various input quantities such as the inner rotor speed calculated by the decoupler are obtained from the deviation values ​​of the current active power, reactive power, and frequency. The input deviations are then superimposed on the real-time values ​​of each input quantity to achieve multivariate control after decoupling.

[0023] The beneficial effects of this application are as follows: Applying a permanent magnet doubly-fed generator set to the energy release process of a high-temperature adiabatic compressed air energy storage system solves the problem of continuous variation in air thermodynamic parameters during energy release, broadens the feasible domain for turbine speed control, and improves system operating efficiency. Based on this, a BP neural network inverse model is used to decouple the strongly coupled relationship between input and output variables in the energy release process, achieving multivariate control of the energy release process. Attached Figure Description

[0024] Figure 1 This is a system schematic diagram of the object controlled by the method described in this application;

[0025] Figure 2This is the control logic diagram of the method described in this application. Detailed Implementation

[0026] The technical solution of this application will now be described in detail with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0027] like Figure 2 As shown, this invention provides a multivariable control method for a permanent magnet doubly-fed generator set used in a high-temperature adiabatic compressed air energy storage system, comprising the following steps:

[0028] S11: Initialize the backpropagation neural network inverse model structure parameters, select 3 hidden layers and 15 nodes in each hidden layer;

[0029] S12: Construct training and testing sets for the inverse BP neural network model using simulation data and actual field data. Train and validate the inverse neural network model. Calculate the network output during forward propagation and update the weights and biases of each node using gradient descent during backpropagation. Use the root mean square error and the square of the correlation coefficient as the objective function of the genetic algorithm to optimize the parameters of the inverse BP neural network model. Repeat the above steps until the root mean square error of the training and testing sets is <1% and the square of the correlation coefficient is >0.99.

[0030] Specifically, such as Figure 1 As shown, the output variable is represented as:

[0031] Y1 = P

[0032] Y2 = Q

[0033] Y3 = f

[0034] Where P is the active power of the generator, Q is the reactive power of the generator, and f is the output frequency of the motor.

[0035] P = u rd i rd +u rq i rq

[0036] Q = u rd i rq -u rq i rd

[0037] As can be seen from the above formula, the active and reactive power output on the rotor side in the synchronous coordinate system is related to the d and q components of the inner rotor excitation voltage and excitation current. Changes in voltage and current on any axis will cause changes in active and reactive power, and there is a strong coupling relationship between active and reactive power.

[0038] The rotor model can be represented as:

[0039]

[0040] Where u and i represent the voltage and current of the synchronous coordinate system, respectively; the subscripts d and q represent the d-axis and q-axis components, respectively; and the subscripts s and r represent the stator and rotor components, respectively. r For the leakage inductance of the rotor, R r The resistance of the rotor is represented by ω, and the angular velocity of rotation is ω. pm Let ω be the angular velocity of the permanent magnet external rotor. i Let ψ be the angular velocity of the inner rotor. in To synthesize magnetic flux in the air gap.

[0041] Under the rated operating conditions of the generator:

[0042]

[0043] ω pm =ω i +ω ire

[0044]

[0045] Where s represents the generator slip, P s P0 represents the generator's rated power, and ω represents the rotor output power. ire Indicates the frequency of the inner rotor winding current;

[0046] Therefore, the multivariable coupled control problem of the high-temperature adiabatic compressed air energy release process can be expressed as:

[0047]

[0048] The input variables are represented as: U = (u1, u2, u3, u4) T =(i gd i gq ,ω ire ,μ) T u1, u2, u3, u4 represent the current i of the inner rotor of the permanent magnet doubly fed generator in the dq synchronous shaft system, respectively. rd0 and i rq0 The frequency ω of the inner rotor winding current ire And the valve opening μ of the air turbine;

[0049] The output variables are: Y1 = P, Y2 = Q, Y3 = f.

[0050] S2: Based on the system characteristics, build a control logic framework and use the feedback output of the BP neural network inverse model decoupler to achieve multivariate regulation of multiple control variables.

[0051] Furthermore, in step S2, a control logic framework is built based on the system characteristics, including:

[0052] S21: Select the rotor winding current frequency ω of the permanent magnet doubly fed generator. ire The current i of the inner rotor under the dq synchronous shaft system gd and i gq The valve opening μ of the air turbine is taken as the input variable, and the active power P, reactive power Q and output frequency f of the generator are selected as the output variables.

[0053] S22: Determine the coupling relationship between input and output variables based on system characteristics.

[0054] S23: Utilize the feedback output of the BP neural network inverse model decoupler to achieve multivariate regulation of multiple control variables.

[0055] Furthermore, the coupling relationship between the input and output variables specifically includes:

[0056] The valve opening degree μ affects the airflow into the turbine, which in turn affects the internal rotor speed n. i The current i of the inner rotor in the dq synchronous shaft system gd and i gq It affects the generator's active power P, reactive power Q, and also the internal rotor speed n. i The internal rotor speed n i With the frequency ω of the inner rotor winding current ire Both input and output quantities affect the generator output frequency f. Analysis shows that there is a strong coupling relationship between the input and output quantities.

[0057] Furthermore, the multivariate regulation includes:

[0058] By using a BP neural network inverse model decoupler, the deviation values ​​of various input quantities such as the inner rotor speed calculated by the decoupler are obtained from the deviation values ​​of the current active power, reactive power, and frequency. The input deviations are then superimposed on the real-time values ​​of each input quantity to achieve multivariate control after decoupling.

[0059] The above description is merely a preferred embodiment of the present invention; other effective embodiments are also possible. Any effective improvements proposed by those skilled in the art based on the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A multivariable control method for a permanent magnet doubly-fed generator set, characterized in that, For high-temperature insulated compressed air energy storage systems, the following steps are included: S1: Train the inverse model of the BP neural network as a decoupler to achieve decoupling of input and output variables; S2: Based on the system characteristics, build a control logic framework and use the feedback output of the BP neural network inverse model decoupler to achieve multivariate regulation of multiple control variables.

2. The multivariable control method for a permanent magnet doubly-fed generator set according to claim 1, characterized in that, The high-temperature adiabatic compressed air energy storage system includes a compressor, an air storage device, an air turbine, and a permanent magnet doubly fed generator connected in sequence. The permanent magnet doubly fed generator includes a stator and two rotors, an inner rotor and an outer rotor. The outer rotor is a permanent magnet rotor with permanent magnets attached to both its inner and outer surfaces. The entire generator is divided into two motors, an inner motor and an outer permanent magnet rotor, and an outer permanent magnet rotor and a stator, respectively. The stator is directly connected to the power grid, and the inner rotor is connected to the power grid through a power converter.

3. The multivariable control method for a permanent magnet doubly-fed generator set according to claim 1, characterized in that, Step S1 specifically includes: S11: Initialize the structural parameters of the BP neural network inverse model, including selecting the number of hidden layers and the number of hidden layer nodes; S12: Construct training and testing sets for the inverse BP neural network model using simulation data and actual field data. Train and validate the inverse neural network model. Calculate the network output during forward propagation and update the weights and biases of each node using gradient descent during backpropagation. Use the root mean square error and the square of the correlation coefficient as the objective function of the genetic algorithm to optimize the parameters of the inverse BP neural network model. Repeat the above steps until the root mean square error of the training and testing sets is <1% and the square of the correlation coefficient is >0.

99.

4. The multivariable control method for a permanent magnet doubly-fed generator set according to claim 3, characterized in that, The output variable is represented as: Y1 = P Y2 = Q Y3 = f Where P is the active power of the generator, Q is the reactive power of the generator, and f is the output frequency of the motor. P=u rd and rd +in rq and rq Q=u rd i rq -u rq i rd As shown in the above formula, the active and reactive power output on the rotor side in the synchronous coordinate system is related to the d and q components of the inner rotor excitation voltage and excitation current. Changes in voltage and current on any axis will cause changes in active and reactive power. There is a strong coupling relationship between active and reactive power. The rotor model is represented as: Where u and i represent the voltage and current of the synchronous coordinate system, respectively; the subscripts d and q represent the d-axis and q-axis components, respectively; and the subscripts s and r represent the stator and rotor components, respectively. r For the leakage inductance of the rotor, R r The resistance of the rotor is represented by ω, and the angular velocity of rotation is ω. pm Let ω be the angular velocity of the permanent magnet external rotor. i Let ψ be the angular velocity of the inner rotor. in To synthesize magnetic flux in the air gap; Under the rated operating conditions of the generator: Where s represents the generator slip, P s P0 represents the generator's rated power, and ω represents the rotor output power. ire Indicates the frequency of the inner rotor winding current; Therefore, the multivariable coupled control problem of the high-temperature adiabatic compressed air energy release process can be expressed as: The input variables are represented as: U = (u1, u2, u3, u4) T =(i gd i gq ,ω ire ,μ) T u1, u2, u3, u4 represent the current i of the inner rotor of the permanent magnet doubly fed generator in the dq synchronous shaft system, respectively. rd0 and i rq0 The frequency ω of the inner rotor winding current ire And the valve opening μ of the air turbine; The output variables are: Y1 = P, Y2 = Q, Y3 = f.

5. The multivariable control method for a permanent magnet doubly-fed generator set according to claim 4, characterized in that, In step S2, the control logic framework is built according to the system characteristics, specifically including: S21: Select the rotor winding current frequency ω of the permanent magnet doubly fed generator. ire The current i of the inner rotor under the dq synchronous shaft system gd and i gq The valve opening μ of the air turbine is taken as the input variable, and the active power P, reactive power Q and output frequency f of the generator are selected as the output variables. S22: Determine the coupling relationship between input and output variables based on system characteristics; S23: Utilize the feedback output of the BP neural network inverse model decoupler to achieve multivariate regulation of multiple control variables.

6. The multivariable control method for a permanent magnet doubly-fed generator set according to claim 5, characterized in that, The coupling relationship between the input and output variables specifically includes: the valve opening degree μ affects the airflow into the turbine, which in turn affects the internal rotor speed n. i The current i of the inner rotor in the dq synchronous shaft system gd and i gq It affects the generator's active power P, reactive power Q, and also the internal rotor speed n. i The internal rotor speed n i With the frequency ω of the inner rotor winding current ire They collectively affect the generator output frequency f.

7. A multivariable control method for a permanent magnet doubly-fed generator set according to claim 5, characterized in that, The multivariate control includes: using a BP neural network inverse model decoupler, calculating the inner rotor speed and the deviation values ​​of each input quantity calculated by the decoupler based on the deviation values ​​of the current active power, reactive power, and frequency, and superimposing the input deviations onto the real-time values ​​of each input quantity to achieve multivariate control after decoupling.

Images