Power grid control method based on neural network and current synchronization UI droop coefficient

Abstract

The disclosed embodiment provides a grid control method based on a neural network and a current-synchronized U-I droop coefficient, which relates to the field of grid control technology. This method controls the inverter of the distribution substation microgrid by designing a voltage loop and a current loop, and controls the q-axis current through an RBF neural network, thereby achieving consistency in the output current phase angle of each distributed generation unit DG and suppressing the circulation problem between the DG units in the substation. In addition, a dual current inner loop control of the output current and the grid current is adopted, and both the output current outer loop and the filter capacitor current inner loop adopt a quasi-PR current control strategy. This achieves effective tracking of a given current, avoids the problems of poor anti-interference performance and tracking performance of high-frequency signals of traditional PI controllers, improves the stability of the system, ensures the synchronous grid-connected operation of the distribution substation inverter, effectively suppresses harmonics, and improves the operating efficiency and power quality of the system.

Description

Technical Field

[0001] The present disclosure relates to the field of power grid control technology, and in particular to a power grid control method based on a neural network and a current synchronization UI droop coefficient. Background Art

[0002] Large-scale renewable energy poses a huge challenge to the safe and stable operation of power systems and distribution substations. Due to the low inertia of distributed power systems composed of a high proportion of renewable energy and inverters, traditional phasor synchronous control strategies have problems such as power-angle stability, multi-substation interconnection control, and adaptation to diverse loads. Droop control is mainly used in microgrids. Since line impedance is difficult to determine due to various factors, traditional UI droop control is ineffective. At the same time, with the development of power electronics technology, the number of nonlinear elements in power systems has gradually increased, resulting in a gradual decline in power quality and a significant increase in harmonic content. Among them, the harmonics of 6n±1 (n is a natural number) are particularly significant. The presence of harmonic currents not only affects UI droop control, but also affects the distribution of active power and reactive power. Therefore, the suppression of harmonic currents is particularly important.

[0003] Among the existing technical solutions, some distribute active power and reactive power according to the inverter capacity ratio. Although this solves the power angle stability problem, the system output voltage is seriously distorted under nonlinear load conditions. Some evenly distribute reactive power by adjusting the droop coefficient. However, a large droop coefficient will cause a large deviation between the inverter output voltage and the bus rated voltage, causing voltage drop and affecting system stability. Some PI controllers cannot achieve precise control of harmonics.

[0004] Therefore, it is difficult to achieve synchronous grid-connected operation and harmonic suppression of inverters in distribution substations with current technical solutions. Summary of the Invention

[0005] The present disclosure proposes a power grid control method, device, equipment and storage medium based on a neural network and a current synchronous UI droop coefficient.

[0006] To solve the above technical problems, the technical solutions adopted by this disclosure are as follows:

[0007] According to a first aspect of the present disclosure, a power grid control method based on a neural network and a current synchronization UI droop coefficient is provided, the method comprising:

[0008] Use synchronous timing phasor measurement units to synchronously measure the voltage and current data of each node of distributed power supply units in multiple substations;

[0009] The current is synchronously controlled using a radial basis function (RBF) neural network. The expression for current control is:

[0010]

[0011] Among them, k p and k i are the proportional and integral parameters of current synchronous control respectively. In RBF neural network control, x=[x c ] T is the network input, x c is the input of the cth neuron, [x c ] T is the transposed matrix of the input of the c-th neuron, y=[y c ] T is the output of the hidden layer of the network, y c is the output of the cth neuron in the hidden layer, [y c ] T is the transposed matrix of the output of the cth neuron in the hidden layer;

[0012] The weights of the RBF neural network are w=[w1,w2,w3,...,w m ] T , m is the number of inputs, T is the calculation of the transposed matrix, and the output of the RBF neural network is:

[0013] y t =w T x=w1x1+w2x2+w3x3+...+w m x m ,

[0014] Through the RBF neural network algorithm, the q-axis current i q To a given value i qref Approximation is performed until the error is eliminated;

[0015] When the system operation state suddenly changes, i qref -i q ≠0, the current synchronization control module participates in the regulation, and the current adopts RBF neural network for current synchronization control, and continuously adjusts θ based on the current control expression. ref The size of the q =0, the system reaches a stable state after adjustment;

[0016] An inductor-capacitor-inductor (LCL) filter is used to filter the high-frequency power supply units in the stable substation area before connecting them to the grid or load. The harmonic compensation control strategy adopts a double closed-loop structure with an output current or grid current outer loop and a filter capacitor current inner loop. Both the output current outer loop and the filter capacitor current inner loop adopt a quasi-proportional-resonant PR current control strategy.

[0017] The grid-connected current outer loop is composed of a proportional control branch, a fundamental wave control branch, and a harmonic control branch in parallel, wherein the harmonic suppression branch is used to suppress each harmonic component. When the current error is greater than the threshold, the 5th, 7th, 11th, and 13th harmonics with greater harmonics are preferentially suppressed.

[0018] In some implementations of the first aspect, the method further includes:

[0019] When I i >(I i +I j ) / 2, indicating that the equivalent impedance Z i <Z j , the current deviation is positive, and a droop parameter adjustment value greater than 0 is obtained through the PI regulator, which increases the droop parameter of the converter and improves the equivalent impedance Z of the converter. i , and finally achieve Z i =Z j , to achieve the current sharing effect; when I i <(I i +I j ) / 2, indicating that the equivalent impedance Z i >Z j , the current deviation is negative, and a droop parameter adjustment value less than 0 is obtained through the PI regulator, which reduces the droop parameter of the converter and reduces the equivalent impedance Z of the converter. i , and finally achieve Z i =Z j ; The droop coefficient after adaptive adjustment is γ i * =γ i +σ i As long as there is a current deviation ΔI, the droop coefficient will be continuously adjusted until ΔI is zero, where I i is the current value related to node i, I j is the current value associated with node j, γ i is the original droop coefficient of node i, σ i is the corresponding correction coefficient of node i.

[0020] In some implementations of the first aspect, the grid current outer loop corresponds to two working modes. When the input current I h is the load current I L When the input current I h When it is 0, the system harmonics are controlled and suppressed.

[0021] In some implementations of the first aspect, both the output current outer loop and the filter capacitor current inner loop adopt a quasi-PR current control strategy, and the transfer function of the quasi-PR controller is expressed as:

[0022]

[0023] Among them, K p is the proportional gain, k nr is the resonant gain, ω nc is the cutoff frequency, ω n is the natural angular frequency, and s is a complex variable.

[0024] In some implementations of the first aspect, the current error signal of the current outer loop is summed through proportional control and proportional R control of each harmonic, and each separated harmonic is individually R-controlled. The harmonic controller only performs steady-state gain control on integer fundamental frequencies.

[0025] Among them, the expressions of the fundamental wave control branch and the harmonic wave control branch are:

[0026]

[0027] Among them, K p is the proportional gain, k nr is the resonant gain, ω nc is the cutoff frequency, ω n is the natural angular frequency, and s is a complex variable.

[0028] In some implementations of the first aspect, the method further includes:

[0029] A neural network algorithm is used to extract the harmonic components of the system, where the harmonics in the system are expressed as:

[0030]

[0031] Where t represents the time variable, b is the fundamental component, w is the angular frequency, w = 2πf, f is the frequency of the signal, ε is the harmonic order, A ε , ψ ε are the amplitude and phase of the ε-th harmonic respectively.

[0032] In some implementations of the first aspect, in the neural network model, the input matrix is ​​represented as:

[0033] C=[c1(t),c2(t),c3(t),L,c 2n (t)]=[sin(wt),cos(wt),sin(2wt),cos(2wt),L,sin(nwt),cos(nwt)],

[0034] The weight matrix is ​​expressed as:

[0035] W=[w1(t),w2(t),w3(t),L,w 2n (t)]=[A1 sinψ1,A1cosψ1,A2 sinψ2,A2 cosψ2,L,A n sinψ n ,A n cosψ n ],

[0036] The output vector of the neural network is the product of the input matrix and the weight matrix Get the error from the neural network model y(t) is the theoretical calibration value. After the error e(t) is processed by the least mean square (LMS) algorithm, the weight matrix W is readjusted based on the principle of minimizing the error target G. The error objective function is expressed as:

[0037]

[0038] Each harmonic is extracted through the neural network model, and the amplitude and phase of each harmonic are expressed as:

[0039]

[0040] According to a second aspect of the present disclosure, a power grid control device based on a neural network and a current synchronization UI droop coefficient is provided, the device comprising:

[0041] The synchronous timing phasor measurement unit is used to synchronously measure the voltage and current data of each node of the distributed power supply units in multiple substations;

[0042] The current is synchronously controlled using a radial basis function (RBF) neural network. The expression for current control is:

[0043]

[0044] Among them, k p and k i are the proportional and integral parameters of current synchronous control respectively. In RBF neural network control, x=[x c ] T is the network input, x c is the input of the cth neuron, [x c ] T is the transposed matrix of the input of the c-th neuron, y=[y c ] T is the output of the hidden layer of the network, y c is the output of the cth neuron in the hidden layer, [y c ]T is the transposed matrix of the output of the cth neuron in the hidden layer;

[0045] The weights of the RBF neural network are w=[w1,w2,w3,...,w m ] T , m is the number of inputs, T is the calculation of the transposed matrix, and the output of the RBF neural network is:

[0046] y t =w T x=w1x1+w2x2+w3x3+...+w m x m ,

[0047] Through the RBF neural network algorithm, the q-axis current i q To a given value i qref Approximation is performed until the error is eliminated;

[0048] The current synchronization control module is used when the system operating state suddenly changes and i qref -i q ≠0, the current is controlled by RBF neural network, and θ is continuously adjusted based on the current control expression. ref The size of the q =0, the system reaches a stable state after adjustment;

[0049] The inductor-capacitor-inductor (LCL) filter is used to filter high frequencies of distributed power supply units in a stable substation before connecting them to the grid or load. The harmonic compensation control strategy adopts a dual closed-loop structure with an output current or grid current outer loop and a filter capacitor current inner loop. Both the output current outer loop and the filter capacitor current inner loop adopt a quasi-proportional-resonant PR current control strategy.

[0050] Among them, the grid-connected current outer loop is composed of a proportional control branch, a fundamental wave control branch, and a harmonic control branch in parallel, among which the harmonic suppression branch is used to suppress each harmonic component. When the current error is greater than the threshold, the 5th, 7th, 11th, and 13th harmonics with greater harmonics are preferentially suppressed.

[0051] According to a third aspect of the present disclosure, an electronic device is provided, including:

[0052] at least one processor;

[0053] and a memory communicatively coupled to the at least one processor;

[0054] It is characterized in that the memory stores instructions that can be executed by at least one processor, and the instructions are executed by at least one processor to implement the method of the first aspect of the present disclosure.

[0055] According to a fourth aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the method according to the first aspect of the present disclosure is implemented.

[0056] The beneficial effects of the grid control method, device, equipment and storage medium provided by the present disclosure based on neural network and current synchronization UI droop coefficient are as follows: by designing the voltage loop and the current loop to jointly control the inverter of the distribution substation microgrid, and controlling the q-axis current through the RBF neural network, the consistency of the output current phase angle of each distributed generation unit DG is achieved, and the circulation problem between the DG units in the substation is suppressed. In addition, dual current inner loop control of output current and grid current is adopted, and the output current outer loop and the filter capacitor current inner loop both adopt quasi-PR current control strategy. Effective tracking of the given current is achieved, avoiding the problems of poor anti-interference performance and tracking performance of high-frequency signals of the traditional PI controller, improving the stability of the system, ensuring the synchronous grid-connected operation of the distribution substation inverter, and effectively suppressing harmonics, thereby improving the operating efficiency and power quality of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the following briefly introduces the drawings required for use in the embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0058] Figure 1 A flow chart of a power grid control method based on a neural network and a current synchronization UI droop coefficient according to an embodiment of the present disclosure is shown;

[0059] Figure 2 A microgrid structure diagram of a distribution station area provided by an embodiment of the present disclosure is shown;

[0060] Figure 3 A control block diagram provided by an embodiment of the present disclosure is shown;

[0061] Figure 4 A control block diagram of an inverter harmonic compensation system provided by an embodiment of the present disclosure is shown;

[0062] Figure 5 A block diagram of current synchronization control based on an RBF neural network provided by an embodiment of the present disclosure is shown;

[0063] Figure 6 A control block diagram of an adaptive droop coefficient provided by an embodiment of the present disclosure is shown;

[0064] Figure 7 A model diagram of harmonic extraction based on a neural network provided by an embodiment of the present disclosure is shown;

[0065] Figure 8 A block diagram of a power grid control device based on a neural network and a current synchronization UI droop coefficient provided by an embodiment of the present disclosure is shown;

[0066] Figure 9 A block diagram of an exemplary electronic device capable of implementing embodiments of the present disclosure is shown. DETAILED DESCRIPTION

[0067] The features and exemplary embodiments of various aspects of the present disclosure will be described in detail below. In order to make the purposes, technical solutions and advantages of the present disclosure clearer, the present disclosure will be further described in detail below in conjunction with the accompanying drawings and Examples. It should be understood that the specific embodiments described herein are only configured to explain the present disclosure and are not configured to limit the present disclosure. For those skilled in the art, the present disclosure can be implemented without the need for some of these specific details. The following description of the embodiments is merely to provide a better understanding of the present disclosure by illustrating examples of the present disclosure.

[0068] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, the elements defined by the phrase "comprising..." do not exclude the presence of other identical elements in the process, method, article, or device comprising the elements.

[0069] In the present disclosure, in order to solve the above technical problems, the voltage loop and the current loop are designed to jointly control the inverter of the distribution substation microgrid, and the q-axis current is controlled by the radial basis function (RBF) neural network, so as to achieve the consistency of the output current phase angle of each distributed generation unit (DG), and suppress the circulation problem between the DG units in the substation. In addition, a dual current inner loop control of the output current and the grid current is adopted, and the output current outer loop and the filter capacitor current inner loop both adopt a quasi-proportional-resonant (PR) current control strategy. Effective tracking of the given current is achieved, avoiding the problems of poor anti-interference performance and tracking performance of high-frequency signals of the traditional proportional-integral (PI) controller, improving the stability of the system, ensuring the synchronous grid-connected operation of the distribution substation inverter, and effectively suppressing harmonics, thereby improving the operating efficiency and power quality of the system. The technical solution provided by the embodiment of the present disclosure is described below in conjunction with the accompanying drawings.

[0070] Figure 1 A flow chart of a power grid control method based on a neural network and a current synchronization UI droop coefficient according to an embodiment of the present disclosure is shown.

[0071] like Figure 1 As shown, the grid control method based on neural network and current synchronous UI droop coefficient may include:

[0072] S101, using a synchronous timing phasor measurement unit to synchronously measure the voltage and current data of each node of the distributed power supply units in multiple substations;

[0073] After the phasor measurement unit obtains the precise time signal from the central clock source, the sampling frequency is determined. The sampling frequency can be directly obtained from the database (different sampling frequencies can be set for different substations, such as small photovoltaic substations: sampling frequency 1000Hz; large wind power substations: sampling frequency 2000Hz to 5000Hz; complex substations (including energy storage and photovoltaics): sampling frequency 2000Hz to 5000Hz; key fault detection substations: sampling frequency 5000Hz or higher).

[0074] Based on the determined sampling frequency, multiple substations are synchronously measured to obtain the voltage and current data of each node of the distributed power supply unit. In order to ensure that the phasor measurement unit can accurately obtain the voltage and current data, it is necessary to ensure that the state of the phasor measurement unit is stable. The stability of the phasor measurement unit can be evaluated through various physical data. Specifically, the response time T of each phasor measurement unit at each node of the distributed power supply unit can be obtained. x, synchronization accuracy B x And phasor measurement error TVE (one or more phasor measurement units can be deployed on the power unit node. If multiple phasor measurement units are deployed, the voltage and current data of each node finally uploaded to the data center are the average values ​​measured by multiple phasor measurement units), where T x It is the time from when the phasor measurement unit receives the instruction to collect voltage and current data to when it starts collecting data (in ms). x The absolute value of the difference between the actual time of the phasor measurement unit and the standard time of the central clock source (in μs) is used. The calculation of TVE is based on the actual measured voltage or current amplitude (in this embodiment, the actual measured voltage amplitude V is selected). m , in V) and the measured phase angle θ m (unit is angle) is calculated, the specific formula is:

[0075]

[0076] Among them, V t is the true amplitude of the current or voltage. In this embodiment, the true amplitude of the voltage is selected (ideally, the precise value of the voltage is measured by high-precision equipment under laboratory conditions and used as a comparison benchmark). t is the true phase angle (ideally, the exact value of the phase angle, which is measured by high-precision equipment under laboratory conditions and used as a comparison benchmark).

[0077] TVE combines amplitude error and phase angle error, using a square root of the sum of squares formula to comprehensively assess the overall measurement error of a phasor measurement unit. A smaller TVE indicates that the PMU's measurement is closer to the true value, and the measurement stability and accuracy are higher. TVE is typically expressed as a percentage. For example, a TVE of 1% indicates a combined error of 1% between the measurement result and the true value. Under steady-state conditions, a TVE of less than 1% is generally required; under dynamic conditions, a TVE of no more than 3% is typically permitted to ensure that the PMU can accurately monitor changes in the power system.

[0078] T x 、B x And the analysis basis of TVE as the state of the phasor measurement unit, specifically, based on the temperature drift ΔV of the environment in which the phasor measurement unit is located temp (ΔV temp=S*ΔT, where S is the temperature sensitivity coefficient, which in this embodiment is expressed as the voltage change per degree Celsius, with the unit of V / °C, and ΔT is the change in ambient temperature, usually expressed in degrees Celsius) to determine the TVE allowable deviation ΔTVE (the temperature drift is compared with each temperature drift interval stored in the database to determine the temperature drift interval corresponding to the calculated temperature drift, thereby obtaining the ΔTVE corresponding to the temperature drift interval from the database). The stability evaluation value can be calculated using the following formula:

[0079]

[0080] Where a is T stored in the database x The weight factor, b is the B stored in the database x The weight factor of TVE is stored in the database, c is the weight factor of TVE deviation, T c To request the response time, B c To require synchronization accuracy, C TVE is the required phasor measurement error, and δ is the stability evaluation value.

[0081] When the stability assessment value is greater than the required threshold, it indicates that the current state of the phasor measurement unit is abnormal and an early warning should be issued. If only one phasor measurement unit is deployed on the power supply unit node, then while issuing an early warning for the phasor measurement unit, an early warning should also be issued that the voltage and current data of the node cannot be measured by the phasor measurement unit. If multiple phasor measurement units are deployed, voltage and current data can be collected through other phasor measurement units whose stability assessment values ​​are not greater than the required threshold. If and only if all phasor measurement units are abnormal, an early warning should be issued that the voltage and current data of the node cannot be measured by the phasor measurement unit.

[0082] For a node with multiple phasor measurement units deployed, if all phasor measurement units of the node are normal within an evaluation cycle, then the actually measured voltage amplitude in the above TVE calculation formula is replaced by the voltage average of multiple phasor measurement units, and the response time and synchronization accuracy in the stability evaluation value are replaced by the average response time and average synchronization accuracy. The calculated stability evaluation value is recorded as the averaged stability evaluation value.

[0083] If and only if the averaged stability evaluation value is greater than the required threshold, an early warning of a phasor measurement unit abnormality is issued, and the stability evaluation value is used to check each unit one by one to determine the faulty phasor measurement unit, and other normal phasor measurement units are used to collect voltage and current data.

[0084] And in the next evaluation cycle, all phasor measurement units of the node are evaluated one by one to see if they are normal. If they are all normal, the averaged stability evaluation value is used to perform the evaluation in the next cycle.

[0085] S102, current synchronous control using radial basis function (RBF) neural network;

[0086] Among them, the expression of current control is:

[0087]

[0088] Among them, k p and k i are the proportional and integral parameters of current synchronous control respectively. In RBF neural network control, x=[x c ] T is the network input, x c is the input of the cth neuron, [x c ] T is the transposed matrix of the input of the c-th neuron, y=[y c ] T is the output of the hidden layer of the network, y c is the output of the cth neuron in the hidden layer, [y c ] T is the transposed matrix of the output of the cth neuron in the hidden layer;

[0089] The weights of the RBF neural network are w=[w1,w2,w3,...,w m ] T , m is the number of inputs, T is the calculation of the transposed matrix, and the output of the RBF neural network is:

[0090] y t =w T x=w1x1+w2x2+w3x3+...+w m x m ,

[0091] Through the RBF neural network algorithm, the q-axis current i q To a given value i qref Approximation is performed until the error is eliminated;

[0092] S103, when the system running state suddenly changes, i qref -i q ≠0, the current synchronization control module participates in the regulation, and the current uses the RBF neural network for current synchronization control, and continuously adjusts θ based on the current control expression. ref The size of the q=0, the system reaches a stable state after adjustment;

[0093] S104: Use an inductor-capacitor-inductor (LCL) filter to perform high-frequency filtering on the distributed power supply units in the stable substation area, and then connect them to the grid or load. The harmonic compensation control strategy adopts a current double closed-loop structure with an output current or grid current outer loop and a filter capacitor current inner loop. Both the output current outer loop and the filter capacitor current inner loop adopt a quasi-proportional-resonant PR current control strategy.

[0094] Among them, the grid-connected current outer loop is composed of a proportional control branch, a fundamental wave control branch, and a harmonic control branch in parallel, among which the harmonic suppression branch is used to suppress each harmonic component. When the current error is greater than the threshold, the 5th, 7th, 11th, and 13th harmonics with greater harmonics are preferentially suppressed.

[0095] The grid control method based on neural network and current synchronization UI droop coefficient provided by the present disclosure jointly controls the inverter of the distribution substation microgrid by designing voltage loop and current loop, and controls the q-axis current through RBF neural network, thereby achieving consistency of the output current phase angle of each distributed generation unit DG and suppressing the circulation problem between the DG units in the substation. In addition, dual current inner loop control of output current and grid current is adopted. The output current outer loop and the filter capacitor current inner loop both adopt quasi-PR current control strategy, which realizes effective tracking of the given current, avoids the problems of poor anti-interference performance and tracking performance of high-frequency signals of traditional PI controllers, improves the stability of the system, ensures the synchronous grid-connected operation of the distribution substation inverter, effectively suppresses harmonics, and improves the operating efficiency and power quality of the system.

[0096] When I i >(I i +I j ) / 2, indicating that the equivalent impedance Z i <Z j , the current deviation is positive, and a droop parameter adjustment value greater than 0 is obtained through the PI regulator, which increases the droop parameter of the converter and improves the equivalent impedance Z of the converter. i , and finally achieve Z i =Z j , to achieve the current sharing effect; when I i <(I i +I j ) / 2, indicating that the equivalent impedance Z i >Z j , the current deviation is negative, and a droop parameter adjustment value less than 0 is obtained through the PI regulator, which reduces the droop parameter of the converter and reduces the equivalent impedance Z of the converter. i , and finally achieve Z i =Z j; The droop coefficient after adaptive adjustment is γ i * =γ i +σ i As long as there is a current deviation ΔI, the droop coefficient will be continuously adjusted until ΔI is zero, where I i is the current value related to node i, I j is the current value associated with node j, γ i is the original droop coefficient of node i, σ i is the corresponding correction coefficient of node i.

[0097] In some embodiments, the above-mentioned grid current outer loop corresponds to two working modes. When the input current I h is the load current I L When the input current I h When it is 0, the system harmonics are controlled and suppressed.

[0098] In some embodiments, both the output current outer loop and the filter capacitor current inner loop adopt a quasi-PR current control strategy. The transfer function of the quasi-PR controller is expressed as:

[0099]

[0100] Among them, K p is the proportional gain, k nr is the resonant gain, ω nc is the cutoff frequency, ω n is the natural angular frequency, and s is a complex variable.

[0101] In some embodiments, the current error signal of the current outer loop is summed through proportional control and proportional R control of each harmonic, and each separated harmonic is subjected to independent R control. The harmonic controller only performs gain steady-state control on integer fundamental frequencies.

[0102] Among them, the expressions of the fundamental wave control branch and the harmonic wave control branch are:

[0103]

[0104] Among them, K p is the proportional gain, k nr is the resonant gain, ω nc is the cutoff frequency, ω n is the natural angular frequency, and s is a complex variable.

[0105] In some embodiments, Figure 1 The method may further include:

[0106] A neural network algorithm is used to extract the harmonic components of the system, where the harmonics in the system are expressed as:

[0107]

[0108] Where t represents the time variable, b is the fundamental component, w is the angular frequency, w = 2πf, f is the frequency of the signal, ε is the harmonic order, A ε , ψ ε are the amplitude and phase of the ε-th harmonic respectively.

[0109] In some embodiments, in a neural network model, the input matrix can be represented as:

[0110] C=[c1(t),c2(t),c3(t),L,c 2n (t)]=[sin(wt),cos(wt),sin(2wt),cos(2wt),L,sin(nwt),cos(nwt)],

[0111] The weight matrix is ​​expressed as:

[0112] W=[w1(t),w2(t),w3(t),L,w 2n (t)]=[A1 sinψ1,A1cosψ1,A2 sinψ2,A2 cosψ2,L,A n sinψ n ,A n cosψ n ],

[0113] The output vector of the neural network is the product of the input matrix and the weight matrix Get the error from the neural network model y(t) is the theoretical calibration value. After the error e(t) is processed by the least mean square (LMS) algorithm, the weight matrix W is readjusted based on the principle of minimizing the error target G. The error objective function is expressed as:

[0114]

[0115] Each harmonic is extracted through the neural network model, and the amplitude and phase of each harmonic are expressed as:

[0116]

[0117] Figure 2 A microgrid structure diagram of a distribution station area provided according to an embodiment of the present disclosure is shown.

[0118] like Figure 2As shown in the figure, the distribution substation microgrid is composed of distributed power sources (DGs) and their corresponding loads. Each distributed power source (DG) is connected to the DC bus through a power electronic inverter and an LCL filter. Under the control of a fixed-frequency synchronization signal, it is connected to the common bus (Point of Common Coupling, PCC) to supply power to local loads or be incorporated into the AC power grid through the substation distribution transformer.

[0119] Since the lines connected to each module are inconsistent, the difference in line parameters will cause the phase angle of the output current between the DG units in each substation to be different. The deviation of the phase angle will cause the circulation between each substation, reducing the operation efficiency of the entire microgrid. To solve the above problems, this paper proposes a current synchronization UI droop control strategy based on RBF neural network and current synchronization UI amplitude adaptive droop coefficient control. The control block diagram is shown in the figure. Figure 3 As shown in the figure, a single distributed power source DG1 is analyzed. d 、i q is the dq axis current component of the output current of the DG1 unit in the station area; i 01 is the output current of DG1 unit; u 01 、U 01 are the no-load output voltage and amplitude of DG1 respectively; u PCC is the connection point voltage, γ i * is the adaptive droop control coefficient, AC represents the AC grid, θ ref 、U ref are the phase angle and amplitude reference values ​​of the output voltage, u dref 、u qref Voltage reference value of the dq axis of the voltage loop.

[0120] This control strategy mainly consists of three parts: a voltage-current dual closed-loop control section, a current synchronization control module, and a UI amplitude droop control module. In the current synchronization control module, an RBF neural network is used to approximate the q-axis current to eliminate errors. Based on this control module, the phase angles of the output currents of each DG unit are made consistent, thereby suppressing the circulating current between the DGs in each substation, effectively improving the power distribution between the DGs, improving the control performance of the system, and improving the operating efficiency. In the UI amplitude droop control module, since the line impedance in actual applications is difficult to determine due to various factors, there are significant drawbacks in configuring the droop coefficient based solely on the capacity of each DG in the substation. Therefore, this paper adaptively adjusts the droop coefficient to achieve power balancing and reasonable distribution of active and reactive power.

[0121] The harmonic compensation control strategy of the DG inverter in the distribution area is explained by taking the DG1 unit in the distribution area as an example. Figure 4The control block diagram of the inverter harmonic compensation system designed in this paper is given. The figure shows the current double inner loop control structure diagram of harmonic suppression, in which the output value of the voltage loop is given by Figure 3 The voltage loop in the .

[0122] The inverter first uses LCL filter for high frequency filtering, and then connects to the grid or load. The control strategy of harmonic compensation uses the output current (or grid current i1) outer loop and the filter capacitor current (i C ) The current double closed-loop structure of the inner loop. Both the output current outer loop and the filter capacitor current inner loop adopt the quasi-PR current control strategy. It can achieve effective tracking of the given current, avoid the problems of poor anti-interference performance and tracking performance of high-frequency signals of the traditional PI controller, and improve the stability of the system. Among them, the filter capacitor current inner loop can suppress the peak resonance in the LCL and improve the reliability and stability of the DG unit. The composition of the grid current outer loop is different from the traditional harmonic detection circuit. It is composed of a proportional control branch, a fundamental control branch, and a harmonic control branch in parallel. The harmonic suppression branch can suppress each harmonic component. When the current error is large, it can be selected to give priority to suppressing the more harmful 5th, 7th, 11th, and 13th harmonics, corresponding to the switching quantities K5, K7, and K 11 , K 13 At the same time, this paper designs a method for extracting harmonic components based on neural network, which has a simple structure and fast dynamic response. In addition, the outer loop of the grid current is designed with two working modes, corresponding to the input current I h Two different values. When the input current I h is the load current I L When the input current I h When it is 0, the main focus is on controlling and suppressing the system's harmonics to improve the current quality of the inverter.

[0123] In order to solve the problems of synchronous grid-connected operation and harmonic suppression of distribution substation inverters, this paper adopts a voltage and current dual closed-loop control strategy for distribution substation microgrid inverters based on synchronous timing and current synchronization UI amplitude droop control scheme. The current synchronization control uses RBF neural network to approximate the q-axis current to eliminate errors. Based on this control module, the phase angle of the output current of each DG unit is made consistent, thereby suppressing the circulating current between the DGs in each substation, effectively improving the power distribution between the DGs, improving the control performance of the system, and improving the operating efficiency. In UI amplitude droop control, since the line impedance in actual application is difficult to determine due to various factors, there are significant drawbacks in configuring the droop coefficient based solely on the capacity of each DG in the substation. Therefore, this paper adaptively adjusts the droop coefficient to achieve power balancing and reasonable distribution of active and reactive power with the adaptive droop coefficient. The output current of each DG unit is successfully synchronized, and the stable operation of the output voltage in both grid-connected and island modes is ensured.

[0124] To address current tracking, harmonic detection, and suppression, a harmonic compensation and suppression strategy that eliminates the need for harmonic detection is proposed. This strategy utilizes dual inner current loop control of both the output and grid currents, effectively achieving harmonic compensation and suppression. Both the output current outer loop and the filter capacitor current inner loop employ a quasi-PR current control strategy. This effectively tracks the given current, avoiding the poor interference rejection and high-frequency signal tracking performance of traditional PI controllers, thereby improving system stability. The grid current outer loop consists of a parallel combination of proportional control, fundamental control, and harmonic control. The harmonic control branch extracts harmonic components using a harmonic component extraction module, which suppresses each harmonic component by switching each harmonic branch on and off. Furthermore, a neural network-based harmonic component extraction method is designed, featuring a simple structure and fast dynamic response. Furthermore, two operating modes are designed for the grid current outer loop, allowing the system to flexibly select the operating mode based on the load capacity of the inverter in the substation. This effectively suppresses system harmonics, improves waveform quality, and enhances inverter efficiency.

[0125] Under the control strategy proposed in this article, the grid-connected inverter in the distribution station area can obtain stable and reliable output voltage regardless of whether it is grid-connected or islanded. The system can flexibly select the operating mode according to the load capacity. The total harmonic distortion rate of the voltage is kept within a small range, and the THD value is below 3%, effectively improving the waveform quality of the inverter.

[0126] In a specific example, since the lines connected to each module are inconsistent, the difference in line parameters will cause the phase angle of the output current between the DG units in each substation to be different. The deviation of the phase angle will cause the circulation between each substation, reducing the operation efficiency of the entire microgrid. To solve the above problem, this paper proposes a current synchronization UI droop control strategy based on RBF neural network and current synchronization UI amplitude adaptive droop coefficient control. The control block diagram is shown in the figure. Figure 3 As shown in the figure, a single distributed power supply DG1 is analyzed. The control strategy mainly consists of three parts: voltage and current double closed-loop control part, current synchronization control module and UI amplitude droop control module. d 、i q is the dq axis current component of the output current of the DG1 unit in the station area; i 01 is the output current of DG1 unit; u 01 、U 01 are the no-load output voltage and amplitude of DG1 respectively; u PCC is the connection point voltage, γ i * is the adaptive droop control coefficient, AC represents the AC grid, θ ref 、U ref are the phase angle and amplitude reference values ​​of the output voltage, u dref 、u qref Voltage reference value of the dq axis of the voltage loop.

[0127] This design uses PMU (phasor measurement unit) control with satellite-synchronized timing to synchronize voltage and current data at each node in the DG unit area, obtaining synchronized voltage and current status information for the entire area. By reading PMU data from each node, the central controller obtains real-time information on the entire area's line status, eliminating the need for complex power flow calculations. This improves the reliability of the area's distribution network control, facilitates real-time adjustments to the operating status of DG units, simplifies system scheduling and control complexity, and enhances response speed and stability.

[0128] Since the phase angles of the output currents of the various DG units are consistent, that is, the phase angles of the output currents of the various DG units relative to the d-axis are 0, the q-axis current reference signal i qref =0 is set as the control target, and the current is synchronously controlled using the RBF neural network. The expression of current control is:

[0129]

[0130] In the formula, k p and k i are the proportional and integral parameters of current synchronous control respectively. In RBF neural network control, x=[x c] T is the network input, x c is the input of the cth neuron, [x c ] T is the transposed matrix of the input of the c-th neuron, y=[y c ] T is the output of the hidden layer of the network, y c is the output of the cth neuron in the hidden layer, [y c ] T is the transposed matrix of the output of the cth neuron in the hidden layer. The weights of the RBF neural network are defined as w=[w1,w2,w3,...,w m ] T , so the output of the RBF neural network is:

[0131] y t =w T x=w1x1+w2x2+w3x3+...+w m x m ,

[0132] By approximating the q-axis current through the RBF neural network, the RBF neural network control can improve the dynamic response of the system and the fault tolerance of the control system. The current synchronization control block diagram based on the RBF neural network is shown in the figure. Figure 5 As shown. Through the RBF neural network algorithm in the figure, the q-axis current i q To a given value i qref Approximation is performed until all errors are eliminated and control performance is improved.

[0133] Since the DG units in each area are synchronized by Beidou satellite / GPS, the rising edge of the Beidou satellite / GPS timing pulse is required to be consistent with the d-axis of the two-phase rotating dq coordinate system. In order to synchronize the output current of each DG unit, the i of the DG units in the area is required to be q = 0. When the system operation state undergoes a sudden change, i qref -i q ≠0, the current synchronization control module participates in the regulation, and the current uses the RBF neural network for current synchronization control, and continuously adjusts θ through the current control expression. ref The size of the q =0.

[0134] When the system reaches a stable state after adjustment, i q When the control is adjusted to a given value of 0, the phase angle of the output current of each DG unit in the area is 0, and the loss generated in the line is ignored. i Active output power P of the unit i and reactive power Qi Can be obtained respectively:

[0135]

[0136] Where U PCC is the voltage amplitude corresponding to the common connection point, θ is the corresponding phase angle, I 0i It is a distributed power source DG in the distribution area i The output current amplitude of the unit.

[0137] The voltage vector relationship shows that when the system is in steady state, the phase angle of the output current of each DG unit in the substation is zero, and the output power factor of the DG unit is consistent with the load power factor. At this time, circulating current between the systems is almost non-existent, which can effectively improve the power distribution between DG units. In addition, because the current synchronization control module uses an RBF neural network control algorithm, it effectively improves the system's dynamic response and fault tolerance, while also reducing the impact of reference signal fluctuations on output voltage distortion.

[0138] The UI amplitude droop control module is based on the DG no-load output voltage amplitude U 01 The voltage amplitude reference signal is solved by the d-axis current amplitude to participate in the system regulation. The output of each DG unit in the substation is reasonably distributed according to the droop coefficient to ensure the stable operation of the output voltage. q =0, i d It is the power frequency component amplitude of the output current of the DG unit in the distribution station area. Therefore, the UI amplitude droop control can be designed as:

[0139] U ref =U 01 -γi d ,

[0140] Generally speaking, the output of each DG unit in the distribution area is distributed in inverse proportion to its capacity. According to the UI amplitude droop control formula, if the UI of each DG unit in the distribution area is 01 are exactly the same. When the influence of line impedance is ignored, the value of the droop coefficient should be inversely proportional to the current flowing through the line in each substation. However, in actual applications, the line impedance is difficult to determine due to various factors. Therefore, this paper makes an adaptive adjustment to the droop coefficient. Figure 6 The control block diagram of the adaptive droop coefficient is given. i >(I i +I j ) / 2, indicating that the equivalent impedance Z i <Z j , the current deviation is positive, and a droop parameter adjustment value greater than 0 is obtained through the PI regulator, which increases the droop parameter of the converter and improves the equivalent impedance Z of the converter. i, and finally achieve Z i =Z j , to achieve the current sharing effect. When I i <(I i +I j ) / 2, indicating that the equivalent impedance Z i >Z j , the current deviation is negative, and a droop parameter adjustment value less than 0 is obtained through the PI regulator, which reduces the droop parameter of converter i and reduces the equivalent impedance Z of the converter. i , and finally achieve Z i =Z j The droop coefficient after adaptive adjustment is γ i * =γ i +σ i As long as there is a current deviation ΔI, the droop coefficient will be continuously adjusted until ΔI is zero, where I i is the current value related to node i, I j is the current value associated with node j, γ i is the original droop coefficient of node i, σ i is the corresponding correction coefficient of node i. The introduction of the adaptive droop coefficient can reasonably divide the power of two different converter systems and achieve synchronous regulation.

[0141] Take the DG1 unit in the distribution area as an example. Figure 4 The control block diagram of the inverter harmonic compensation is given. First, an LCL filter is used for high-frequency filtering, and then connected to the grid or load. The control strategy of harmonic compensation uses the output current (or grid current i1) outer loop and the filter capacitor current (i C ) The current double closed-loop structure of the inner loop. Both the output current outer loop and the filter capacitor current inner loop adopt the quasi-PR current control strategy. It can achieve effective tracking of the given current, avoid the problems of poor anti-interference performance and tracking performance of high-frequency signals of the traditional PI controller, and improve the stability of the system. Among them, the filter capacitor current inner loop can suppress the peak resonance in the LCL and improve the reliability and stability of the DG unit. The composition of the grid current outer loop is different from the traditional harmonic detection circuit. It is composed of a proportional control branch, a fundamental control branch, and a harmonic control branch in parallel. The harmonic suppression branch can suppress each harmonic component. When the current error is large, it can be selected to give priority to suppressing the more harmful 5th, 7th, 11th, and 13th harmonics, corresponding to the switching quantities K5, K7, and K 11 , K 13 In addition, the outer loop of the grid current is designed with two working modes, corresponding to the input current I h Two different values. When the input current I h is the load current I LWhen the input current I h When it is 0, the main focus is on controlling and suppressing the system's harmonics to improve the current quality of the inverter.

[0142] Based on this design concept, both the output current outer loop and the filter capacitor current inner loop adopt a quasi-PR current control strategy. This effectively tracks the given current, avoiding the poor anti-interference performance and high-frequency signal tracking performance of traditional PI controllers, and improving system stability. The transfer function of the quasi-PR controller can be expressed as:

[0143]

[0144] Where K p is the proportional gain, k nr is the resonant gain, ω nc is the cutoff frequency, ω n is the natural angular frequency, and s is a complex variable.

[0145] Through the control box Figure 4 It can be seen that the current error signal of the current outer loop is summed through proportional control and R control of each harmonic, and each separated harmonic is individually R controlled. The harmonic controller only performs high-gain steady-state control on integer fundamental frequencies. The amplitude and phase margin of other frequency signals are low, thus suppressing the adverse effects of other frequencies and improving the control accuracy of each harmonic. The expressions of the fundamental control branch and the harmonic control branch are: Among them, K p is the proportional gain, k nr is the resonant gain, ω nc is the cutoff frequency, ω n is the natural angular frequency, and s is a complex variable.

[0146] This design can adaptively limit the current of harmonics by controlling the control switch K after each harmonic component extraction module. i To determine the compensation of each harmonic, in normal state or when the error is small, the switches of each harmonic branch are all closed, which can achieve full compensation of each harmonic. Set a limit value. If the current error signal is large, due to the error exceeding the limit, the 5th, 7th, 11th, and 13th harmonics in the output current of the DG inverter, which are more harmful, are compensated first. At this time, only K5, K7, and K 11 , K 13 Wait for the control switch to be closed.

[0147] In addition, the outer loop of the grid current is designed with two working modes, corresponding to the input current I hThe two different values ​​of the selector switch position can be flexibly selected according to the output capacity of the inverter, so as to flexibly suppress and compensate the harmonics of the system. Mode 1: When the load on the inverter is small, the selector switch is switched to mode 1, so that the system input current I h is the load current I L In this mode, the harmonic compensation of the load current can be realized directly without the need for a harmonic detection module, thus simplifying the control structure. Mode 2: When the inverter carries a large load, the switch is selected to switch to mode 2, so that the input current I h When the current error signal is 0, the current limiting of the harmonics is adaptively adjusted according to the value of the current error signal, and the harmonics of the system are flexibly controlled and suppressed, thereby improving the current quality of the inverter.

[0148] This design uses a linear neural network algorithm to extract the harmonic components of the system. The harmonics in the system can be expressed as:

[0149]

[0150] Where t represents the time variable, b is the fundamental component, w is the angular frequency, w = 2πf, f is the frequency of the signal, ε is the harmonic order, A ε , ψ ε are the amplitude and phase of the ε-th harmonic respectively.

[0151] Figure 7 The model diagram of harmonic extraction based on neural network is given. In the neural network model, the input matrix can be expressed as:

[0152] C=[c1(t),c2(t),c3(t),L,c 2n (t)]=[sin(wt),cos(wt),sin(2wt),cos(2wt),L,sin(nwt),cos(nwt)],

[0153] The weight matrix can be expressed as:

[0154] W=[w1(t),w2(t),w3(t),L,w 2n (t)]=[A1sinψ1,A1cosψ1,A2 sinψ2,A2cosψ2,L,A n sinψ n ,A n cosψ n ]

[0155] Therefore, the output vector of the neural network is the product of the input matrix and the weight matrix:

[0156]

[0157] The error can be obtained from the neural network model y(t) is the theoretical calibration value. After the error e(t) is processed by the LMS algorithm, the weight matrix W is readjusted according to the principle of minimizing the error target G. The error objective function can be expressed as:

[0158]

[0159] Therefore, the harmonics are extracted through the neural network model, and the amplitude and phase of each harmonic can be expressed as:

[0160]

[0161] The above is an introduction to the method embodiment. The following further illustrates the disclosed solution through an apparatus embodiment.

[0162] Figure 8 A block diagram of a power grid control device based on a neural network and a current synchronous UI droop coefficient according to an embodiment of the present disclosure is shown.

[0163] like Figure 8 As shown, the device includes:

[0164] The synchronous timing phasor measurement unit 801 is used to synchronously measure the voltage and current data of each node of the distributed power supply units in multiple substations;

[0165] The current is synchronously controlled using a radial basis function (RBF) neural network. The expression for current control is:

[0166]

[0167] Among them, k p and k i are the proportional and integral parameters of current synchronous control respectively. In RBF neural network control, x=[x c ] T is the network input, x c is the input of the cth neuron, [x c ] T is the transposed matrix of the input of the c-th neuron, y=[y c ] T is the output of the hidden layer of the network, y c is the output of the cth neuron in the hidden layer, [y c ] T is the transposed matrix of the output of the cth neuron in the hidden layer;

[0168] The weights of the RBF neural network are w=[w1,w2,w3,...,w m ] T, m is the number of inputs, T is the calculation of the transposed matrix, and the output of the RBF neural network is:

[0169] y t =w T x=w1x1+w2x2+w3x3+...+w m x m ,

[0170] Through the RBF neural network algorithm, the q-axis current i q To a given value i qref Approximation is performed until the error is eliminated;

[0171] The current synchronization control module 802 is used to control the current when the system operation state suddenly changes and i qref -i q ≠0, the current is controlled by RBF neural network, and θ is continuously adjusted based on the current control expression. ref The size of the q =0, the system reaches a stable state after adjustment;

[0172] The inductor-capacitor-inductor LCL filter 803 is used to perform high-frequency filtering on the distributed power supply units in the stable substation area before connecting to the grid or load. The control strategy for harmonic compensation adopts a current double closed-loop structure with an output current or grid current outer loop and a filter capacitor current inner loop. Both the output current outer loop and the filter capacitor current inner loop adopt a quasi-proportional-resonant PR current control strategy.

[0173] Among them, the grid-connected current outer loop is composed of a proportional control branch, a fundamental wave control branch, and a harmonic control branch in parallel, among which the harmonic suppression branch is used to suppress each harmonic component. When the current error is greater than the threshold, the 5th, 7th, 11th, and 13th harmonics with greater harmonics are preferentially suppressed.

[0174] In some embodiments, when I i >(I i +I j ) / 2, indicating that the equivalent impedance Z i <Z j , the current deviation is positive, and a droop parameter adjustment value greater than 0 is obtained through the PI regulator, which increases the droop parameter of the converter and improves the equivalent impedance Z of the converter. i , and finally achieve Z i =Z j , to achieve the current sharing effect; when I i <(I i +I j ) / 2, indicating that the equivalent impedance Z i >Zj , the current deviation is negative, and a droop parameter adjustment value less than 0 is obtained through the PI regulator, which reduces the droop parameter of the converter and reduces the equivalent impedance Z of the converter. i , and finally achieve Z i =Z j ; The droop coefficient after adaptive adjustment is γ i * =γ i +σ i As long as there is a current deviation ΔI, the droop coefficient will be continuously adjusted until ΔI is zero, where I i is the current value related to node i, I j is the current value associated with node j, γ i is the original droop coefficient of node i, σ i is the corresponding correction coefficient of node i.

[0175] In some embodiments, the above-mentioned grid current outer loop corresponds to two working modes. When the input current I h is the load current I L When the input current I h When it is 0, the system harmonics are controlled and suppressed.

[0176] In some embodiments, both the output current outer loop and the filter capacitor current inner loop adopt a quasi-PR current control strategy. The transfer function of the quasi-PR controller is expressed as:

[0177]

[0178] Among them, K p is the proportional gain, k nr is the resonant gain, ω nc is the cutoff frequency, ω n is the natural angular frequency, and s is a complex variable.

[0179] In some embodiments, the current error signal of the current outer loop is summed through proportional control and proportional R control of each harmonic, and each separated harmonic is subjected to independent R control. The harmonic controller only performs gain steady-state control on integer fundamental frequencies.

[0180] Among them, the expressions of the fundamental wave control branch and the harmonic wave control branch are:

[0181]

[0182] Among them, K p is the proportional gain, k nr is the resonant gain, ω nc is the cutoff frequency, ωn is the natural angular frequency, and s is a complex variable.

[0183] In some embodiments, a neural network algorithm may be used to extract the harmonic components of the system, where the harmonics in the system are expressed as:

[0184]

[0185] Where t represents the time variable, b is the fundamental component, w is the angular frequency, w = 2πf, f is the frequency of the signal, ε is the harmonic order, A ε , ψ ε are the amplitude and phase of the ε-th harmonic respectively.

[0186] In some embodiments, in a neural network model, the input matrix is ​​represented as:

[0187] C=[c1(t),c2(t),c3(t),L,c 2n (t)]=[sin(wt),cos(wt),sin(2wt),cos(2wt),L,sin(nwt),cos(nwt)],

[0188] The weight matrix is ​​expressed as:

[0189] W=[w1(t),w2(t),w3(t),L,w 2n (t)]=[A1 sinψ1,A1cosψ1,A2 sinψ2,A2 cosψ2,L,A n sinψ n ,A n cosψ n ],

[0190] The output vector of the neural network is the product of the input matrix and the weight matrix Get the error from the neural network model y(t) is the theoretical calibration value. After the error e(t) is processed by the least mean square (LMS) algorithm, the weight matrix W is readjusted based on the principle of minimizing the error target G. The error objective function is expressed as:

[0191]

[0192] Each harmonic is extracted through the neural network model, and the amplitude and phase of each harmonic are expressed as:

[0193]

[0194] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0195] Figure 9 A block diagram of an exemplary electronic device capable of implementing embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are provided as examples only and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0196] The device 900 includes a computing unit 901 that can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 902 or a computer program loaded from a storage unit 908 into a random access memory (RAM) 903. Various programs and data required for the operation of the device 900 can also be stored in the RAM 903. The computing unit 901, the ROM 902, and the RAM 903 are connected to each other via a bus 904. An input / output (I / O) interface 905 is also connected to the bus 904.

[0197] Various components in the device 900 are connected to the I / O interface 905, including an input unit 906, such as a keyboard, a mouse, etc.; an output unit 907, such as various types of displays, speakers, etc.; a storage unit 908, such as a magnetic disk, an optical disk, etc.; and a communication unit 909, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 909 allows the device 900 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0198] The computing unit 901 may be a variety of general and / or specialized processing components with processing and computing capabilities. Some examples of the computing unit 901 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 901 performs the various methods and processes described above, such as Figure 1 For example, in some embodiments, Figure 1The method in the embodiment of the present invention may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as the storage unit 908. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 900 via the ROM 902 and / or the communication unit 909. When the computer program is loaded into the RAM 903 and executed by the computing unit 901, the above-described Figure 1 One or more steps of the method.

[0199] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system comprising at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0200] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0201] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0202] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0203] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0204] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact through a communication network. The client-server relationship arises through computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, a server in a distributed system, or a server integrated with a blockchain.

[0205] It should be understood that the various forms of processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions of this disclosure can be achieved, and this document is not limited here.

[0206] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A power grid control method based on neural network and current synchronization UI droop coefficient, characterized in that: The method comprises: Use synchronous timing phasor measurement units to synchronously measure the voltage and current data of each node of distributed power supply units in multiple substations; The current is synchronously controlled using a radial basis function (RBF) neural network. The expression for current control is: Among them, k p and k i are the proportional and integral parameters of current synchronous control respectively. In RBF neural network control, x=[x c ] T is the network input, x c is the input of the cth neuron, [x c ] T is the transposed matrix of the input of the c-th neuron, y=[y c ] T is the output of the hidden layer of the network, y c is the output of the cth neuron in the hidden layer, [y c ] T is the transposed matrix of the output of the cth neuron in the hidden layer; The weights of the RBF neural network are w=[w1,w2,w3,...,w m ] T , m is the number of inputs, T is the calculation of the transposed matrix, and the output of the RBF neural network is: y t =w T x=w1x1+w2x2+w3x3+...+w m x m , Through the RBF neural network algorithm, the q-axis current i q To a given value i qref Approximation is performed until the error is eliminated; When the system operation state suddenly changes, i qref -i q ≠0, the current synchronization control module participates in the regulation, and the current adopts RBF neural network for current synchronization control, and continuously adjusts θ based on the current control expression. ref The size of the q =0, the system reaches a stable state after adjustment; An inductor-capacitor-inductor (LCL) filter is used to filter the high-frequency power supply units in the stable substation area before connecting them to the grid or load. The harmonic compensation control strategy adopts a double closed-loop structure with an output current or grid current outer loop and a filter capacitor current inner loop. Both the output current outer loop and the filter capacitor current inner loop adopt a quasi-proportional-resonant PR current control strategy. The grid-connected current outer loop is composed of a proportional control branch, a fundamental wave control branch, and a harmonic control branch in parallel, wherein the harmonic control branch is used to suppress each harmonic component. When the current error is greater than the threshold, the 5th, 7th, 11th, and 13th harmonics with greater harmonics are preferentially suppressed.

2. The method according to claim 1, characterized in that The method further comprises: When I i >(I i +I j ) / 2, indicating that the equivalent impedance Z i <Z j , the current deviation is positive, and a droop parameter adjustment value greater than 0 is obtained through the PI regulator, which increases the droop parameter of the converter and improves the equivalent impedance Z of the converter. i , and finally achieve Z i =Z j , to achieve the current sharing effect; when I i <(I i +I j ) / 2, indicating that the equivalent impedance Z i >Z j , the current deviation is negative, and a droop parameter adjustment value less than 0 is obtained through the PI regulator, which reduces the droop parameter of the converter and reduces the equivalent impedance Z of the converter. i , and finally achieve Z i =Z j ; The droop coefficient after adaptive adjustment is γ i * =γ i +σ i As long as there is a current deviation ΔI, the droop coefficient will be continuously adjusted until ΔI is zero, where I i is the current value related to node i, I j is the current value associated with node j, γ i is the original droop coefficient of node i, σ i is the corresponding correction coefficient of node i.

3. The method according to claim 1, characterized in that The grid current outer loop corresponds to two working modes. h is the load current I L When the input current I h When it is 0, the system harmonics are controlled and suppressed.

4. The method according to claim 1, wherein The output current outer loop and the filter capacitor current inner loop both adopt the quasi-PR current control strategy. The transfer function of the quasi-PR controller is expressed as: Among them, K p is the proportional gain, k nr is the resonant gain, ω nc is the cutoff frequency, ω n is the natural angular frequency, and s is a complex variable.

5. The method according to claim 1, wherein The current error signal of the current outer loop is summed through proportional control and proportional R control of each harmonic. The separated harmonics are individually R-controlled. The harmonic controller only performs gain steady-state control on integer fundamental frequencies. Among them, the expressions of the fundamental wave control branch and the harmonic wave control branch are: Among them, K p is the proportional gain, k nr is the resonant gain, ω nc is the cutoff frequency, ω n is the natural angular frequency, and s is a complex variable.

6. The method according to claim 1, wherein The method further comprises: A neural network algorithm is used to extract the harmonic components of the system, where the harmonics in the system are expressed as: Where t represents the time variable, b is the fundamental component, w is the angular frequency, w = 2πf, f is the frequency of the signal, ε is the harmonic order, A ε , ψ ε are the amplitude and phase of the ε-th harmonic respectively.

7. The method according to claim 6, characterized in that In the neural network model, the input matrix is ​​represented as: C[c1(t),c2(t),c3(t),L,c 2n (t)]s[sin(wt),cos(wt),sin(2wt),cos(2wt),L,sin(nwt),cos(nwt)] The weight matrix is ​​expressed as: W=[w1(t),w2(t),w3(t),L,w 2n (t)]=[A1 sinψ1,A1cosψ1,A2 sinψ2,A2 cosψ2,L,A n sinψ n ,TO n cosψ n ], The output vector of the neural network is the product of the input matrix and the weight matrix Get the error from the neural network model y(t) is the theoretical calibration value. After the error e(t) is processed by the least mean square (LMS) algorithm, the weight matrix W is readjusted based on the principle of minimizing the error target G. The error objective function is expressed as: Each harmonic is extracted through the neural network model, and the amplitude and phase of each harmonic are expressed as:

8. A power grid control device based on a neural network and a current synchronization UI droop coefficient, used to implement the method according to any one of claims 1 to 7, characterized in that: The device comprises: The synchronous timing phasor measurement unit is used to synchronously measure the voltage and current data of each node of the distributed power supply units in multiple substations; The current is synchronously controlled using a radial basis function (RBF) neural network. The expression for current control is: Among them, k p and k i are the proportional and integral parameters of current synchronous control, respectively. In RBF neural network control, x=[x c ] T is the network input, x c is the input of the cth neuron, [x c ] T is the transposed matrix of the input of the c-th neuron, y=[y c ] T is the output of the hidden layer of the network, y c is the output of the cth neuron in the hidden layer, [y c ] T is the transposed matrix of the output of the cth neuron in the hidden layer; The weights of the RBF neural network are w=[w1,w2,w3,...,w m ] T , m is the number of inputs, T is the calculation of the transposed matrix, and the output of the RBF neural network is: y t =w T x=w1x1+w2x2+w3x3+...+w m x m , Through the RBF neural network algorithm, the q-axis current i q To a given value i qref Approximation is performed until the error is eliminated; The current synchronization control module is used when the system operating state suddenly changes and i qref -i q ≠0, the current is controlled by RBF neural network, and θ is continuously adjusted based on the current control expression. ref The size of the q =0, the system reaches a stable state after adjustment; The inductor-capacitor-inductor (LCL) filter is used to filter high frequencies of distributed power supply units in a stable substation before connecting them to the grid or load. The harmonic compensation control strategy adopts a dual closed-loop structure with an output current or grid current outer loop and a filter capacitor current inner loop. Both the output current outer loop and the filter capacitor current inner loop adopt a quasi-proportional-resonant PR current control strategy. Among them, the grid-connected current outer loop is composed of a proportional control branch, a fundamental wave control branch, and a harmonic control branch in parallel. The harmonic control branch is used to suppress each harmonic component. When the current error is greater than the threshold, the 5th, 7th, 11th, and 13th harmonics with greater harmonics are preferentially suppressed.

9. An electronic device, characterized in that: include: at least one processor; and a memory communicatively coupled to the at least one processor; It is characterized in that the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 7.

Images