Multi-degree-of-freedom wind power system based on short-term prediction

Abstract

The application provides a multi-degree-of-freedom wind power generation system based on short-term wind power prediction, which comprises a short-term wind power prediction module, a wind power generator control module and a multi-degree-of-freedom permanent magnet synchronous wind power generator; the short-term wind power prediction module is used for predicting the change of wind direction and wind speed in a short term based on the relationship between wind speed and wind direction by a Gaussian mixture model; the wind power generator control module is used for controlling the operation of the wind power generator by a model predictive control strategy based on a long short-term memory network and the prediction result of the short-term wind power prediction module; and the multi-degree-of-freedom permanent magnet synchronous wind power generator is used for converting wind power into electric energy.The application establishes a multi-degree-of-freedom efficient wind power generation system based on short-term wind power prediction; when the wind speed and wind direction change, the multi-degree-of-freedom wind power generator can be deflected around the rotor shaft so as to keep consistent with the wind direction, thereby replacing the yaw system of the traditional wind power generator; and a complete efficient wind power generation system composed of wind power prediction results, a control system and a new type of wind power generator is formed.

Description

Technical Field

[0001] This invention belongs to the field of clean energy technology, and specifically relates to a short-term predictive multi-degree-of-freedom power generation system for energy conversion management and optimization. Background Technology

[0002] Wind energy technology is a high-tech field involving more than a dozen disciplines and specialties, including meteorology, aerodynamics, structural mechanics, computer technology, electronic control technology, materials science, chemistry, electromechanical engineering, electrical engineering, and environmental science. It is a complex technology. Harsh and variable environmental conditions place high demands on large-scale wind power technology. Furthermore, grid connection technology for large-scale wind power is still under development, and a series of issues continue to constrain its progress.

[0003] The efficient and stable operation of wind turbines is crucial for large-scale power generation. Wind direction and speed in nature are constantly changing and unpredictable, placing high demands on the control system of wind turbines and necessitating improved reliability. The most effective way to enhance control system reliability is to establish predictive models to create a predictive control system that accurately predicts wind turbine power, enabling the controller to better manage the operation of the wind turbine.

[0004] Based on prediction results, wind forecasting models can be divided into two types: point prediction and probabilistic prediction. Mainstream wind forecasting models primarily use point prediction, predicting the expected value of wind data at a future point in time. However, due to the strong volatility and randomness of wind resources and the inherent limitations of the prediction models themselves, point prediction errors are inevitable. Point prediction cannot provide information about the uncertainties associated with the prediction results. Therefore, probabilistic prediction of wind data, which can quantitatively reflect the uncertainties of wind data, is receiving increasing attention from researchers. Traditional forecasting system models only focus on predicting a single type of wind data, and their wind turbine operation is passive and lagging.

[0005] Multi-degree-of-freedom (MDF) power generation system models can effectively improve the shortcomings of traditional models, enhance the control accuracy of transmission mechanisms, and reduce floor space and cost. Based on existing conditions, the dynamic characteristics of the multi-degree-of-freedom generator rotor are analyzed, and a mathematical model of the rotor under multi-degree-of-freedom conditions is established. Furthermore, the electromagnetic, thermal, and solid-state fields of the multi-degree-of-freedom permanent magnet motor are studied, along with multi-physics coupling analysis of the motor. Taking the electromagnetic structure as the research object, the temperature rise and stress-strain of the designed structure can be obtained through magnetothermal-structure interaction (MTCH) simulation.

[0006] Multi-degree-of-freedom (MDF) generators are developing rapidly, and their research methods and theories are constantly evolving. Current research largely focuses on the generator itself, with few researchers concentrating on applying MDF generators to other applications. Even in existing research on applying MDF generators to wind power generation, the maximum deflection angle is only 30°, which does not meet the operational requirements of wind turbines. The goal of this research is to develop a MDF generator that can be widely used in wind power generation and achieve 360° deflection. Summary of the Invention

[0007] To address the aforementioned problems, this invention establishes a multi-degree-of-freedom (MDF) high-efficiency wind power generation system based on short-term wind forecasting. It utilizes a short-term wind forecasting model to obtain changes in wind direction and speed over 24 hours, and uses the forecast results to guide the real-time operation of an LSTM-based MPC (Multi-Degree-of-Freedom Control) system. When wind speed and direction change, the MDF wind turbine can deflect around its rotor shaft to align with the wind direction, replacing the traditional yaw system. This research represents the first time a complete high-efficiency wind power generation system has been established, consisting of wind forecasting results, a control system, and a novel wind turbine. Specifically, the technical solution proposed in this invention is as follows:

[0008] A multi-degree-of-freedom wind power generation system based on short-term wind forecasting, the multi-degree-of-freedom wind power generation system comprising:

[0009] The short-term wind forecasting module is based on a Gaussian mixture model and predicts the changes in wind direction and wind speed in the short term by using the relationship between wind speed and wind direction.

[0010] A wind turbine control module, wherein the wind turbine control module is based on a model predictive control strategy using a long short-term memory network and uses the prediction results of the short-term wind prediction module to control the operation of the wind turbine;

[0011] A multi-degree-of-freedom permanent magnet synchronous wind turbine generator that converts wind power into electrical energy.

[0012] Optionally, it also includes an analysis module, which performs finite element and magnetic field modeling and analysis based on analysis software.

[0013] Optionally, the long short-term memory network includes an input gate, a forget gate, and an output gate structure.

[0014] Optionally, it also includes a position calculation module, which is used to determine the position of the angle between the front normal vector of the generator blade rotation plane and the geodetic coordinate system.

[0015] Optionally, it also includes a rolling optimization module, which obtains the next control sequence through a long short-term memory network based on the deflection angular velocity and the predicted output.

[0016] Optionally, it also includes an optimal path logic judgment module, which determines whether to continue adjusting the task path and obtains a given deflection angular velocity r by inputting the adjustment angle, wind speed and prevailing wind direction into the optimal path logic judgment module.

[0017] Optionally, the multi-degree-of-freedom permanent magnet synchronous wind turbine includes a multi-degree-of-freedom permanent magnet synchronous generator, which includes a stator, a rotor, and windings.

[0018] Optionally, the stator includes a rotating stator and a deflecting stator; and / or the winding includes rotating windings and deflecting windings.

[0019] Optionally, the rotor is equipped with a spherical permanent magnet for rotation and a ring-shaped permanent magnet for deflection.

[0020] Optionally, the multi-degree-of-freedom permanent magnet synchronous wind turbine can rotate around its rotor shaft at any angle to align it with the wind direction.

[0021] This invention establishes a multi-degree-of-freedom (MDF) high-efficiency wind power generation system based on short-term wind forecasting. It utilizes a short-term wind forecasting model to obtain changes in wind direction and speed over a 24-hour period and uses the forecast results to guide the real-time operation of an LSTM-based Multiprocessor Control (MPC) system. When wind speed and direction change, the MDF wind turbine can deflect around its rotor shaft to align with the wind direction, replacing the traditional yaw system. This research represents the first time a complete high-efficiency wind power generation system has been developed, comprising wind forecasting results, a control system, and a novel wind turbine. Attached Figure Description

[0022] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0023] Figure 1 This is a schematic diagram of the structure of a multi-degree-of-freedom wind power generation system based on short-term wind power prediction proposed in one embodiment of the present invention;

[0024] Figure 2 This is a schematic diagram of the LSTM memory block structure proposed in one embodiment of the present invention;

[0025] Figure 3 This is a schematic diagram of the LSTM-MPC control strategy proposed in one embodiment of the present invention;

[0026] Figure 4 This is a schematic diagram of the multi-degree-of-freedom deflection type PMSWG wind turbine structure proposed in one embodiment of the present invention;

[0027] Figure 5 This is a comparison diagram of the phase voltage of phase A winding under different deflection angles, as proposed in one embodiment of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] This invention proposes a multi-degree-of-freedom wind power generation system based on short-term wind forecasting, the multi-degree-of-freedom wind power generation system comprising:

[0030] The short-term wind forecasting module is based on a Gaussian mixture model and predicts the changes in wind direction and wind speed in the short term by using the relationship between wind speed and wind direction.

[0031] A wind turbine control module, wherein the wind turbine control module is based on a model predictive control strategy using a long short-term memory network and uses the prediction results of the short-term wind prediction module to control the operation of the wind turbine;

[0032] A multi-degree-of-freedom permanent magnet synchronous wind turbine generator that converts wind power into electrical energy.

[0033] Specifically, for the short-term wind forecasting module, a Gaussian Mixture Model (GMM) is first used to consider the relationship between wind speed and direction, and the changes in wind direction and speed over the next 24 hours are obtained through short-term wind condition forecasting. The GMM is used to consider the dependency between wind speed and direction over the next 24 hours to complete the wind probability forecast. Unlike traditional models that only focus on forecasting a single type of wind data, this invention considers the relationship between wind speed and direction in wind forecasting to improve the model's accuracy.

[0034] Gaussian models are a commonly used parameter estimation method, which can be divided into Single Gaussian Models (SGM) and Gaussian Mixture Models (GMMs). GMMs are an extension of SGMs and can smoothly approximate probability density functions of any shape. Furthermore, a GMM can be viewed as a model composed of K SGMs, where the K sub-models are the hidden variables of a mixture model. Due to its multiple models, GMMs are finely partitioned and suitable for modeling complex objects.

[0035] Probabilistic prediction models for wind data can be divided into parametric and non-parametric models. Parametric models assume that the data conforms to a specific distribution based on prior knowledge before prediction; non-parametric models do not make assumptions but directly calculate the distribution function through data-driven methods. Since wind data does not perfectly conform to a specific distribution, the prediction error of parametric models is usually high. Although non-parametric models have better prediction accuracy, they cannot provide specific fitting formulas, resulting in poor interpretability. GMM (Geometric Matrix Modeling) is a parametric model, but it does not require assumptions about the distribution of wind data, similar to non-parametric models; this effectively solves the shortcomings of parametric models. At the same time, GMM provides clear solution formulas, simplifying the interpretation of GMM predictions. Therefore, GMM combines the strong interpretability of both parametric and non-parametric models, without requiring prior assumptions about the data.

[0036] For the wind turbine control module, this invention proposes a model predictive control strategy based on a long short-term memory network (LSTM) network, utilizing forward-looking wind power prediction results to guide real-time wind turbine operation. Simultaneously, a real-time operation guidance method for the wind turbine based on wind power prediction results is proposed. LSTM is introduced into MPC control to address the rolling optimization process. Compared to traditional MPC, LSTM, as a special type of recurrent neural network, can learn long-term dependencies and enhance long-term memory capabilities; therefore, the LSTM-MPC strategy can more efficiently and accurately complete the operation control of the wind turbine.

[0037] Recurrent Neural Networks (RNNs) are networks with memory capabilities that can pass information from one time step to the next. LSTM (Laser-Switch Transformer) is a special type of RNN that uses constant-error looping in its memory cells to learn long-term memory. An LSTM has a memory block structure containing memory cells and three gates: an input gate, a forget gate, and an output gate. Its gate signals are as follows: Figure 2 As shown. The forget gate determines whether to retain the information in the current memory cell; the input gate determines whether to allow the current input information to be added to the memory cell; and the output gate determines whether the information in the current memory cell should be output to the next memory cell of the LSTM. Compared with traditional model predictive control global optimization, MPC uses distributed rolling optimization. The control process is divided into several parts, continuously using real-time information to perform online feedback correction on the running trajectory, increasing the robustness of the control system.

[0038] Figure 3The LSTM-MPC control strategy diagram is given. The alignment angle of this invention is the angle between the front normal vector of the generator blade rotation plane and the geodetic coordinate system. The prevailing wind direction is calculated by the weighted average method as shown in equation (2). At the initial moment, the position of the alignment angle is determined by the alignment angle and position calculation module. The adjustment angle, wind speed and prevailing wind direction are input into the optimal path logic judgment module to determine whether to continue adjusting the task path and obtain the given deflection angular velocity r. Because there is no control information in the initial stage of the previous step, this prediction model does not work now. Only the given deflection angular velocity r is input into the rolling optimization module, and the rolling optimization result is solved by the LSTM neural network. Then the control sequence U(0) is obtained, and the deflection angular velocity controller is input to drive the shaft to generate the deflection velocity y0. Then y0 is input into the alignment angle and position calculation module to obtain the alignment angle at time T1. U(0) is input into the prediction model to obtain the prediction output (y) of the next step I. m1 y m2 , ..., y mi ), and based on the actual wind speed y0 and the predicted output y mi The final prediction output (y) of step I is obtained as follows. p1 y p2 , ..., y pi Then the obtained deflection angular velocity r and the final predicted output input y are used. pi In the rolling optimization module, the next control sequence U(1) is obtained through the LSTM network.

[0039] For multi-degree-of-freedom permanent magnet synchronous wind turbines, the turbine is deflected around its rotor shaft to align with the wind direction, thus replacing the yaw system of traditional wind turbines. When the wind direction fluctuates, the multi-degree-of-freedom permanent magnet synchronous wind turbine can rotate freely 360° around its rotor shaft to align with the wind direction. This is a non-contact electromagnetic transmission method, rather than the traditional contact gear mechanical transmission method used in yaw systems. Therefore, the proposed wind turbine not only reduces costs and failure rates but also improves wind alignment accuracy and reduces response time.

[0040] Compared to megawatt-scale wind power generation, small distributed wind turbines have relatively fewer site restrictions and are mainly installed in remote mountainous areas, islands, pastoral areas, and other regions with limited electricity access. These areas account for a very high proportion of the world's total land area. Therefore, distributed small wind power generation is gradually becoming the mainstay of wind power supply and is an effective distributed generation solution. In traditional small wind turbines, a yaw system is used to keep the generator aligned with the wind direction. However, yaw systems are costly and have a high failure rate. This invention proposes a multi-degree-of-freedom yaw type PMSWG, such as... Figure 4 As shown, when the wind fluctuates, the generator can rotate 360° around its rotor shaft to align with the wind direction, unlike the yaw system of traditional wind turbines.

[0041] A multi-degree-of-freedom (DOF) PMSWG model wind turbine consists of a rotor, horizontal shaft, bearings, bearing housings, tower, base, coupling, multi-DOF permanent magnet synchronous generator (PMSG), controller, and frequency converter. The support structure comprises a base and tower, with the PMSSG, controller, and frequency converter mounted on the top of the tower. The bearing housings and bearings are fixed to the horizontal shaft. The rotor, including blades, hub, and reinforcing components, is mounted on a gearbox. The PMSSG consists of a stator, rotor, and windings. The stator includes a rotating stator and a deflection stator. The generator's rotation and deflection share a single rotor, which is equipped with two types of permanent magnets: spherical permanent magnets for rotation and toroidal permanent magnets for deflection. The spherical permanent magnets are alternately polarized along the diameter directions (N and S), while the toroidal permanent magnets are magnetized along the diameter direction. The windings are further divided into rotating windings and deflection windings. When the generator rotates, the spherical permanent magnets interact with the rotating windings, and the magnetic induction lines generated by the magnetic poles cut through the rotating windings. According to the law of electromagnetic induction, a current is generated in the winding, causing the generator to produce electrical energy. When the generator deflects, the deflection winding is powered by an external power source. The deflection winding on the stator interacts with the deflection permanent magnet on the rotor shaft to complete the deflection, thus realizing the generator's deflection around the shaft.

[0042] Furthermore, this invention includes an analysis module that uses Ansoft Maxwell software for finite element and magnetic field modeling and analysis, better revealing the dynamic characteristics of the system. The material properties, boundary conditions, and excitation sources of the model are established. The solver parameters are: step size 0.0005s, number of steps 100, operation time 0.05s, and nonlinear residual 0.0001.

[0043] In torque calculation, assuming the flux linkage is constant and there is no loss before and after the virtual displacement, the total energy stored in the solution domain can be estimated, as shown in the following formula:

[0044]

[0045] In the formula, V is the volume of the magnetic field, H is the magnetic field strength, and B is the magnetic induction intensity.

[0046] During deflection, the controller directs external power to energize the generator, allowing it to operate simultaneously in both rotation and deflection states, ensuring normal power generation even during deflection. When external power is applied, it injects current into multiple coils in the same direction, generating deflection torque. Simultaneously, opposite currents are applied to other coils, causing them to repel and attract each other according to the principle of like and unlike poles.

[0047] Example:

[0048] Transient analysis of a multi-degree-of-freedom permanent magnet synchronous generator with deflection

[0049] The generator's deflection at 0°, 180°, and 360° was analyzed using an experimental platform. Since phase current can be calculated from phase voltage, and the experimental oscilloscope cannot directly measure induced voltage and flux linkage, only the changes in phase voltage at different deflection angles were compared. The oscilloscope results showed excessive harmonics; therefore, a low-pass filter in MATLAB was used to eliminate these harmonics. The stator has three-phase windings; taking phase A as an example, the phase voltage changes before and after deflection are as follows: Figure 5 As shown in the figure, the phase voltage changes with the same trend before and after deflection. Harmonics have a relatively small impact on voltage, but overall they exhibit a sinusoidal change over time; therefore, the proposed generator can be said to operate normally within the maximum deflection range.

[0050] Wind tunnel tests to assess the performance of wind power generation systems

[0051] The effectiveness of the proposed wind power system was verified through wind tunnel experiments compared with those of a conventional wind turbine. The maximum angular velocity of the generator is 2° / s, with an allowable alignment error of [-5°, +5°]. When the alignment error is within this range, the generator is considered aligned with the wind direction and no deflection is required. When the alignment error is greater than 5°, the wind turbine deflects around its rotor shaft to align with the wind direction. The time interval for wind prediction results is 1 minute, and the wind period will run for 10 minutes. Therefore, throughout the control period, the prevailing wind direction θ is calculated based on the weighted average method using measured wind direction and wind speed data. d The wind speed and direction data during wind operation can be expressed as:

[0052] Wind speed V i =V1, V2, ..., V n ;

[0053] Wind direction θi = θ1, θ2,..., θ n ;

[0054] To obtain the prevailing wind direction, the weighted average of the wind direction data for the entire period is calculated as follows:

[0055]

[0056] Among them, w i This represents the weighting value for wind direction data. From a wind energy perspective, wind force is directly proportional to the cube of wind speed. Therefore, w i Using the third power of wind speed, that is:

[0057]

[0058] It is necessary to determine whether the wind direction is stable during wind operation. When the wind direction changes significantly in a short period of time, the prevailing wind direction is not significant. Calculate the wind direction data during wind operation, and when max(θ) i )-min(θi When the wind direction is ≥40°, the wind direction is considered unstable and the generator itself is not aligned with the wind. Otherwise, the wind direction is considered stable.

[0059] For traditional wind turbines, wind tunnels generate wind based on historical wind data. For multi-degree-of-freedom wind turbines, wind tunnel operation is performed based on point predictions of wind data, and extreme cases are determined based on probability predictions. When the confidence level of wind speed conditions is greater than 50%, if the wind speed at a future moment exceeds 18 m / s for more than 5 minutes; or exceeds 25 m / s for more than 3 minutes; or exceeds 30 m / s for more than 1 minute, the wind turbine will shut down.

[0060] The microcontroller inputs wind forecast results from one sampling period in advance and real-time information on the turbine shaft position into the controller. The controller can proactively decide whether wind direction and specific operations are needed a point in time. The wind tunnel test environment can only generate airflow in one direction, thus simulating only the operation of the wind turbine generator under windy conditions. The experiment compared the power and efficiency of two types of wind turbines, finding that the power generation curves of the two generators were similar, but the real-time power of the proposed multi-degree-of-freedom (MPC) wind turbine generator was significantly higher than that of the traditional generator. This is because the MPC system can accurately predict changes in wind data for the next period, effectively guiding the operation of the wind turbine, thus solving the wind variation problem better than the traditional lagging yaw system. Furthermore, multi-degree-of-freedom wind power generation has significant advantages in terms of cost and response time.

[0061] This invention establishes a multi-degree-of-freedom (MDF) high-efficiency wind power generation system based on short-term wind forecasting. It utilizes a short-term wind forecasting model to obtain changes in wind direction and speed over a 24-hour period and uses the forecast results to guide the real-time operation of an LSTM-based Multiprocessor Control (MPC) system. When wind speed and direction change, the MDF wind turbine can deflect around its rotor shaft to align with the wind direction, replacing the traditional yaw system. This research represents the first time a complete high-efficiency wind power generation system has been developed, comprising wind forecasting results, a control system, and a novel wind turbine.

[0062] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A multi-degree-of-freedom wind power generation system based on short-term wind forecasting, characterized in that, The multi-degree-of-freedom wind power generation system includes: The short-term wind forecasting module is based on a Gaussian mixture model and predicts the changes in wind direction and wind speed in the short term by using the relationship between wind speed and wind direction. A wind turbine control module, wherein the wind turbine control module is based on a model predictive control strategy using a long short-term memory network and uses the prediction results of the short-term wind prediction module to control the operation of the wind turbine; A multi-degree-of-freedom permanent magnet synchronous wind turbine generator that converts wind power into electrical energy; Throughout the control period, the prevailing wind direction was calculated using a weighted average method based on measured wind direction and speed data. ,in in, The weighting value for wind direction data, For wind direction, Wind speed and direction data are expressed as follows: wind speed ; wind direction ; Using the cube of the wind speed, i.e.: Calculate wind direction data during wind operation, when If the wind direction is unstable, the generator itself is not aligned with the wind; otherwise, the wind direction is stable.

2. The multi-degree-of-freedom wind power generation system based on short-term wind forecasting according to claim 1, characterized in that, It also includes an analysis module, which performs finite element and magnetic field modeling and analysis based on analysis software.

3. The multi-degree-of-freedom wind power generation system based on short-term wind forecasting according to claim 1, characterized in that, The Long Short-Term Memory (LSTM) network includes an input gate, a forget gate, and an output gate structure.

4. The multi-degree-of-freedom wind power generation system based on short-term wind forecasting according to claim 1, characterized in that, It also includes a position calculation module, which is used to determine the position of the angle between the front normal vector of the generator blade rotation plane and the geodetic coordinate system.

5. The multi-degree-of-freedom wind power generation system based on short-term wind forecasting according to any one of claims 1-4, characterized in that, It also includes a rolling optimization module, which obtains the next control sequence through a long short-term memory network based on the deflection angular velocity and the predicted output.

6. The multi-degree-of-freedom wind power generation system based on short-term wind forecasting according to any one of claims 1-4, characterized in that, It also includes an optimal path logic judgment module, which determines whether to continue adjusting the task path based on the input of the adjustment angle, wind speed, and prevailing wind direction, and obtains the given deflection angular velocity. r .

7. The multi-degree-of-freedom wind power generation system based on short-term wind forecasting according to claim 1, characterized in that, The multi-degree-of-freedom permanent magnet synchronous wind turbine includes a multi-degree-of-freedom permanent magnet synchronous generator, which includes a stator, a rotor, and windings.

8. The multi-degree-of-freedom wind power generation system based on short-term wind forecasting according to claim 7, characterized in that, The stator includes a rotating stator and a deflecting stator; and / or the winding includes rotating windings and deflecting windings.

9. The multi-degree-of-freedom wind power generation system based on short-term wind forecasting according to claim 7, characterized in that, The rotor is equipped with a spherical permanent magnet for rotation and a ring-shaped permanent magnet for deflection.

10. The multi-degree-of-freedom wind power generation system based on short-term wind forecasting according to claim 6, characterized in that, The multi-degree-of-freedom permanent magnet synchronous wind turbine can rotate around the rotor shaft at any angle to align it with the wind direction.

Images