Floating type fan load shedding control method and system

Abstract

According to the load shedding control method and system for the floating type draught fan, a unified variable-pitch controller and an independent variable-pitch controller are comprehensively applied to control of the floating type draught fan, and the particle fitness function corresponding to the parameters of the unified variable-pitch controller or the parameters of the independent variable-pitch controller is calculated; calculating a current fitness function corresponding to each particle in a particle swarm corresponding to the current unified variable pitch controller parameter or the independent variable pitch controller parameter; setting a particle swarm corresponding to the adjusted fitness function or the current fitness function according to the adjusted fitness function or the current fitness function, and outputting a target unified variable pitch controller parameter or a target independent variable pitch controller parameter; and the output of the particle swarm optimization algorithm is optimized again by adopting the Bayesian algorithm, and the particle swarm optimization algorithm and the Bayesian optimization algorithm are combined, so that the parameter setting precision and efficiency are further improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of floating wind turbine control, and in particular relates to a floating wind turbine load reduction control method and system. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] As an important member of wind power, offshore wind power generation plays a key role in achieving sustainable development and alleviating the energy crisis. Since floating wind turbines OFWT are installed on floating platforms, they are easily affected by the coupling of multiple factors such as wind, waves, and currents. Load reduction technology has become a hot topic in current research. Existing floating wind turbine load reduction technologies are mainly divided into three categories: platform and mooring-based optimization design, structural control-based vibration suppression methods, and servo control-based load suppression. Among them, the servo-control-based variable pitch technology has attracted widespread attention due to its significant load reduction effect and the absence of the need to add additional physical structures.

[0004] Wind turbine pitch control is mainly divided into two modes: unified pitch control CPC and independent pitch control IPC. In CPC mode, the control signals of the pitch angles of the three blades of the wind turbine are consistent; while IPC can adjust the pitch angle independently for each blade. IPC can effectively overcome the problem of uneven distribution of wind turbine loads in time and space caused by environmental factors such as tower shadow effect and wind shear effect, and thus has higher control flexibility and efficiency.

[0005] The unified pitch controller CPC controls the output power by setting the same pitch angle for all blades. This method is simple to operate, but has the following obvious disadvantages under conditions such as non-uniform loads and non-uniform wind fields:

[0006] (1) It cannot adapt to non-uniform wind fields and cannot adjust the pitch angle according to the specific load of each blade, which easily leads to uneven distribution of blade load and increases structural fatigue;

[0007] (2) CPC control accuracy is limited, and the wind turbine output power tends to fluctuate with wind speed, affecting operating efficiency. Based on CPC, the independent pitch controller IPC introduces an additional signal that reflects the load difference of each blade to achieve independent adjustment of the pitch angle of each blade, thereby more effectively coping with uneven wind field conditions and load distribution on the wind rotor.

[0008] However, the application of independent pitch controller IPC also has the following limitations:

[0009] (1) The floating wind turbine system is complex, with multiple factors such as water, air, and structure coupled with each other. In addition, the nonlinearity and strong uncertainty of the system itself make the setting of control parameters challenging;

[0010] (2) The setting of controller parameters is crucial to the safe and stable operation of the fan system. Improper parameter settings may affect system performance or even lead to serious consequences such as system crash. Therefore, how to safely and reliably select controller parameters to ensure that the system can operate stably and efficiently under various working conditions is a key issue that needs to be solved urgently. Summary of the invention

[0011] In order to overcome the shortcomings of the above-mentioned prior art, the present invention provides a floating wind turbine load reduction control method and system, which comprehensively applies a unified pitch controller and an independent pitch controller to the control of a floating wind turbine, and uses a particle swarm optimization algorithm and a safe Bayesian algorithm to adjust the parameters of the unified pitch controller or the independent pitch controller, thereby improving the accuracy and efficiency of parameter setting.

[0012] In order to achieve the above object, the present invention adopts the following technical solution:

[0013] In a first aspect, the present invention provides a floating wind turbine load reduction control method, comprising:

[0014] The difference between the actual speed of the wind turbine and the reference speed is used as the input of the unified pitch controller to obtain the common pitch angle of the blades;

[0015] The flapping moment of the blade load is used as the input of the independent pitch controller to obtain the additional pitch angle of each blade;

[0016] The obtained common pitch angle of the blades and the additional pitch angles of the individual blades are superimposed to control the floating wind turbine;

[0017] Wherein, the fitness function of the particle corresponding to the unified pitch controller parameter or the independent pitch controller parameter is calculated, the pitch controller parameter or the independent pitch controller parameter is adjusted, and the current fitness function corresponding to each particle in the particle group corresponding to the adjusted unified pitch controller parameter or the independent pitch controller parameter is calculated;

[0018] According to the current fitness function, the particle swarm corresponding to the current fitness function is adjusted, and target unified pitch controller parameters or target independent pitch controller parameters are output;

[0019] According to the output target unified pitch controller parameters or target independent pitch controller parameters, a safe Bayesian algorithm is used to perform iterative optimization to obtain final target unified pitch controller parameters or target independent pitch controller parameters.

[0020] In a second aspect, the present invention provides a floating wind turbine load reduction control system, comprising:

[0021] A first control module is configured to: use the difference between the actual speed of the wind turbine and the reference speed as an input of a unified pitch controller to obtain a common pitch angle of the blades;

[0022] A second control module is configured to: use the flapping moment of the blade load as an input of the independent pitch controller to obtain an additional pitch angle of each blade;

[0023] A third control module is configured to: superimpose the obtained common pitch angle of the blades and the additional pitch angles of each blade to control the floating wind turbine;

[0024] Wherein, the fitness function of the particle corresponding to the unified pitch controller parameter or the independent pitch controller parameter is calculated, the pitch controller parameter or the independent pitch controller parameter is adjusted, and the current fitness function corresponding to each particle in the particle group corresponding to the adjusted unified pitch controller parameter or the independent pitch controller parameter is calculated;

[0025] According to the current fitness function, the particle swarm corresponding to the current fitness function is adjusted, and target unified pitch controller parameters or target independent pitch controller parameters are output;

[0026] According to the output target unified pitch controller parameters or target independent pitch controller parameters, a safe Bayesian algorithm is used to perform iterative optimization to obtain final target unified pitch controller parameters or target independent pitch controller parameters.

[0027] In a third aspect, the present invention provides an electronic device comprising a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method described in the first aspect is performed.

[0028] In a fourth aspect, the present invention provides a computer-readable storage medium for storing computer instructions, wherein when the computer instructions are executed by a processor, the method described in the first aspect is performed.

[0029] In a fifth aspect, the present invention provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method described in the first aspect is implemented.

[0030] One or more of the above technical solutions have the following beneficial effects:

[0031] In the present invention, a unified pitch controller and an independent pitch controller are comprehensively applied to the control of a floating wind turbine. The fitness function of particles corresponding to the unified pitch controller parameters or the independent pitch controller parameters is calculated, and the current fitness function corresponding to each particle in the particle swarm corresponding to the current unified pitch controller parameters or the independent pitch controller parameters is calculated; the particle swarm corresponding to the unified pitch controller parameters or the independent pitch controller parameters is adjusted according to the adjusted fitness function or the current fitness function, and the target unified pitch controller parameters or the target independent pitch controller parameters are output; the output of the particle swarm optimization algorithm is optimized again using the Bayesian algorithm, and the particle swarm optimization algorithm and the Bayesian optimization algorithm are combined to further improve the accuracy and efficiency of parameter adjustment.

[0032] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0034] Figure 1 The figure is an overall block diagram of the floating wind turbine pitch control based on PSO-BO in the first embodiment of the present invention;

[0035] Figure 2 The CPC variable gain PI controller in the first embodiment of the present invention;

[0036] Figure 3 Schematic diagram of the PSO optimization algorithm in the first embodiment of the present invention;

[0037] Figure 4 This is a flow chart of controller parameters optimized using PSO-BO in Embodiment 1 of the present invention. DETAILED DESCRIPTION

[0038] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0039] It should be noted that the terms used herein are for describing specific embodiments only and are not intended to be limiting of exemplary embodiments according to the present invention.

[0040] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0041] Embodiment 1

[0042] This embodiment discloses a floating wind turbine load reduction control method, including:

[0043] The difference between the actual speed of the wind turbine and the reference speed is used as the input of the unified pitch controller to obtain the common pitch angle of the blades;

[0044] The flapping moment of the blade load is used as the input of the independent pitch controller to obtain the additional pitch angle of each blade;

[0045] The obtained common pitch angle of the blades and the additional pitch angles of the individual blades are superimposed to control the floating wind turbine;

[0046] Wherein, the fitness function of the particle corresponding to the unified pitch controller parameter or the independent pitch controller parameter is calculated, the pitch controller parameter or the independent pitch controller parameter is adjusted, and the current fitness function corresponding to each particle in the particle group corresponding to the adjusted unified pitch controller parameter or the independent pitch controller parameter is calculated;

[0047] According to the current fitness function, the particle swarm corresponding to the current fitness function is adjusted, and target unified pitch controller parameters or target independent pitch controller parameters are output;

[0048] According to the output target unified pitch controller parameters or target independent pitch controller parameters, a safe Bayesian algorithm is used to perform iterative optimization to obtain final target unified pitch controller parameters or target independent pitch controller parameters.

[0049] Combine the following Figure 1-Figure 3 A floating wind turbine load reduction control method proposed in this embodiment is described in detail.

[0050] 1. Floating wind turbine pitch control strategy

[0051] The structure of floating wind turbine pitch control system is as follows: Figure 1 As shown, the floating wind turbine in this embodiment adopts the high-fidelity FAST nonlinear model developed by the National Renewable Energy Laboratory of the United States for wind turbine simulation. Its control includes CPC and IPC. CPC compares the actual speed ω of the wind turbine with the reference speed ω r The difference between the two is taken as input, and the common pitch angle β of the three blades is adjusted. c , so that the wind turbine maintains the rated speed. Since the offshore wind speed of the floating wind turbine is generally higher than the rated wind speed of the wind turbine, the control strategy is constant torque control, that is, the output power of the floating wind turbine is achieved at the rated power by adjusting the speed.

[0052] IPC takes the flapping moments M1, M2, and M3 reflecting the loads of the three blades as input, calculates the additional pitch angles of the three blades respectively, and adds them to the common pitch angle β output by CPC c, realizing independent pitch control. This control strategy can not only maintain the rated power output, but also effectively reduce the wind turbine load and achieve the purpose of load reduction.

[0053] When performing IPC control, the classic Coleman transformation and Coleman inverse transformation are used to transform the blade coordinates, as shown in formula (1) and formula (2) respectively. Among them, the Coleman transformation converts the moment load on the three blades at different positions in space into the pitch moment M at the same position of the cabin. tilt and yaw moment M yaw , and the Coleman inverse transform converts the controller's corresponding output pitch angle β tilt and the yaw pitch angle β yaw Converted into the actual pitch angles Δβ1, Δβ2, and Δβ3 of the three blades of the wind turbine.

[0054]

[0055] Among them, M1, M2, and M3 are the loads on the three blades during rotation. Formula (1) converts the physical quantities in these three rotations into physical quantities in a fixed coordinate system. The wind speed change is low-frequency in this fixed coordinate system, so the load response of the blade is also a low-frequency quantity and will not be affected by the rotation sampling. Formula (2) is the inverse transformation for generating the increments of the pitch angles of the three blades when independent pitch is varied.

[0056] The PI controller of CPC is a special variable gain PI controller GSPI, and its control block diagram is as follows: Figure 2 Specifically:

[0057] β c =K P (β c )·e(t)+K I (β c )·∫e(t) (3)

[0058] The variable gain is reflected in the parameter K of the PI controller. P (β c ), K I (β c ) and the controller output β c This is because when the wind turbine is at a constant wind speed, the pitch angle is a constant value, and the power fluctuation caused by a small range of disturbances in the pitch angle is related to the constant wind speed, which is called the sensitivity of power to the pitch angle. In order to eliminate the influence of the pitch angle on the sensitivity under different wind speeds, GSPI is introduced:

[0059]

[0060] In the formula, KP (β c =0), K I (β c =0) is the PI parameter when the pitch angle is 0, GK(β c ) is a gain related to the pitch angle:

[0061]

[0062] Where k is the fitting coefficient, which is approximately 6.3°.

[0063] Through formula (3) to formula (5), CPC based on GSPI is realized, which is combined with IPC to form a complete floating wind turbine pitch control strategy, making the control effect more flexible and efficient.

[0064] 2. PSO-based controller parameter tuning strategy

[0065] PSO searches for the optimal solution by simulating the process of multiple "particles" moving in the search space. Each particle represents a candidate solution and has two properties: position and speed. In each iteration, the particle updates its speed and position based on its own experience and group information, gradually approaching the optimal solution. Specifically, the particle tracks two "extreme values": one is the optimal position in the particle's own history (individual optimality), and the other is the optimal position experienced by all particles in the group (global optimality). During the search process, the particles adjust their movement direction and speed based on the information of these two optimal positions, thereby exploring and utilizing them globally, and finally converging to the optimal solution of the objective function.

[0066] The basic search mechanism of the PSO algorithm is as follows Figure 3 As shown. Each individual represents a pair of PI controller parameters, and multiple such individuals constitute a population. Among them, point A represents the global optimal position, point B represents the individual optimal position of a particle, and point C represents the current position of the particle. At the next moment, the moving direction of the particle is composed of three components: the optimal direction of the group (red arrow), the optimal direction of the individual (green arrow), and the moving direction of the particle itself (brown arrow), and the position moves to point C'. As the iteration proceeds, the optimal position of the group and the optimal position of the individual may be continuously updated, eventually guiding the particle swarm to converge to the global optimal solution. The flowchart of the PSO algorithm is shown in Figure 4 As shown in the figure, the specific steps of applying it to the OFWT load shedding controller parameter optimization are as follows:

[0067] Step 1: Initialize the population x, including the initial position and initial velocity.

[0068] Step 2: Calculate the fitness of each individual and select the position of the individual with the best fitness as the individual optimal Optimal with Group

[0069] The fitness calculation method is:

[0070]

[0071] Where x = [x1 x2 x3 x4] T =[Pwr Myb Surge TTDspFA] T They represent power fluctuation, blade flapping moment, platform longitudinal motion, and tower top fore-and-aft displacement respectively, and a=[a1 a2 a3 a4] is the corresponding coefficient.

[0072] Step 3: Update the particle velocity With location The update criterion is shown in formula (7):

[0073]

[0074] Among them, w, c1, and c2 are the inertia weight, individual learning factor, and group learning factor of the particle itself, respectively, corresponding to the three components in the above update iteration process. r1 and r2 represent random numbers introduced to increase the randomness of particle exploration.

[0075] In order to balance the exploration and utilization of particles in the iterative process, the three weight coefficients w, c1, and c2 are changed as follows: in the early stage of the iteration, emphasis is placed on exploration, w and c1 are larger, and c2 is smaller; in the later stage, emphasis is placed on utilization to converge as quickly as possible, making w and c1 smaller, and c2 larger.

[0076] Step 4: Calculate the fitness of the particle at the new position using the same method as formula (6), and update the individual optimum and the group optimum.

[0077] Step 5: Repeat steps 3 and 4 until the termination condition is met and output the global optimal value.

[0078] The above steps must meet speed and position constraints. The speed constraint limits the search step of each iteration of the particle to prevent the span from being too large; the position constraint defines the search space of the particle to ensure that it searches for the best solution within a given range. The termination condition in step 5 can be set according to actual needs. Common termination conditions include the maximum number of iterations, fitness function convergence, etc.

[0079] 3. BO-based controller parameter tuning strategy

[0080] BO is a global optimization algorithm for expensive black box functions, especially for objective functions that are computationally expensive or cannot be expressed explicitly. The core idea is to use a proxy model (usually a Gaussian process GP) to approximate the objective function. First, the proxy model constructs the probability distribution of the objective function based on the known finite sample data, and on this basis predicts the possible value of the objective function at future sampling points and its uncertainty. Next, the next sampling point is selected by maximizing an acquisition function (AF), which balances the exploration of unknown areas and the use of information in known areas. After each sampling, the BO algorithm updates the proxy model to make it more accurately reflect the objective function, thereby gradually approaching the global optimal solution. The flowchart of the BO algorithm is shown below. Figure 4 As shown in the figure, the specific working steps in the optimization of load shedding controller applied to OFWT are:

[0081] Step 1: Initialize the sample initial point to obtain a set of known parameter-objective function value pairs D = {(x i ,y i )|i=1,2,…,n}, where x i is the parameter vector, i.e. the fan PI controller parameters obtained by initialization, y i =f(x i ) is the corresponding objective function value, and the calculation method is the same as the calculation of fitness in the particle swarm method, that is, formula (6).

[0082] Step 2: Build a GP model based on known points (i.e., explored points) to obtain the prior probability distribution of the objective function value y corresponding to any parameter x under known conditions D, that is, build a GP model between the PI controller parameters and the objective function y, which can be specifically described as a mean function m(x) and a covariance function (kernel function) k(x,x'), which can be recorded as:

[0083]

[0084] in,

[0085]

[0086] It is usually assumed that the objective function value follows a multivariate normal distribution: The mean function m(x) is usually selected as zero, and the covariance function k(x, x') describes the correlation between different parameters, that is, the similarity of the influence of different PI controller parameters on the objective function of the wind turbine represented.

[0087] Step 3: Optimize the acquisition function and select the next pair of PI parameters as the test point; calculate the posterior probability distribution of each parameter point in the search space according to the GP model established in the previous step, and then select the next test point x through strategies such as the upper confidence bound (UCB) and probability increment (PI).next , to maximize the acquisition function x next =argmax u(x|D), where u(x|D) is the sampling function calculated according to the current GP model.

[0088] UCB is a collection function that selects the next sampling point based on the weighted prediction mean and standard deviation, balancing exploration and exploitation through the parameter β, as shown in formula (10).

[0089] UCB(x)=μ(x)+β·σ(x) (10)

[0090] Where μ(x) and σ(x) are the mean and variance at the prediction point, respectively, and β is a trade-off factor. A larger β increases the exploration effort, while a smaller β tends to favor development.

[0091] Use the point with the largest UCB(x) as the next sampling point:

[0092]

[0093] The purpose of the PI sampling function is to select the function that maximizes the current optimal value f(x best ) Improved point, calculate the probability that the value of the sampling point exceeds the optimal solution:

[0094]

[0095] Pick the next point by maximizing PI:

[0096]

[0097] Step 4: Evaluate the objective function formula (6) at the new test point x next The function value at , and the newly obtained x next —Add known data to the function value pair and update the GP model.

[0098] Step 5: Repeat steps 3 and 4 until the termination condition is met and output the global optimal value.

[0099] Generally speaking, the effective sampling function in step 3 needs to meet the following requirements: (1) The sampling function value of the explored area should be low to avoid repeated sampling; (2) The sampling value of the high uncertainty area (wide confidence interval) should be higher to encourage the exploration of unknown areas; (3) For minimization problems such as floating wind turbine load shedding, the sampling value of the low mean area should be higher to guide the algorithm to converge efficiently to the global optimal solution.

[0100] 4. PSO-BO controller parameter tuning strategy based on safety constraints

[0101] The controller parameter tuning methods based on PSO and BO mentioned above, although they perform well in the final optimization results, ignore the safety issues in the optimization process. For example, insufficient initial samples or excessive exploration during the iteration process may cause the system to enter an unsafe zone or even crash. Therefore, it is necessary to add safety threshold constraints in the BO process to set a safety boundary for the system optimization to avoid touching the critical point that causes the system to crash during the exploration process.

[0102] Since PSO is better at global search, and secure Bayesian optimization is better at precise tuning and has higher security requirements for initial values, PSO can be used to perform an initial global search in the parameter search space, retaining several parameter points with better performance as the output of the PSO stage and providing them to Bayesian optimization. The goal is to provide a high-quality initial point for BO.

[0103] The high-quality sample points generated in the PSO stage are then used as the initial data for Bayesian optimization to train the Gaussian process (GP) proxy model; then the acquisition function (such as PI, UCB) in Bayesian optimization is used, and a safety threshold is added to select the next sampling point for iteration, thereby ensuring the safety and efficiency of the parameters obtained by optimization.

[0104] The specific operation steps of the PSO-BO controller parameter tuning strategy based on safety constraints are similar to the strategy in the previous part, with the main difference being steps one and three. In step one, when initializing the sample initial point, in order to ensure the safety of the initial point, the optimization result output by PSO is used as the initial condition D for constructing the initial GP model; in step three, when selecting the next test point in the optimization acquisition function, a safety threshold constraint is introduced, so that the next sampling point selected not only considers the expected improvement of the optimization target, but also ensures that the new parameters will not cause the system to exceed the safe operating range, thereby achieving the overall performance improvement of parameter tuning while ensuring safety. The introduced safety threshold represents the critical threshold for safe operation of the wind turbine in this application, which can be expressed as:

[0105] μ(x)-α·σ(x)≥threshold (14)

[0106] In the formula, μ(x) and σ(x) represent the mean and variance predicted by the GP model at the evaluation point, respectively, and α is the confidence factor, which is related to the confidence level. For example, a confidence factor of 1.96 corresponds to a 95% confidence interval. "Threshold" represents the safety threshold, which indicates the safety margin allowed in each step of the search process to ensure that the new evaluation point selected during the optimization process is safe. "Threshold" combines the uncertainty of the prediction to provide a soft constraint, that is, when selecting a new evaluation point, not only the predicted mean but also the predicted variance is considered to avoid selecting points with high predicted mean but large uncertainty, thereby improving safety and robustness. Due to the complex and changeable offshore environment, the parameter tuning of the variable pitch load reduction controller of floating wind turbines faces huge challenges. On the one hand, the inherent strong nonlinearity, uncertainty and complex load distribution characteristics of floating wind turbines significantly increase the difficulty of parameter tuning; on the other hand, improper parameter settings can easily cause system instability, threaten the safe operation of wind turbines, and even shorten their service life.

[0107] This embodiment aims to use particle swarm and safe Bayesian optimization technology to achieve safe and efficient tuning of floating wind turbine variable pitch load shedding controller parameters, and significantly improve the load shedding effect while ensuring system safety.

[0108] This embodiment has the following significant advantages:

[0109] 1. Improve controller performance: Safety Bayesian optimization can effectively balance exploration and utilization while ensuring safety, and quickly search for the optimal parameter combination, thereby maximizing the load reduction effect of the controller and reducing fatigue damage to wind turbine blades and towers.

[0110] 2. Reduce parameter tuning risks: By introducing safety constraints, safe Bayesian optimization can avoid entering unsafe areas during parameter exploration, effectively reduce the trial and error risks during the tuning process, and effectively ensure the safe operation of the system.

[0111] 3. Strong system adaptability: Safety Bayesian optimization can continuously update the Gaussian process model based on real-time data, accurately reflecting the dynamic changes of the system, thereby enhancing the controller's adaptability and robustness to complex and changeable sea conditions and operating conditions.

[0112] Embodiment 2

[0113] The purpose of this embodiment is to provide a floating wind turbine load reduction control system, including:

[0114] A first control module is configured to: use the difference between the actual speed of the wind turbine and the reference speed as an input of a unified pitch controller to obtain a common pitch angle of the blades;

[0115] A second control module is configured to: use the flapping moment of the blade load as an input of the independent pitch controller to obtain an additional pitch angle of each blade;

[0116] A third control module is configured to: superimpose the obtained common pitch angle of the blades and the additional pitch angles of each blade to control the floating wind turbine;

[0117] Wherein, the fitness function of the particle corresponding to the unified pitch controller parameter or the independent pitch controller parameter is calculated, the pitch controller parameter or the independent pitch controller parameter is adjusted, and the current fitness function corresponding to each particle in the particle group corresponding to the adjusted unified pitch controller parameter or the independent pitch controller parameter is calculated;

[0118] According to the current fitness function, the particle swarm corresponding to the current fitness function is adjusted, and target unified pitch controller parameters or target independent pitch controller parameters are output;

[0119] According to the output target unified pitch controller parameters or target independent pitch controller parameters, a safe Bayesian algorithm is used to perform iterative optimization to obtain final target unified pitch controller parameters or target independent pitch controller parameters.

[0120] In further embodiments, there is also provided:

[0121] An electronic device includes a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method described in Embodiment 1 is performed. For the sake of brevity, no further description is given here.

[0122] It should be understood that in this embodiment, the processor may be a central processing unit CPU, and the processor may also be other general-purpose processors, digital signal processors DSP, application-specific integrated circuits ASIC, off-the-shelf programmable gate arrays FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0123] The memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A portion of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0124] A computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, the method described in embodiment 1 is completed.

[0125] The method in the first embodiment can be directly embodied as a hardware processor, or a combination of hardware and software modules in the processor. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware. To avoid repetition, it will not be described in detail here.

[0126] A computer program product includes a computer program, and when the computer program is executed by a processor, the method described in the first embodiment is implemented.

[0127] The present invention also provides at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. The computer program product includes computer executable instructions, such as instructions included in a program module, which are executed in a device on a real or virtual processor of the target to perform the process / method as described above. Typically, program modules include routines, programs, libraries, objects, classes, components, data structures, etc. that perform specific tasks or implement specific abstract data types. In various embodiments, the functions of program modules can be combined or divided between program modules as needed. Machine executable instructions for program modules can be executed in local or distributed devices. In distributed devices, program modules can be located in local and remote storage media.

[0128] The computer program code for implementing the method of the present invention can be written in one or more programming languages. These computer program codes can be provided to the processor of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that the program code, when executed by the computer or other programmable data processing device, causes the function / operation specified in the flow chart and / or block diagram to be implemented. The program code can be executed completely on a computer, partially on a computer, as an independent software package, partially on a computer and partially on a remote computer or completely on a remote computer or server.

[0129] In the context of the present invention, computer program codes or related data may be carried by any appropriate carrier to enable a device, apparatus or processor to perform the various processes and operations described above. Examples of carriers include signals, computer readable media, etc. Examples of signals may include electrical, optical, radio, acoustic or other forms of propagation signals, such as carrier waves, infrared signals, etc.

[0130] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with this embodiment can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0131] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A floating wind turbine load reduction control method, characterized in that: include: The difference between the actual speed of the wind turbine and the reference speed is used as the input of the unified pitch controller to obtain the common pitch angle of the blades; The flapping moment of the blade load is used as the input of the independent pitch controller to obtain the additional pitch angle of each blade; The obtained common pitch angle of the blades and the additional pitch angles of the individual blades are superimposed to control the floating wind turbine; Wherein, the fitness function of the particle corresponding to the unified pitch controller parameter or the independent pitch controller parameter is calculated, the pitch controller parameter or the independent pitch controller parameter is adjusted, and the current fitness function corresponding to each particle in the particle group corresponding to the adjusted unified pitch controller parameter or the independent pitch controller parameter is calculated; According to the current fitness function, the particle swarm corresponding to the current fitness function is adjusted, and target unified pitch controller parameters or target independent pitch controller parameters are output; According to the output target unified pitch controller parameters or target independent pitch controller parameters, a safe Bayesian algorithm is used to perform iterative optimization to obtain final target unified pitch controller parameters or target independent pitch controller parameters.

2. A floating wind turbine load reduction control method according to claim 1, characterized in that: According to the current fitness function, the particle swarm corresponding to the current fitness function is adjusted, including: Determine the optimal position of the particle swarm corresponding to the current fitness function; The particle swarm corresponding to the adjusted fitness function or the current fitness function is adjusted according to the optimal position of the particle swarm.

3. A floating wind turbine load reduction control method according to claim 1, characterized in that: The fitness function is composed of power fluctuation, blade flapping moment, platform surge motion and tower top fore-and-aft displacement.

4. A floating wind turbine load reduction control method according to claim 1, characterized in that: According to the output target unified pitch controller parameters or target independent pitch controller parameters, the safe Bayesian algorithm is used for iterative optimization, specifically: The high-quality samples obtained based on the particle swarm optimization algorithm are used as the initial data for Bayesian optimization; Constructing a proxy model of unified pitch controller parameters or independent pitch controller parameters; Based on the constructed surrogate model, the posterior probability distribution of each parameter in the search space is calculated and the next test point is selected; The function value of the objective function at the new test point is calculated, and the agent model is updated to output the final target unified pitch controller parameters or target independent pitch controller parameters.

5. A floating wind turbine load reduction control method as claimed in claim 4, characterized in that: When optimizing the acquisition function to select the next test point, a safety threshold constraint is introduced, and the safety threshold constraint is a critical threshold for the safe operation of the wind turbine.

6. A floating wind turbine load reduction control method as claimed in claim 3, characterized in that: The next trial point is selected by a confidence upper bound or a probability increment sampling function.

7. A floating wind turbine load reduction control system, characterized in that: include: A first control module is configured to: use the difference between the actual speed of the wind turbine and the reference speed as an input of a unified pitch controller to obtain a common pitch angle of the blades; A second control module is configured to: use the flapping moment of the blade load as an input of the independent pitch controller to obtain an additional pitch angle of each blade; A third control module is configured to: superimpose the obtained common pitch angle of the blades and the additional pitch angles of each blade to control the floating wind turbine; Wherein, the fitness function of the particle corresponding to the unified pitch controller parameter or the independent pitch controller parameter is calculated, the pitch controller parameter or the independent pitch controller parameter is adjusted, and the current fitness function corresponding to each particle in the particle group corresponding to the adjusted unified pitch controller parameter or the independent pitch controller parameter is calculated; According to the current fitness function, the particle swarm corresponding to the current fitness function is adjusted, and target unified pitch controller parameters or target independent pitch controller parameters are output; According to the output target unified pitch controller parameters or target independent pitch controller parameters, a safe Bayesian algorithm is used to perform iterative optimization to obtain final target unified pitch controller parameters or target independent pitch controller parameters.

8. An electronic device, characterized in that: The method comprises a memory and a processor and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method according to any one of claims 1 to 6 is completed.

9. A computer-readable storage medium, characterized in that: Used to store computer instructions, which, when executed by a processor, complete the method described in any one of claims 1 to 6.

10. A computer program product, characterized in that The invention comprises a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 6.

Images