A method and system for maximum wind energy capture control of a wind turbine

Abstract

The present invention discloses a method and system for maximum wind energy capture control of a wind turbine. The inverse model of the wind turbine is online identified by using a BP neural network; the wind turbine is controlled in real time, the aerodynamic torque of the wind turbine is estimated online, and the physical relationship between the aerodynamic torque of the wind turbine and the wind speed is iteratively solved by using the Newton-Raphson method to obtain the wind speed estimate value in real time; the reference speed is calculated according to the wind speed estimate value; the reference speed, the wind turbine speed at the previous moment, the generator electromagnetic torque at the previous moment, and the estimated value of the aerodynamic torque are used as the inputs of the inverse model of the wind turbine to calculate the torque control amount of the inverse model of the wind turbine at the current moment; the closed-loop torque control amount is calculated by using a BP neural network controller; the torque control amount of the inverse model of the wind turbine at the current moment and the closed-loop torque control amount are added to obtain the combined electromagnetic torque control amount, and the combined torque setting target is realized through a converter, so as to realize the maximum wind energy capture control of the wind turbine. The logic is simple and easy to implement, the response is fast, and the stability is good.

Description

Technical Field

[0001] The invention belongs to the technical field of wind power generation, and particularly relates to a maximum wind energy capture control method and system for a wind turbine generator set. Background Art

[0002] Improving the wind energy utilization efficiency is an eternal topic in the development of wind power. The amount of energy captured by a wind turbine generator set from wind energy mainly depends on the wind turbine speed, wind speed and pitch angle. Pitch control is generally used to limit the speed of the wind turbine to prevent failures caused by overspeed and excessive aerodynamic noise. Below the rated wind speed, pitch control is generally not enabled. At this time, the power coefficient is a single-valued function of the tip speed ratio, that is, each wind speed below the rated value corresponds to an optimal wind turbine speed, making the wind energy utilization efficiency the highest. The purpose of the maximum wind energy capture control of a wind turbine generator set is to adjust the electromagnetic torque of the generator, so that the wind turbine speed changes towards the optimal speed value, thereby achieving maximum wind energy capture.

[0003] The maximum wind energy capture control of a wind turbine generator set mainly includes the optimal torque method, the tip speed ratio method and the hill climbing method. The tip speed ratio method calculates the reference speed according to the measured wind speed value. The common wind speed sensor is installed in the nacelle, but it can only measure a single point. Wind shear, wake effect, wind speed turbulence characteristics and the inertia of the anemometer all make it difficult for the wind speed sensor to measure an accurate wind speed signal. Different installation positions result in different wind speed signals, and the measured signal cannot accurately reflect the effective wind speed acting on the entire wind turbine. Therefore, the tip speed ratio method is rarely used in large wind turbine generator sets. The hill climbing method, also known as the perturbation observation method, does not depend on system parameters. It adjusts the wind turbine speed towards the maximum power by comparing the relationship between the output powers of the generators. When the moment of inertia of the wind turbine is large, the tracking speed of the hill climbing method is slow. Therefore, it is mainly applied to small wind turbine generator sets. The optimal torque method is the mainstream method for maximum power point tracking of large wind turbine generator sets. The traditional optimal torque method is based on the steady-state optimal curve and ignores the dynamic processes of different steady-state operating points. From the perspective of the control system, the traditional optimal torque method exhibits significant nonlinear and large inertia characteristics, that is, the wind turbine response lags severely at low wind speeds, and the degree of response lag decreases with the increase of wind speed. On the other hand, the wind has strong randomness, and there is no absolute steady state in the actual working conditions. The wind turbine generator set is also in a dynamic process at all times. The optimal torque method based on the steady-state assumption will surely cause energy loss. Therefore, designing a non-linear maximum wind energy capture control method considering the unsteady state of the unit is of great significance for improving the dynamic response performance of wind energy capture and the wind energy utilization efficiency of wind turbine generator sets. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a maximum wind energy capture control method and system for a wind turbine generator set in view of the deficiencies in the above-mentioned prior art, so as to solve the technical problems of low dynamic response performance of wind energy capture and low wind energy utilization efficiency of wind turbine generator sets.

[0005] The present invention adopts the following technical solutions:

[0006] A maximum wind energy capture control method for a wind turbine, comprising the following steps:

[0007] S1. Online identify the inverse model of the wind turbine using a BP neural network;

[0008] S2. Conduct real-time control on the wind turbine, read the wind turbine speed and the electromagnetic torque signal of the generator, online estimate the aerodynamic torque of the wind turbine, and use the Newton-Raphson method to iteratively solve the physical relationship between the aerodynamic torque of the wind turbine and the wind speed to obtain the wind speed estimate in real time;

[0009] S3. Design a reference model and calculate the reference speed based on the wind speed estimate obtained in step S2;

[0010] S4. Use the estimated value of the aerodynamic torque of the wind turbine, the reference speed obtained in step S3, the wind turbine speed at the previous moment obtained in step S2, and the electromagnetic torque signal of the generator at the previous moment as the inputs of the inverse model of the wind turbine obtained in step S1, and calculate the torque control amount of the inverse model of the wind turbine at the current moment;

[0011] S5. Subtract the wind turbine speed obtained in step S2 from the reference speed obtained in step S3 to obtain the speed control error;

[0012] S6. Based on the speed control error obtained in step S5, use a BP neural network controller to calculate the closed-loop torque control amount;

[0013] S7. Add the torque control amount of the inverse model of the wind turbine at the current moment obtained in step S4 to the closed-loop torque control amount obtained in step S6 to obtain the combined electromagnetic torque control amount, and realize the combined torque setting target through an inverter to achieve the maximum wind energy capture control of the wind turbine.

[0014] Specifically, in step S1, the online identification of the inverse model of the wind turbine using a BP neural network is specifically as follows:

[0015] Operate the wind turbine under the traditional torque control strategy; read the wind turbine speed signal a and the electromagnetic torque signal b of the generator in real time; combine the discrete state space equation of the wind turbine drive train, and based on the real-time wind turbine speed signal a and the electromagnetic torque signal b of the generator, use an unscented Kalman filter to perform online unbiased estimation of the aerodynamic torque of the wind turbine to obtain the estimated value c of the aerodynamic torque of the wind turbine; use a BP neural network to perform online identification of the inverse model of the controlled wind turbine, use the difference between the electromagnetic torque b(k) at the current moment and the neural network output as the error to train the weights, and use the trained neural network as the inverse model of the wind turbine.

[0016] Furthermore, the model structure of the controlled wind turbine is:

[0017] a(k) = f[a(k - 1), b(k), b(k - 1), c(k)]

[0018] Where f represents a function, a(k) is the rotational speed at the current moment, a(k - 1) is the rotational speed at the previous moment, b(k) is the electromagnetic torque at the current moment, b(k - 1) is the electromagnetic torque at the previous moment, and c(k) is the aerodynamic torque at the current moment.

[0019] Furthermore, the output of the wind turbine inverse model is expressed as:

[0020] Net[a d (k), a(k - 1), b(k - 1), c(k)]

[0021] Where a(k) is the rotational speed at the current moment, a(k - 1) is the rotational speed at the previous moment, b(k - 1) is the electromagnetic torque at the previous moment, and c(k) is the aerodynamic torque at the current moment.

[0022] Specifically, in step S2, the wind speed estimated value d satisfies the following relationship:

[0023]

[0024] Where ρ is the incoming wind density; R is the wind turbine radius; C T (λ) is the wind turbine torque coefficient, λ is the ratio of the tip speed to the incoming wind speed; c is the wind turbine aerodynamic torque.

[0025] Furthermore, the online estimation of the wind turbine aerodynamic torque is specifically:

[0026] First, read the wind turbine rotational speed signal a and the generator electromagnetic torque signal b, and then use the unscented Kalman filter to online estimate the wind turbine aerodynamic torque c.

[0027] Specifically, in step S3, the reference model is specifically:

[0028]

[0029] Where λ opt is the optimal tip speed ratio, is the reference model gain; e is the time constant, a d (s) is the Laplace transform of the reference rotational speed, d(s) is the Laplace transform of the wind speed estimated value, and s is the Laplace operator.

[0030] Specifically, in step S4, the torque control quantity g(k) of the wind turbine inverse model at the current moment is:

[0031] g(k) = Net[a d (k), a(k - 1), b(k - 1), c(k)]

[0032] Among them, a d (k) is the reference speed, a(k - 1) is the wind turbine speed at the previous moment, b(k - 1) is the electromagnetic torque of the generator at the previous moment, and c(k) is the estimated value of the aerodynamic torque.

[0033] Specifically, step S6 is specifically as follows:

[0034] The neural network controller takes the reference speed a d (k) obtained in step S3, the wind turbine speed a(k - 1) at the previous moment obtained in step S2, and the closed-loop torque control amount i(k - 1) at the previous moment as inputs, takes the closed-loop torque control amount i(k) as the output, and online trains the weight of the neural network controller with the speed control error h(k) as the target.

[0035] In a second aspect, an embodiment of the present invention provides a maximum wind energy capture control system for a wind turbine, including:

[0036] An identification module that online identifies the inverse model of the wind turbine using a BP neural network;

[0037] An estimation module that performs real-time control on the wind turbine, reads the wind turbine speed and the electromagnetic torque signal of the generator, online estimates the aerodynamic torque of the wind turbine, iteratively solves the physical relationship between the aerodynamic torque of the wind turbine and the wind speed using the Newton-Raphson method, and obtains the wind speed estimation value in real time; designs a reference model and calculates the reference speed based on the wind speed estimation value;

[0038] A calculation module that takes the estimated value of the aerodynamic torque of the wind turbine, the reference speed obtained by the estimation module, the wind turbine speed at the previous moment, and the electromagnetic torque signal of the generator at the previous moment as the inputs of the inverse model of the wind turbine obtained by the identification module, and calculates the torque control amount of the inverse model of the wind turbine at the current moment;

[0039] A network module that subtracts the reference speed obtained by the estimation module from the wind turbine speed to obtain the speed control error, and calculates the closed-loop torque control amount using a BP neural network controller;

[0040] A control module that adds the torque control amount of the inverse model of the wind turbine at the current moment obtained by the calculation module to the closed-loop torque control amount obtained by the network module to obtain the combined electromagnetic torque control amount, and realizes the combined torque setting target through a converter to achieve the maximum wind energy capture control of the wind turbine.

[0041] Compared with the prior art, the present invention has at least the following beneficial effects:

[0042] A maximum wind energy capture control method for a wind turbine, which includes two control loops, namely a control loop based on the inverse model of a BP neural network and a BP neural network control loop based on the backpropagation of the rotational speed error; the method uses a BP neural network to first identify the inverse model, and then designs the inverse model torque control quantity according to the inverse model. Its logic is simple and easy to implement. It directly conducts control without relying on the lagging rotational speed error, has a fast response and good stability. Since the inverse model includes the non-steady-state characteristics of the wind turbine, the dynamic response performance of the torque control quantity designed based on this model is better than that of the traditional optimal torque control based on the steady-state assumption; this method selects a simple and reliable linear reference model and reasonable model gains, which can not only ensure the steady-state control accuracy, but also ensure that the maximum power point tracking of the wind turbine has a constant tracking bandwidth like a linear system in a wide range of wind speeds; the BP neural network control loop based on the backpropagation of the rotational speed error backpropagates the rotational speed error to the hidden layer and the input layer to correct the weights. Its essence is a high-precision feedback control with high control accuracy. Combining the BP neural network based on inverse model identification and the BP neural network controller based on error minimization ensures both stability and steady-state accuracy indicators.

[0043] Furthermore, a BP neural network is used to online identify the inverse model of the wind turbine. The non-linear approximation ability of the neural network is very strong, which is very suitable for the maximum power point tracking system of the wind turbine with significant non-linear characteristics. Its logic is simple, easy to implement, and has high identification accuracy;

[0044] Furthermore, the model structure of the controlled wind turbine is set as a function of the current rotational speed with respect to the rotational speed at the previous moment, the current aerodynamic torque, the electromagnetic torque at the current moment and the previous moment. This model structure has few input and output variables and is capable of fully reflecting the non-linear and time-varying characteristics of the dynamic response of the wind turbine rotor speed;

[0045] Furthermore, taking the rotational speed signals at the current moment and the previous moment, the electromagnetic torque signal at the previous moment and the estimated value of the current aerodynamic torque as the inputs of the inverse model, and the current torque as the output of the inverse model, the relationship between the input and output is identified through BP neural network identification. Then, as long as the desired rotational speed is given, and the rotational speed at the previous moment, the electromagnetic torque signal at the previous moment and the estimated value of the current aerodynamic torque are detected, the required inverse model torque control quantity can be easily calculated through the identified inverse model;

[0046] Furthermore, using the relationship between the wind speed and the aerodynamic torque, the estimation of the wind speed is equivalently converted into the solution of a non-linear equation, and the wind speed estimation is directly realized by using the solution method of the non-linear equation. Its estimation cost is low and there is no need to add a wind speed sensor;

[0047] Furthermore, an unscented Kalman filter is used to achieve unbiased estimation of the aerodynamic torque, which has low estimation cost and does not require additional sensors. At the same time, the unscented Kalman filter is more suitable for state estimation of nonlinear systems than the traditional Kalman filter;

[0048] Furthermore, the gain of the reference model can ensure that the reference speed value a d (k) and the wind speed estimated value d(k) satisfy the optimal tip speed ratio state, thereby ensuring the steady-state accuracy of the maximum power point tracking control. In addition, the first-order linear inertial reference model is simple, reliable, and easy to select, and can ensure that the maximum power point tracking of the wind turbine has a constant tracking bandwidth like a linear system in a wide range of wind speeds.

[0049] Furthermore, the inverse model torque control quantity of the wind turbine is calculated by the inverse model identified by the BP neural network. This control quantity directly controls without relying on the lagging speed error, and has a fast response speed. Because the inverse model includes the non-steady-state characteristics of the wind turbine, the dynamic response performance of the torque control quantity designed based on this model will be significantly better than the optimal torque control based on the steady-state assumption;

[0050] Furthermore, the BP neural network controller based on the minimum speed error feeds back the speed error to the hidden layer and the input layer to correct the weights. In essence, it is a high-precision feedback control, which has high control precision, very low dependence on the model, and strong adaptability to environmental factors and unit parameters.

[0051] It can be understood that the beneficial effects of the second aspect above can be referred to the relevant descriptions in the first aspect above, and will not be elaborated here.

[0052] In summary, the present invention makes full use of the advantages of the nonlinear approximation of the neural network, has a fast response speed and good stability; has high control precision, low dependence on the model, strong adaptability to environmental factors and unit parameters, and at the same time ensures the stability and steady-state accuracy indicators.

[0053] Next, through the drawings and embodiments, the technical solutions of the present invention will be further described in detail. Description of the Drawings

[0054] Figure 1 is the method flow chart of the present invention;

[0055] Figure 2 is the flow chart for obtaining the inverse model of the controlled unit;

[0056] Figure 3 is the real-time control block diagram of the system. Detailed Embodiments

[0057] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0058] In the description of the present invention, it should be understood that the terms "comprising" and "including" indicate the presence of the described features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0059] It should also be understood that the terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an", and "the" are intended to include the plural forms.

[0060] It should be further understood that the term " / and" as used in the specification of the present invention and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations. For example, A and / or B can represent three cases: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this article generally indicates an "or" relationship between the contextually related objects.

[0061] It should be understood that although the terms first, second, third, etc. may be used in the embodiments of the present invention to describe preset ranges, etc., these preset ranges should not be limited to these terms. These terms are only used to distinguish the preset ranges from each other. For example, without departing from the scope of the embodiments of the present invention, the first preset range may also be referred to as the second preset range, and similarly, the second preset range may also be referred to as the first preset range.

[0062] Depending on the context, the word "if" as used herein can be interpreted as "when" or "while" or "in response to determining" or "in response to detecting". Similarly, depending on the context, the phrase "if determined" or "if detecting (stated condition or event)" can be interpreted as "when determined" or "in response to determining" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)".

[0063] Various structural schematic diagrams according to the disclosed embodiments of the present invention are shown in the accompanying drawings. These figures are not drawn to scale, where for the purpose of clear expression, some details are enlarged and some details may be omitted. The shapes of various regions and layers shown in the figures, as well as their relative sizes and positional relationships, are merely exemplary. In practice, there may be deviations due to manufacturing tolerances or technical limitations, and those skilled in the art can design regions / layers with different shapes, sizes, and relative positions according to actual needs.

[0064] The present invention provides a maximum wind energy capture control method for a wind turbine, which uses a BP neural network to online identify the inverse model of the wind turbine; performs real-time control on the wind turbine, online estimates the aerodynamic torque of the wind turbine, uses the Newton-Raphson method to iteratively solve the physical relationship between the aerodynamic torque of the wind turbine and the wind speed, and obtains the wind speed estimate value in real time; designs a reference model, and calculates the reference speed based on the wind speed estimate value; uses the reference speed, the wind turbine speed at the previous moment, the generator electromagnetic torque at the previous moment, and the estimated value of the aerodynamic torque as the input of the inverse model of the wind turbine, and calculates the torque control amount of the inverse model of the wind turbine at the current moment; subtracts the wind turbine speed from the reference speed to obtain the speed control error, and uses a BP neural network controller to calculate the closed-loop torque control amount; adds the torque control amount of the inverse model of the wind turbine at the current moment to the closed-loop torque control amount to obtain the combined electromagnetic torque control amount, and realizes the combined torque setting target through an inverter, so as to achieve the maximum wind energy capture control of the wind turbine; combines the BP neural network based on inverse model identification and the BP neural network controller based on minimum error, which ensures the stability and steady-state accuracy index.

[0065] Please refer to Figure 1 , a maximum wind energy capture control method for a wind turbine of the present invention includes the following steps:

[0066] S1. Use a BP neural network to online identify the inverse model of the wind turbine;

[0067] Please refer to Figure 2 , and the identification process is as follows:

[0068] S101. Operate the wind turbine under the traditional torque control strategy;

[0069] S102. Read the wind turbine speed signal a and the generator electromagnetic torque signal b in real time;

[0070] S103. Combine the discrete state space equation of the wind turbine drive train, and based on the real-time wind turbine speed signal a and the generator electromagnetic torque signal b, use the unscented Kalman filter to online and unbiasedly estimate the aerodynamic torque of the wind turbine, and denote the estimated value of the aerodynamic torque of the wind turbine as c;

[0071] Generally, the state space equation of the continuous transmission chain can be obtained first, and then it can be converted into a discrete model by using linear system theory.

[0072] Advantages of the unscented Kalman filter in realizing unbiased estimation of aerodynamic torque: low cost and no need to add additional sensors; the unscented Kalman filter is more suitable for state estimation of nonlinear systems than the traditional Kalman filter.

[0073] S104. Use a BP neural network to online identify the inverse model of the controlled unit;

[0074] Based on the basic aerodynamic and structural theories of wind turbines, the model structure of the controlled unit is determined as:

[0075] a(k) = f[a(k - 1), b(k), b(k - 1), c(k)]

[0076] where f represents a function.

[0077] The rotational speed a(k) at the current moment is related to the rotational speed a(k - 1) at the previous moment, the electromagnetic torque b(k) at the current moment, the electromagnetic torque b(k - 1) at the previous moment, and the aerodynamic torque c(k) at the current moment. Generally, the adjustment of a(k) is achieved by changing b(k);

[0078] Conversely, given the desired wind turbine rotational speed a d(k) and the current state information a(k - 1), b(k - 1), c(k), the corresponding electromagnetic torque control quantity at the current moment can be obtained. Use a BP neural network to identify this mapping relationship, and take the difference between the electromagnetic torque b(k) at the current moment and the neural network output as the error to train the weights. The trained neural network is the inverse model of the wind turbine, and this inverse model is denoted as Net. The output of Net is expressed as: Net[a d (k), a(k - 1), b(k - 1), c(k)].

[0079] S2. Perform real-time control on the wind turbine. First, read the wind turbine rotational speed signal a and the generator electromagnetic torque signal b, use the unscented Kalman filter to online estimate the wind turbine aerodynamic torque c, and based on the physical relationship between the wind turbine aerodynamic torque c and the wind speed, use the Newton - Raphson method to iteratively solve this physical relationship to obtain the wind speed estimate value d in real time;

[0080] Please refer to Figure 3, the real-time wind speed estimation value is used to calculate the reference speed through a linear reference model. The control objective of the entire control system is that the desired wind turbine speed can be kept consistent with the speed calculated by the linear reference model. The inverse model of the BP neural network trained in S1 detects the wind turbine speed signal at the previous moment, the current moment's estimated wind turbine aerodynamic torque signal estimated online by the unscented Kalman filter, the electromagnetic torque signal at the previous moment, and the reference speed signal at the current moment to calculate the inverse model torque control amount. This control amount is essentially an open-loop control with good stability but insufficient anti-interference ability. The BP neural network controller makes up for this defect. It takes the wind turbine speed signal at the previous moment, the reference speed signal at the current moment, and the closed-loop torque control amount at the previous moment as inputs, uses the speed error as the target to correct the neural network weights, and calculates the closed-loop torque control amount. Its essence is a speed closed-loop control with strong anti-interference ability. The inverse model torque control amount and the closed-loop torque control amount are added to obtain the total torque control amount, and the converter realizes the torque setting requirement.

[0081] The estimation of the real-time wind speed is essentially to solve the non-linear equation of the wind turbine aerodynamic torque c and the wind speed estimation value d:

[0082]

[0083] where ρ is the incoming wind density; R is the wind turbine radius; C T (λ) is the wind turbine torque coefficient, is the ratio of the tip speed to the incoming wind speed; the wind turbine torque coefficient characteristic C T (λ) can be obtained by the steady-state aerodynamic calculation of the wind turbine.

[0084] S3. Design a reference model. According to the wind speed estimation value d(k) obtained in step S2, calculate the reference speed a d (k);

[0085] The reference model selects a typical first-order inertial model:

[0086]

[0087] where λ opt is the optimal tip speed ratio, which is determined by the wind turbine structure and aerodynamic characteristics and is generally provided by the complete machine manufacturer. The reference model gain can ensure that the reference speed value a d (k) and the wind speed estimation value d(k) satisfy the optimal tip speed ratio state under the steady-state wind speed, thereby ensuring the steady-state accuracy of the maximum wind energy capture control; e is the time constant, which determines the dynamic response speed of the maximum power point tracking, and this value should be reasonably selected according to the size of the wind turbine.

[0088] Advantages of choosing the first-order inertial model:

[0089] ①The reference model is simple and reliable;

[0090] ②Ensure the steady-state control accuracy;

[0091] ③The traditional maximum power point tracking of wind turbines is non-linear, and the dynamic tracking effect at low wind speeds is poor. Selecting a linear reference model in the model reference adaptive control can ensure that the maximum power point tracking of wind turbines has a constant tracking bandwidth like a linear system over a wide range of wind speeds.

[0092] S4. Call the Net model identified in step S1, and use the reference speed a d (k) obtained in step S3, the previous wind turbine rotor speed a(k - 1) obtained in step S2, the previous generator electromagnetic torque b(k - 1), and the estimated aerodynamic torque value c(k) as the inputs of the Net model to calculate the inverse model torque control quantity at the current moment;

[0093] The control quantity is denoted as:

[0094] g(k) = Net[a d (k), a(k - 1), b(k - 1), c(k)]

[0095] S5. Subtract the wind turbine rotor speed a(k) obtained in step S2 from the reference speed a d (k) obtained in step S3 to obtain the speed control error h(k);

[0096] S6. Use the BP neural network controller to calculate the closed-loop torque control quantity i(k);

[0097] The neural network controller uses the reference speed a d (k) obtained in step S3, the previous wind turbine rotor speed a(k - 1) obtained in step S2, and the previous closed-loop torque control quantity i(k - 1) as inputs, takes the closed-loop torque control quantity i(k) as the output, and online trains the weights with the speed control error h(k) as the target.

[0098] S7. Add the inverse model torque control quantity g(k) obtained in step S4 to the closed-loop torque control quantity h(k) calculated by the BP neural network controller in step S6 to obtain the combined electromagnetic torque control quantity, and achieve the combined torque setting target through the converter.

[0099] In another embodiment of the present invention, a maximum wind energy capture control system for a wind turbine is provided. This system can be used to implement the above-mentioned maximum wind energy capture control method for a wind turbine. Specifically, the maximum wind energy capture control system for a wind turbine includes an identification module, an estimation module, a calculation module, a network module, and a control module.

[0100] Among them, the identification module uses a BP neural network to identify the inverse model of the wind turbine online;

[0101] The estimation module performs real-time control on the wind turbine, reads the wind turbine speed and the electromagnetic torque signal of the generator, estimates the aerodynamic torque of the wind turbine online, and uses the Newton-Raphson method to iteratively solve the physical relationship between the aerodynamic torque of the wind turbine and the wind speed to obtain the wind speed estimate in real time; a reference model is designed, and the reference speed is calculated based on the wind speed estimate;

[0102] The calculation module uses the estimated value of the aerodynamic torque of the wind turbine, the reference speed obtained by the estimation module, the wind turbine speed at the previous moment, and the electromagnetic torque signal of the generator at the previous moment as the input of the inverse model of the wind turbine obtained by the identification module, and calculates the torque control amount of the inverse model of the wind turbine at the current moment;

[0103] The network module subtracts the reference speed obtained by the estimation module from the wind turbine speed to obtain the speed control error, and uses a BP neural network controller to calculate the closed-loop torque control amount;

[0104] The control module adds the torque control amount of the inverse model of the wind turbine at the current moment obtained by the calculation module to the closed-loop torque control amount obtained by the network module to obtain the combined electromagnetic torque control amount, and realizes the combined torque setting target through the converter to achieve the maximum wind energy capture control of the wind turbine.

[0105] Since the neural network has a strong non-linear approximation ability and is suitable for the maximum power point tracking system of wind turbines with strong non-linearity, the BP neural network is used to identify the inverse model first, and then the torque control amount of the inverse model is designed according to the inverse model. Its logic is simple and easy to implement. It does not perform control directly based on the lagging speed error, has a fast response and good stability. Because this inverse model includes the non-steady-state characteristics of the wind turbine, the dynamic response performance of the torque control amount designed based on this model is better than that of the traditional optimal torque control based on steady-state assumptions. However, due to the limited number of samples, the accuracy of the identified inverse model may be limited, resulting in limited control accuracy and insufficient anti-interference ability. The BP neural network controller based on the minimum speed error feeds the speed error back to the hidden layer and the input layer to correct the weights. Its essence is a high-precision feedback control. Although the response is slow, the control accuracy is high. Combining the BP neural network based on inverse model identification and the BP neural network controller based on minimum error ensures both stability and steady-state accuracy indicators.

[0106] In summary, for the maximum wind energy capture control method and system of a wind turbine in the present invention, the non-linear approximation advantage of the neural network is fully utilized, and a BP design network inverse model controller and a BP neural network controller based on the backpropagation of the rotational speed error are introduced into the maximum power point tracking system of the wind turbine with significant non-linear characteristics. The inverse model torque control quantity is calculated by the inverse model of the wind turbine identified by the BP neural network. This control quantity directly controls without relying on the lagging rotational speed error, and has a fast response speed and good stability. The BP neural network controller based on the backpropagation of the rotational speed error has high control precision, is highly dependent on the model, has strong adaptability to environmental factors and unit parameters. Combining the two ensures both stability and steady-state accuracy indicators.

[0107] Those skilled in the art should understand that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0108] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, and the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0109] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0110] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, so that a series of operation steps are executed on the computer or other programmable apparatus to produce a computer-implemented process, thereby the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in one process or multiple processes and / or one block or multiple blocks. Figure 1 one process or multiple processes and / or Figure 1 one block or multiple blocks.

[0111] The above is only to illustrate the technical idea of the present invention, and the protection scope of the present invention cannot be limited thereby. Any modifications made on the basis of the technical solution according to the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.

Claims

1. A method for maximum wind energy capture control of a wind turbine, characterized in that, it includes the following steps: S1. Use a BP neural network to online identify the inverse model of the wind turbine. Specifically, using a BP neural network to online identify the inverse model of the wind turbine is as follows: Operate the wind turbine under the traditional torque control strategy; read the wind turbine rotor speed signal a and the generator electromagnetic torque signal b in real time; based on the real-time wind turbine rotor speed signal a and the generator electromagnetic torque signal b, utilize the unscented Kalman filter to perform an online unbiased estimation of the wind turbine aerodynamic torque in combination with the discrete state space equation of the wind turbine drive train, and obtain the estimated value c of the wind turbine aerodynamic torque; utilize the BP neural network to perform an online identification of the inverse model of the controlled wind turbine, and use the difference between the electromagnetic torque at the current moment and the neural network output as the error to train the weights, and use the trained neural network as the inverse model of the wind turbine; S2. Conduct real-time control on the wind turbine, read the wind wheel speed and the electromagnetic torque signal of the generator, online estimate the aerodynamic torque of the wind wheel, and use the Newton-Raphson method to iteratively solve the physical relationship between the aerodynamic torque of the wind wheel and the wind speed to obtain the wind speed estimate value in real time; S3. Design a reference model and calculate the reference speed based on the wind speed estimate value obtained in step S2; S4. Use the estimated value of the aerodynamic torque of the wind wheel, the reference speed obtained in step S3, the wind wheel speed at the previous moment obtained in step S2, and the electromagnetic torque signal of the generator at the previous moment as the input of the inverse model of the wind turbine obtained in step S1 to calculate the torque control amount of the inverse model of the wind turbine at the current moment; S5. Subtract the reference speed obtained in step S3 from the wind wheel speed obtained in step S2 to obtain the speed control error; S6. Based on the rotational speed control error obtained in step S5, use a BP neural network controller to calculate the closed-loop torque control quantity. The neural network controller uses the reference rotational speed obtained in step S3 , the wind turbine rotational speed at the previous moment obtained in step S2 and the closed-loop torque control quantity i(k - 1) at the previous moment as inputs, with the closed-loop torque control quantity i(k) as the output, and online train the weights of the neural network controller with the rotational speed control error h(k) as the target; S7. Add the torque control amount of the inverse model of the wind turbine at the current moment obtained in step S4 to the closed-loop torque control amount obtained in step S6 to obtain the combined electromagnetic torque control amount, and achieve the combined torque setting target through the converter to realize the maximum wind energy capture control of the wind turbine.

2. The method for maximum wind energy capture control of a wind turbine according to claim 1, characterized in that, the model structure of the controlled wind turbine is: where f represents a function, is the rotational speed at the current moment, is the rotational speed at the previous moment, is the electromagnetic torque at the current moment, is the electromagnetic torque at the previous moment, is the pneumatic torque at the current moment.

3. The method for maximum wind energy capture control of a wind turbine according to claim 1, characterized in that, the output of the inverse model of the wind turbine is expressed as: Among them, is the rotational speed at the previous moment, is the electromagnetic torque at the previous moment, is the pneumatic torque at the current moment.

4. The method for maximum wind energy capture control of a wind turbine according to claim 1, characterized in that, in step S2, the wind speed estimate value d satisfies the following relationship: wherein, is the incoming flow wind density; is the wind turbine radius; is the wind turbine torque coefficient, is the ratio of the tip speed to the incoming flow wind speed; c is the aerodynamic torque of the wind turbine.

5. The method for maximum wind energy capture control of a wind turbine according to claim 4, characterized in that, online estimating the aerodynamic torque of the wind wheel is specifically: First, read the wind wheel speed signal a and the electromagnetic torque signal b of the generator, and then use the unscented Kalman filter to online estimate the aerodynamic torque c of the wind wheel.

6. The method for maximum wind energy capture control of a wind turbine according to claim 1, characterized in that, in step S3, the reference model is specifically: wherein, is the optimal tip speed ratio, is the reference model gain; e is the time constant, is the Laplace transform of the reference rotational speed, is the Laplace transform of the estimated wind speed value, is the Laplace operator.

7. The method for maximum wind energy capture control of a wind turbine according to claim 1, characterized in that, In step S4, the torque control amount of the wind turbine inverse model at the current moment is as follows: Among them, is the reference speed, is the wind turbine speed at the previous moment, is the electromagnetic torque of the generator at the previous moment, is the estimated value of the aerodynamic torque.

8. A maximum wind energy capture control system for a wind turbine, characterized in that, based on the method for maximum wind energy capture control of a wind turbine according to any one of claims 1 to 7, it includes: an identification module that uses a BP neural network to online identify the inverse model of the wind turbine; an estimation module that conducts real-time control on the wind turbine, reads the wind wheel speed and the electromagnetic torque signal of the generator, online estimates the aerodynamic torque of the wind wheel, uses the Newton-Raphson method to iteratively solve the physical relationship between the aerodynamic torque of the wind wheel and the wind speed to obtain the wind speed estimate value in real time; designs a reference model and calculates the reference speed based on the wind speed estimate value; A calculation module, taking the estimated value of the aerodynamic torque of the wind turbine rotor, the reference speed obtained by the estimation module, the wind turbine rotor speed at the previous moment, and the generator electromagnetic torque signal at the previous moment as the inputs of the inverse model of the wind turbine obtained by the identification module, calculates the torque control quantity of the inverse model of the wind turbine at the current moment; A network module subtracts the reference speed obtained by the estimation module from the wind turbine rotor speed to obtain a speed control error, and calculates a closed-loop torque control quantity by using a BP neural network controller; A control module adds the torque control quantity of the inverse model of the wind turbine at the current moment obtained by the calculation module to the closed-loop torque control quantity obtained by the network module to obtain a combined electromagnetic torque control quantity, and realizes the combined torque setting target through a converter to achieve the maximum wind energy capture control of the wind turbine.

Images