Multivariable cooperative control method and system of wind generating set

Abstract

The invention discloses a multivariable cooperative control method and system for a wind generating set, and relates to the field of wind generating sets, and the system comprises a fusion module which is used for collecting the wind speed, rotating speed, pitch angle, cabin vibration and power grid voltage signals of the wind generating set, and carrying out the normalization processing of each signal, so as to construct a multi-dimensional data matrix; the analysis module is used for receiving the multi-dimensional data matrix, mining a nonlinear association relationship among variables in the multi-dimensional data matrix and outputting a quantized coupling coefficient table and an association direction identifier; according to the multi-variable cooperative control method, the nonlinear incidence relation between the variables is deeply mined by accurately collecting and optimizing the multiple types of operation signals, scientific disassembly of the total power adjustment target is achieved, the accuracy of multi-variable cooperative control is effectively improved, and the influence of variable coupling interference on the adjustment effect is greatly reduced.

Description

Technical Field

[0001] This invention relates to the field of wind turbine generator technology, specifically to a multivariable collaborative control method and system for wind turbine generators. Background Technology

[0002] Wind turbine generators capture wind energy through blades and convert it into mechanical energy, which is then used to drive a generator to produce electricity via a transmission system. The control system adjusts the blade angle and rotation speed to adapt to changes in wind speed, achieving stable and efficient power output.

[0003] The invention patent application with application number 202210412210.6 discloses a wind turbine diagnostic method. The application aims to solve the problem that "wind turbines are often under complex operating conditions under alternating loads, and their components are also affected to varying degrees or even damaged. Therefore, the condition monitoring and real-time fault diagnosis of wind turbines are of great significance for monitoring and preventing damage to various components and reducing operation and maintenance costs."

[0004] However, while existing technologies can intelligently control various parameters of wind turbine generators, they do not consider which variables are coupled or have insufficient adjustment precision, resulting in poor coordinated control of various parameters of wind turbine generators.

[0005] To address this, we propose a multivariable collaborative control method and system for wind turbine generator sets. Summary of the Invention

[0006] In view of the above-mentioned shortcomings of the existing technology, the present invention provides a multivariable cooperative control method and system for wind turbine generator sets, which can effectively solve the problems of the existing technology.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions; This invention discloses a multivariable cooperative control system for wind turbine generator sets, comprising: The system comprises the following modules: a fusion module, which collects wind speed, rotational speed, pitch angle, nacelle vibration, and grid voltage signals from the wind turbine generator set, and normalizes these signals to construct a multidimensional data matrix; a parsing module, which receives the multidimensional data matrix, mines the nonlinear relationships between variables within the matrix, and outputs a quantified coupling coefficient table and correlation direction indicators; an allocation module, which obtains the coupling coefficient table and applies it along with the current operating threshold of the generator set to decompose the total power regulation target into independent control sub-targets for pitch angle, rotational speed, and excitation current; a generation module, which introduces variable coupling compensation factors into each control sub-target, constructs time-varying parameter control equations, and generates real-time control commands for pitch regulation, rotational speed regulation, and excitation regulation; a scheduling module, which receives real-time control commands, sorts them according to the response priority of the corresponding actuators, and controls the timing and amplitude of the actions of the pitch drive, converter, and braking system through a timing allocation strategy; and a correction module, which collects the actual output parameters after the actions of each actuator, calculates the deviation from the control sub-targets, and feeds the deviation values ​​back to the generation module to update the control equation parameters. The fusion module is interconnected with the parsing module via a wireless network. The parsing module is interconnected with the allocation module via a wireless network. The allocation module is interconnected with the generation module via a wireless network. The generation module is interconnected with the scheduling module via a wireless network. The scheduling module is interconnected with the correction module via a wireless network.

[0008] Furthermore, when the fusion module normalizes each signal, it follows the following rules: ; In the formula: This is the normalized output value of the i-th type of signal; This represents the original acquired value of the i-th type of signal; For this signal in the most recent sampling period The minimum value within; For this signal in the most recent sampling period The maximum value within; , These are the preset upper and lower limits of the normalized output value, respectively; The sampling period The multidimensional data matrix is ​​dynamically adjusted based on the current operating conditions of the unit. The dimension of the multidimensional data matrix is ​​m×n, where m is the number of signal categories and n is the number of sampling points in a single sampling period.

[0009] Furthermore, in the stage of the analysis module for mining the nonlinear correlation between variables, a mutual information calculation method based on an improved kernel function is adopted. The improved kernel function is: ; In the formula: To improve the kernel function output value; Let be the Euclidean distance between the i-th type variable and the j-th type variable; This is a dynamic kernel width parameter; Let be the time series fluctuation coefficients of the i-th type variable and the j-th type variable at time t; Each coupling coefficient in the output coupling coefficient table ; In the formula: Let be the coupling coefficient between the i-th type of variable and the j-th type of variable; Let be the mutual information value between the i-th type variable and the j-th type variable; The time-series correlation weights between the i-th type of variable and the j-th type of variable; This represents the self-information value of the i-th type of variable; This represents the self-information value of the j-th type of variable; The association direction identifier uses a bidirectional array. express, =1 indicates that the i-th type of variable has a positive correlation with the j-th type of variable. =-1 indicates that the i-th type of variable has a negative correlation with the j-th type of variable. =0 indicates that the two types of variables have no direct correlation; the direction of correlation is determined by the cross-core covariance matrix. The positive or negative sign of the trace is determined.

[0010] Furthermore, when the allocation module decomposes the total power adjustment target into independent control sub-targets, it first constructs a variable correlation influence matrix M based on the coupling coefficient table. M has a dimension of 3×3, corresponding to three types of control variables: pitch angle, rotational speed, and excitation current. The matrix elements... Where p,q∈{1,2,3}, The coupling coefficient is the output of the parsing module. Weights are used to control the interaction effects between variables; Based on the safety constraints corresponding to the current operating condition threshold of the unit, the target is decomposed through a hierarchical target allocation strategy and matrix solving. The mapping relationship between the total power regulation target and each control sub-target satisfies the following formula: ; In the formula: The sub-targets are pitch angle control, speed control, and excitation current control. This is the inverse of the correlation influence matrix M; , , Assign power weights to pitch angle, rotational speed, and excitation current; The target for total power regulation; The sum of the off-diagonal elements in the p-th row of the correlation influence matrix; The safety constraint correction vector.

[0011] Furthermore, the variable coupling compensation factor in the generation module is a time-varying parameter that changes dynamically with time, and its value is jointly determined by the coupling coefficient, variable deviation rate, and unit operating status evaluation value output by the analysis module. ; In the formula: This is the output value of the k-th type of control instruction; , , These are the dynamic proportional coefficient, dynamic integral coefficient, and dynamic differential coefficient. Let be the real-time deviation value of the k-th type of control sub-objective; The current moment; for The real-time deviation value of the k-th type of control sub-target at time k; This is the coupling compensation factor for the k-th type of control command relative to the j-th type of associated variable; Let be the real-time deviation value of the j-th type of associated variable; The ; In the formula: Let be the coupling coefficient between the control variable corresponding to the k-th type of control instruction and the j-th type of associated variable; Let be the real-time deviation rate of the j-th type of associated variable; This is a real-time operating status assessment value for the unit; This is the deviation sensitivity adjustment coefficient; This is a dynamic correction item.

[0012] Furthermore, the timing allocation strategy in the scheduling module is constructed based on the response characteristic parameters of the actuator and the priority coefficient of the control command; The response characteristic parameters of the actuator include response delay time, upper limit of action rate, load tolerance threshold, and interaction interference coefficient, and the priority coefficient. ; In the formula: These are preset weighting coefficients, all of which are positive numbers and their sum is 1; The deviation weights corresponding to the control commands; This represents the maximum coupling coefficient output by the parsing module. Assign safety priority weights to the generating units; During the timing allocation process, priority coefficients are used first. All real-time control commands are sorted in descending order, and execution windows are allocated to high-priority commands. Then, an interference matrix is ​​established based on the interaction interference coefficient of the actuators. The mutual exclusion intervals of each actuator's actions are determined according to the interference matrix, so that the timing difference of actions of adjacent actuators is not less than the interference avoidance time calculated based on the interference matrix, and the superposition value of the action amplitude corresponding to all control commands executed at the same time does not exceed the preset safety threshold determined based on the unit safety design standard. The start time and duration of the execution window are dynamically adjusted by the target response speed of the control command, the upper limit of the action rate of the actuator, and the current load status. The action amplitude is adaptively limited according to the load tolerance threshold of the actuator.

[0013] Furthermore, in the deviation calculation stage of the correction module, the deviation value between the actual output parameter and the control sub-target is decomposed into static deviation components, dynamic deviation components, and coupled deviation components, and the comprehensive deviation value is calculated using the following formula: ; In the formula: This represents the comprehensive deviation value of the k-th type of control sub-objective; , , These are static deviation weights, dynamic deviation weights, and coupling deviation weights. The actual output parameters for the k-th type of control sub-target; This is the setpoint for the k-th type of control sub-target; This represents the rate of change of the actual output parameters; To control the set rate of change of sub-targets; This represents the actual output value of the j-th type of association variable; The target value for the j-th type of association variable; The comprehensive deviation value After being fed back to the generation module, the time-varying parameter control equations are updated synchronously. , , and .

[0014] On the other hand, a multivariable collaborative control method for wind turbine generator sets includes: The system collects and normalizes wind speed, rotational speed, pitch angle, nacelle vibration, and grid voltage signals from wind turbine generators to construct a multi-dimensional data matrix that integrates multi-source operational information. It then mines the nonlinear correlations between variables in the multi-dimensional data matrix, outputting a quantified coupling coefficient table and a bidirectional array indicating the direction of these correlations. Based on the coupling coefficient table, it constructs a variable correlation influence matrix. Combining this with the unit's operating condition safety constraints, it decomposes the total power regulation target into independent control sub-targets using a hierarchical target allocation strategy. It introduces a time-varying variable coupling compensation factor to construct time-varying control equations and generate real-time control commands for three types of regulation. Based on the actuator response characteristics and control command priority coefficients, it controls the timing and amplitude of each actuator's actions using a timing allocation strategy. Finally, it collects the actual output parameters after the actuators' actions, decomposes the deviation components, calculates the comprehensive deviation value, and synchronously feeds it back to the control equations to update the equation parameters.

[0015] Compared with the known prior art, the technical solution provided by this invention has the following beneficial effects: This invention provides a multivariable collaborative control method and system for wind turbine generators. During execution, this method and system accurately collect and optimize various operating signals, deeply explore the nonlinear relationships between variables, and scientifically decompose the total power regulation target, effectively improving the accuracy of multivariable collaborative control and significantly reducing the impact of variable coupling interference on the regulation effect. Simultaneously, it introduces time-varying parameters and coupling compensation factors to construct control equations, combining dynamic coefficients to adapt to different operating conditions, accelerating the unit's response speed and enhancing the real-time performance and flexibility of power regulation. Furthermore, by rationally prioritizing control command execution, it avoids interaction interference between actuators, ensuring that the timing and amplitude of actions adapt to the unit's operating state and improving operational stability. Additionally, it relies on deviation feedback to dynamically correct control parameters, continuously optimizing the control strategy, reducing static, dynamic, and coupling deviations, lowering overshoot risk and operating losses, balancing unit operating efficiency and safety constraints, and adapting to complex and variable wind speeds and grid conditions, thereby improving the overall operating performance and reliability of the wind turbine generator. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0017] Figure 1 This is a schematic diagram of a multivariable collaborative control system for a wind turbine generator set. Figure 2 This is a flowchart illustrating a multivariable collaborative control method for wind turbine generators. Detailed Implementation

[0018] 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, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0019] The present invention will be further described below with reference to embodiments.

[0020] Example: This embodiment provides a multivariable cooperative control system for a wind turbine generator set, such as... Figure 1 As shown, it includes: The fusion module is used to collect wind speed, rotational speed, pitch angle, nacelle vibration and grid voltage signals of the wind turbine generator set, and normalize each signal to construct a multi-dimensional data matrix. When the fusion module normalizes each signal, it follows the following rules: ; In the formula: This is the normalized output value of the i-th type of signal; This represents the original acquired value of the i-th type of signal; For this signal in the most recent sampling period The minimum value within; For this signal in the most recent sampling period The maximum value within; , These are the preset upper and lower limits of the normalized output value, respectively; The above formula calculates the difference between the original acquired value of the i-th type of signal and the maximum value in the most recent sampling period, and then normalizes and scales it to the preset upper and lower limit range. At the same time, the sampling period can be dynamically adjusted according to the current operating conditions of the unit. This not only eliminates the impact of the difference in the dimensions of different types of signals on the subsequent construction of the multidimensional data matrix, but also adapts to the changing operating conditions of the unit, ensuring the consistency and timeliness of the data. Sampling period The multidimensional data matrix is ​​dynamically adjusted based on the current operating conditions of the unit. The dimensions of the multidimensional data matrix are m×n, where m is the number of signal categories and n is the number of sampling points in a single sampling period. The parsing module is used to receive a multidimensional data matrix, mine the nonlinear correlation between variables in the multidimensional data matrix, and output a quantized coupling coefficient table and correlation direction identifier. In the analysis module's stage of uncovering nonlinear relationships between variables, a mutual information calculation method based on an improved kernel function is adopted. The improved kernel function is as follows: ; In the formula: To improve the kernel function output value; Let be the Euclidean distance between the i-th type variable and the j-th type variable; This is a dynamic kernel width parameter; Let be the time series fluctuation coefficients of the i-th type variable and the j-th type variable at time t; The above formula uses Euclidean distance to measure the basic correlation between two types of variables. By introducing a kernel width parameter and a time-series fluctuation coefficient that change dynamically over time, the kernel function is improved. This not only retains the advantage of the kernel function in capturing nonlinear relationships, but also adapts to the dynamic changes of variables in real time, making the output value of the kernel function more consistent with the actual correlation characteristics of variables in unit operation. Each coupling coefficient in the output coupling coefficient table ; In the formula: Let be the coupling coefficient between the i-th type of variable and the j-th type of variable; Let be the mutual information value between the i-th type variable and the j-th type variable; The time-series correlation weights between the i-th type of variable and the j-th type of variable; This represents the self-information value of the i-th type of variable; This represents the self-information value of the j-th type of variable; The above formula combines the ratio of mutual information value to self-information value between variables and incorporates temporal correlation weight to quantify the coupling coefficient. It fully considers the information correlation strength between variables and also takes into account the correlation characteristics in the temporal dimension, so that the calculation results can accurately reflect the degree of coupling between the i-th type variable and the j-th type variable. The association direction identifier uses a two-way array. express, =1 indicates that the i-th type of variable has a positive correlation with the j-th type of variable. =-1 indicates that the i-th type of variable has a negative correlation with the j-th type of variable. =0 indicates that the two types of variables have no direct correlation; the direction of correlation is determined by the cross-core covariance matrix. The sign of the trace is determined; Among them, based on the improved kernel function mentioned above, the mutual information value between the i-th type variable and the j-th type variable is... : ; In the formula: Let be the kernel covariance matrix of the i-th type of variable; Let be the kernel covariance matrix of the j-th type of variable; Let be the cross-kernel covariance matrix between the i-th type variable and the j-th type variable; Represents the determinant of a matrix; This represents the k-th sampled value of the i-th type of variable; The mean of the i-th type of variable during the sampling period; This represents the number of sampling points within the sampling period. The above formula constructs the kernel covariance matrix and the cross kernel covariance matrix based on the improved kernel function, and derives the mutual information value through the ratio operation of the matrix determinants. It not only utilizes the ability of the kernel function to handle nonlinear relationships, but also realizes the accurate calculation of mutual information between variables through matrix operations, providing a reliable foundation for solving the coupling coefficients in the future. It is calculated by weighting the standard deviations of the two types of variables at the current time t; It is calculated by the average rate of change of the two types of variables in the most recent sampling period; The allocation module is used to obtain the coupling coefficient table, and apply the coupling coefficient table and the current operating condition threshold of the unit to decompose the total power regulation target into independent control sub-targets of pitch angle, speed and excitation current. When the allocation module decomposes the total power regulation target into independent control sub-targets, it first constructs a variable correlation influence matrix M based on the coupling coefficient table. M has a dimension of 3×3, corresponding to three types of control variables: pitch angle, speed, and excitation current. The matrix elements... Where p,q∈{1,2,3}, The coupling coefficient is the output of the parsing module. Weights are used to control the interaction effects between variables; Based on the safety constraints corresponding to the current operating condition threshold of the unit, the target is decomposed through a hierarchical target allocation strategy and matrix solving. The mapping relationship between the total power regulation target and each control sub-target satisfies the following formula: ; In the formula: The sub-targets are pitch angle control, speed control, and excitation current control. This is the inverse of the correlation influence matrix M; , , Assign power weights to pitch angle, rotational speed, and excitation current; The target for total power regulation; The sum of the off-diagonal elements in the p-th row of the correlation influence matrix; The safety constraint correction vector; The above formula first constructs a 3×3 variable correlation influence matrix based on the coupling coefficient and the interaction influence weight, which accurately describes the interaction between the three types of control variables: pitch angle, speed, and excitation current. Then, it combines the power allocation weight, the total power adjustment target, and the sum of the off-diagonal elements of the correlation influence matrix, and achieves the initial decomposition of the total target through matrix inversion. At the same time, it introduces a safety constraint correction vector composed of the deviation between the safe operating boundary of the control variables and the actual value, which not only ensures that the allocation of control sub-targets matches the variable correlation characteristics, but also ensures that each sub-target is within the preset safety range, taking into account both adjustment accuracy and operational safety. in, The value range is [0.1, 0.9]. The larger the value is when the synergistic effect of the p-th type of control variable and the q-th type of control variable can significantly improve the unit's operating efficiency, the smaller the value is when the interaction between the two may cause operational risks or weaken the regulation accuracy. Each component is calculated from the deviation between the preset safe operating range boundary value and the current actual value of the corresponding control variable, so that... ∈ , ∈ , ∈ , This is a preset safety threshold that is dynamically updated based on the unit's design parameters and real-time operating conditions. , , All are positive numbers, and their sum is 1; The generation module is used to introduce variable coupling compensation factors into each control sub-objective, construct time-varying parameter control equations, and generate real-time control commands for pitch regulation, speed regulation, and excitation regulation. The variable coupling compensation factor in the generation module is a time-varying parameter that changes dynamically with time. Its value is determined by the coupling coefficient, variable deviation rate, and unit operating status evaluation value output by the analysis module. ; In the formula: is the output value of the kth type of control command, where k corresponds to pitch regulation, speed regulation, and excitation regulation, respectively. , , These are the dynamic proportional coefficient, dynamic integral coefficient, and dynamic differential coefficient. Let be the real-time deviation value of the k-th type of control sub-objective; The current moment; for The real-time deviation value of the k-th type of control sub-target at time k; This is the coupling compensation factor for the k-th type of control command relative to the j-th type of associated variable; Let be the real-time deviation value of the j-th type of associated variable; The above formula applies the core logic of PID control, introducing dynamically changing proportional, integral, and derivative coefficients to adapt to the dynamic situations of deviation magnitude, static deviation accumulation, and deviation change rate, respectively. At the same time, it innovatively adds a cross-variable compensation term based on coupling coefficient, incorporating the real-time deviation of the associated variable into the calculation of the current control command, thereby achieving an organic combination of single-variable precise adjustment and multi-variable collaborative compensation. It can not only quickly respond to the deviation of a single control sub-target, but also effectively offset the interference caused by the coupling of other variables. in, The value range is a preset [0.1, 5]. The larger the value, the more likely it is to be when the generator set needs to respond quickly to power regulation requirements. The smaller the value, the more stable the unit's operating conditions; The value range is preset to [0.01, 1]. The value is larger when the static deviation of the k-th type of control sub-target persists or the cumulative deviation is large, and smaller when the deviation decreases rapidly, approaches the control sub-target set value, or there is an overcompensation risk. The value range is preset to [0.05, 2]. When the deviation change rate of the k-th type of control sub-target... The larger the value, the greater the risk of overshoot when the deviation is large or the unit has an overshoot risk; the smaller the value, the smaller the value when the deviation rate is flat and the operating condition fluctuation is small. ; In the formula: Let be the coupling coefficient between the control variable corresponding to the k-th type of control instruction and the j-th type of associated variable; Let be the real-time deviation rate of the j-th type of associated variable; This is a real-time operating status assessment value for the unit; This is the deviation sensitivity adjustment coefficient; This is a dynamic correction item; The above formula is based on the coupling coefficient between variables, combined with the real-time deviation rate of the associated variables and the unit operating status evaluation value. The deviation sensitivity adjustment coefficient is used to adapt the compensation intensity under different deviation scenarios. At the same time, a dynamic correction term that changes with the operating condition characteristics is introduced, so that the coupling compensation factor can respond in real time to the correlation strength of variables, the operating status of the unit and the deviation changes. This ensures the pertinence of the coupling compensation and enhances its adaptability to the dynamic operating conditions of the unit. in, The value range is [0.1, 1], which is calculated by comprehensively considering wind speed level, nacelle vibration amplitude, and grid voltage stability, and follows the principle that the more stable the wind speed, the smaller the vibration, and the more stable the voltage. The closer to 1; The preset value range is [0.5, 5]. The value is larger when the allowable deviation threshold of the corresponding control sub-target of the unit is smaller and the tracking accuracy of the main control target needs to be prioritized; the value is smaller when a certain deviation is allowed for the control sub-target and coupling compensation needs to be strengthened to adapt to the requirements of multi-variable collaborative control. The characteristic period representing the unit's operating condition is specifically calculated adaptively from the real-time wind speed fluctuation period. The scheduling module is used to receive real-time control commands, sort them according to the response priority of the corresponding actuators, and control the timing and amplitude of the pitch drive, converter and braking system through a timing allocation strategy. The timing allocation strategy in the scheduling module is constructed based on the response characteristic parameters of the actuator and the priority coefficient of the control command; The response characteristic parameters of the actuator include response delay time, upper limit of action rate, load tolerance threshold, cross-interference coefficient, and priority coefficient. ; In the formula: These are preset weighting coefficients, all of which are positive numbers and their sum is 1; The deviation weights corresponding to the control commands; This represents the maximum coupling coefficient output by the parsing module. Assign safety priority weights to the generating units; It should be noted that: The value of 'a' is larger when the real-time deviation corresponding to the control command is closer to or exceeds the preset deviation threshold, and the deviation needs to be corrected quickly to ensure adjustment accuracy; the value of 'a' is smaller when the real-time deviation is much lower than the preset deviation threshold, the control sub-target approaches stability, and there is no need to prioritize the response to the deviation. When the maximum value of the coupling coefficient output by the analytical module is larger, and the coupling interference between variables has a more significant impact on the adjustment effect, and the coupling risk needs to be avoided, the value of b should be larger; when the maximum value of the coupling coefficient is small, the mutual interference between variables is weak, and the impact on the coordinated control can be ignored, the value of b should be smaller. When safety-related parameters such as engine room vibration and grid voltage are closer to the preset safety limits, the higher the safety risk of unit operation, the larger the value of c; when safety-related parameters are far from the safety limits, the unit is in stable operating conditions and the safety risk is low, the value of c is smaller. The above formula integrates the deviation weight, maximum coupling coefficient, and unit safety priority weight corresponding to the integrated control command. The priority coefficient is calculated by weighting the preset weight coefficients that sum to 1. This not only reflects the urgency of the control command deviation and the importance of variable coupling, but also highlights the core position of unit safety, providing a basis for priority determination for the timing allocation of subsequent control commands. During the timing allocation process, priority coefficients are used first. All real-time control commands are sorted in descending order, and execution windows are allocated to high-priority commands. Then, an interference matrix is ​​established based on the interaction interference coefficient of the actuators. The mutual exclusion intervals of each actuator's actions are determined according to the interference matrix, so that the timing difference of actions of adjacent actuators is not less than the interference avoidance time calculated based on the interference matrix, and the superposition value of the action amplitude corresponding to all control commands executed at the same time does not exceed the preset safety threshold determined based on the unit safety design standard. The start time and duration of the execution window are dynamically adjusted by the target response speed of the control command, the upper limit of the action rate of the actuator, and the current load status. The action amplitude is adaptively limited according to the load tolerance threshold of the actuator. in, The value range is [0,1]. The value is larger when the real-time deviation is closer to or exceeds the preset deviation threshold, and smaller when the real-time deviation is much lower than the preset deviation threshold and the control sub-target approaches stability. The value range is [0,1]. The value is larger when the safety-related parameters such as engine room vibration and grid voltage are closer to the preset safety limit, and smaller when the safety-related parameters are far from the safety limit and the unit is in a stable operating condition. The interaction interference coefficient of the actuators is obtained by quantifying and calculating the normalized product of the sum of the action amplitudes of the two associated actuators and the difference in response delay; the interference avoidance time is determined by dividing the interaction interference coefficient by the upper limit of the action rate of the actuators and combining the maximum value of the response delay times of the two actuators. The correction module is used to collect the actual output parameters after the action of each actuator, calculate the deviation with the control sub-target, and feed the deviation value back to the generation module to update the control equation parameters. In the deviation calculation stage of the correction module, the deviation between the actual output parameters and the control sub-target is decomposed into static deviation components, dynamic deviation components, and coupled deviation components. The comprehensive deviation value is calculated using the following formula: ; In the formula: This represents the comprehensive deviation value of the k-th type of control sub-objective; , , These are static deviation weights, dynamic deviation weights, and coupling deviation weights. The actual output parameters for the k-th type of control sub-target; This is the setpoint for the k-th type of control sub-target; This represents the rate of change of the actual output parameters; To control the set rate of change of sub-targets; This represents the actual output value of the j-th type of association variable; The target value for the j-th type of association variable; The above formula decomposes the comprehensive deviation into static deviation, dynamic deviation and coupling deviation. By quantifying the influence of each deviation component through preset weight coefficients, it comprehensively reflects the deviation between the actual output of the control sub-target and the set value under the influence of numerical value, rate of change and related variables, and provides a comprehensive and accurate basis for the correction of subsequent control parameters. Overall deviation value After being fed back to the generation module, the time-varying parameter control equations are updated synchronously. , , and ; in, ; In the formula: For the revised , , and ; For proportional gain adjustment, 0 < <1; The preset maximum error threshold; For integral adjustment of gain, 0 < <0.8; For integration variables; For differential adjustment of gain, 0 < <0.6; The sampling period; For coupling adjustment gain, 0 < <0.5; For associated error parameters; for The preset maximum threshold; It should be noted that: When the unit has high requirements for steady-state operation accuracy, and the persistent static deviation will significantly affect the power generation efficiency or operational stability. The larger the value, the more important it is to prioritize dynamic response or coupled disturbances when operating conditions fluctuate drastically and static deviations have a negligible impact on control performance. The smaller the value; When the unit is under dynamic operating conditions such as sudden changes in wind speed and fluctuations in grid voltage, dynamic deviations can easily lead to overshoot or equipment impact. The larger the value, the less impact dynamic deviation has on operational safety and adjustment accuracy when the operating conditions are stable, the rate of change is gradual, and the dynamic deviation has a smaller impact. The smaller the value; When the coupling coefficient between variables is large and the coupling deviation significantly interferes with the adjustment accuracy of the current control sub-target, The larger the value, the better; when the coupling coefficient is small, the mutual interference between variables is weak, and the impact of coupling deviation on the control effect is negligible. The smaller the value; The above formula is based on the comprehensive deviation value and its derived parameters such as integral and rate of change. It corrects the dynamic proportional, integral, differential coefficients and coupling compensation factors respectively, so that the control equation parameters can be dynamically adjusted according to the deviation. This not only ensures the pertinence of the deviation correction, but also improves the adaptive capability and adjustment accuracy of the control command to changes in the unit's operating status. The fusion module interacts with the parsing module via a wireless network. The parsing module interacts with the allocation module via a wireless network. The allocation module interacts with the generation module via a wireless network. The generation module interacts with the scheduling module via a wireless network. The scheduling module interacts with the correction module via a wireless network.

[0021] In this embodiment, the fusion module collects wind speed, rotational speed, pitch angle, nacelle vibration, and grid voltage signals from the wind turbine generator set. It normalizes these signals to construct a multidimensional data matrix. The parsing module then receives the multidimensional data matrix, mines the nonlinear relationships between variables, and outputs a quantified coupling coefficient table and correlation direction identifiers. The allocation module further obtains the coupling coefficient table and, using it along with the current operating condition thresholds of the generator set, decomposes the total power regulation target into independent control sub-targets for pitch angle, rotational speed, and excitation current. The generation module simultaneously introduces variable coupling compensation factors for each control sub-target, constructs time-varying parameter control equations, and generates real-time control commands for pitch regulation, rotational speed regulation, and excitation regulation. The scheduling module receives these real-time control commands, sorts them according to the response priority of the corresponding actuators, and controls the timing and amplitude of the pitch drive, converter, and braking system actions through a timing allocation strategy. Finally, the correction module collects the actual output parameters after the actions of each actuator, calculates the deviation from the control sub-targets, and feeds the deviation values ​​back to the generation module to update the control equation parameters.

[0022] In the above embodiments, the system can accurately capture the correlation characteristics of multiple parameters of wind turbine operation, scientifically allocate power regulation targets, dynamically optimize control commands, rationally schedule execution actions and correct deviations in real time, effectively improve power regulation accuracy and response speed, reduce the impact of operating condition fluctuations, reduce equipment operation risks, enhance unit operation stability and coordination, and improve power generation efficiency.

[0023] Example 2: At the implementation level, based on Example 1, this example refers to... Figure 2 A more detailed description is provided of the multivariable cooperative control system for a wind turbine generator set in Example 1: A multivariable cooperative control method for wind turbine generator sets includes: The wind speed, rotational speed, pitch angle, nacelle vibration, and grid voltage signals of the wind turbine generator are collected, normalized, and a multi-dimensional data matrix is ​​constructed to integrate multi-source operating information. Mine the nonlinear relationships between variables in a multidimensional data matrix and output a table of quantified coupling coefficients and a bidirectional array that identifies the direction of the relationships; Based on the coupling coefficient table, a variable correlation influence matrix is ​​constructed. Combined with the safety constraints of unit operating conditions, the total power regulation target is decomposed into independent control sub-targets through a hierarchical target allocation strategy. By introducing a time-varying variable coupling compensation factor, a time-varying control equation is constructed, generating real-time control commands for three types of regulation; Based on the response characteristics of the actuators and the priority coefficient of the control commands, the timing and amplitude of the actions of each actuator are controlled by a timing allocation strategy. The actual output parameters after the actuator moves are collected, the deviation components are decomposed and the comprehensive deviation value is calculated, and the feedback is synchronously fed back to the control equation to update the equation parameters.

[0024] In summary, the systems and methods described in the above embodiments, during execution, accurately collect and optimize multiple types of operating signals, deeply explore the nonlinear correlations between variables, and scientifically decompose the total power regulation target, effectively improving the accuracy of multi-variable collaborative control and significantly reducing the impact of variable coupling interference on the regulation effect. Simultaneously, they introduce time-varying parameters and coupling compensation factors to construct control equations, combining dynamic coefficients to adapt to different operating conditions, accelerating the unit's response speed, and enhancing the real-time performance and flexibility of power regulation. Furthermore, by rationally prioritizing the execution of control commands, they avoid interactive interference between actuators, ensuring that the timing and amplitude of actions adapt to the unit's operating state and improving operational stability. Additionally, they rely on deviation feedback to dynamically correct control parameters, continuously optimizing the control strategy, reducing static, dynamic, and coupling deviations, lowering overshoot risk and operating losses, balancing unit operating efficiency and safety constraints, and adapting to complex and variable wind speeds and grid conditions, thereby improving the overall operating performance and reliability of the wind turbine generator set.

[0025] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multivariable cooperative control system for a wind turbine generator set, characterized in that, include: The fusion module is used to collect wind speed, rotational speed, pitch angle, nacelle vibration and grid voltage signals of the wind turbine generator set, and normalize each signal to construct a multi-dimensional data matrix. The parsing module is used to receive a multidimensional data matrix, mine the nonlinear correlation between variables in the multidimensional data matrix, and output a quantized coupling coefficient table and correlation direction identifier. The allocation module is used to obtain the coupling coefficient table, and apply the coupling coefficient table and the current operating condition threshold of the unit to decompose the total power regulation target into independent control sub-targets of pitch angle, speed and excitation current. The generation module is used to introduce variable coupling compensation factors into each control sub-objective, construct time-varying parameter control equations, and generate real-time control commands for pitch regulation, speed regulation, and excitation regulation. The scheduling module is used to receive real-time control commands, sort them according to the response priority of the corresponding actuators, and control the timing and amplitude of the pitch drive, converter and braking system through a timing allocation strategy. The correction module is used to collect the actual output parameters after the actions of each actuator, calculate the deviation from the control sub-target, and feed the deviation value back to the generation module to update the control equation parameters.

2. The multivariable cooperative control system for a wind turbine generator set according to claim 1, characterized in that, When the fusion module normalizes each signal, it follows the following rules: ； In the formula: This is the normalized output value of the i-th type of signal; This represents the original acquired value of the i-th type of signal; For this signal in the most recent sampling period The minimum value within; For this signal in the most recent sampling period The maximum value within; , These are the preset upper and lower limits of the normalized output value, respectively; The sampling period The multidimensional data matrix is ​​dynamically adjusted based on the current operating conditions of the unit. The dimension of the multidimensional data matrix is ​​m×n, where m is the number of signal categories and n is the number of sampling points in a single sampling period.

3. The multivariable cooperative control system for a wind turbine generator set according to claim 1, characterized in that, In the stage of the analysis module for mining the nonlinear correlation between variables, a mutual information calculation method based on an improved kernel function is adopted. The improved kernel function is: ； In the formula: To improve the kernel function output value; Let be the Euclidean distance between the i-th type variable and the j-th type variable; This is a dynamic kernel width parameter; Let be the time series fluctuation coefficients of the i-th type variable and the j-th type variable at time t; Each coupling coefficient in the output coupling coefficient table ; In the formula: Let be the coupling coefficient between the i-th type of variable and the j-th type of variable; Let be the mutual information value between the i-th type variable and the j-th type variable; The time-series correlation weights between the i-th type of variable and the j-th type of variable; Let be the self-information value of the i-th type of variable; Let be the self-information value of the j-th type variable; The association direction identifier uses a bidirectional array. express, =1 indicates that the i-th type of variable has a positive correlation with the j-th type of variable. =-1 indicates that the i-th type of variable has a negative correlation with the j-th type of variable. =0 indicates that the two types of variables have no direct correlation; the direction of correlation is determined by the cross-core covariance matrix. The positive or negative sign of the trace is determined.

4. The multivariable cooperative control system for a wind turbine generator set according to claim 1, characterized in that, When the allocation module decomposes the total power adjustment target into independent control sub-targets, it first constructs a variable correlation influence matrix M based on the coupling coefficient table. M has a dimension of 3×3 and corresponds to three types of control variables: pitch angle, speed, and excitation current. The matrix elements... Where p,q∈{1,2,3}, The coupling coefficient is the output of the parsing module. Weights are used to control the interaction effects between variables; Based on the safety constraints corresponding to the current operating condition threshold of the unit, the target is decomposed through a hierarchical target allocation strategy and matrix solving. The mapping relationship between the total power regulation target and each control sub-target satisfies the following formula: ； In the formula: The sub-targets are pitch angle control, speed control, and excitation current control. This is the inverse of the correlation influence matrix M; , , Assign power weights to pitch angle, rotational speed, and excitation current; The target for total power regulation; The sum of the off-diagonal elements in the p-th row of the correlation influence matrix; The safety constraint correction vector.

5. A multivariable cooperative control system for a wind turbine generator set according to claim 1, characterized in that, The variable coupling compensation factor in the generation module is a time-varying parameter that changes dynamically with time. Its value is determined by the coupling coefficient, variable deviation rate, and unit operating status evaluation value output by the analysis module. ； In the formula: This is the output value of the k-th type of control instruction; , , These are the dynamic proportional coefficient, dynamic integral coefficient, and dynamic differential coefficient. Let be the real-time deviation value of the k-th type of control sub-objective; The current moment; for The real-time deviation value of the k-th type of control sub-target at time k; This is the coupling compensation factor for the k-th type of control command relative to the j-th type of associated variable; denoted as the real-time deviation value of the j-th type of associated variable.

6. A multivariable cooperative control system for a wind turbine generator set according to claim 5, characterized in that, The ; In the formula: Let be the coupling coefficient between the control variable corresponding to the k-th type of control instruction and the j-th type of associated variable; Let be the real-time deviation rate of the j-th type of associated variable; This is a real-time operating status assessment value for the unit; This is the deviation sensitivity adjustment coefficient; This is a dynamic correction item.

7. A multivariable cooperative control system for a wind turbine generator set according to claim 1, characterized in that, The timing allocation strategy in the scheduling module is constructed based on the response characteristic parameters of the actuator and the priority coefficient of the control command; The response characteristic parameters of the actuator include response delay time, upper limit of action rate, load tolerance threshold, and interaction interference coefficient, and the priority coefficient. ; In the formula: These are preset weighting coefficients, all of which are positive numbers and their sum is 1; The deviation weights corresponding to the control commands; This represents the maximum coupling coefficient output by the parsing module. Assign safety priority weights to the generating units; During the timing allocation process, priority coefficients are used first. All real-time control commands are sorted in descending order, and execution windows are allocated to high-priority commands. Then, an interference matrix is ​​established based on the interaction interference coefficient of the actuators. The mutual exclusion intervals of each actuator's actions are determined according to the interference matrix, so that the timing difference of actions of adjacent actuators is not less than the interference avoidance time calculated based on the interference matrix, and the superposition value of the action amplitude corresponding to all control commands executed at the same time does not exceed the preset safety threshold determined based on the unit safety design standard. The start time and duration of the execution window are dynamically adjusted by the target response speed of the control command, the upper limit of the action rate of the actuator, and the current load status. The action amplitude is adaptively limited according to the load tolerance threshold of the actuator.

8. A multivariable cooperative control system for a wind turbine generator set according to claim 1, characterized in that, In the deviation calculation stage of the correction module, the deviation between the actual output parameter and the control sub-target is decomposed into static deviation components, dynamic deviation components, and coupled deviation components, and the comprehensive deviation value is calculated using the following formula: ； In the formula: This represents the comprehensive deviation value of the k-th type of control sub-objective; , , These are static deviation weights, dynamic deviation weights, and coupling deviation weights. The actual output parameters for the k-th type of control sub-target; This is the setpoint for the k-th type of control sub-target; This represents the rate of change of the actual output parameters; To control the set rate of change of sub-targets; This represents the actual output value of the j-th type of association variable; The target value for the j-th type of association variable; The comprehensive deviation value After being fed back to the generation module, the time-varying parameter control equations are updated synchronously. , , and .

9. A multivariable cooperative control system for a wind turbine generator set according to claim 1, characterized in that, The fusion module is interconnected with the parsing module via a wireless network. The parsing module is interconnected with the allocation module via a wireless network. The allocation module is interconnected with the generation module via a wireless network. The generation module is interconnected with the scheduling module via a wireless network. The scheduling module is interconnected with the correction module via a wireless network.

10. A multivariable cooperative control method for a wind turbine generator set, wherein the method is an implementation method of a multivariable cooperative control system for a wind turbine generator set as described in any one of claims 1-9, characterized in that, include: The wind speed, rotational speed, pitch angle, nacelle vibration, and grid voltage signals of the wind turbine generator are collected, normalized, and a multi-dimensional data matrix is ​​constructed to integrate multi-source operating information. Mine the nonlinear relationships between variables in a multidimensional data matrix and output a table of quantified coupling coefficients and a bidirectional array that identifies the direction of the relationships; Based on the coupling coefficient table, a variable correlation influence matrix is ​​constructed. Combined with the safety constraints of unit operating conditions, the total power regulation target is decomposed into independent control sub-targets through a hierarchical target allocation strategy. By introducing a time-varying variable coupling compensation factor, a time-varying control equation is constructed, generating real-time control commands for three types of regulation; Based on the response characteristics of the actuators and the priority coefficient of the control commands, the timing and amplitude of the actions of each actuator are controlled by a timing allocation strategy. The actual output parameters after the actuator moves are collected, the deviation components are decomposed and the comprehensive deviation value is calculated, and the feedback is synchronously fed back to the control equation to update the equation parameters.

Images