A method, management device and storage medium for suppressing subsynchronous oscillation of a doubly-fed wind farm based on intelligent learning and predictive control

Abstract

The application discloses a double-fed wind farm subsynchronous oscillation suppression method based on intelligent learning and predictive control. First, based on double-fed wind farm dominant oscillation mode analysis, double-fed wind turbine rotor side converter outer ring damping control prediction model and inner ring damping control prediction model are analyzed; a damping optimization function approximator is constructed based on a robust model predictive control, and a double-fed wind turbine outer ring damping optimization control module and an inner ring damping optimization control module are established; then, based on Q learning optimization, optimal control laws of the double-fed wind turbine outer ring damping optimization control module and the inner ring damping optimization control module are solved, and the optimal control laws are stored based on data cubes; finally, the double-fed wind turbine outer ring damping optimization control module and the inner ring damping optimization control module are incorporated into the inner and outer ring control of the rotor side converter, and double-fed wind farm subsynchronous oscillation suppression is realized. The application considers the uncertain influence of the control system parameters, can realize double-fed wind farm subsynchronous oscillation suppression, and avoids oscillation propagation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power system subsynchronous oscillation suppression, in particular to a subsynchronous oscillation suppression method for a doubly-fed wind farm based on intelligent learning and predictive control, a management device and a storage medium. BACKGROUND

[0002] Under the background of global energy crisis, countries around the world are actively promoting new energy power generation represented by wind power to replace fossil energy, in order to alleviate energy supply crisis and cope with environmental deterioration. China has vast territory and abundant wind energy resources, and large-scale wind farms are concentrated in the northwest and north of China, far away from the load center, and need to use series capacitors to compensate for the voltage and stability problems caused by long-distance power transmission. However, the addition of series capacitors may cause subsynchronous oscillation problems in doubly-fed wind farms. At present, subsynchronous oscillation problems have occurred many times worldwide, seriously threatening the safety and stability of wind farms and grid-connected systems.

[0003] At present, wind farm subsynchronous oscillation suppression strategies can be divided into two categories according to different suppression methods, namely suppression by using additional power electronic devices and suppression by improving wind turbine control strategies. Among them, the additional power electronic device suppression will increase the additional power electronic equipment, leading to higher system complexity and operating cost, while the improvement of wind turbine control strategy is simple and easy to operate, with faster response speed and greater practical significance. Among the wind turbine control strategies, additional damping control is a common and effective control method, and there are many existing methods, such as adding subsynchronous current to the rotor current inner loop, adding damping filter to the rotor side converter, and adding virtual inductance to the rotor side current loop, etc.

[0004] However, the existing methods have weak robustness, and the additional damping control involved lacks a correction link for control parameters, which cannot accurately suppress subsynchronous oscillation when the system parameters change greatly. SUMMARY

[0005] The purpose of the present application is to overcome the shortcomings of the prior art and provide a doubly-fed wind farm subsynchronous oscillation suppression method based on intelligent learning and predictive control, comprising the following steps:

[0006] Step one: according to the established doubly-fed wind turbine rotor side converter outer loop damping control prediction model and inner loop damping control prediction model, determine the control parameter variation range, so that the outer loop damping control prediction model and the inner loop damping control prediction model meet the requirements of doubly-fed wind farm subsynchronous oscillation suppression;

[0007] Step two: construct a damping optimization function approximator based on a robust model predictive control, and establish a doubly-fed wind turbine outer loop damping optimization control model and an inner loop damping optimization control model based on the outer loop damping control prediction model and the inner loop damping control prediction model according to the damping optimization function approximator;

[0008] Step three: based on Q-learning optimization to solve the optimal control law of the outer ring damping optimization control model and the inner ring damping optimization control model of the doubly-fed wind turbine;

[0009] Step four: incorporating the constructed outer ring damping optimization control model and the inner ring damping optimization control model of the doubly-fed wind turbine into the rotor-side converter for inner and outer ring control through the optimal control law, to realize the subsynchronous oscillation suppression of the doubly-fed wind farm.

[0010] Further, the control parameter variation range is determined according to the established outer ring damping control prediction model and the inner ring damping control prediction model of the doubly-fed wind turbine rotor-side converter, comprising:

[0011] Based on the dominant oscillation mode analysis of the doubly-fed wind farm, the outer ring power damping control compensation instruction and the inner ring current damping control compensation instruction are constructed, and the outer ring power damping control compensation instruction is represented as:

[0012]

[0013] In the formula, ΔP and ΔQ are active power damping control compensation instructions and reactive power damping control compensation instructions respectively; K P1 and K P2 are outer ring active power and reactive power proportional regulator adjustment coefficients respectively; K I1 and K I2 are outer ring active power and reactive power integral regulator integral coefficients respectively; p is a differential operator; z1 and z2 are intermediate variables, and the inner ring current damping control compensation instruction is represented as:

[0014]

[0015] In the formula, ΔI rd and ΔI rq are active power inner ring current damping control compensation instructions and reactive power inner ring current damping control compensation instructions respectively; K P3 and K P4 are inner ring current active power and reactive power proportional regulator adjustment coefficients respectively; K I3 and K I4 are inner ring active current and reactive current integral regulator integral coefficients respectively; T1 and T2 are inner ring active current and reactive current time constants respectively; i rd and i rq are inner ring active current and reactive current respectively.

[0016] Further, the doubly-fed wind turbine outer ring damping optimization control model and the inner ring damping optimization control model based on the outer ring damping control prediction model and the inner ring damping control prediction model are established according to the damping optimization function approximator, comprising the following processes:

[0017] Based on the state space expression of the outer loop damping control prediction model and the inner loop damping control prediction model of the rotor side converter of the doubly-fed wind turbine, the outer loop damping optimization control module and the inner loop damping optimization control module of the doubly-fed wind turbine are established, and a robust optimization is used to construct a damping optimization function approximator, and the state space equation of the prediction model is:

[0018]

[0019] In the formula, is a state variable; is a control variable; is an output variable; and is a coefficient matrix, k is a discrete time variable, and the damping optimization function approximator based on robust optimization is expressed as:

[0020]

[0021] In the formula, J N is an objective function, which represents the optimal control amount of the system under the maximum change of the control parameter; N is a time step; r is a prediction step; W and M are symmetric positive definite weighting matrices of state vectors and control vectors; S is an optimal coefficient matrix of the control vector; U max is a control vector boundary.

[0022] Further, the optimal control law based on the Q-learning optimization solving the outer loop damping optimization control model and the inner loop damping optimization control model of the doubly-fed wind turbine includes the following processes:

[0023] The solving model of the outer loop damping optimization control module and the inner loop damping optimization control module of the doubly-fed wind turbine based on Q-learning is expressed as:

[0024] R(s,a)=maxD -1

[0025]

[0026] a=C(k)

[0027] In the formula, R max (s,a) represents the reward value obtained by the intelligent agent after taking action a in state s, N x is the number of elements in the state vector, and the variable β is an adaptive change factor which changes with the value of R max (s,a), and the parameters β max , β min and R max (s,a) are boundary values. For the optimal control law F expressed based on robust optimization solution, the gradient of the objective function is used to determine F, I is a unit vector, and the origin O represents the stable state of the control system. The Q-learning action value function update strategy is expressed as:

[0028]

[0029] In the formula, α is a learning rate used to represent the degree of action value function update, γ is a learning factor of Q-learning, A act is an action set, s' represents a next time state corresponding to state s, a series of invariable ellipsoids and corresponding control laws obtained by robust optimization solution are stored and management of calling is realized based on a simple offline database based on data cubes, and the results of multiple iterations are stored based on multi-dimensional data cube sorting.

[0030] A management device for suppressing subsynchronous oscillation of a doubly-fed wind farm based on predictive control, comprising:

[0031] A control parameter determination module is configured to determine a control parameter variation range based on the established outer loop damping control prediction model and inner loop damping control prediction model of the rotor-side converter of the doubly-fed wind turbine, so that the outer loop damping control prediction model and inner loop damping control prediction model meet the requirement of suppressing subsynchronous oscillation of the doubly-fed wind farm.

[0032] A prediction model optimization module is configured to construct a damping optimization function approximator based on a robust model predictive control, and establish an outer loop damping optimization control model and an inner loop damping optimization control model of the doubly-fed wind turbine based on the outer loop damping control prediction model and the inner loop damping control prediction model of the damping optimization function approximator.

[0033] An optimal control law solving module is configured to solve optimal control laws of the outer loop damping optimization control model and the inner loop damping optimization control model of the doubly-fed wind turbine based on Q-learning optimization.

[0034] A subsynchronous oscillation suppression execution module is configured to incorporate the constructed outer loop damping optimization control model and inner loop damping optimization control model of the doubly-fed wind turbine into the rotor-side converter to perform inner and outer loop control through the optimal control laws, so as to realize suppression of subsynchronous oscillation of the doubly-fed wind farm.

[0035] Preferably, the control parameter determination module comprises a control parameter calculation module and a control module; the control parameter calculation module is configured to determine a control parameter variation range based on the established outer loop damping control prediction model and inner loop damping control prediction model of the rotor-side converter of the doubly-fed wind turbine.

[0036] The control module is configured to control the outer loop damping control prediction model and inner loop damping control prediction model based on the control parameters, so that the outer loop damping control prediction model and inner loop damping control prediction model meet the requirement of suppressing subsynchronous oscillation of the doubly-fed wind farm.

[0037] Preferably, the prediction model optimization module comprises a damping optimization function approximator module and an optimization control model establishment module; the damping optimization function approximator module is used to construct a damping optimization function approximator based on a robust model prediction control; the optimization control model establishment module is used to establish an outer ring damping optimization control model based on an outer ring damping control prediction model and an inner ring damping optimization control model based on an inner ring damping control prediction model according to the established damping optimization function approximator.

[0038] Preferably, the optimal control law solving module comprises a Q learning module, which is used to generate a double-fed wind turbine outer ring damping optimization control model and an inner ring damping optimization control model solving model based on Q learning, and obtain an optimal control law according to the solving model.

[0039] Preferably, the optimal control law solving module comprises a Q learning module, which is used to generate a double-fed wind turbine outer ring damping optimization control model and an inner ring damping optimization control model solving model based on Q learning, and obtain an optimal control law according to the solving model.

[0040] A computer readable recording medium stores computer executable instructions, wherein the computer executable instructions, when executed by a processor, cause the processor to perform the suppression method of any one of claims 1-4.

[0041] The present application has the following advantages: (1) the present application can realize double-fed wind farm subsynchronous oscillation suppression, avoid oscillation propagation, and effectively improve the safety and stability characteristics of the wind farm grid-connected system by constructing a double-fed wind turbine outer ring damping optimization control module and an inner ring damping optimization control module.

[0042] (2) the present application considers the uncertain influence of control system parameters, and effectively improves the optimization calculation convergence speed of the additional damping controller by combining model prediction control with Q learning, thereby guaranteeing the effectiveness and practicality of the additional damping controller. BRIEF DESCRIPTION OF DRAWINGS

[0043] Figure 1 It is a flow chart of the double-fed wind farm subsynchronous oscillation suppression method based on intelligent learning and prediction control.

[0044] Figure 2 It is a schematic diagram of a data cube simple offline database.

[0045] Figure 3 It is a double-fed wind turbine subsynchronous oscillation waveform diagram. DETAILED DESCRIPTION

[0046] The technical solutions of the present application will be further described in detail below with reference to the accompanying drawings, but the protection scope of the present application is not limited to the following description.

[0047] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention; that is, the described embodiments are only a part of the embodiments of the invention, and not all of them. The components of the embodiments of the invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0048] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention. It should be noted that relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0049] Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0050] The features and performance of the present invention will be further described in detail below with reference to embodiments.

[0051] Example 1

[0052] like Figure 1 As shown, a method for suppressing subsynchronous oscillations in a doubly-fed wind farm based on intelligent learning and predictive control includes the following steps:

[0053] Step 1: Establish prediction models for the outer loop damping control and the inner loop damping control of the rotor-side converter of the doubly fed wind turbine, and determine the range of control parameter variations.

[0054] Step 2: Construct a damping optimization function approximator based on robust model predictive control, and establish an outer loop damping optimization control module and an inner loop damping optimization control module for the doubly fed wind turbine;

[0055] Step 3: Based on Q-learning optimization, solve for the optimal control laws of the outer loop damping optimization control module and the inner loop damping optimization control module of the doubly fed wind turbine, and store the optimal control laws based on the data cube;

[0056] Step four: incorporate the constructed outer ring damping optimization control module and inner ring damping optimization control module of the doubly-fed wind turbine into the inner and outer ring control of the rotor side converter, to realize the subsynchronous oscillation suppression of the doubly-fed wind farm.

[0057] The outer ring damping control prediction model and the inner ring damping control prediction model of the doubly-fed wind turbine rotor side converter are established, and the control parameter variation range is determined, including:

[0058] Based on the analysis of the dominant oscillation mode of the doubly-fed wind farm, the outer ring power damping control compensation instruction and the inner ring current damping control compensation instruction are constructed, and the outer ring power damping control compensation instruction is expressed as:

[0059]

[0060] In the formula, ΔP and ΔQ are active power damping control compensation instructions and reactive power damping control compensation instructions respectively; K P1 and K P2 are the adjustment coefficients of the outer ring active power and reactive power proportional adjuster; K I1 and K I2 are the integral coefficients of the outer ring active power and reactive power integral adjuster; p is the differential operator; z1 and z2 are intermediate variables, and the inner ring current damping control compensation instruction is expressed as:

[0061]

[0062] In the formula, ΔI rd and ΔI rq are active power inner ring current damping control compensation instructions and reactive power inner ring current damping control compensation instructions respectively; K P3 and K P4 are the adjustment coefficients of the inner ring current active power and reactive power proportional adjuster; K I3 and K I4 are the integral coefficients of the inner ring active current and reactive current integral adjuster; T1 and T2 are the time constants of the inner ring active current and reactive current respectively; i rd and i rq are the inner ring active current and reactive current.

[0063] The outer ring damping optimization control module and the inner ring damping optimization control module of the doubly-fed wind turbine are constructed based on the robust model prediction control to construct the damping optimization function approximator, including the following processes:

[0064] Based on the state space expression of the outer loop damping control prediction model and the inner loop damping control prediction model of the rotor side converter of the doubly-fed wind turbine, the outer loop damping optimization control module and the inner loop damping optimization control module of the doubly-fed wind turbine are established, and a robust optimization is used to construct a damping optimization function approximator, and the state space equation of the prediction model is:

[0065]

[0066] wherein, is a state variable; is a control variable; is an output variable; and is a coefficient matrix, k is a discrete time variable, and the damping optimization function approximator based on the robust optimization is expressed as:

[0067]

[0068] wherein, J N is an objective function, which represents the optimal control amount of the system under the maximum change of the control parameter; N is a time step; r is a prediction step; W and M are symmetric positive definite weighting matrices of the state vector and the control vector; S is an optimal coefficient matrix of the control vector; U max is a control vector boundary.

[0069] The optimal control law of the outer loop damping optimization control module and the inner loop damping optimization control module of the doubly-fed wind turbine based on the Q-learning optimization is solved, and the optimal control law is stored based on the data cube, including the following processes:

[0070] The solution model of the outer loop damping optimization control module and the inner loop damping optimization control module of the doubly-fed wind turbine based on the Q-learning is expressed as:

[0071] R(s,a)=maxD -1

[0072]

[0073] a=C(k)

[0074] wherein, R max (s,a) represents the reward value obtained by the intelligent agent after taking action a in state s, N x is the number of elements in the state vector, ρ is a probability subject to a normal distribution, parameters μ and σ respectively represent the mathematical expectation and the standard deviation of the search range and the correction step, and the variable β is an adaptive change factor, which changes with the value of R max (s,a), parameters β max , β min and R max (s,a) are boundary values, The gradient of the objective function is used to determine F, I is a unit vector, and the origin O represents the stable state of the control system. The Q-learning action value function update strategy is expressed as:

[0075]

[0076] In the formula, a is a learning rate used to represent the degree of action value function update, γ is a learning factor of Q-learning, A act is an action set, s' represents a next time state corresponding to state s, a series of invariable ellipsoids and corresponding control laws obtained by robust optimization solution are stored and called by using a simple offline database based on a data cube, and the results of multiple iterations are stored based on multi-dimensional data cube sorting.

[0077] Specifically, the specific embodiment is as follows:

[0078] Step 1: Based on the dominant oscillation mode analysis, a rotor-side converter outer loop damping control prediction model and an inner loop damping control prediction model of a doubly-fed wind turbine are established, and the control parameter variation boundary is determined.

[0079] The rotor-side converter outer loop damping control prediction model and the inner loop damping control prediction model of the doubly-fed wind turbine are expressed as: based on the dominant oscillation mode analysis of the doubly-fed wind farm, an outer loop power damping control compensation instruction and an inner loop current damping control compensation instruction are constructed to reduce the negative damping effect of the rotor loop. The outer loop power damping control compensation instruction is expressed as:

[0080]

[0081] In the formula, ΔP and ΔQ are active power damping control compensation instructions and reactive power damping control compensation instructions respectively; K P1 and K P2 are outer loop active power and reactive power proportional regulator adjustment coefficients respectively; K I1 and K I2 are outer loop active power and reactive power integral regulator integral coefficients respectively; p is a differential operator; z1 and z2 are intermediate variables, and the inner loop current damping control compensation instruction is expressed as:

[0082]

[0083] In the formula, ΔI rd and ΔI rq are active power inner loop current damping control compensation instructions and reactive power inner loop current damping control compensation instructions respectively; K P3 and K P4 are inner loop current active power and reactive power proportional regulator adjustment coefficients respectively; K I3 and K I4are integral coefficients of inner loop active current and reactive current integral regulator respectively; T1 and T2 are time constants of inner loop active current and reactive current respectively; i rd and i rq are inner loop active current and reactive current respectively.

[0084] Step 2: Construct a damping optimization function approximator based on robust model predictive control, and establish an outer loop damping optimization control module and an inner loop damping optimization control module of the doubly-fed wind turbine.

[0085] The damping optimization function approximator constructed based on the robust model predictive control is specifically expressed as follows: based on the state space expression of the outer loop damping control prediction model and the inner loop damping control prediction model of the rotor-side converter of the doubly-fed wind turbine, an outer loop damping optimization control module and an inner loop damping optimization control module of the doubly-fed wind turbine are established, and a robust optimization is used to construct the damping optimization function approximator, and the state space equation of the prediction model is as follows:

[0086]

[0087] In the formula, x is a state variable; is a control variable; is an output variable; and are coefficient matrices, k is a discrete time variable, and the damping optimization function approximator based on the robust optimization is expressed as follows:

[0088]

[0089]

[0090] In the formula, J N is an objective function, which represents the optimal control amount of the system under the maximum change of the control parameter; N is a time step; r is a prediction step; W and M are symmetric positive definite weighting matrices of state vectors and control vectors; S is an optimal coefficient matrix of the control vector; U max is a control vector boundary.

[0091] Step 3: based on Q-learning optimization, the optimal control law of the outer loop damping optimization control module and the inner loop damping optimization control module of the doubly-fed wind turbine is solved, and the optimal control law is stored based on data cube.

[0092] Step 3 is specifically expressed as follows: based on Q-learning optimization, the optimal control law of the outer loop damping optimization control module and the inner loop damping optimization control module of the doubly-fed wind turbine is solved, and the optimal control law is stored based on data cube, and the solution model of the outer loop damping optimization control module and the inner loop damping optimization control module of the doubly-fed wind turbine based on Q-learning is expressed as follows:

[0093] R(s,a)=maxD-1 (9)

[0094]

[0095] a=C(k) (16)

[0096] In the formula, R max (s, a) represents the reward value obtained by the intelligent agent after taking action a in state s, N x is the number of elements in the state vector, p is a probability subject to a normal distribution, parameters mu and sigma are mathematical expectation and standard deviation representing a search range and a correction step respectively, and variable beta is an adaptive change factor which changes with the value of R max (s, a), parameters beta max , beta min and R max (s, a) are boundary values, the optimal control law F expressed based on robust optimization solution is represented by the gradient of the objective function, I is a unit vector, and the origin O represents a stable state of the control system, and the Q-learning action value function update strategy is expressed as:

[0097]

[0098] In the formula, alpha is a learning rate used to represent the action value function update degree, gamma is a learning factor of Q-learning, A act is an action set, s' represents a next time state corresponding to state s, and a series of invariable ellipsoids and corresponding control laws obtained by robust optimization solution are stored and managed by calling based on a simple offline database of data cubes, as shown in Figure 6. Figure 2 As shown in Figure 6, the multiple iteration results are stored based on multi-dimensional data cube sorting, so that data can be stored in real time, queried in real time and transmitted in real time.

[0099] Step 4: Based on the above steps, the constructed double-fed wind turbine outer ring damping optimization control module and inner ring damping optimization control module are integrated into the rotor side converter inner and outer ring control to realize double-fed wind farm subsynchronous oscillation suppression.

[0100] The double-fed wind farm subsynchronous oscillation phenomenon through series compensation grid connection is suppressed by the application, and the suppression effect is as shown in Figure 7. Figure 3 The embodiment shows that the double-fed wind farm subsynchronous oscillation suppression method based on intelligent learning and predictive control can effectively suppress the subsynchronous oscillation phenomenon of the double-fed wind farm, effectively avoid oscillation propagation, and improve the stability of the wind farm grid connection system.

[0101] Specifically, the application further provides a management device for double-fed wind farm subsynchronous oscillation suppression based on predictive control, comprising:

[0102] The control parameter determination module is configured to determine a control parameter variation range based on the established outer loop damping control prediction model and the inner loop damping control prediction model of the rotor-side converter of the doubly-fed wind turbine, so that the outer loop damping control prediction model and the inner loop damping control prediction model meet the requirement of subsynchronous oscillation suppression of the doubly-fed wind farm.

[0103] The prediction model optimization module is configured to construct a damping optimization function approximator based on robust model predictive control, and establish an outer loop damping optimization control model and an inner loop damping optimization control model of the doubly-fed wind turbine based on the outer loop damping control prediction model and the inner loop damping control prediction model.

[0104] The optimal control law solving module is configured to solve the optimal control law of the outer loop damping optimization control model and the inner loop damping optimization control model of the doubly-fed wind turbine based on Q-learning optimization.

[0105] The subsynchronous oscillation suppression execution module is configured to incorporate the constructed outer loop damping optimization control model and inner loop damping optimization control model of the doubly-fed wind turbine into the rotor-side converter for inner and outer loop control through the optimal control law, so as to realize subsynchronous oscillation suppression of the doubly-fed wind farm.

[0106] The control parameter determination module comprises a control parameter calculation module and a control module. The control parameter calculation module is configured to determine a control parameter variation range based on the established outer loop damping control prediction model and the inner loop damping control prediction model of the rotor-side converter of the doubly-fed wind turbine.

[0107] The control module is configured to control the outer loop damping control prediction model and the inner loop damping control prediction model based on the control parameter, so that the outer loop damping control prediction model and the inner loop damping control prediction model meet the requirement of subsynchronous oscillation suppression of the doubly-fed wind farm.

[0108] The prediction model optimization module comprises a damping optimization function approximator module and an optimization control model establishment module. The damping optimization function approximator module is configured to construct a damping optimization function approximator based on robust model predictive control. The optimization control model establishment module is configured to establish an outer loop damping optimization control model of the doubly-fed wind turbine based on the outer loop damping control prediction model, and an inner loop damping optimization control model based on the inner loop damping control prediction model.

[0109] The optimal control law solving module comprises a Q-learning module. The Q-learning module is configured to generate a doubly-fed wind turbine outer loop damping optimization control model and an inner loop damping optimization control model solving model based on Q-learning, and obtain the optimal control law based on the solving model.

[0110] The control law storage module stores the control law based on a data cube and realizes control law calling management.

[0111] A computer-readable recording medium stores computer-executable instructions, wherein the computer-executable instructions, when executed by a processor, cause the processor to perform the suppression method of any one of claims 1-5.

[0112] Those skilled in the art will understand that embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) containing computer usable program code. The solutions in the embodiments of the present application can be implemented in various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0113] The present application is described with reference to flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus produce an apparatus that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in the flow or multiple flows and / or blocks Figure 1 The functions specified in the flow or multiple flows and / or blocks

[0114] These computer program instructions can also be stored in a computer-readable memory that can direct the computer or other programmable data processing apparatus to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including instruction apparatus, which implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in the flow or multiple flows and / or blocks Figure 1 The functions specified in the flow or multiple flows and / or blocks

[0115] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, so that a series of operation steps are performed on the computer or other programmable data processing apparatus to produce a computer-implemented process, so that the instructions executed on the computer or other programmable data processing apparatus provide a process for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in the flow or multiple flows and / or blocksFigure 1 the steps of the functions specified in the one or more blocks.

[0116] While the preferred embodiments of the application have been described, additional variations and modifications can be employed, as will be appreciated by those of ordinary skill in the art once advised of the essential inventive concepts. Therefore, the scope of the application should be determined not with reference to the preferred embodiments, but with reference to the appended claims and equivalents thereof.

[0117] It will be clearly understood that, although a number of forms and modes have been disclosed, still many variations and modifications thereof can be made without departing from the spirit and scope of the application. Accordingly, it is intended to include all such variations and modifications as fall within the scope of the application and equivalents thereof.

Claims

1. A method for suppressing subsynchronous oscillations in a doubly-fed wind farm based on intelligent learning and predictive control, characterized in that, Includes the following steps: Step 1: Based on the established outer-loop damping control prediction model and inner-loop damping control prediction model of the rotor-side converter of the doubly-fed wind turbine, determine the range of control parameter variation so that the outer-loop damping control prediction model and inner-loop damping control prediction model meet the requirements for suppressing subsynchronous oscillations in the doubly-fed wind farm. Step 2: Construct a damping optimization function approximator based on robust model predictive control, and establish the outer loop damping optimization control model and inner loop damping optimization control model of the doubly fed wind turbine based on the outer loop damping control prediction model and the inner loop damping control prediction model according to the damping optimization function approximator. Step 3: Based on Q-learning optimization, solve for the optimal control law of the outer loop damping optimization control model and the inner loop damping optimization control model of the doubly-fed wind turbine; Step 4: Integrate the constructed outer-loop damping optimization control model and inner-loop damping optimization control model of the doubly-fed wind turbine into the rotor-side converter and perform inner and outer loop control through the optimal control law to achieve suppression of subsynchronous oscillations in the doubly-fed wind farm.

2. The method for suppressing subsynchronous oscillations in a doubly-fed wind farm based on intelligent learning and predictive control according to claim 1, characterized in that, The determination of the control parameter variation range based on the established outer-loop damping control prediction model and inner-loop damping control prediction model of the doubly-fed wind turbine rotor-side converter includes: Based on the dominant oscillation mode analysis of a doubly-fed induction generator (DFIG) wind farm, an outer-loop power damping control compensation command and an inner-loop current damping control compensation command are constructed. The outer-loop power damping control compensation command is expressed as follows: In the formula, ΔP and ΔQ are the active power damping control compensation command and the reactive power damping control compensation command, respectively; K P1 and K P2 These are the adjustment coefficients for the proportional regulators of the outer loop active power and reactive power, respectively; K I1 and K I2 These are the integral coefficients of the outer loop active and reactive power integral regulators, respectively; p is the differential operator; z1 and z2 are intermediate variables, and the inner loop current damping control compensation command is expressed as: In the formula, ΔI rd and ΔI rq These are the active power inner loop current damping control compensation command and the reactive power inner loop current damping control compensation command, respectively; K P3 and K P4 These are the adjustment coefficients for the active and reactive power proportional regulators of the inner loop current, respectively; K I3 and K I4 These are the integral coefficients of the inner-loop active and reactive current integral regulators, respectively; T1 and T2 are the time constants of the inner-loop active and reactive currents, respectively; i rd andi rq These are the active current and reactive current of the inner loop, respectively.

3. The method for suppressing subsynchronous oscillations in a doubly-fed wind farm based on intelligent learning and predictive control according to claim 1, characterized in that, The process of establishing the doubly-fed wind turbine outer-loop damping optimization control model and inner-loop damping optimization control model based on the damping optimization function approximator includes the following steps: Based on the state-space representation of the outer-loop damping control prediction model and the inner-loop damping control prediction model of the rotor-side converter of the doubly-fed induction generator (DFIG) wind turbine, an outer-loop damping optimization control module and an inner-loop damping optimization control module are established. A robust optimization method is used to construct a damping optimization function approximator. The state-space equation of the prediction model is as follows: In the formula, For state variables; For control variables; For output variables; and Let k be the coefficient matrix and k be the discrete-time variable. The damping optimization function approximator based on robust optimization is expressed as: In the formula, J N Let be the objective function, representing the optimal control input of the system under the condition of maximum change in control parameters; N is the time step; r is the prediction step; W and M are the symmetric positive definite weighting matrices of the state vector and control vector, respectively; S is the optimal coefficient matrix of the control vector; U max This is to control the boundary of the vector.

4. The method for suppressing subsynchronous oscillations in a doubly-fed wind farm based on intelligent learning and predictive control according to claim 1, characterized in that, The optimal control law obtained by solving the external and internal damping optimization control models of the doubly-fed wind turbine based on Q-learning optimization includes the following process: The solution models for the Q-learning-based external loop damping optimization control module and the internal loop damping optimization control module of the doubly-fed wind turbine are expressed as follows: R(s,a)=maxD -1 a=C(k) In the formula, R max (s, a) represents the reward value obtained by the agent after taking action a in state s, N x R is the number of elements in the state vector, and β is the adaptive change factor, which varies with R. max The value of (s, a) varies, and the parameter β... max β min and R max (s, a) are boundary values. The optimal control law F is represented by the robust optimization solution. The gradient of the objective function is used to determine F, I is a unit vector, the origin O represents the steady state of the control system, and the Q-learning action value function update strategy is expressed as follows: In the formula, α is the learning rate used to characterize the degree of action-value function update, γ is the learning factor for Q-learning, and A act It is an action set, where s' represents the next time state corresponding to state s. A series of invariant ellipsoids and corresponding control laws obtained from robust optimization are stored and managed using a simple offline database based on data cubes. The results of multiple iterations are stored in a sorted manner based on multidimensional data cubes.

5. A management device for suppressing subsynchronous oscillations in a doubly-fed wind farm based on predictive control, characterized in that, include: The control parameter determination module is used to determine the range of control parameter variation based on the established outer ring damping control prediction model and inner ring damping control prediction model of the rotor-side converter of the doubly fed wind turbine, so that the outer ring damping control prediction model and inner ring damping control prediction model meet the requirements for suppressing subsynchronous oscillations in the doubly fed wind farm. The predictive model optimization module is used to construct a damping optimization function approximator based on robust model predictive control. Based on the damping optimization function approximator, it establishes an outer-loop damping optimization control model and an inner-loop damping optimization control model for the doubly fed wind turbine based on the outer-loop damping control predictive model and the inner-loop damping control predictive model. The optimal control law solving module is used to solve the optimal control law of the outer loop damping optimal control model and the inner loop damping optimal control model of the doubly fed wind turbine based on Q-learning optimization. The subsynchronous oscillation suppression execution module for doubly-fed wind farms is used to integrate the constructed outer-loop damping optimization control model and inner-loop damping optimization control model of the doubly-fed wind turbine into the rotor-side converter and perform inner and outer loop control through the optimal control law to achieve subsynchronous oscillation suppression of the doubly-fed wind farm.

6. The management device for suppressing subsynchronous oscillations in a doubly-fed wind farm based on predictive control according to claim 5, characterized in that, The control parameter determination module includes a control parameter calculation module and a control module; the control parameter calculation module is used to determine the range of control parameter variation based on the established outer ring damping control prediction model and inner ring damping control prediction model of the rotor-side converter of the doubly fed wind turbine. The control module is used to control the outer loop damping control prediction model and the inner loop damping control prediction model according to the control parameters, so that the outer loop damping control prediction model and the inner loop damping control prediction model meet the requirements for suppressing subsynchronous oscillations in a doubly-fed wind farm.

7. A management device for suppressing subsynchronous oscillations in a doubly-fed wind farm based on predictive control, as described in claim 5, is characterized in that... The prediction model optimization module includes a damping optimization function approximator module and an optimization control model establishment module. The damping optimization function approximator module is used to construct a damping optimization function approximator based on robust model predictive control. The optimization control model establishment module is used to establish an outer-loop damping optimization control model for a doubly-fed wind turbine based on an outer-loop damping control prediction model and an inner-loop damping optimization control model based on an inner-loop damping control prediction model, based on the established damping optimization function approximator.

8. The management device for suppressing subsynchronous oscillations in a doubly-fed wind farm based on predictive control according to claim 5, characterized in that, The optimal control law solving module includes a Q-learning module, which is used to generate solution models for the doubly fed wind turbine outer loop damping optimization control module and inner loop damping optimization control module based on Q-learning, and obtain the optimal control law based on the solution models.

9. A management device for suppressing subsynchronous oscillations in a doubly-fed wind farm based on predictive control, as described in claim 5, is characterized in that... It also includes a control law storage module, which uses a simple offline database based on data cubes for storage and implements control law call management.

10. A computer-readable recording medium storing computer-executable instructions, wherein, The computer-executable instructions, when executed by the processor, cause the processor to perform the suppression method as described in any one of claims 1-4.

Images