Inertia damping adaptive control method, system and device, storage medium and product

Abstract

The invention discloses an inertia damping adaptive control method, system and device, a storage medium and a product. The method comprises the following steps: constructing a virtual synchronous generator mathematical model; analyzing the power grid frequency deviation and the relationship between the power grid frequency deviation change rate and the virtual inertia variable quantity and the virtual damping variable quantity; establishing an initial fuzzy rule based on the relationship; initializing the fuzzy neural network model by using the initial fuzzy rule, and training the initialized fuzzy neural network model by using the sample data set to obtain a target model; adding the target model into the virtual synchronous generator to obtain a virtual synchronous generator based on the fuzzy neural network; and performing adaptive control on the virtual inertia coefficient and the virtual damping coefficient by using the virtual synchronous generator based on the fuzzy neural network. According to the method, the frequency deviation and the frequency deviation change rate of the power grid are used as the input of the fuzzy neural network model, and the method has better frequency response characteristics during load change and island grid connection.

Description

Technical Field

[0001] The present invention belongs to the technical field of virtual synchronous generators (VSG), and in particular relates to a method, system, device, storage medium and product for adaptively controlling inertia damping of a virtual synchronous generator based on a fuzzy neural network. Background Art

[0002] Since the direct-drive permanent magnet wind power system (D-PMSWG) eliminates vulnerable parts such as gearboxes, it greatly reduces maintenance costs, and the direct-drive permanent magnet wind power system has strong fault ride-through capability, so it has been widely used. The permanent magnet synchronous generator plays an important role in maintaining the stable operation of the wind power system with its own damping and inertia. However, with the increasing penetration rate of wind power, the wind power system is connected to the power grid through the power electronic inverter interface, which makes the operation of the wind power system lack the support of inertia and damping. The virtual synchronous generator technology is used to make the grid-connected inverter simulate the operation mechanism of the synchronous generator, so that the inverter produces inertia and damping similar to the synchronous generator, which can improve the anti-interference ability of the grid-connected inverter, enhance the stability of the wind power system, and provide a certain voltage and frequency support to the power grid. Therefore, the virtual synchronous generator control strategy has received the attention and attention of many scholars and is widely used in new energy power generation.

[0003] Although the grid-connected inverter using the virtual synchronous generator control strategy can have external characteristics and rotor inertia characteristics similar to those of the synchronous generator, the conventional virtual synchronous generator control strategy uses a fixed moment of inertia, which cannot take into account rapidity while ensuring system stability. In fact, the virtual inertia coefficient J (or virtual moment of inertia) of the grid-connected inverter can adaptively select a more appropriate value according to the actual application scenario. In the virtual synchronous generator control, the virtual inertia coefficient J determines the speed of change of the angular velocity ω, and the virtual damping coefficient D (or virtual damping) determines the amount of change of the angular velocity ω, thereby affecting the change of active power. When the value of the virtual inertia coefficient J is too small, although the system response time is reduced, it will not be able to suppress power oscillations due to the lack of inertia characteristics; when the value of the virtual inertia coefficient J is too large, although the system can effectively suppress power oscillations, the response time is increased. Similarly, the selection of the virtual damping coefficient D will also affect the power oscillation suppression performance.

[0004] In most studies related to the adaptive control of virtual synchronous generators, the influence of the virtual damping coefficient D is always ignored, and only the adaptive control of the virtual inertia coefficient J of the virtual synchronous generator is considered. For example, Reference 1 (Zhang Tao, Zheng Jiaqi, Wang Fudong, et al. VSG rotation inertia adaptive algorithm based on fuzzy control [J]. Power Electronics Technology, 2021, 55(01): 40-44.) proposed an inertia adaptive control strategy for virtual synchronous generators, adopted a fuzzy algorithm to train the rotation inertia, output the adaptive inertia to the virtual synchronous generator control, and added the voltage and current double loops, ignoring the virtual damping coefficient, and requiring artificial design of the membership function, which cannot be automatically generated. For example, Reference 2 (Yang Zebin, Wang Mingtao, Sun Xiaodong. Bearingless asynchronous motor control based on adaptive fuzzy neural network [J]. Transactions of the Chinese Society of Agricultural Engineering, 2014, 30(02): 78-86.) proposed an adaptive fuzzy neural network control strategy for bearingless asynchronous motors, but the mathematical models of bearingless asynchronous motors and virtual synchronous generators are different, so the training objects are different. For example, patent document 3 (publication number CN113469332A) discloses a virtual synchronous generator inertia damping adaptive control method based on fuzzy neural network. Although the virtual rotational inertia and virtual damping coefficient of the virtual synchronous generator are taken into account at the same time, it is based on the system's rotor angular frequency change rate dω / dt and rotor angular frequency deviation Δω as input, and the initial fuzzy rules are created based on the relationship between the system's rotor angular frequency change rate and rotor angular frequency deviation and the virtual inertia and virtual damping coefficient, respectively. It is more inclined to island operation, does not take the grid frequency into account, and has poor grid-connected operation performance.

[0005] The traditional virtual synchronous generator inertia damping dual adaptive control needs to consider the changes of the virtual inertia coefficient J and the virtual damping coefficient D in different intervals, which increases the complexity of the system. When subjected to multiple disturbances, integral saturation is prone to occur and cannot meet the requirements of fast frequency response. Summary of the invention

[0006] The purpose of the present invention is to provide an inertia damping adaptive control method, system, device, storage medium and product to solve at least one of the problems that traditional control technology needs to consider the changes of virtual inertia coefficient J and virtual damping coefficient D in different intervals, and when subjected to multiple disturbances, it is easy to have integral saturation and cannot meet the demand for fast frequency response.

[0007] The present invention solves the above technical problems through the following technical solutions: an inertia damping adaptive control method for a direct-drive permanent magnet wind power system, comprising:

[0008] Construct a mathematical model of a virtual synchronous generator;

[0009] According to the virtual synchronous generator mathematical model, the relationship between the grid frequency deviation and the grid frequency deviation change rate and the virtual inertia change and virtual damping change is analyzed;

[0010] Establishing initial fuzzy rules according to the relationship between the grid frequency deviation and the grid frequency deviation change rate and the virtual inertia change and virtual damping change;

[0011] Constructing a sample data set; wherein each sample in the sample data set includes a power grid frequency deviation, a power grid frequency deviation change rate, a virtual inertia change, and a virtual damping change;

[0012] Constructing a fuzzy neural network model, initializing the fuzzy neural network model using the initial fuzzy rules, and training the initialized fuzzy neural network model using the sample data set to obtain a target model;

[0013] Adding the target model to the virtual synchronous generator to obtain the virtual synchronous generator based on fuzzy neural network;

[0014] The virtual synchronous generator based on fuzzy neural network is used to perform adaptive control of virtual inertia coefficient and virtual damping coefficient.

[0015] Furthermore, the expression of the virtual synchronous generator mathematical model is:

[0016]

[0017] Δω=ω-ω ref ;

[0018] Where, J represents the virtual inertia coefficient of the virtual synchronous generator, D represents the virtual damping coefficient of the virtual synchronous generator, ω represents the mechanical angular velocity of the virtual synchronous generator, ω ref represents the rated angular velocity of the virtual synchronous generator, T m , T e They represent the mechanical torque and electromagnetic torque of the virtual synchronous generator respectively.

[0019] Furthermore, the relationship between the grid frequency deviation and the grid frequency deviation change rate and the virtual inertia change and virtual damping change specifically includes:

[0020] When Δf>0, and When , the change of virtual inertia increases, and the change of virtual damping increases;

[0021] When Δf>0, and When , the change of virtual inertia decreases, and the change of virtual damping increases;

[0022] When Δf<0, and When , the change of virtual inertia increases, and the change of virtual damping increases;

[0023] When Δf<0, and When , the change of virtual inertia decreases, and the change of virtual damping increases;

[0024] Where Δf represents the grid frequency deviation, Indicates the rate of change of grid frequency deviation.

[0025] Based on the same concept, the present invention also provides a virtual synchronous generator control method for a direct-drive permanent magnet wind power system, wherein the virtual synchronous generator includes a virtual speed regulator, a virtual excitation regulator, and a target model obtained by using the inertia damping adaptive control method as described above; the control method includes:

[0026] Obtain the output active power P of the inverter bridge and the virtual mechanical power P obtained by using the virtual speed regulator m , according to the virtual mechanical power P m Calculate the power difference with the output active power P;

[0027] Obtain the grid frequency deviation Δf and the grid frequency deviation change rate According to the grid frequency deviation Δf, the grid frequency deviation change rate and the target model, obtaining a virtual inertia change ΔJ and a virtual damping change ΔD;

[0028] Calculating an angular velocity deviation Δω according to the power difference, the virtual inertia change ΔJ and the virtual damping change ΔD;

[0029] According to the obtained rated angular velocity ω ref and the angular velocity deviation Δω to calculate the phase angle θ;

[0030] Get the stator current i of the permanent magnet synchronous generator L , using the virtual excitation regulator to obtain the stator winding terminal voltage U0, according to the stator current i L , phase angle θ and stator winding terminal voltage U0 to calculate the reference voltage u ref .

[0031] Based on the same concept, the present invention provides a direct-drive permanent magnet wind power system, including a wind turbine, a permanent magnet synchronous generator, a rectifier bridge, an inverter bridge, a modulator, a driver and a controller, wherein the wind turbine is connected to the permanent magnet synchronous generator, the permanent magnet synchronous generator is connected to the rectifier bridge, the rectifier bridge is connected to the inverter bridge via a DC capacitor, and the inverter bridge is connected to an AC load via a filter circuit; the controller is connected to the modulator, the modulator is connected to the driver, and the driver is connected to a power switch in the inverter bridge;

[0032] The controller includes an acquisition unit, a calculation unit, a virtual synchronous generator and a voltage and current loop; the acquisition unit is used to acquire the output voltage and grid-connected current of the inverter bridge; the calculation unit is used to calculate the output active power P of the inverter bridge according to the output voltage and grid-connected current; the virtual synchronous generator is used to calculate the reference voltage u using the virtual synchronous generator control method as described above ref The voltage and current loop is used according to the reference voltage u ref Get three-phase voltage instructions;

[0033] The modulator is used to generate a modulation wave under the action of the three-phase voltage instruction;

[0034] The driver is used to control the opening and closing of the power switch of the inverter bridge under the drive of the modulation wave.

[0035] Based on the same concept, the present invention provides an electronic device, including a memory, a processor, and a computer program / instruction stored in the memory, wherein the processor executes the computer program / instruction to implement the inertia damping adaptive control method as described above or the virtual synchronous generator control method as described above.

[0036] Based on the same concept, the present invention provides a computer-readable storage medium having a computer program / instruction stored thereon, which, when executed by a processor, implements the inertia damping adaptive control method as described above or the virtual synchronous generator control method as described above.

[0037] Based on the same concept, the present invention provides a computer program product, including a computer program / instruction, which implements the above-mentioned inertia damping adaptive control method or the above-mentioned virtual synchronous generator control method when executed by a processor.

[0038] Beneficial Effects

[0039] Compared with the prior art, the advantages of the present invention are:

[0040] The present invention takes into account both the virtual inertia coefficient and the virtual damping coefficient. Since the fuzzy neural network model integrates the self-learning ability of the neural network and the fuzzy reasoning ability of the fuzzy control, there is no need to consider the different changes and mutual constraints of the virtual inertia coefficient and the virtual damping coefficient in different intervals, and there is no need to manually design the membership function.

[0041] Compared with the traditional inertia damping dual adaptive control strategy, the present invention uses the frequency deviation of the power grid and the frequency deviation change rate as the input of the fuzzy neural network model, and has better frequency response characteristics when the load changes and the island is connected to the grid;

[0042] The output active power of the present invention has a better ability to track the trajectory of a given value; the three-phase output current of the virtual synchronous generator transitions more smoothly when the grid-connected state and the output power set value change. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solution of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only one embodiment of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0044] Figure 1 is a flow chart of an inertia damping adaptive control method in an embodiment of the present invention;

[0045] Figure 2 is a power angle characteristic curve of a virtual synchronous generator in an embodiment of the present invention;

[0046] Figure 3 It is a diagram of the architecture of the fuzzy neural network model (ANFIS) in an embodiment of the present invention;

[0047] Figure 4 is a membership function diagram of input variables when the virtual inertia coefficient is adaptively adjusted in an embodiment of the present invention;

[0048] Figure 5 is a membership function diagram of input variables when the virtual damping coefficient is adaptively adjusted in an embodiment of the present invention;

[0049] Figure 6 is the fuzzy inference control surface of the virtual inertia change in the embodiment of the present invention;

[0050] Figure 7 is the fuzzy inference control surface of the virtual damping variation in the embodiment of the present invention;

[0051] Figure 8 is a training error curve of the fuzzy neural network model in an embodiment of the present invention;

[0052] Fig. 9 is a control block diagram of ANFIS+virtual synchronous generator (VSG) in an embodiment of the present invention;

[0053] Fig.10 is a control block diagram of a direct-drive permanent magnet wind power system in an embodiment of the present invention;

[0054] Fig.11 is a frequency variation curve when the load increases in the island operation mode in an embodiment of the present invention;

[0055] Fig.12 is a frequency variation curve when the island operation mode is converted to the grid-connected mode in an embodiment of the present invention;

[0056] Fig.13 is a curve showing a change in output active power when the load increases in an island operation mode in an embodiment of the present invention;

[0057] Fig.14 is an output active power variation curve when the island operation mode is converted to the grid-connected mode in an embodiment of the present invention;

[0058] Fig.15 is a curve of active power change of VSG output when the active power given value changes in the embodiment of the present invention;

[0059] Fig.16 is a system frequency variation curve when the active power given value changes in the embodiment of the present invention;

[0060] Fig.17 It is the traditional dual adaptive control output three-phase current in the embodiment of the present invention;

[0061] Fig.18 It is the ANFIS-VSG dual adaptive control output three-phase current of the present invention in the embodiment of the present invention;

[0062] Fig.19 is a curve showing a change in virtual inertia coefficient in an embodiment of the present invention;

[0063] Fig. 20 It is a virtual damping coefficient variation curve in the embodiment of the present invention. DETAILED DESCRIPTION

[0064] The following is a clear and complete description of the technical solutions in the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0065] The technical solution of the present application is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0066] Example 1

[0067] like Figure 1 As shown, an inertia damping adaptive control method for a direct-drive permanent magnet wind power system provided by an embodiment of the present invention comprises the following steps:

[0068] Step 1: Construct a mathematical model of a virtual synchronous generator.

[0069] In a specific implementation of the present invention, the expression of the virtual synchronous generator mathematical model is:

[0070]

[0071] Δω=ω-ω ref (2)

[0072] Where, J represents the virtual inertia coefficient of the virtual synchronous generator, D represents the virtual damping coefficient of the virtual synchronous generator, ω represents the mechanical angular velocity of the virtual synchronous generator, ω ref represents the rated angular velocity of the virtual synchronous generator, T m , T e They represent the mechanical torque and electromagnetic torque of the virtual synchronous generator respectively, and Δω represents the angular velocity deviation.

[0073] Step 2: Analyze the grid frequency deviation Δf and the grid frequency deviation change rate based on the virtual synchronous generator mathematical model The relationship between the virtual inertia change ΔJ and the virtual damping change ΔD.

[0074] Figure 2 The power angle characteristic curve of the virtual synchronous generator is shown, where δ represents the power angle of the virtual synchronous generator (referring to the phase difference between the generator output voltage and the grid voltage vector, reflecting the synchronous state of the generator in the grid), and P represents the output active power of the virtual synchronous generator. According to the power angle characteristic curve, the changes of power and angular velocity in the four intervals on the power angle characteristic curve have different trends, and the virtual inertia change and virtual damping change will also change accordingly, thereby obtaining the relationship between the grid frequency deviation and the grid frequency deviation change rate and the virtual inertia change and virtual damping change shown in Table 1.

[0075] Table 1 Selection principles of virtual inertia change and virtual damping change under different conditions

[0076]

[0077] Step 3: Establish initial fuzzy rules based on the relationship between the grid frequency deviation and the grid frequency deviation change rate and the virtual inertia change and virtual damping change.

[0078] The value ranges of the grid frequency deviation, the grid frequency deviation change rate, the virtual inertia change, and the virtual damping change are divided into 7 segments, and the corresponding fuzzy values ​​of the 7 segments are NB (negative large), NM (negative medium), NS (negative small), ZE (zero), PS (positive small), PM (positive medium), and PB (positive large). Different initial fuzzy rules are obtained according to the relationship between the grid frequency deviation and the grid frequency deviation change rate and the virtual inertia change and the virtual damping change. Exemplarily, if the grid frequency deviation is PB and the grid frequency deviation change rate is PB, the virtual inertia change is PS and the virtual damping change is PS, that is, the initial fuzzy rule is {PB, PB, PS, PS}.

[0079] Step 4: construct a sample data set; wherein each sample in the sample data set includes a grid frequency deviation, a grid frequency deviation change rate, a virtual inertia change, and a virtual damping change.

[0080] Through simulation, the virtual inertia change and virtual damping change at different grid frequency deviations and grid frequency deviation change rates are obtained, thereby constructing a sample data set; among which, the grid frequency deviation and the grid frequency deviation change rate are input quantities, and the virtual inertia change and virtual damping change are output quantities.

[0081] Grid frequency is the most intuitive variable in the power system that reflects the system's operating status. Frequency changes directly reflect the imbalance of the system's active power, so the control of grid frequency deviation and frequency deviation change rate is easier to understand intuitively. Angular frequency is another form of frequency representation, but for engineering applications, grid frequency (Hz) is more common and easier to operate and understand.

[0082] Grid frequency is a parameter actually used by various protection devices and dispatching systems in the power system. Most responses and dispatching controls of the system are based on grid frequency. Therefore, directly using grid frequency as control input can better match the control and protection strategies of the existing system. In actual working conditions, changes in grid frequency usually reflect sudden changes in system load or power generation, and this change will be directly reflected in the deviation change of frequency, thus providing a more direct and effective input for fuzzy neural networks.

[0083] The frequency deviation change can capture the dynamic changes of the grid frequency more quickly. This sensitivity is essential for quickly adjusting the inertia and damping characteristics of the virtual synchronous generator, so that it can better cope with system disturbances. Although the change of angular frequency is related to frequency, its physical meaning is more complicated. Compared with the frequency deviation change, angular frequency cannot directly reflect the fluctuation of electrical parameters in system operation.

[0084] Using grid frequency deviation and frequency deviation change rate as input reduces the conversion steps from angular frequency to frequency, simplifies the structure of the fuzzy neural network controller, reduces the computational complexity, and improves the real-time performance of the controller.

[0085] Step 5: Construct a fuzzy neural network model, initialize the fuzzy neural network model using the initial fuzzy rules, and train the initialized fuzzy neural network model using the sample data set to obtain the target model.

[0086] like Figure 3 As shown in the figure, the fuzzy neural network model includes an input layer, a membership function layer, a fuzzy reasoning layer, a normalization layer and an output layer connected in sequence; wherein x1 and x2 represent input quantities, A1, A2, B1 and B2 represent language value variables, w1 and w2 represent rule applicability, It represents the ratio of the applicability of the first and second rules to the sum of the applicability of all rules. and represents the node function, and y represents the output. In the membership function layer, if the membership function is a bell-shaped function (or Gaussian function), its specific expression is:

[0087]

[0088] Among them, x represents the input quantity, a, b, c represent parameters, and μ represents the membership function value.

[0089] Since the fuzzy neural network model combines the self-learning ability of neural network and the fuzzy reasoning ability of fuzzy control, there is no need to consider the mutual constraints of virtual inertia coefficient J and virtual damping coefficient D, nor is there any need to manually design the membership function. The variation law of virtual inertia coefficient J and virtual damping coefficient D can be learned according to the sample data set, and the trained fuzzy neural network model can be used as the target model for virtual synchronous generator control. The fuzzy neural network model outputs the virtual inertia change and virtual damping change in real time, adjusts the frequency deviation of the virtual synchronous generator during load switching online, and reduces the active power deviation.

[0090] Step 6: Add the target model to the virtual synchronous generator to obtain the virtual synchronous generator based on fuzzy neural network.

[0091] Step 7: Use the virtual synchronous generator based on fuzzy neural network to perform adaptive control of the virtual inertia coefficient and the virtual damping coefficient.

[0092] In order to verify the effectiveness of the method of the present invention, Matlab / simulink is used for simulation. The present invention adopts a fuzzy neural network model, and uses a neural network to automatically extract fuzzy rules and fuzzy membership functions for fuzzy control, so that the model becomes an adaptive fuzzy neural network model, and uses the BP back propagation algorithm and the least squares method to learn the relationship between the input and output, extract the corresponding information from the sample data set, and establish the corresponding fuzzy rules, so that the trained fuzzy neural network model can better simulate the relationship between the desired or actual input and output. When learning, the adaptive fuzzy neural network model automatically calculates the learning error according to the actual output value and the expected output value, and automatically adjusts the parameters of the fuzzy neural network model through the error back propagation algorithm to meet the goal of minimizing the error. The adjusted parameters mainly include: weight, Gaussian function center and width.

[0093] Since the back propagation algorithm requires the gradient information of each neuron's transfer function, when the neuron input is too large, the corresponding gradient value of the independent variable at that point is too small, and the weight and threshold cannot be adjusted smoothly. Therefore, before training, the input needs to be normalized to ensure convergence.

[0094] The fuzzy neural network model trained by the present invention is a first-order Sugeno model, the fuzzy subsets of the input quantity are 7, the membership functions are all Gaussian functions, the allowable error is 0.005, and the number of training times is 500. The input quantity includes the frequency deviation Δf and the frequency deviation change rate The output quantities are the virtual inertia change ΔJ and the virtual damping change ΔD. The input and output quantities in the sample data set are obtained from the PID simulation control experiment. Figure 4 and Figure 5The input membership function of ANFIS is shown, and the corresponding fuzzy rules can be directly generated after training without referring to the inherent fuzzy control rule table. The membership function is used to describe the degree to which a variable in fuzzy logic belongs to a certain set, and its value range is usually between 0 and 1. In fuzzy logic, there are multiple membership function models. The membership function of the present invention is a bell-shaped function or a Gaussian function. inAmfB is a specific membership function model. inAmfB represents the Bth membership function (mfB) of the Ath input (inA). The Bth membership function may be a bell-shaped function. For example, in2mf4 represents the 4th membership function (mf4) of the 2nd input (in2). In practical applications, inAmfB may be used to describe how the input variables in the fuzzy rules affect the output results through the membership function. Each membership function has its own parameters (such as center position and width). Through these membership functions, the fuzzy neural network can process and analyze uncertain or fuzzy data. This membership function is usually used to describe the change process of a variable from completely not belonging to a set to completely belonging to a set. Its function form can be linear or nonlinear, depending on the application scenario and requirements.

[0095] In the design of fuzzy logic controllers (such as PID controllers), the proper selection of membership functions is crucial to the performance of the control system. Different membership functions can provide different control fineness, thus affecting the response speed and stability of the system.

[0096] This embodiment uses the Gaussian function as the membership function. Since the Gaussian function has good smoothness and continuity, it can provide a smooth output response when the input variable changes. This characteristic is very important in the control process because it can prevent the system from producing violent fluctuations under small changes in the input signal and ensure the stability of the controller output. The curve shape of the Gaussian function can effectively and more accurately describe the uncertainty and fuzziness of the input variable. It has a more delicate response to the gradual change of the input variable and is suitable for dealing with those control problems with strong fuzziness and nonlinearity, so that the fuzzy control system can better cope with the complex dynamic characteristics in the power system. The symmetry and parameter adjustment of the Gaussian function give it better generalization ability. Even in the case of diverse or irregular input data distribution, it can adapt to various input distributions by adjusting the mean and standard deviation parameters, which enables the system to flexibly respond to various working conditions and maintain good control effects. The two core parameters of the Gaussian function - mean (center position) and standard deviation (width) - can be adjusted relatively simply to optimize the response speed and accuracy of the control system; by adjusting these two parameters, the system can improve its adaptability to different disturbance conditions while ensuring the control accuracy, simplifying the design and tuning process of the controller. Compared with some other types of membership functions (such as trigonometric functions and trapezoidal functions), the calculation of Gaussian functions is relatively simple, and its derivatives have good mathematical properties, which helps to achieve efficient gradient calculation and parameter optimization in neural networks, thereby improving the computational efficiency and real-time performance of the entire fuzzy neural network controller. Gaussian functions show stronger robustness in the presence of noise. Due to its smooth curve form, the system can better resist the influence of external disturbances or measurement noise, ensuring that the controller can still provide stable control performance in a complex power grid environment.

[0097] The defuzzification method adopted in the present invention is the weighted average method, which is relatively simple to calculate compared with the commonly used median method, weakens the role of the extreme value membership point, and makes the result of clarification more accurate, making the control surface of the output virtual inertia change ΔJ and virtual damping change ΔD smoother, achieving better control effect. Figure 6 and Figure 7 shown.

[0098] The weighted average method is a simple and efficient defuzzification method. The calculation process only involves the sum of the product of the weight and membership value of the fuzzy output, and the calculation complexity is low. Compared with other complex defuzzification methods, such as the maximum membership method or the center of gravity method, the use of the weighted average method can greatly reduce the burden of real-time calculation, especially in scenarios where the control system needs to respond quickly, it can improve the calculation efficiency and real-time performance of the system. The weighted average method can generate a smooth control signal by weighting the output of all fuzzy rules. This smooth transition characteristic helps to avoid violent oscillation or jitter of the control system caused by sudden changes in the output of some rules, ensures that the output of the system is more stable, and reduces discontinuities in the control response. Since the weighted average method considers the output of all rules, not just the results of the strongest rule, it can more accurately reflect the behavioral characteristics of complex systems. In this way, even when the input changes greatly, the system can still comprehensively consider the influence of multiple fuzzy rules to ensure that the output results have high adaptability and accuracy. In the weighted average method, the weight of each fuzzy rule is determined by the degree of membership, which can reflect the contribution of different rules to the final decision; rules with high membership will have greater weights. This feature allows the weighted average method to automatically adjust the importance of each rule according to the input situation, ensuring that the most relevant rules dominate the final output, while the influence of irrelevant rules on the results is weakened. Unlike the method of selecting only a single output value, the weighted average method combines the outputs of multiple fuzzy rules, which can reduce the control error caused by the failure of a single rule or excessive deviation. Therefore, the weighted average method has better robustness and fault tolerance, can alleviate the random disturbances and uncertainties in the system to a certain extent, and improve the stability and reliability of the overall control system. The weighted average method can provide a more linear defuzzification result, especially suitable for control scenarios where continuous input produces continuous output. This linear characteristic makes the weighted average method very suitable for dynamic control systems, which is conducive to the system maintaining a stable control effect in a large range of input changes, that is, it has linear output controllability.

[0099] Figure 8 The error of the fuzzy neural network model is shown. After 500 training times, the training error value has dropped below 0.025°, which improves the accuracy of the ANFIS-VSG inertia damping dual adaptive control.

[0100] Example 2

[0101] The embodiment of the present invention provides a virtual synchronous generator control method for a direct-drive permanent magnet wind power system, wherein the virtual synchronous generator includes a virtual speed regulator, a virtual excitation regulator, and a target model obtained by the inertia damping adaptive control method in embodiment 1 of the present application; the virtual speed regulator is used to provide a virtual mechanical power P m, in order to achieve the regulation of the inverter bridge output active power; the virtual excitation regulator is used to obtain the stator winding terminal voltage U0 to achieve the stability of the inverter bridge output voltage and the regulation of reactive power balance. Considering that the equivalent output resistance of the inverter bridge main circuit is very small, in order to simplify the power calculation, the output active power P of the inverter bridge is used to replace the stator virtual electromagnetic power P e .like Fig. 9 As shown, the virtual synchronous generator control method includes the following steps:

[0102] Step 1: Obtain the output active power P of the inverter bridge and the virtual mechanical power P obtained by using the virtual speed regulator m , according to the virtual mechanical power P m The power difference is calculated by comparing the output active power P.

[0103] Step 2: Obtain the grid frequency deviation Δf and the grid frequency deviation change rate According to the grid frequency deviation Δf, the grid frequency deviation change rate And the target model, the virtual inertia change ΔJ and the virtual damping change ΔD are obtained.

[0104] Step 3: Calculate the angular velocity deviation Δω based on the power difference, the virtual inertia change ΔJ and the virtual damping change ΔD (see formula (1) and formula (2)).

[0105] Step 4: According to the obtained rated angular velocity ω ref And the phase angle θ is calculated from the angular velocity deviation Δω.

[0106] Step 5: Get the stator current i of the permanent magnet synchronous generator L , use the virtual excitation regulator to obtain the stator winding terminal voltage U0, according to the stator current i L , phase angle θ and stator winding terminal voltage U0 to calculate the reference voltage u ref .

[0107] Figure 4 In the equation, s represents the Laplace operator, ω represents the angular velocity, E represents the induced electromotive force of the stator winding, and R a represents the filter resistance (equivalent to the stator armature resistance), X t represents filter reactance and j represents an imaginary unit.

[0108] Example 3

[0109] like Fig.10As shown, a direct-drive permanent magnet wind power system provided by an embodiment of the present invention includes a wind turbine, a permanent magnet synchronous generator, a rectifier bridge, an inverter bridge, a modulator, a driver and a controller. The wind turbine is connected to the permanent magnet synchronous generator, the permanent magnet synchronous generator is connected to the rectifier bridge, the rectifier bridge is connected to the inverter bridge through a DC capacitor, and the inverter bridge is connected to the AC load through an LCL filtering circuit; the controller is connected to the modulator, the modulator is connected to the driver, and the driver is connected to the power switch in the inverter bridge. Figure 5 In, e abc 、u abc and i abc They are equivalent to the induced electromotive force, output terminal voltage and grid-connected current of the permanent magnet synchronous generator respectively. PCC represents the common coupling point, C represents the filter capacitor, L1 and L2 represent the filter inductor, and R i Indicates the internal resistance of the filter circuit. abc and i abc After Park transformation, the output active power P and reactive power Q of the grid-connected inverter bridge can be obtained by calculation.

[0110] The controller includes an acquisition unit, a calculation unit, a virtual synchronous generator and a voltage and current loop; the acquisition unit is used to obtain the output voltage u of the inverter bridge abc and grid current i abc ; The calculation unit is used to calculate the output voltage u abc and grid current i abc Calculate the output active power P of the inverter bridge (replace the stator virtual electromagnetic power P with the output active power P e ); The virtual synchronous generator is used to calculate the reference voltage u using the virtual synchronous generator control method of Example 2 of the present application ref ; The voltage and current loop is used to calculate the current according to the reference voltage u ref Get the three-phase voltage command U abc_ref , so that the output voltage of the inverter bridge responds quickly and achieves accurate tracking of the three-phase voltage instructions.

[0111] The modulator is used to generate a modulation wave under the action of the three-phase voltage command. In this embodiment, the modulator uses a sinusoidal pulse width modulation (SPWM) module to convert the three-phase voltage command U output by the voltage and current loop into abc_ref Generate a pulse width modulation signal to drive the power switch of the inverter bridge to complete the three-phase power output. The driver is used to control the opening and closing of the power switch of the inverter bridge under the drive of the modulation wave to achieve three-phase power output control.

[0112] In order to verify the correctness and feasibility of the present invention, a Fig.10 The control model shown in the figure and the main simulation parameters are shown in Table 2.

[0113] Table 2 Main parameters of system simulation

[0114] Parameter name Specific value <![CDATA[DC voltage U dc > 700V <![CDATA[The effective value U of the single-phase line voltage abc > 380V <![CDATA[The inductive reactance L of the filter circuit is L = L1 + L2]]> 1mH <![CDATA[Internal resistance R of the filter circuit i > 0.5Ω Filter circuit filter capacitor C 40μF <![CDATA[Rated angular velocity ω ref > 314rad / s <![CDATA[Frequency deviation feedback coefficient k f > 2000 <![CDATA[Active and reactive power deviation feedback coefficient k p > 0.0005

[0115] Condition 1: During the simulation setting of 0 to 0.6s, the system initially runs without load and is in island operation mode. A 15kW load is added at 0.2s and connected to the grid at 0.5s. This process studies the system load change and the frequency stability problem when the system switches from island operation mode to grid operation mode. The simulation comparison results with the traditional dual adaptive method (i.e., document 1 in the background technology) are shown as follows: Fig.11 and Fig.12 shown.

[0116] analyze Fig.11 It can be seen that the dynamic performance of the system of the present invention is improved. When the system is in the island operation mode at 0.2s, a large-capacity load of 10kW is put into operation, causing the frequency to increase. Compared with the traditional dual adaptive method, the present invention restores the frequency to stability at a faster speed, and recovers to 50Hz at 0.26s, while the traditional dual adaptive method adjusts to 50Hz at 0.29s; at the same time, the present invention limits the oscillation amplitude to a smaller range, and the frequency oscillation amplitude is 0.11Hz smaller than that of the traditional dual adaptive method. The present invention effectively suppresses the interference of the large-capacity load to the system in a short time.

[0117] Depend on Fig.12 It can be seen that when the system enters the grid-connected operation mode from the island operation mode at 0.5s, the system frequency drops slightly. The frequency of the present invention drops by 0.11Hz and recovers to stability at 0.6s, while the frequency of the traditional dual adaptive method drops by 0.15Hz and does not recover to stability until 0.63s. It can be seen that when the system in the island operation mode is connected to the grid, the present invention can also reduce frequency fluctuations and speed up frequency response speed, and has stronger anti-interference ability.

[0118] Depend on Fig.13 and Fig.14 It can be seen that when the system is loaded from the island mode at 0.2s, the present invention achieves active power balance at a speed only 0.002s ahead of the traditional dual adaptive method; and when the island is connected to the grid at 0.5s, the VSG output active power of the present invention and the traditional dual adaptive method decreases almost at the same time, but the active power of the present invention decreases slightly less and recovers slightly faster. It can be seen that the power-frequency characteristics of the present invention in the island mode are slightly better than those of the traditional dual adaptive method.

[0119] Condition 2: At 0.6s, the output active power reference value is set to 10kW, and at 1s, the output active power set value is suddenly changed to -10kW. This process detects the dynamic tracking performance of the controller. The simulation comparison results with the traditional dual adaptive method (i.e., document 1 in the background technology) are as follows: Fig.15 and Fig.16 shown.

[0120] like Fig.15 As shown, in order to verify the dynamic tracking performance of the system of the present invention, the given active power is set to 10kW when the system runs to 0.6s, and the given active power is set to -10kW at 1s. From the simulation results, it can be seen that during the period of 0.6s to 1.45s, the error between the given value of the present invention and the actual value when they are stable is less than 0.05kW, while the error between the given value of the traditional dual adaptive method and the actual value when they are stable is less than 1kW. It can be seen that the present invention has better tracking of the given active power, higher control accuracy, and faster tracking speed.

[0121] analyze Fig.16 It can be seen that when the output active power setting value changes, the system frequency is disturbed and changes accordingly. At 0.6s, compared with the traditional dual adaptive method, the overshoot of the present invention is reduced by 0.06Hz, and the response time is shortened by 55.26%; at 1s, the frequency overshoot of the present invention is reduced by 0.03Hz, and the response time is shortened by 20.83%. It can be seen that when the system output active power setting value changes, the present invention has better power-frequency performance.

[0122] Fig.17 and Fig.18 The output current comparison between the conventional dual adaptive method and the method of the present invention is shown. When the island is connected to the grid at 0.5s, and the given active power is changed at 0.6s and 1.0s, compared with the conventional dual adaptive method, the maximum current amplitude of the present invention changes less, which is 10A, while the maximum current amplitude change of the conventional dual adaptive method reaches 25A. It can be seen that the output current of the present invention can reach a stable state more smoothly.

[0123] Fig.19 and Fig. 20 The changes of virtual inertia coefficient and virtual damping coefficient under the coordinated adaptive control strategy are given. Fig.19 and Fig. 20 It can be seen that when the operating conditions change at 0.2s, 0.5s, 0.6s and 1.0s, the virtual inertia coefficient and the virtual damping coefficient are adaptively adjusted accordingly; when the load is put into operation at 0.2s, the system frequency increases, and the virtual inertia coefficient and the virtual damping coefficient of the VSG produce a larger adjustment compared with other time points; when the island is connected to the grid at 0.5s and the output active power setting value is changed at 0.6s and 1s, the virtual inertia coefficient and the virtual damping coefficient have continuous and smooth characteristics, which can make the inverter bridge of the wind power system stabilize to the target value as soon as possible during the adjustment process and reduce the active power overshoot.

[0124] Practice shows that when frequency deviation and frequency deviation change rate are used as control inputs, fuzzy neural networks can more accurately adjust the output inertia and damping characteristics of virtual synchronous generators, thereby better maintaining the system frequency stability and improving the dynamic response effect. The change of frequency deviation can provide an opportunity to predict the dynamic characteristics of the system in advance, so that the controller can make adjustments in advance before the system undergoes large fluctuations, thereby avoiding excessive frequency fluctuations. When the output power set value changes, the present invention can quickly track the set value through adaptively adjusted inertia and damping parameters to ensure the accuracy and stability of the output active power. In addition, the nonlinear processing capability of fuzzy neural networks makes the control strategy of the present invention particularly suitable for complex and time-varying power grid environments, thereby improving the robustness of the system. Through this intelligent and dynamically adjusted control mechanism, the virtual synchronous generator can maintain the stability of the current transition process and the stable operation of the overall system even in the face of frequent power grid fluctuations or power adjustment requirements in the grid-connected state, thereby effectively enhancing the operational reliability and adaptability of the grid-connected power generation system.

[0125] Example 4

[0126] An embodiment of the present invention also provides an electronic device, which includes: a memory, a processor, and a computer program / instructions stored in the memory, and the processor executes the computer program / instructions to implement the inertia damping adaptive control method or the virtual synchronous generator control method in the embodiment of the present application.

[0127] Although not shown, the electronic device includes a processor, which can perform various appropriate operations and processes according to the program and / or data stored in the read-only memory (ROM) or the program and / or data loaded from the storage portion into the random access memory (RAM). The processor can be a multi-core processor, or it can include multiple processors. In some embodiments, the processor can include a general main processor and one or more special coprocessors, such as a central processing unit, a graphics processing unit (GPU), a neural network processor (NPU), a digital signal processor (DSP), etc. In RAM, various programs and data required for device operation are also stored. The processor, ROM and RAM are connected to each other via a bus. The input / output (I / O) interface is also connected to the bus.

[0128] The processor and memory are used together to execute the program / instructions stored in the memory. When the program / instructions are executed by the computer, the methods, steps or functions described in the above embodiments can be implemented.

[0129] Although not shown, an embodiment of the present invention further provides a computer-readable storage medium having a computer program / instruction stored thereon, which, when executed by a processor, implements the inertia damping adaptive control method or the virtual synchronous generator control method in the embodiment of the present application.

[0130] Storage media in embodiments of the present invention include permanent and non-permanent, removable and non-removable items that can be implemented by any method or technology to store information. Examples of storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission medium that can be used to store information that can be accessed by a computing device.

[0131] Readable storage media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include temporary computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0132] Although not shown, an embodiment of the present invention further provides a computer program product, including: a computer program / instruction, which, when executed by a processor, implements the inertia damping adaptive control method or the virtual synchronous generator control method in the embodiment of the present application.

[0133] What is disclosed above is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or modifications within the technical scope disclosed in the present invention, which should be covered within the protection scope of the present invention.

Claims

1. An inertia damping adaptive control method for a direct-drive permanent magnet wind power system, characterized in that: The method comprises: Construct a mathematical model of a virtual synchronous generator; According to the virtual synchronous generator mathematical model, the relationship between the grid frequency deviation and the grid frequency deviation change rate and the virtual inertia change and virtual damping change is analyzed; Establishing initial fuzzy rules according to the relationship between the grid frequency deviation and the grid frequency deviation change rate and the virtual inertia change and virtual damping change; Constructing a sample data set; wherein each sample in the sample data set includes a power grid frequency deviation, a power grid frequency deviation change rate, a virtual inertia change, and a virtual damping change; Constructing a fuzzy neural network model, initializing the fuzzy neural network model using the initial fuzzy rules, and training the initialized fuzzy neural network model using the sample data set to obtain a target model; Adding the target model to the virtual synchronous generator to obtain the virtual synchronous generator based on fuzzy neural network; The virtual synchronous generator based on fuzzy neural network is used to perform adaptive control of virtual inertia coefficient and virtual damping coefficient.

2. The inertia damping adaptive control method according to claim 1, characterized in that: The expression of the virtual synchronous generator mathematical model is: Give = oh-oh ref ; Where, J represents the virtual inertia coefficient of the virtual synchronous generator, D represents the virtual damping coefficient of the virtual synchronous generator, ω represents the mechanical angular velocity of the virtual synchronous generator, ω ref represents the rated angular speed of the virtual synchronous generator, T m , T e They represent the mechanical torque and electromagnetic torque of the virtual synchronous generator respectively.

3. The inertia damping adaptive control method according to claim 1, characterized in that: The relationship between the grid frequency deviation and the grid frequency deviation change rate and the virtual inertia change and virtual damping change specifically includes: When Δf>0, and When , the change of virtual inertia increases, and the change of virtual damping increases; When Δf>0, and When , the change of virtual inertia decreases, and the change of virtual damping increases; When Δf<0, and When , the change of virtual inertia increases, and the change of virtual damping increases; When Δf<0, and When , the change of virtual inertia decreases, and the change of virtual damping increases; Where Δf represents the grid frequency deviation, Indicates the rate of change of grid frequency deviation.

4. The inertia damping adaptive control method according to claim 1, characterized in that: The fuzzy neural network model comprises an input layer, a membership function layer, a fuzzy reasoning layer, a normalization layer and an output layer which are connected in sequence; in the membership function layer, a Gaussian function is used as the membership function.

5. The inertia damping adaptive control method according to claim 1, characterized in that: The fuzzy neural network model is defuzzified by using a weighted average method.

6. A virtual synchronous generator control method for a direct-drive permanent magnet wind power system, characterized in that: The virtual synchronous generator includes a virtual speed regulator, a virtual excitation regulator, and a target model obtained by using the inertia damping adaptive control method according to any one of claims 1 to 5; the control method includes: Obtain the output active power P of the inverter bridge and the virtual mechanical power P obtained by using the virtual speed regulator m , according to the virtual mechanical power P m Calculate the power difference with the output active power P; Obtain the grid frequency deviation Δf and the grid frequency deviation change rate According to the grid frequency deviation Δf, the grid frequency deviation change rate and the target model, obtaining a virtual inertia change ΔJ and a virtual damping change ΔD; Calculating an angular velocity deviation Δω according to the power difference, the virtual inertia change ΔJ and the virtual damping change ΔD; According to the obtained rated angular velocity ω ref and the angular velocity deviation Δω to calculate the phase angle θ; Get the stator current i of the permanent magnet synchronous generator L , using the virtual excitation regulator to obtain the stator winding terminal voltage U0, according to the stator current i L , phase angle θ and stator winding terminal voltage U0 to calculate the reference voltage u ref .

7. A direct-drive permanent magnet wind power system, the system comprising a wind turbine, a permanent magnet synchronous generator, a rectifier bridge, an inverter bridge, a modulator, a driver and a controller, the wind turbine is connected to the permanent magnet synchronous generator, the permanent magnet synchronous generator is connected to the rectifier bridge, the rectifier bridge is connected to the inverter bridge via a DC capacitor, the inverter bridge is connected to an AC load via a filter circuit; the controller is connected to the modulator, the modulator is connected to the driver, and the driver is connected to a power switch in the inverter bridge; characterized in that: The controller includes an acquisition unit, a calculation unit, a virtual synchronous generator and a voltage and current loop; the acquisition unit is used to acquire the output voltage and grid-connected current of the inverter bridge; the calculation unit is used to calculate the output active power P of the inverter bridge according to the output voltage and grid-connected current; The virtual synchronous generator is used to calculate the reference voltage u using the virtual synchronous generator control method as claimed in claim 6 ref The voltage and current loop is used according to the reference voltage u ref Get three-phase voltage instructions; The modulator is used to generate a modulation wave under the action of the three-phase voltage instruction; The driver is used to control the opening and closing of the power switch of the inverter bridge under the drive of the modulation wave.

8. An electronic device comprising a memory, a processor, and a computer program / instruction stored in the memory, characterized in that: The processor executes the computer program / instructions to implement the inertia damping adaptive control method according to any one of claims 1 to 5 or the virtual synchronous generator control method according to claim 6.

9. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instruction is executed by a processor, the inertia damping adaptive control method according to any one of claims 1 to 5 or the virtual synchronous generator control method according to claim 6 is implemented.

10. A computer program product comprising a computer program / instructions, characterized in that: When the computer program / instruction is executed by a processor, the inertia damping adaptive control method according to any one of claims 1 to 5 or the virtual synchronous generator control method according to claim 6 is implemented.

Images