RBFNN estimator-based grid-connected inverter system model-free integral sliding mode control method

Abstract

The invention provides a model-free integral sliding mode control method for a grid-connected inverter system based on an RBFNN estimator, and the method comprises the following steps: firstly, taking a related filter and a grid-connected impedance parameter in a main circuit of a single-phase LCL type grid-connected inverter as unknown quantities, and building a super-local model; secondly, designing an RBFNN estimator which is used for estimating an unknown part in the hyper-local model; thirdly, designing a control law based on an RBFNN estimator and an integral sliding mode control method so as to enhance the robustness of the system, improve the dynamic response speed of the system and effectively suppress buffeting in sliding mode control; and finally, designing an online updating method for the weight and the central value of the neural network of the estimator. According to the method, a model-free control strategy is provided for the grid-connected inverter system, the robustness of the grid-connected inverter system for dealing with uncertain disturbance in actual operation can be enhanced, it is guaranteed that the grid-connected current accurately tracks the reference value of the grid-connected current, and the method has remarkable engineering practical value.

Description

Technical Field

[0001] The present invention belongs to the technical field of photovoltaic-based new energy power generation, and relates to a model-free integral sliding mode control method for a grid-connected inverter system based on an RBFNN estimator. Background Art

[0002] In recent years, research on photovoltaic-based renewable energy power generation technologies has become a hot topic both domestically and internationally. As a key interface device connecting the photovoltaic array to the power grid, the performance of the photovoltaic grid-connected inverter directly impacts the system's power quality, conversion efficiency, and operational stability.

[0003] At present, the control methods used in existing grid-connected inverter systems include: backstepping control, PR control, Although the above methods can control the grid-connected current to track its reference value, they all rely on the accurate mathematical model of the main circuit of the grid-connected inverter. However, the photovoltaic grid-connected inverter system faces complex and changeable working conditions in actual operation, such as filter parameter perturbations, time-varying grid impedance and other unmodeled dynamic changes. These factors make it difficult to obtain an accurate system model.

[0004] Traditional controllers designed based on fixed models often experience significant degradation in control performance when faced with system parameter changes and external disturbances. Although sliding mode control has attracted attention for its strong robustness to matching uncertainties and disturbances, its design usually needs to be based on the system nominal model, and the inherent high-frequency chattering problem still needs to be solved. At the same time, traditional sliding mode has limited direct processing capabilities for modeling errors.

[0005] Therefore, developing an advanced control method that does not rely on precise mathematical models, can adaptively handle unknown dynamics and strong uncertainties of the system, and has high-precision tracking capabilities and strong robustness is urgently needed and of great significance for improving the operating performance of photovoltaic grid-connected inverters under various complex working conditions. Summary of the Invention

[0006] In order to overcome the dependence on precise mathematical models, the present invention provides a model-free integral sliding mode control method for a grid-connected inverter system based on an RBFNN estimator, which solves the problem that the traditional grid-connected inverter system control method based on fixed model design has degraded control performance and weak error handling capability when dealing with filter perturbations, grid impedance time-varying and other unmodeled dynamic factors in actual operation.

[0007] The present invention regards all the filter parameters, grid impedance and unmodeled dynamics in the single-phase LCL type grid-connected inverter system as nonlinear unknown dynamics, designs an RBFNN estimator, uses the RBFNN estimator to estimate the above nonlinear unknown dynamics online, and designs a grid-connected current controller in combination with an integral sliding mode control method based on an improved reaching law. A method is proposed that deeply integrates the adaptive intelligent estimation capability of the RBFNN estimator with the strong robustness and high-precision tracking of the integral sliding mode control, thereby providing a high-performance and high-reliability model-free control method for photovoltaic grid-connected inverter systems.

[0008] According to the above technical concept, the present invention specifically adopts the following technical solutions:

[0009] A model-free integral sliding mode control method for a grid-connected inverter system based on an RBFNN estimator includes the following steps:

[0010] Step 1: Establish a mathematical model of the main circuit of a single-phase LCL grid-connected inverter system and its hyperlocal model; including:

[0011] (1)

[0012] Where, is the inductance on the grid-connected inverter side; Output current for grid-connected inverter; is the current time; is the control law, and ; Input DC voltage to the grid-connected inverter; is the capacitor voltage of the grid-connected inverter; It is the filter capacitor of grid-connected inverter; is the grid-connected current; is the grid-side inductance of the grid-connected inverter; is the grid-connected impedance of the grid-connected inverter; is the grid voltage;

[0013] The mathematical model of the main circuit of the above grid-connected inverter system is further written as:

[0014] (2)

[0015] Where, Grid-connected current The third derivative of Grid-connected current The first derivative of ; is the grid voltage The second derivative of is the first parameter to be estimated; is the second unknown parameter;

[0016] Considering unknown disturbances , write (2) as:

[0017] (3)

[0018] Where:

[0019] (4)

[0020] make:

[0021] (5)

[0022] Where, is the first unknown parameter; is an unknown disturbance and satisfies , For unknown disturbance The upper bound of , which is a positive real number;

[0023] The hyperlocal model of the main circuit of the grid-connected inverter system is obtained as follows:

[0024] (6)

[0025] Step 2: Based on the hyperlocal model of the main circuit of the grid-connected inverter system, design a first RBFNN estimator and a second RBFNN estimator; wherein the first RBFNN estimator is used to estimate the first unknown parameter , the second RBFNN estimator is used to estimate the second unknown parameter ;

[0026] Specifically, the first RBFNN estimator has 3 neurons in the input layer, 15 neurons in the hidden layer, and 1 neuron in the output layer; the second RBFNN estimator has the same structure as the first RBFNN estimator.

[0027] Design the first RBFNN estimator input layer as:

[0028] (7)

[0029] in: ; ;

[0030] Where, is the grid-connected current error vector; is the grid-connected current error; is the grid current error The first derivative of ; is the grid current error The second derivative of is the transpose of the matrix; is the given value of grid-connected current; is the grid-connected current given value The first derivative of ; is the grid-connected current given value The second derivative of Grid-connected current The second derivative of Grid-connected current The reference amplitude of is the grid voltage angular frequency;

[0031] The second RBFNN estimator input layer is the same as the first RBFNN estimator input layer.

[0032] The hidden layers of the first RBFNN estimator and the second RBFNN estimator are designed as follows:

[0033] (8)

[0034] Where, is the first radial basis function, which is used as the first RBFNN estimator. The output of hidden nodes; is the exponential operation; For the first RBFNN estimator The radial basis function center vector of hidden nodes; For the first RBFNN estimator The width of the radial basis function of the hidden nodes; is the hidden node of the RBFNN estimator; is the second radial basis function, which serves as the second RBFNN estimator The output of hidden nodes; For the second RBFNN estimator The radial basis function center vector of hidden nodes; For the second RBFNN estimator The width of the radial basis function of the hidden nodes.

[0035] Design the first RBFNN estimator output and the second RBFNN estimator output layer as follows:

[0036] (9)

[0037] in:

[0038] (10)

[0039] Where, The first unknown parameter estimated value of; For the first RBFNN estimator The weights between hidden nodes and outputs; is the radial basis function of the first RBFNN estimator; The transpose of the weight matrix connecting the hidden layer and the output layer of the first RBFNN estimator; is the radial basis function matrix of the hidden layer of the first RBFNN estimator; The second unknown parameter estimated value of; For the second RBFNN estimator The weights between hidden nodes and outputs; is the radial basis function of the second RBFNN estimator; The transpose of the weight matrix connecting the hidden layer and the output layer for the second RBFNN estimator; is the radial basis function matrix of the hidden layer of the second RBFNN estimator; The weight matrix connecting the hidden layer and the output layer of the first RBFNN estimator; The weight matrix connecting the hidden layer and the output layer for the second RBFNN estimator.

[0040] Step 3: Based on the first RBFNN estimator, design an online adaptive law for the first weight matrix; based on the second RBFNN estimator, design an online adaptive law for the second weight matrix;

[0041] Specifically, the first weight matrix online adaptive law and the second weight matrix online adaptive law are both RBFNN estimator weight update rates. The update rate improves the estimator performance by continuously iteratively updating the weight matrix, thereby enabling the estimator to obtain more accurate estimated values.

[0042] The expressions for designing the online adaptive law of the first weight matrix and the online adaptive law of the second weight matrix are:

[0043] (11)

[0044] Where, is the first weighted adaptive law coefficient, and ; is the sliding surface; is the second parameter of the sliding surface; is the second weighted adaptive law coefficient, and .

[0045] Step 4: Design a first online update method and a second online update method; wherein the first online update method is used to update the radial basis function center value of the first RBFNN estimator, and the second online update method is used to estimate the radial basis function center value of the second RBFNN estimator.

[0046] Step 4 specifically includes the following steps:

[0047] set up is a vector With the first center vector The best matching index value, is a vector With the second center vector The subscript value of the best match; where the first center vector For the first RBFNN estimator The central vector of the activation function at the nth iteration; the second central vector For the second RBFNN estimator The center vector of the activation function at the nth iteration;

[0048] When the first central vector , the second central vector When performing the nth iteration, according to the Euclidean minimum distance criterion, we can get:

[0049] (12)

[0050] Where, The subscript value corresponding to the minimum value of the arg min function about f at the nth moment; The subscript value corresponding to the minimum value of the arg min function about g at the nth moment; Calculate the value of the function minimization; is the grid-connected current error vector at the nth moment;

[0051] Based on equation (12), the first online update method and the second online update method are designed respectively, and their expressions are as follows:

[0052] (13)

[0053] Where, For the first RBFNN estimator The center vector of the activation function at the n+1th iteration; For the first RBFNN estimator The center vector of the activation function at the nth iteration; is the learning rate of the radial basis center value of the RBFNN estimator with respect to f; For the second RBFNN estimator The center vector of the activation function at the n+1th iteration; For the second RBFNN estimator The center vector of the activation function at the nth iteration; is the learning rate of the radial basis center value of the RBFNN estimator with respect to g; is the hidden node of the RBFNN estimator.

[0054] Step 5: Designing a grid-connected inverter system control law based on the first RBFNN estimator and the second RBFNN estimator; including:

[0055] Select sliding surface as follows:

[0056] (14)

[0057] Where, are all parameters to be designed in the sliding surface, and are all positive numbers;

[0058] The state equation composed of the grid current tracking error and its derivative is expressed as:

[0059] (15)

[0060] Where, is the grid current error The first derivative of ; for The first derivative of ; for The first derivative of ; is the grid-connected current given value The third derivative of

[0061] In order to reduce chattering, the reaching law is selected as:

[0062] (16)

[0063] Where, is the law of approach; is the reaching law gain, which is a positive real number; is adjustable gain; is a symbolic function;

[0064] Among them, the design has adjustable gain for:

[0065] (17)

[0066] Where, is the amplitude coefficient of the adjustable gain function; is the width coefficient of the adjustable gain function; is the center coefficient of the adjustable gain function; are all positive real numbers;

[0067] Thus, the control law of the grid-connected inverter is obtained as:

[0068] (18)

[0069] Among them, the symbol function The expression is as follows:

[0070]

[0071] Preferably, in order to further eliminate the chattering of the grid-connected inverter system, the above sign function Replace with a smooth function :

[0072]

[0073] Where, It is a positive number that adjusts the smoothness.

[0074] Compared with the initial symbol function , the improved smoothing function The smoothness adjustment parameters can be used to make the step jump improvement smoother, thereby further reducing the jitter of the system output grid-connected current waveform.

[0075] The following is a process for proving the stability of the control law of the grid-connected inverter system of the present invention.

[0076] A common proof process of control theory in this technical field includes: defining a Lyapunov function, taking its derivative, proving the negative definiteness of the derivative, and concluding that the control system is stable.

[0077] To facilitate subsequent stability proof, first define the following variables:

[0078] Assume that there are ideal weights such that the following holds:

[0079] (19)

[0080] in: ;

[0081] Where, is the output of the first ideal RBFNN estimator; is the transpose of the first optimal weight vector; is the approximation error of the first ideal RBFNN estimator, and its upper bound is , is a positive real number that satisfies ; is the output of the second ideal RBFNN estimator; is the transpose of the second optimal weight vector; is the approximation error of the second ideal RBFNN estimator, and its upper bound is , is a positive real number that satisfies .

[0082] The difference between the first ideal RBFNN estimator weight matrix and the estimated weight matrix, and the difference between the second ideal RBFNN estimator weight matrix and the estimated weight matrix are defined respectively. The two expressions are as follows:

[0083] (20)

[0084] Where, is the weight bias of the first RBFNN estimator; is the ideal weight of the first RBFNN estimator; is the actual weight of the first RBFNN estimator; is the weight bias of the second RBFNN estimator; is the ideal weight of the second RBFNN estimator; is the actual weight of the second RBFNN estimator;

[0085] Define the first ideal RBFNN estimator output error respectively , the second ideal RBFNN estimator output error , the two expressions are as follows:

[0086] (twenty one)

[0087] Where, is the transpose of the actual weights of the first RBFNN estimator; is the transpose of the ideal weights of the first RBFNN estimator; is the transpose of the second RBFNN estimator weight bias; is the estimated value of the second RBFNN estimator; is the transpose of the actual weights of the second RBFNN estimator; is the transpose of the ideal weights of the second RBFNN estimator; is the weight bias of the second RBFNN estimator.

[0088] After defining the above variables, we will formally begin to explain the proof process.

[0089] First, define the Lyapunov function , whose expression is:

[0090] (twenty two)

[0091] Next, calculate the Lyapunov function The first derivative with respect to time, and prove its negative definiteness:

[0092] (twenty three)

[0093] Sliding surface The first derivative with respect to time is calculated as follows:

[0094] (twenty four)

[0095] Substituting formula (23) into formula (22) yields:

[0096] (25)

[0097] Since the approximation error can be limited to be small enough, if we take , then we can get .

[0098] Therefore, existence , , making .

[0099] The above analysis shows that the sliding mode control method for grid-connected inverter based on RBFNN estimator proposed in the present invention can ensure the stability of the system.

[0100] The beneficial effects of the present invention are:

[0101] 1. The present invention provides a control method based on a hyperlocal model, which is essentially a model-free control method. By using an RBFNN estimator to online estimate the nonlinear dynamic part of the system parameters and the modeling uncertainty when the filter parameters are disturbed during actual operation, the present invention does not require knowledge of the actual parameters of the grid-connected inverter system or the disturbances to which the grid-connected inverter system is subjected during actual operation. This method can enhance the robustness of the grid-connected inverter system to cope with uncertain disturbances during actual operation, ensure that the grid-connected current accurately tracks its reference value, and has significant engineering practical value.

[0102] 2. The present invention does not need to obtain sample data, thus eliminating the need for an offline training process. The online training process of the RBFNN estimator adopted in the present invention can be achieved through the weight adaptation law and the center value update method;

[0103] 3. By combining the integral sliding mode control method based on the improved reaching law, the present invention can significantly reduce the steady-state error and suppress chattering, so that the grid-connected inverter system has strong robustness, enhances the ability of the grid-connected inverter system to cope with interference, and can significantly improve the dynamic and steady-state performance of the grid-connected inverter system. BRIEF DESCRIPTION OF THE DRAWINGS

[0104] Figure 1 The first unknown parameter in this embodiment Schematic diagram of RBFNN estimator;

[0105] Figure 2 For this embodiment, the second unknown parameter Schematic diagram of RBFNN estimator;

[0106] Figure 3 : is a control block diagram of a grid-connected inverter based on an RBFNN estimator in this embodiment;

[0107] Figure 4 The waveforms of grid-connected current and grid voltage under two control methods are shown;

[0108] Figure 5 Schematic diagram for comparison of grid-connected current tracking errors under two control methods;

[0109] Figure 6 When the grid current reference amplitude is When the current drops from 15A to 10A, the grid-connected current waveforms under the two control methods are shown;

[0110] Figure 7 When the inverter side filter inductor is Shi You Reduced to 0.7 Grid-connected current waveforms under two control methods;

[0111] Figure 8 For the filter capacitor Shi You Increased to 1.1 Grid-connected current waveforms under two control methods;

[0112] Figure 9 The grid-side filter inductor is Shi You Reduced to 0.7 Grid-connected current waveforms under two control methods;

[0113] Figure 10 is the first unknown parameter in this embodiment Schematic diagram of the comparison between the estimated value and the true value waveform;

[0114] Figure 11 The second unknown parameter in this embodiment Schematic diagram of the comparison between the estimated value and the true value waveform. DETAILED DESCRIPTION

[0115] The technical solution of the present invention is described below in conjunction with the accompanying drawings and implementation methods.

[0116] Example

[0117] A model-free integral sliding mode control method for a grid-connected inverter system based on an RBFNN estimator includes the following steps:

[0118] Step 1: Establish a mathematical model of the main circuit of a single-phase LCL grid-connected inverter system and its hyperlocal model; including:

[0119] The mathematical model of the main circuit of the grid-connected inverter system is as follows:

[0120]

[0121] The hyperlocal model of the main circuit of the grid-connected inverter system is as follows:

[0122]

[0123] Step 2: Based on the hyperlocal model of the main circuit of the grid-connected inverter system, design a first RBFNN estimator and a second RBFNN estimator; wherein the first RBFNN estimator is used to estimate the first unknown parameter , the second RBFNN estimator is used to estimate the second unknown parameter ;

[0124] Specifically, the first RBFNN estimator has 3 neurons in the input layer, 15 neurons in the hidden layer, and 1 neuron in the output layer; the second RBFNN estimator has the same structure as the first RBFNN estimator.

[0125] Design the first RBFNN estimator input layer as:

[0126]

[0127] in: ; ;

[0128] Where, is the grid-connected current reference amplitude; is the grid voltage angular frequency;

[0129] The second RBFNN estimator input layer is the same as the first RBFNN estimator input layer.

[0130] The expressions for designing the hidden layer of the first RBFNN estimator and the hidden layer of the second RBFNN estimator are:

[0131]

[0132] Design the expressions of the first RBFNN estimator output layer and the second RBFNN estimator output layer respectively:

[0133]

[0134] in:

[0135]

[0136] Step 3: Based on the first RBFNN estimator, design an online adaptive law for the first weight matrix; based on the second RBFNN estimator, design an online adaptive law for the second weight matrix;

[0137] The expressions of the first weight matrix online adaptive law and the second weight matrix online adaptive law are:

[0138]

[0139] Step 4: Design a first online update method and a second online update method; wherein the first online update method is used to update the radial basis function center value of the first RBFNN estimator, and the second online update method is used to estimate the radial basis function center value of the second RBFNN estimator;

[0140] The expressions of the first online update method and the second online update method are respectively:

[0141]

[0142] Step 5: Based on the first RBFNN estimator and the second RBFNN estimator, design the grid-connected inverter system control law, which is expressed as follows:

[0143]

[0144] in:

[0145]

[0146] Where, It is a positive number that adjusts the smoothness.

[0147] In order to verify the effectiveness and feasibility of the method proposed in the present invention, the first control method and the second control method are respectively used for comparison; among them, the first control method adopts the integral sliding mode control method (ISMC), and the second control method adopts the model-free integral sliding mode control method (ISMC&RBF) of the grid-connected inverter system based on the RBFNN estimator provided by the present invention.

[0148] Two sets of single-phase grid-connected inverter simulation models were built on the MATLAB / Simulink platform. The main parameters involved are shown in Table 1.

[0149] Table 1 Electrical parameters and control parameters

[0150]

[0151] Depend on Figure 4 It can be seen that the grid-connected current phases corresponding to the two control methods are both consistent with the common coupling point voltage The phases are consistent, so the grid current phase under the two control strategies can accurately track the phase of the common coupling point voltage.

[0152] Depend on Figure 5 As can be seen from Table 2, compared with the first control method, the grid-connected current tracking error and total harmonic distortion rate of this embodiment are significantly reduced.

[0153] Table 2 Comparison of total harmonic distortion of grid-connected current

[0154]

[0155] Depend on Figure 6 It can be seen that when the inverter input power changes, that is, when the light and temperature change, the above two control methods can make the grid-connected current track its reference value, but according to the local enlarged diagram, compared with the first control method, this embodiment can make the grid-connected current have smaller jitter and better tracking effect, that is, the grid-connected inverter system under this embodiment can better adapt to light changes and temperature changes.

[0156] Depend on Figure 7 It can be seen that when the filter inductance on the inverter side suddenly changes at 0.1s, both control methods can make the grid-connected current track its reference value. However, according to the local enlarged diagram, compared with the first control method, this embodiment makes the grid-connected current have smaller jitter and better tracking effect, indicating that this embodiment has stronger robustness to changes in the inverter side filter inductance parameters.

[0157] Depend on Figure 8It can be seen that when the filter capacitor changes in 0.1s, both control methods can make the grid-connected current track its reference value. However, according to the local enlarged diagram, compared with the first control method, this embodiment makes the grid-connected current have smaller jitter and better tracking effect, indicating that this embodiment has stronger robustness to changes in filter capacitor parameters.

[0158] Depend on Figure 9 It can be seen that when the grid-side filter inductance suddenly changes at 0.1s, both control methods can make the grid-connected current track its reference value. However, according to the local enlarged diagram, compared with the first control method, this embodiment can make the grid-connected current have smaller jitter and better tracking effect, indicating that this embodiment has stronger robustness to changes in grid-side inductance parameters.

[0159] Depend on Figure 10 and Figure 11 It can be seen that the first unknown parameter estimated in this embodiment is and the second unknown parameter The ability to fully track its actual value indicates that the estimator in this embodiment has a good estimation effect and high feasibility.

Claims

1. A model-free integral sliding mode control method for a grid-connected inverter system based on an RBFNN estimator, characterized in that: The steps include: Design a mathematical model of the main circuit of the grid-connected inverter system and its hyperlocal model; Based on the hyperlocal model of the main circuit of the grid-connected inverter system, a first RBFNN estimator and a second RBFNN estimator are designed; wherein the first RBFNN estimator is used to estimate the first unknown parameter , the second RBFNN estimator is used to estimate the second unknown parameter ; Designing a first online update method and a second online update method; wherein the first online update method is used to update the radial basis function center value of the first RBFNN estimator, and the second online update method is used to estimate the radial basis function center value of the second RBFNN estimator; Based on the first RBFNN estimator, the first weight matrix online adaptive law is designed; Based on the second RBFNN estimator, the second weight matrix online adaptive law is designed; Based on the first RBFNN estimator and the second RBFNN estimator, a control law for the grid-connected inverter system is designed.

2. The model-free integral sliding mode control method for a grid-connected inverter system based on an RBFNN estimator according to claim 1, characterized in that: The mathematical model of the main circuit of the grid-connected inverter system is as follows: ； Where, is the inductance on the grid-connected inverter side; Output current for grid-connected inverter; is the current time; is the control law, and ; Input DC voltage to the grid-connected inverter; is the capacitor voltage of the grid-connected inverter; It is the filter capacitor of grid-connected inverter; is the grid-connected current; is the grid-side inductance of the grid-connected inverter; is the grid-connected impedance of the grid-connected inverter; is the grid voltage.

3. The model-free integral sliding mode control method for a grid-connected inverter system based on an RBFNN estimator according to claim 2, characterized in that: The super-local model of the main circuit of the grid-connected inverter system is as follows: ； Where, is the first unknown parameter; is the second unknown parameter.

4. The model-free integral sliding mode control method for a grid-connected inverter system based on an RBFNN estimator according to claim 1, characterized in that: The first RBFNN estimator has 3 neurons in the input layer, 10 to 20 neurons in the hidden layer, and 1 neuron in the output layer; The second RBFNN estimator has the same structure as the first RBFNN estimator.

5. The model-free integral sliding mode control method for a grid-connected inverter system based on an RBFNN estimator according to claim 4, characterized in that: The first RBFNN estimator input layer is: ； in: ; ; Where, is the grid-connected current error vector; is the grid-connected current error; is the first-order derivative of the grid-connected current error; is the second-order derivative of the grid current error; is the transpose of the matrix; is the given value of grid-connected current; is the first-order derivative of the grid-connected current given value; Grid-connected current The first derivative of ; is the second-order derivative of the grid-connected current given value; Grid-connected current The second derivative of is the grid-connected current reference amplitude; is the grid voltage angular frequency.

6. The model-free integral sliding mode control method for a grid-connected inverter system based on an RBFNN estimator according to claim 4, characterized in that: The hidden layer of the first RBFNN estimator is: ； Where, is the first radial basis function, which is used as the first RBFNN estimator. The output of hidden nodes; is the exponential operation; For the first RBFNN estimator The radial basis function center vector of hidden nodes; For the first RBFNN estimator The width of the radial basis function of the hidden nodes; is the hidden node of the RBFNN estimator; is the second radial basis function, which serves as the second RBFNN estimator The output of hidden nodes; For the second RBFNN estimator The radial basis function center vector of hidden nodes; For the second RBFNN estimator The width of the radial basis function of the hidden nodes.

7. The model-free integral sliding mode control method for a grid-connected inverter system based on an RBFNN estimator according to claim 4, characterized in that: The first RBFNN estimator output layer is: ； in: ； Where, The first unknown parameter estimated value of; For the first RBFNN estimator The weights between hidden nodes and outputs; is the radial basis function of the first RBFNN estimator; The transpose of the weight matrix connecting the hidden layer and the output layer of the first RBFNN estimator; is the radial basis function matrix of the hidden layer of the first RBFNN estimator; The second unknown parameter estimated value of; For the second RBFNN estimator The weights between hidden nodes and outputs; is the radial basis function of the second RBFNN estimator; The transpose of the weight matrix connecting the hidden layer and the output layer for the second RBFNN estimator; is the radial basis function matrix of the hidden layer of the second RBFNN estimator.

8. The model-free integral sliding mode control method for a grid-connected inverter system based on an RBFNN estimator according to claim 1, characterized in that: The expressions of the first weight matrix online adaptive law and the second weight matrix online adaptive law are respectively: ； Where, The weight matrix connecting the hidden layer and the output layer of the first RBFNN estimator; is the first weighted adaptive law coefficient; is the sliding surface; is the second parameter of the sliding surface; The weight matrix connecting the hidden layer and the output layer for the second RBFNN estimator; is the second weighted adaptive law coefficient; is the radial basis function matrix of the hidden layer of the first RBFNN estimator; is the radial basis function matrix of the hidden layer of the second RBFNN estimator; For the control law.

9. The model-free integral sliding mode control method for a grid-connected inverter system based on an RBFNN estimator according to claim 1, characterized in that: The expressions of the first online update method and the second online update method are respectively: ； in: ； Where, For the first RBFNN estimator The center vector of the activation function at the n+1th iteration; For the first RBFNN estimator The center vector of the activation function at the nth iteration; is the learning rate of the radial basis center value of the first RBFNN estimator; is the grid-connected current error vector at the nth moment; The subscript value corresponding to the minimum value of the argmin function of f at the nth moment; For the second RBFNN estimator The center vector of the activation function at the n+1th iteration; For the second RBFNN estimator The center vector of the activation function at the nth iteration; The learning rate of the radial basis center value of the second RBFNN estimator; The subscript value corresponding to the minimum value of the arg min function about g at the nth moment; is the hidden node of the RBFNN estimator.

10. The model-free integral sliding mode control method for a grid-connected inverter system based on an RBFNN estimator according to claim 1, characterized in that: The expression of the grid-connected inverter control law is: ； in: ； Where, is the grid-connected current given value The third derivative of is the third parameter of the sliding surface; is the first parameter of the sliding surface; is the reaching law gain; is adjustable gain; is a smooth function; is the smoothness adjustment parameter, which is a positive number; is the control law; is the second parameter of the sliding surface; is the grid-connected current error; is the first-order derivative of the grid-connected current error; is the second-order derivative of the grid current error; The second unknown parameter estimated value of; The first unknown parameter estimated value of; is the sliding surface.

Images