Adaptive Optimization Control Method and Control System for Active Damper of Grid Converter

Abstract

The present invention relates to the field of smart grid technology, and discloses an adaptive optimization control method and control system for an active damper of a grid-type converter. The system collects the voltage resonance component at the PCC, simulates impedance in the resonance frequency range to damp the resonance, so as to suppress the transfer characteristics formed by the resonance through the PCC voltage, and specifically constructs a virtual resistor in the voltage control loop and designs a virtual resistor adaptive optimization algorithm so that its output impedance maintains a pure resistance characteristic in the frequency band where the grid-type converter is prone to resonance instability, thereby providing resonance damping for the grid, thereby solving the resonance problem of the grid-type equipment at the grid-connected access point. The method of the present invention utilizes the active support capability of the grid-type converter for the access point voltage, reshapes the virtual resistor on the control structure of the grid-type converter, designs a virtual resistor adaptive optimization algorithm, and solves the wide-band resonance problem of the voltage at the access point caused by the resonance instability of the grid-type converter when the grid impedance increases.

Description

Technical Field

[0001] The present invention belongs to the technical field of smart grids, and in particular to a grid-connected system, and specifically proposes an adaptive optimization control method and control system for an active damper of a grid-connected converter, which is suitable for a grid-connected system containing grid-following and grid-connected converters. Background Art

[0002] As a high-quality flexible adjustment resource and potential active support means, energy storage system has become a strong support for new power systems to cope with grid security challenges. Grid-following energy storage and grid-building energy storage are two important technical routes, which have important application value in improving the stability of the power grid and the absorption capacity of new energy. The main difference between grid-following and grid-building energy storage systems lies in their power supply properties and control methods. The grid-following type is essentially a current source, which cannot provide voltage and frequency support by itself and must rely on the voltage and frequency of the grid. The grid-following converter (PCS) captures the phase information of the grid, measures the phase of the grid connection point (PCC) through the phase-locked loop (PLL), and achieves synchronization with the grid. Then, after coordinate transformation and PWM modulation, the control signal is formed and fed back to the switch tube. The grid-forming type is essentially a voltage source. The grid-forming converter (PCS) can independently set the voltage parameter signal output voltage and frequency. It can be connected to the grid or operated off the grid. Through the power synchronization control mechanism, the energy storage converter is shaped into the external characteristics of a voltage source. It can independently construct the AC side voltage amplitude and phase without relying on the external AC system, providing strong voltage support for the power system and having strong support capabilities for the power grid.

[0003] With the large-scale access of new energy converters, grid-following and grid-forming converters of varying sizes will be connected to the power grid at the same time. The grid impedance characteristics will show unpredictable wide-range changes depending on the operating conditions. When the grid impedance increases, the grid-following converter is prone to resonant instability, causing wide-band resonance of the voltage at the access point, which will have a serious impact on the safety and stability of the power grid.

[0004] In order to reduce the system resonance instability caused by grid impedance working conditions (such as load changes, different types of power supply access, etc.), on the one hand, the grid-following and grid-forming converters are used in combination, and the stable voltage and frequency support provided by the grid-forming converter is used to reduce the burden of the grid-following converter. At the same time, suitable filters (such as LC filters, LCL filters) are added to the converter design to reduce high-frequency harmonics and suppress resonance. On the other hand, by designing the converter control strategy optimization algorithm, the control parameters of the converter are optimized and adjusted by real-time monitoring of grid impedance changes. For example, active damping control is introduced to actively suppress resonance through control strategies; for example, the control mode of the converter is dynamically switched according to the real-time changes of the grid impedance, from the grid-following mode to the grid-forming mode, etc., to provide stability for the grid-connected system. Summary of the invention

[0005] The purpose of the present invention is to optimize the deficiencies in the prior art and provide an adaptive optimization control method and control system for an active damper of a grid-type converter, so as to realize the adaptive adjustment of the virtual resistance of the active damper function of the grid-type converter, and design a pure resistive compensation link of the virtual impedance for the frequency band where resonance instability is prone to occur in the grid-type converter, thereby significantly enhancing the pure resistive characteristics of the active damper function of the grid-type converter, and being able to effectively suppress the resonance problem of the grid-type converter under a weak power grid.

[0006] According to a first aspect of the present invention, an adaptive optimization control system for an active damper of a grid-type converter is provided, comprising:

[0007] The grid-connected converter adopts the structure of a three-phase LC full-bridge grid-connected converter to access the grid, and the three-phase access position is defined as the grid-connected common coupling point between the grid-connected converter and the grid-following converter; the actual values ​​of the grid-connected active power and reactive power are calculated based on the voltage and current collected at the grid-connected common coupling point;

[0008] The virtual synchronous machine control module includes active and reactive loop control links, which take grid-connected active power and reactive power command values ​​and actual values ​​as inputs, and output inner loop control reference voltage and phase angle;

[0009] The virtual impedance control module is used to output the virtual resistance through the virtual resistance adaptive regulator according to the current and voltage at the grid-connected common coupling point, and send the virtual resistance to the pure resistive compensation link of the virtual impedance to obtain the resonant voltage component, and the resonant voltage component is added to the inner loop control reference voltage output by the virtual synchronous machine control module to obtain the virtual impedance control output d The axis reference voltage, 0 value is subtracted from the resonant voltage component to obtain the virtual impedance control output qaxis reference voltage; wherein the virtual resistance adaptive regulator is set to be a regulator based on the Elman neural network model to achieve adaptive regulation of the virtual resistance;

[0010] The voltage-current dual closed-loop control module includes a voltage closed-loop controller and a current closed-loop controller. The voltage closed-loop controller is based on the d Axis and q Axis reference voltage, d Axis and q The filter capacitor voltage on the axis is used as input and output after closed-loop control d Axis and q The on-axis filter inductor reference current is used as input to combine d Axis and q The filter inductor current on the axis is input to the current control regulator for closed-loop control, and a modulated voltage is output;

[0011] A coordinate transformation module, used for performing coordinate transformation from the dq0 axis to the abc axis according to the modulation voltage;

[0012] The SVPWM modulator is used to generate a driving signal for controlling the power switch tube according to the coordinates of the abc axis.

[0013] According to a second aspect of the present invention, an adaptive optimization control method for an active damper of a grid-type converter is also proposed, the method comprising:

[0014] Collect the voltage and current collected at the grid common coupling point, and calculate the actual value of grid active power and reactive power based on it;

[0015] The actual value and command value of the grid-connected active power are used as input, and the active power control link of the virtual synchronous machine control module is combined with the angular frequency output by the virtual synchronous machine. ω With the virtual angular velocity command value ω ref Output inner loop control reference phase angle θ ;

[0016] The actual value and command value of the grid-connected reactive power are used as input, and the reactive power control link of the virtual synchronous machine control module is combined with the grid-connected point voltage reference value. U ref Output inner loop control reference voltage U ;

[0017] According to the current at the grid common coupling point i pcc and voltage u pcc , output virtual resistance through virtual resistance adaptive regulator R v , and the virtual resistorR v The resonant voltage component is obtained by sending it into the purely resistive compensation link of the virtual impedance;

[0018] The resonant voltage component and the inner loop control reference voltage output by the virtual synchronous machine control module U Adding the virtual impedance control output d The axis reference voltage, 0 value is subtracted from the resonant voltage component to obtain the virtual impedance control output q axis reference voltage; the virtual resistance adaptive regulator adopts an Elman neural network model to achieve adaptive adjustment output of the virtual resistance;

[0019] The virtual impedance control output d Axis and q The reference voltage of the axis and the collected d Axis and q The filter capacitor voltage on the shaft is input into the voltage closed-loop controller and output after closed-loop control d Axis and q On-axis filter inductor reference current;

[0020] After closed loop control, the output d Axis and q On-axis filter inductor reference current and collected d Axis and q The filter inductor current on the axis is input to the current control regulator, and the modulated voltage is output after closed-loop control. u s ;

[0021] The modulation voltage u s Input coordinate transformation module to realize coordinate transformation from dq0 axis to abc axis;

[0022] The coordinates of the abc axis are input into the SVPWM modulator, and the drive signal for controlling the power switch tube is generated according to the coordinates of the abc axis. k pwm .

[0023] By implementing the adaptive optimization control system of the active damper of the grid-type converter proposed in the present invention, utilizing the active support capability of the grid-type converter for the access point voltage, drawing on the working principle of the active damper, a virtual resistance adaptive adjustment strategy is designed on the control structure of the grid-type converter, and a pure resistive compensation link of the virtual impedance is designed for the frequency band where the grid-type converter is prone to resonant instability, which significantly enhances the pure resistive characteristics of the active damper function of the grid-type converter, can effectively suppress the resonance problem of the grid-type converter under a weak power grid, so as to enhance the stability and anti-interference performance of the grid-connected system operation and reduce the greater impact of the resonance on the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a principle block diagram of an adaptive optimization control system of a grid-type converter active damper according to an embodiment of the present invention.

[0025] Figure 2 It is the structural diagram of the Elman neural network.

[0026] Figure 3 Flowchart of the Elman neural network weight adjustment process.

[0027] Figure 4 The total output impedance after introducing virtual resistance before and after the virtual impedance pure resistive compensation link is Z ov ( s )’s Bode plot.

[0028] Figure 5 A schematic diagram of an experimental platform for an adaptive optimization control system of an active damper of a grid-connected converter constructed according to an embodiment of the present invention.

[0029] Figure 6a , 6b 6c are the experimental waveforms and FFT analysis diagrams of grid-connected current and voltage of the grid-forming converter and the grid-following converter when the grid impedance is 2.3mH before compensation.

[0030] Figure 7a , 7b 7c are the experimental waveforms of grid-connected current and voltage of the grid-connected converter and the grid-following converter and the FFT analysis diagram when the grid impedance is 2.3mH after re-grid connection compensation.

[0031] Figure 8a , 8b 8c are the experimental waveforms and FFT analysis diagrams of grid-connected current and voltage of the grid-connected converter and the grid-following converter when the grid impedance is 3mH after re-grid connection compensation. DETAILED DESCRIPTION

[0032] The technical solution of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical solution of the present application, rather than limitations on the technical solution of the present application. In the absence of conflict, the embodiments of the present application and the technical features in the embodiments can be combined with each other. Example

[0033] Combination Figure 1As shown, an embodiment of the present invention provides an adaptive optimization control system for an active damper of a grid-connected converter, designs a parameter adaptive optimization algorithm for the active damper function, and takes a grid-connected system containing grid-following and grid-connected converters as an example, by integrating the active damper function in the grid-connected converter, the system stability is improved.

[0034] As an example implementation, Figure 1 In the control logic of the adaptive optimization control system of the active damper of the grid-type converter shown in the figure, the active supporting capability of the grid-type converter for the access point voltage is utilized, and the working principle of the active damper is referred to, and a virtual resistance adaptive adjustment strategy is designed on the control structure of the grid-type converter. In addition, a pure resistive compensation link of the virtual impedance is designed for the frequency band where the grid-type converter is prone to resonant instability, which significantly enhances the pure resistive characteristics of the active damper function of the grid-type converter, and can effectively suppress the resonance problem of the grid-type converter under weak power grid, so as to enhance the stability and anti-interference performance of the grid-connected system operation and reduce the greater impact of resonance on the power grid.

[0035] See also Figure 1 The adaptive optimization control system of the active damper of the grid-connected converter shown in the figure, wherein the grid-connected converter adopts a three-phase LC full-bridge grid-connected converter structure. Q 1 ~ Q 6 For the power switch tube, L 1 and C They are filter inductor and filter capacitor respectively, which constitute LC filter.

[0036] like Figure 1 The adaptive optimization control system of the active damper of the grid-type converter is provided with a virtual impedance control module, a voltage-current dual closed-loop control module, a coordinate transformation module (dq0 / abc) and a SVPWM modulator.

[0037] The aforementioned virtual synchronous machine control module includes active and reactive loop control links, which take grid-connected active power and reactive power command values ​​and actual values ​​as inputs, respectively, and output inner loop control reference voltage and phase angle.

[0038] The virtual impedance control module is used to output the virtual resistance through the virtual resistance adaptive regulator according to the current and voltage at the grid-connected common coupling point, and send the virtual resistance to the pure resistive compensation link of the virtual impedance to obtain the resonant voltage component, and the resonant voltage component is added to the inner loop control reference voltage output by the virtual synchronous machine control module to obtain the virtual impedance control output d The axis reference voltage, 0 value is subtracted from the resonant voltage component to obtain the virtual impedance control output qaxis reference voltage; wherein, the virtual resistance adaptive regulator is set as a regulator based on the Elman neural network model to achieve adaptive regulation of the virtual resistance.

[0039] The voltage-current dual closed-loop control module includes a voltage closed-loop controller and a current closed-loop controller. The voltage closed-loop controller is based on the d Axis and q Axis reference voltage, d Axis and q The filter capacitor voltage on the axis is used as input and output after closed-loop control d Axis and q The on-axis filter inductor reference current is used as input to combine d Axis and q The filter inductor current on the axis is input to the current control regulator for closed-loop control and outputs a modulated voltage.

[0040] The coordinate transformation module is used to transform the coordinates from the dq0 axis to the abc axis according to the modulation voltage.

[0041] The SVPWM modulator is used to generate a driving signal for controlling the power switch tube according to the coordinates of the abc axis.

[0042] like Figure 1 The principle shown is as follows: the grid impedance Z g , grid voltage v g The three-phase LC full-bridge grid-connected converter is connected to the grid, and the three-phase connection position is defined as the grid-connected common coupling point of the grid-connected converter and the grid-following converter. The actual values ​​of the grid-connected active power and reactive power are calculated based on the voltage and current collected at the grid-connected common coupling point.

[0043] i pcc and u pcc They represent the current and voltage at the grid common coupling point respectively. P ref and Q ref are active power and reactive power command values ​​respectively, P e and Q e are the actual values ​​of grid-connected active power and reactive power, K is the reactive integrator gain. U ref It is the grid connection point voltage reference value.

[0044] Combination Figure 1 The virtual synchronous control module (VSG) shown in FIG. includes reactive power control and active power mitigation. Figure 1As shown, ω represents the angular frequency of the virtual synchronous machine output, ω ref is the virtual angular velocity command value, θ and U They are the inner loop control reference phase angle and voltage output by the virtual synchronous machine respectively.

[0045] Combination Figure 1 The virtual impedance control module shown, R v Represents a virtual resistance, from the current at the grid common coupling point i pcc and voltage u pcc The virtual resistors are sent to the virtual resistor adaptive regulator based on the Elman neural network model for adaptive output. R v Purely resistive compensation link fed into virtual impedance G c ( s ) to obtain the resonant voltage component. Figure 1 As shown, the resonant voltage component is added to the inner loop control reference voltage output by the virtual synchronous machine control module to obtain the virtual impedance control output d The axis reference voltage, 0 value is subtracted from the resonant voltage component to obtain the virtual impedance control output q Axis reference voltage, d Axis reference voltage and q The axis reference voltages are expressed as u cdref and u cqref .

[0046] Combination Figure 1 As shown, the virtual impedance controls the output d Axis and q The reference voltage of the axis is adjusted by the voltage-current dual closed-loop control module to obtain the modulation voltage. u s .

[0047] The modulation voltage u s Input coordinate transformation module, that is Figure 1 The dq0 / abc module shown realizes the coordinate transformation from the dq0 axis to the abc axis.

[0048] Finally, the converted coordinates of the abc axis are input into the SVPWM modulator, and the drive signal for controlling the power switch tube is generated according to the coordinates of the abc axis. k pwm, used to drive the power switch tube of the three-phase LC full-bridge grid-connected converter to turn on.

[0049] Combination Figure 1 As shown in the figure, in the voltage-current dual closed-loop control module, in the voltage closed-loop controller, PI represents the proportional integral controller. In the current closed-loop controller, I represents the integral controller. Figure 1 , u cd and u cq Respectively expressed in d Axis and q The filter capacitor voltage on the axis, u cdref and u cqref They represent the virtual impedance control output d Axis and q The reference voltage of the axis, ωL 1 and ωC They are the inductive reactance and capacitive reactance of the LC filter, collectively referred to as reactance. i Ldref and i Lqref Respectively d Axis and q The filter inductor reference current on the axis, i ld and i lq Respectively d Axis and q The filter inductor current on the axis.

[0050] As an optional embodiment, combining Figure 1 As shown, the control equations of the active and reactive loop control links of the aforementioned virtual synchronous machine control module (VSG) are expressed as:

[0051] ;

[0052] in, J is the moment of inertia, ω is the angular frequency of the virtual synchronous machine output, ω ref is the virtual angular velocity command value, P ref and Q ref are active power and reactive power command values ​​respectively, P e and Q e are the actual values ​​of grid-connected active power and reactive power, K is the reactive integrator gain, θ is the inner loop control reference phase angle output by the virtual synchronous machine, U ref It is the grid connection point voltage reference value.

[0053] As an optional embodiment, combining Figure 1 As shown in the figure, in the voltage-current dual closed-loop control module, the output voltage and current control regulator expressions of the voltage-current dual closed-loop control are: ;

[0054] ;

[0055] In the formula, k up and k ui are the proportional coefficient and integral coefficient of the voltage control regulator, k ip is the proportionality factor of the current control regulator, s is the Laplace operator.

[0056] Figure 1 In the virtual resistor R v Adaptive adjustment is achieved based on the Elman neural network model.

[0057] Figure 2 The Elman neural network model is a feedback neural network, which consists of four layers of neurons: input layer, hidden layer, receiving layer and output layer. The collected current and voltage at the grid common coupling point are input as input quantities, and the hidden layer outputs the current and voltage resonant components at the current moment. i pcc_h , u pcc_h The receiving layer receives feedback signals from the hidden layer to memorize the output value of the hidden layer neurons at the last moment. i pcc_h ’ , u pcc_h ’ After delay and storage, it is fed back to the hidden layer and finally outputs the ideal tracking value through the output layer as a virtual resistor. R v .

[0058] In the embodiment of the present invention, the Elman neural network assumes that the vector composed of the input layer data is u, and the input in the present invention is the current at the PCC i pcc ; Let the output layer vector be y, and the output in the present invention is a virtual resistor Rv ; Let the hidden layer output vector be x, n-dimensional, then the corresponding receiving layer output vector r is also an n-dimensional vector; let w 1 , w 2 , w 3 , are the connection weights from the hidden layer to the output layer, from the input layer to the hidden layer, and from the receiving layer to the hidden layer, respectively.

[0059] set up g (·) is the activation function of the output neuron, f (·) is the activation function of the hidden layer neurons, h (·) is the activation function of the receiving layer; net (·) represents the net input of a layer, A represents the input layer, B represents the receiving layer, and k represents the iteration order.

[0060] Define two functions:

[0061] ;

[0062] ;

[0063] Then the input and output of the nth neuron in the hidden layer are:

[0064] ;

[0065] ;

[0066] The input and output of the neurons in the receiving layer are:

[0067] ;

[0068] ;

[0069] The input and output of the output layer neurons are:

[0070] ;

[0071] ;

[0072] Figure 3 is a flow chart of the weight adjustment process. In the embodiment of the present invention, the Elman neural network uses the following empirical formula to determine the number of neurons in the hidden layer:

[0073] ;

[0074] In the formula, m Indicates the number of input quantities; n Represents the number of neurons in the output layer; δ Represents a constant, with a value of 1 to 10.

[0075] In the embodiment of the present invention, the Levenberg-Marquardt algorithm is used to train the Elman neural network. The error index function of the Levenberg-Marquardt algorithm is:

[0076] ;

[0077] In the formula, p is the number of samples; e i is the system error; y i ’ Indicates the actual output of the network; y i Represents the expected output of the network.

[0078] As an optional embodiment, the Elman neural network adjusts the weights:

[0079] ;

[0080] The calculation formula of weight increment Δw is as follows:

[0081] ;

[0082] In the formula, T is the identity matrix; μ is the learning rate defined, J ( w ) is the Jacobian matrix.

[0083] The Jacobian matrix expression is:

[0084] ;

[0085] exist Figure 1 Introducing a virtual resistor R v After that, the total output impedance Z ov ( s ) is expressed as:

[0086] ;

[0087] In the formula, T A ( s ) is the loop gain, Z o ( s) represents the original equivalent output impedance of the grid-type converter.

[0088] exist Figure 4 It can be seen that the total output impedance of the grid-type converter after the introduction of virtual resistance before adding the virtual impedance pure resistive compensation link is Z ov ( s ) can maintain the characteristics of pure resistance at low frequencies, but in the mid- and high-frequency bands, Z ov ( s ) gradually becomes more sensual. Z ov ( s ) still maintains good pure resistance in the medium and high frequency bands, such as Figure 1 As shown, the control system of the present invention adds a pure resistive compensation link of virtual impedance G c ( s ). Due to the output impedance of the grid-type converter Z o ( s ) itself is a very small impedance, so it can be approximately ignored. In addition, since the impedance output of the grid-type converter is relatively stable, it continues this characteristic, thus obtaining the compensation link G c ( s ) is expressed as:

[0089] ;

[0090] Since the above compensation function is complex and difficult to implement, it is simplified in this embodiment. Considering that the PI controller only has an effect at the fundamental wave and low frequency, it can be approximately considered that k up + k ui / s ≈ k ui ,neglect k up .

[0091] In order to ensure sufficient phase margin, the current loop cutoff frequency is usually much lower than the resonant frequency of the LC filter. Therefore, in the frequency range lower than and equal to the current loop cutoff frequency, the influence of the filter capacitor can be ignored. At this time, the compensation link expression is:

[0092] ;

[0093] Considering that the compensation link expression contains differential terms, it is difficult to implement in actual application and is sensitive to noise. Therefore, in the embodiment of the present invention, a second-order Pade approximate discrete domain function is used to replace the differential terms to digitally implement the proposed compensation link.

[0094] Second-order Pade approximation discrete domain function G pa ( z ) is:

[0095] ;

[0096] Therefore, the digital realization of the compensation function after equivalent substitution is:

[0097] ;

[0098] in, s is the Laplace operator, z is the representation of the transfer function in the discrete domain, and can be converted into the Laplace operator s in the continuous domain.

[0099] from Figure 4 It can be seen that after adding the virtual impedance pure resistive link compensation link, Z ov ( s ) presents good pure resistance in the frequency range lower than or equal to the cut-off frequency, so the virtual impedance pure resistance link compensation link added in the present invention can effectively suppress and damp the resonance in the system. Example

[0100] Combination Figure 1 And the implementation of the adaptive optimization control system of the active damper of the grid-type converter in the above embodiment, the process of using it to realize the adaptive optimization control of the active damper of the grid-type converter includes:

[0101] Collect the voltage and current collected at the grid common coupling point, and calculate the actual value of grid active power and reactive power based on it;

[0102] The actual value and command value of the grid-connected active power are used as input, and the active power control link of the virtual synchronous machine control module is combined with the angular frequency output by the virtual synchronous machine. ω With the virtual angular velocity command value ω ref Output inner loop control reference phase angle θ ;

[0103] The actual value and command value of the grid-connected reactive power are used as input, and the reactive power control link of the virtual synchronous machine control module is combined with the grid-connected point voltage reference value. U ref Output inner loop control reference voltageU ;

[0104] According to the current at the grid common coupling point i pcc and voltage u pcc , output virtual resistance through virtual resistance adaptive regulator R v , and the virtual resistor R v The resonant voltage component is obtained by sending it into the purely resistive compensation link of the virtual impedance;

[0105] The inner loop control reference voltage output by the resonant voltage component and the virtual synchronous machine control module U Adding the virtual impedance control output d The axis reference voltage, 0 value is subtracted from the resonant voltage component to obtain the virtual impedance control output q axis reference voltage; wherein the virtual resistance adaptive regulator adopts the Elman neural network model to realize the adaptive adjustment output of the virtual resistance;

[0106] The virtual impedance control output d Axis and q The reference voltage of the axis and the collected d Axis and q The filter capacitor voltage on the shaft is input into the voltage closed-loop controller and output after closed-loop control d Axis and q On-axis filter inductor reference current;

[0107] After closed loop control, the output d Axis and q On-axis filter inductor reference current and collected d Axis and q The filter inductor current on the axis is input to the current control regulator, and the modulated voltage is output after closed-loop control. u s ;

[0108] The modulation voltage u s Input coordinate transformation module to realize coordinate transformation from dq0 axis to abc axis;

[0109] The coordinates of the abc axis are input into the SVPWM modulator, and the drive signal for controlling the power switch tube is generated according to the coordinates of the abc axis. k pwm . Example

[0110] In order to further verify the feasibility of the adaptive optimization control system of the active damper of the grid-connected converter proposed in this invention and the effectiveness of the proposed compensation method, we built the following Figure 5 The prototypes of the grid-forming converter and the grid-following converter shown in the figure adopt the regenerative grid simulation power supply-61830 to simulate the grid, and simulate different grid conditions by changing the size of the grid impedance. The effective value of the voltage is 220V and the frequency is 50Hz.

[0111] The DC side is powered by DC power supplies 62100H-600S and 62180D-1800. The experimental parameters of the grid-following converter can be set as shown in Table 1, and the experimental parameters of the grid-building converter can be set as shown in Table 2.

[0112] Table 1- Grid-following converter experimental parameters

[0113]

[0114] Table 2- Grid-connected converter experimental parameters

[0115]

[0116] Figure 6a , 6b 6c are the experimental waveforms and FFT analysis diagrams of the grid-connected current and voltage of the grid-forming converter and the grid-following converter when the grid impedance is 2.3mH before compensation. It can be seen from the figure that the grid-forming converter can operate stably under weak grid, but the grid-connected current and PCC voltage of the grid-following converter will be distorted to a certain extent due to the existence of grid impedance. FFT analysis is performed on the grid-connected current of the two converters respectively. At this time, the THD value of the grid-connected current of the grid-forming converter is 3.15%, and the THD value of the grid-connected current of the grid-following converter is 11.77%, and the resonance is mostly concentrated in the 1-2kHz frequency band, and the system stability is insufficient.

[0117] Figure 7a , 7b 7c are the experimental waveforms of grid-connected current and voltage of the grid-connected converter and the grid-following converter and the FFT analysis diagram when the grid impedance is 2.3mH after re-grid-connected compensation. It can be seen from the figure that the grid-connected current of the grid-following converter and the voltage waveform of the grid connection point are significantly improved, while the grid-connected converter can still operate stably. The grid-connected current of the two converters is analyzed by FFT again. The THD value of the grid-connected current of the grid-connected converter is 3.48%, which is basically unchanged. The THD value of the grid-connected current of the grid-following converter drops to 2.75%, and the resonant current content in the 1-2kHz frequency band is significantly reduced. It can be seen that the virtual resistance accuracy of the compensation method proposed in the present invention is high, and it has a good damping effect in the target resonant frequency band, thereby improving the stability of the system.

[0118] Figure 8a , 8b 8c are the experimental waveforms of grid-connected current and voltage of the grid-forming converter and the grid-following converter and the FFT analysis diagram when the grid impedance is 3mH after re-grid compensation. It can be seen from the figure that even under weaker grid conditions, the grid-forming converter can still improve the grid-connected current of the grid-following converter and the voltage waveform of the grid connection point while maintaining its own stable operation, and the grid-connected current of the two converters is subjected to FFT analysis. The THD value of the grid-connected current of the grid-forming converter is 2.22%, and the THD value of the grid-connected current of the grid-following converter drops to 2.58%, and the resonant current content in the 1-2kHz frequency band is significantly reduced. It can be seen that the virtual resistance accuracy of the compensation method proposed in the present invention is high, and it has a good damping effect in the target resonant frequency band, thereby improving the stability of the system.

[0119] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. An adaptive optimization control system for an active damper of a grid-type converter, characterized in that: include: The grid-connected converter adopts the structure of a three-phase LC full-bridge grid-connected converter to access the grid, and the three-phase access position is defined as the grid-connected common coupling point between the grid-connected converter and the grid-following converter; Calculate the actual values ​​of grid-connected active power and reactive power based on the voltage and current collected at the grid-connected common coupling point; The virtual synchronous machine control module includes active and reactive loop control links, which take grid-connected active power and reactive power command values ​​and actual values ​​as inputs, and output inner loop control reference voltage and phase angle; The virtual impedance control module is used to output the virtual resistance through the virtual resistance adaptive regulator according to the current and voltage at the grid-connected common coupling point, and send the virtual resistance to the pure resistive compensation link of the virtual impedance to obtain the resonant voltage component, and the resonant voltage component is added to the inner loop control reference voltage output by the virtual synchronous machine control module to obtain the virtual impedance control output d The axis reference voltage, 0 value is subtracted from the resonant voltage component to obtain the virtual impedance control output q axis reference voltage; wherein the virtual resistance adaptive regulator is set to be a regulator based on the Elman neural network model to achieve adaptive regulation of the virtual resistance; The voltage-current dual closed-loop control module includes a voltage closed-loop controller and a current closed-loop controller. The voltage closed-loop controller is based on the d Axis and q Axis reference voltage, d Axis and q The filter capacitor voltage on the axis is used as input and output after closed-loop control d Axis and q The on-axis filter inductor reference current is used as input to combine d Axis and q The filter inductor current on the axis is input to the current control regulator for closed-loop control, and a modulated voltage is output; A coordinate transformation module, used for performing coordinate transformation from the dq0 axis to the abc axis according to the modulation voltage; SVPWM modulator, used to generate a driving signal for controlling the power switch tube according to the coordinates of the abc axis; The Elman neural network model is a feedback neural network, which consists of four layers of neurons: input layer, hidden layer, receiving layer and output layer; The collected current and voltage at the grid-connected common coupling point are input as input quantities, and the hidden layer outputs the current and voltage resonant components at the current moment. The receiving layer receives feedback signals from the hidden layer to memorize the output value of the neurons in the hidden layer at the previous moment. The output of the neurons in the receiving layer is delayed and stored, and then fed back to the hidden layer. Finally, the ideal tracking value is output through the output layer as the virtual resistance output for adaptive adjustment. The purely resistive compensation link of the virtual impedance is set to compensate the characteristics of the total output impedance of the grid-type converter at medium and high frequencies according to the following compensation function: ； In the formula, k ui is the integral coefficient of the voltage control regulator, L 1 is the inductance of the grid-type converter LC filter, s represents the Laplace operator Among them, the second-order Pade approximate discrete domain function is used to digitally implement the compensation function, specifically: ； Among them, the second-order Pade approximation discrete domain function expression is: ； In the formula, s is the Laplace operator, z is the representation of the transfer function in the discrete domain, and can be converted into the Laplace operator s in the continuous domain.

2. The adaptive optimization control system of the active damper of the grid-type converter according to claim 1, characterized in that: The Elman neural network model uses the following method to determine the number of hidden layer neurons: k : ； In the formula, m Indicates the number of input quantities; n Represents the number of neurons in the output layer; δ Represents a constant, with a value of 1 to 10.

3. The adaptive optimization control system of the active damper of the grid-type converter according to claim 1, characterized in that: The Elman neural network model is trained using the Levenberg-Marquardt algorithm.

4. The adaptive optimization control system of the active damper of the grid-type converter according to claim 1, characterized in that: The control equations of the active and reactive loop control links of the virtual synchronous machine control module are expressed as follows: ； in, J is the moment of inertia, ω is the angular frequency of the virtual synchronous machine output, ω ref is the virtual angular velocity command value, P ref and Q ref are the active power and reactive power command values ​​respectively, P e and Q e are the actual values ​​of grid-connected active power and reactive power, K is the reactive integrator gain, θ is the inner loop control reference phase angle output by the virtual synchronous machine, U ref It is the grid connection point voltage reference value.

5. The adaptive optimization control system of the active damper of the grid-type converter according to claim 1, characterized in that: In the voltage-current dual closed-loop control module, the output voltage and current control regulator expressions of the voltage-current dual closed-loop control are respectively: ； ； In the formula, k up and k ui are the proportional coefficient and integral coefficient of the voltage control regulator, k ip is the proportionality factor of the current control regulator, s is the Laplace operator.

6. An adaptive optimization control method for a grid-type converter active damper based on the adaptive optimization control system of the grid-type converter active damper according to any one of claims 1 to 5, characterized in that: The method comprises: Collect the voltage and current collected at the grid common coupling point, and calculate the actual value of grid active power and reactive power based on it; The actual value and command value of the grid-connected active power are used as input, and the active power control link of the virtual synchronous machine control module is combined with the angular frequency output by the virtual synchronous machine. ω With the virtual angular velocity command value ω ref Output inner loop control reference phase angle θ ; The actual value and command value of the grid-connected reactive power are used as input, and the reactive power control link of the virtual synchronous machine control module is combined with the grid-connected point voltage reference value. U ref Output inner loop control reference voltage U ; According to the current at the grid common coupling point i pcc and voltage u pcc , output virtual resistance through virtual resistance adaptive regulator R v , and the virtual resistor R v The resonant voltage component is obtained by sending it into the purely resistive compensation link of the virtual impedance; The resonant voltage component and the inner loop control reference voltage output by the virtual synchronous machine control module U Adding the virtual impedance control output d The axis reference voltage, 0 value is subtracted from the resonant voltage component to obtain the virtual impedance control output q axis reference voltage; the virtual resistance adaptive regulator adopts an Elman neural network model to achieve adaptive adjustment output of the virtual resistance; The virtual impedance control output d Axis and q The reference voltage of the axis and the collected d Axis and q The filter capacitor voltage on the shaft is input into the voltage closed-loop controller and output after closed-loop control d Axis and q On-axis filter inductor reference current; After closed loop control, the output d Axis and q On-axis filter inductor reference current and collected d Axis and q The filter inductor current on the axis is input to the current control regulator, and the modulated voltage is output after closed-loop control. u s ; The modulation voltage u s Input coordinate transformation module to realize coordinate transformation from dq0 axis to abc axis; The coordinates of the abc axis are input into the SVPWM modulator, and the drive signal for controlling the power switch tube is generated according to the coordinates of the abc axis. k pwm。

Images