A wind heat unit constraint optimal model predictive anti-interference control method, system, device and medium

Abstract

The application discloses a wind heat unit constraint optimal model prediction anti-interference control method, system, equipment and medium, and belongs to the technical field of thermal process automatic control, and the method comprises the following steps: real-time estimation of the observation value of the lumped disturbance of the wind heat unit; the future dynamics of the lumped disturbance in the prediction time domain are predicted by using a disturbance predictor; the observation value and the future dynamics of the lumped disturbance are fused into a wind heat unit state space model to solve a steady-state target sequence, and an infinite prediction time domain constraint optimization problem based on the steady-state target sequence and the wind heat unit state is constructed; a maximum controlled allowable set with disturbance estimation compensation is constructed; the infinite prediction time domain constraint optimization problem is converted into an equivalent quadratic programming type constraint optimization problem for solving, and a wind heat unit anti-interference control law is acquired. The application considers the dual control targets of variable constraint and disturbance suppression, and can effectively utilize the feedforward signal of the lumped disturbance in a future period of time, thereby improving the anti-interference performance of the wind heat unit.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of automatic control of thermal processes, and particularly relates to a wind heat unit constraint optimal model predictive anti-interference control method, system, device and medium. BACKGROUND

[0002] The wind heat unit is a kind of high-efficiency and environmentally-friendly heating technology for directly driving a heat pump system by wind energy. Above the rated wind speed range, the output power of the wind turbine can be maintained at a rated value by adjusting the pitch angle of the wind turbine. The control system has multiple sources of unmeasurable lumped disturbances, including external disturbances (wind speed changes) and internal disturbances (such as parameter perturbations). Most of these disturbances are unmeasurable or difficult to measure accurately. In addition, in order to ensure the safety of the system, there are certain constraints on the changes of the pitch angle and its increment. The simultaneous existence of constraints and disturbances greatly increases the control difficulty.

[0003] Meeting the constraints and suppressing the disturbances are mutually contradictory control objectives: in order to suppress the disturbances, the change amount of the pitch angle needs to be increased, while the constraints limit the change of the pitch angle. Currently, the PID control method is usually used to adjust the pitch angle. The PID indirectly makes the pitch angle meet the constraints through parameter tuning, and compensates for the influence of the disturbances by using the output deviation to eliminate the deviation. On the one hand, the PID controller cannot directly handle variable constraints, and on the other hand, using deviation to eliminate deviation is a post-adjustment, which is not timely in dealing with strong disturbances such as wind speed. Therefore, the PID cannot effectively consider both the variable constraints and the anti-interference dual control objectives. SUMMARY

[0004] The application provides a wind heat unit constraint optimal model predictive anti-interference control method, system, device and medium, which can consider both the variable constraints and the disturbance suppression dual control objectives, and can effectively utilize the feedforward signals of the lumped disturbances in a future period of time to improve the anti-interference performance of the wind heat unit.

[0005] The application provides the following technical solutions:

[0006] In a first aspect, a wind heat unit constraint optimal model predictive anti-interference control method is provided, comprising:

[0007] using the established extended state observer to estimate the observation value of the lumped disturbance of the wind heat unit in real time

[0008] according to the observation value of the lumped disturbance using the established disturbance predictor to predict the future dynamics of the lumped disturbance in the prediction time domain

[0009]

[0010] observed value of the lumped disturbance and future dynamics are fused into the wind heat unit state space model to solve the steady-state target sequence (x ss|k+i , u ss|k+i ) and construct an infinite prediction horizon constraint optimization problem based on the steady-state target sequence and the wind heat unit state;

[0011] Combined with the steady-state target sequence and the dual-mode control strategy, a maximum controlled allowable set with disturbance prediction compensation is constructed;

[0012] Based on the maximum controlled allowable set and the steady-state target sequence, the infinite prediction horizon constraint optimization problem is converted into an equivalent quadratic programming type constraint optimization problem for solving, and the wind heat unit anti-interference control law is obtained.

[0013] Optionally, the expression of the extended state observer is:

[0014]

[0015] wherein, and are the derivatives of z1 and z2 respectively; z1 and z2 are the estimated values of the output power y of the wind turbine and the lumped disturbance d respectively, u is the pitch angle of the wind turbine, A m and B m are model parameters of the wind heat unit state space model; L = [β 01 , β 02 ] T is the gain of the extended state observer.

[0016] Optionally, the disturbance predictor established according to the observed value of the lumped disturbance predicts the future dynamics of the lumped disturbance in the prediction horizon, which is specifically:

[0017] According to the current time k and the past observed lumped disturbance sequence, the training sample D = (d in , d out ) is updated;

[0018]

[0019] wherein, N is the sample size, n p is the dimension of the input variable in the training sample;

[0020] The least square parameters b and α i of the disturbance predictor are fitted by using the updated training sample;

[0021]

[0022] Where E = [1, 1, …, 1] T , H is the kernel function matrix;

[0023] Will As the test sample input, the disturbance predictor predicts the predicted value of the lumped disturbance at the next time k+1 for:

[0024]

[0025] Among them, K a is the kernel function;

[0026] The predicted value based on the aggregate disturbance at time k+1 is continuously updated Until the future dynamics of the lumped disturbance in the prediction time domain are obtained

[0027]

[0028] Among them, n a is the predicted length.

[0029] Optionally, the observation value of the lumped disturbance and future developments Fusion into the state space model of the wind thermal unit to conduct steady-state target sequence (x ss|k+i ,u ss|k+i ) and construct an infinite prediction time domain constraint optimization problem based on the steady-state target sequence and the state of the wind-heat unit. The specific process is as follows:

[0030] At the current time k, the lumped disturbance observation value and future developments Fused to contain the steady-state target sequence (x ss|k+i ,u ss|k+i ) in the state space model of the wind-thermal unit;

[0031]

[0032] where x ss|k+i and u ss|k+i are the steady-state target values ​​of the state variables and the control variables at time k+i when considering disturbances; is the predicted value of the lumped disturbance at time k+i, A, B, B d , C are the model parameters of the discretized state space model of the wind thermal unit; y a|k+i It is the steady-state output that the system can achieve under the constraints of the control variables and the steady-state target sequence; the constraints of the control variables are:

[0033]

[0034] where Δu and are the lower and upper limits of the control variable pitch angle increment, u and are the lower and upper limits of the control variable pitch angle, u k+i is the control variable at k+i, Δu k+i is the increment of the control variable at k+i;

[0035] The steady-state target sequence (x a|k+i , u r ) is dynamically solved by minimizing the Euclidean distance between the achievable steady-state output y r and the set value (y ss|k+i , u ss|k+i );

[0036]

[0037] where and are the steady-state target sequences of the state variable and the input variable, Q a and R a are two weight matrices;

[0038] An infinite prediction horizon constraint optimization problem is constructed based on the steady-state target sequence and the state of the wind heat unit;

[0039]

[0040] where x k+i+1 is the state variable at k+i+1, u k+i is the control variable at k+i, Q and R are the weight matrices of and , respectively.

[0041] Optionally, the steady-state target sequence and the dual-mode control strategy are combined to construct a maximum controlled allowable set with disturbance estimation compensation, and the specific process is as follows:

[0042] The infinite prediction horizon is divided into a transient adjustment zone and a gradual convergence zone, and a dual-mode control law is adopted;

[0043]

[0044] where K is the feedback gain, u k+i is the control variable of the wind heat unit at k+i, c k+i is the control degree of freedom at k+i, n c is the step size of the transient adjustment zone, x ss|k+i and u ss|k+i are the steady-state target values of the state variable and the control variable at k+i when the disturbance is considered, respectively.

[0045] Based on the dual-mode control law, the deviation between the state variable of the air heating unit at k+1 time and the steady-state target value is obtained;

[0046]

[0047] Define a variable The autonomous expression of the air heating unit control model about the variable s k is:

[0048]

[0049] Wherein, is the control freedom sequence starting from k time in the prediction time domain, is the steady-state target sequence of the state variable; A and B are both model parameters of the discrete model of the air heating unit state space model; Ω and Φ are both parameters for simplifying the formula;

[0050] Convert the constraint of the control variable into an expression about the autonomous variable s k ;

[0051]

[0052] Wherein,

[0053] According to the constraint expression converted into the autonomous variable s k , the maximum controlled allowable set with disturbance estimation compensation is constructed.

[0054] Optionally, according to the constraint expression converted into the autonomous variable s k , the maximum controlled allowable set with disturbance estimation compensation is constructed, and the specific process is:

[0055] Step a: define n=0,

[0056] Step b: solve the linear programming problem:

[0057]

[0058] Step c: judge whether the s k obtained in step b satisfies the constraint If not, then n=n+1, and return to step b; if yes, then n is large enough, the algorithm stops, and the maximum controlled allowable set is F·s k ≤t, the maximum controlled allowable set is converted into the form about the control freedom :

[0059]

[0060] wherein, N s , M s and V s are respectively and corresponding weights.

[0061] Optionally, the infinite prediction horizon constrained optimization problem is converted into an equivalent quadratic programming type constrained optimization problem based on the maximum controlled allowable set and the steady-state target sequence, and the wind heat unit anti-interference control law is obtained, specifically:

[0062] The infinite prediction horizon constrained optimization problem is converted into an equivalent quadratic programming type constrained optimization problem by simultaneously combining the wind heat unit state model and the maximum controlled allowable set with disturbance estimation compensation:

[0063]

[0064] wherein, is the control freedom sequence at time k, S cs , S xs , S ss and S us are respectively x k , and corresponding weights, x k is the state variable at time k,

[0065] The equivalent quadratic programming type constrained optimization problem is solved by using a quadratic programming solver to obtain the optimal control freedom sequence

[0066] The first element c k of the optimal control freedom sequence is combined with the dual-mode control law to obtain the anti-interference control law u k at the current time k;

[0067] u k = -K(x k -x ss|k ) + u ss|k-1 +c k

[0068] wherein, x k is the state variable at the current time k, K is the feedback gain, x ss|k is the steady-state target value of the state variable at the current time k, and u ss|k-1 is the steady-state target value of the control variable at time k-1.

[0069] In a second aspect, a wind heat unit constrained optimal model predictive anti-interference control system is provided, comprising:

[0070] a lumped disturbance observation module configured to estimate an observation value of a lumped disturbance of the wind heat unit in real time by using the established extended state observer

[0071] a lumped disturbance prediction module configured to predict future dynamics of the lumped disturbance in a prediction time domain according to the observation value of the lumped disturbance by using the established disturbance predictor

[0072] a constrained optimization problem construction module configured to fuse the observation value of the lumped disturbance and the future dynamics into the state space model of the wind heat unit to solve a steady-state target sequence (x ss|k+i , u ss|k+i ) and construct an infinite prediction time domain constrained optimization problem based on the steady-state target sequence and the state of the wind heat unit

[0073] a maximum controlled allowable set construction module configured to construct a maximum controlled allowable set with disturbance prediction compensation in combination with the steady-state target sequence and the dual-mode control strategy

[0074] an anti-interference control law acquisition module configured to convert the infinite prediction time domain constrained optimization problem into an equivalent quadratic programming type constrained optimization problem based on the maximum controlled allowable set and the steady-state target sequence to solve the problem and acquire an anti-interference control law of the wind heat unit

[0075] In a third aspect, a computer device is provided, comprising a processor and a memory; when the processor executes a computer program stored in the memory, the steps of the wind heat unit constrained optimal model predictive anti-interference control method of any one of the first aspect are implemented.

[0076] In a fourth aspect, a computer readable storage medium is provided for storing a computer program; when the computer program is executed by a processor, the steps of the wind heat unit constrained optimal model predictive anti-interference control method of any one of the first aspect are implemented.

[0077] Compared with the prior art, the present application has the following beneficial effects:

[0078] The application provides a wind heat unit constraint optimal model predictive anti-interference control method, system, device and medium. First, an extended state observer and a disturbance estimator are designed based on invariance principle and least square kernel regression thought respectively to obtain a lumped disturbance estimation dynamic on line, and the disturbance estimation dynamic is organically integrated into a model predictive control framework through dynamic solving of a steady state target sequence. Then, equivalent conversion of infinite constraint inequalities to a limited number of inequalities is realized through derivation of an autonomous form of a control object, conversion of original constraints and construction of a maximum controlled allowable set. Finally, the optimal control action at the current time is solved in combination with a dual-mode control strategy and quadratic programming, so that the application can consider two-dimensional control targets of variable constraint and disturbance suppression, organically introduces a lumped disturbance future dynamic to increase feedforward information of the system, can make the control action more timely, and thus improves the anti-interference capability of the controller. BRIEF DESCRIPTION OF DRAWINGS

[0079] Figure 1 is a wind heat unit constraint optimal model predictive anti-interference control method structure diagram of the application.

[0080] Figure 2 is a wind turbine output power control effect diagram given by example 1 of the application. DETAILED DESCRIPTION

[0081] The application will be further described below in combination with the drawings. The following examples are only used to more clearly illustrate the technical solutions of the application, and cannot be used to limit the protection scope of the application. It should be noted that the terms "comprise" and any variations thereof in the specification and claims of the application are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device comprising a series of steps or units does not have to be limited to the clearly listed steps or units, but can include other steps or units that are not clearly listed or inherent to the process, method, product or device.

[0082] Example 1

[0083] As shown in Figure 1 , a wind heat unit constraint optimal model predictive anti-interference control method comprises the following steps:

[0084] Step S1: Real-time estimation of the observation value of the lumped disturbance of the wind heat unit by using the established extended state observer

[0085] Specifically, the extended state observer that can real-time estimate the lumped disturbance of the wind heat unit is designed according to the invariance principle.

[0086] Step S11: Discretization of the state space model (1) of the wind heat unit into formula (2),

[0087]

[0088] where y, u, d are the wind turbine output power, pitch angle and lumped disturbance, in this embodiment, A m and B m are obtained from step response experiments, B dm = 1.

[0089]

[0090] where A = e Ts , x k , u k and y k represent the discretized values of x, u and y, in this embodiment, the sampling time T s = 1 s.

[0091] Step S12: Establish the extended form of the state space model (1)

[0092]

[0093] Step S13: Design the corresponding extended state observer (ESO) according to the extended form (3):

[0094]

[0095] where z1 and z2 are the estimates of y and d, respectively, L = [β 01 , β 02 ] T is the gain of the ESO. In this embodiment, β 01 = 100, β 02 = 2500.

[0096] Step S14: Estimate the lumped disturbance d k using the designed ESO (4) to obtain the observation value at time k

[0097] Step S2: Use the established disturbance predictor to predict the future dynamics of the lumped disturbance in the prediction horizon based on the observation value of the lumped disturbance

[0098] Specifically, a disturbance predictor that can predict the future dynamics of the lumped disturbance online is established based on the least square kernel regression idea.

[0099] Step S21: Update the training sample D = (d in , d out ) according to the current time k and the past observed sequence of the lumped disturbance; ​

[0100] Specifically, step S211: constructing training sample D=(d in ,d out ).

[0101]

[0102] In this embodiment, N is the sample size, n p is the dimension of input variable in training sample; D is initialized as a 0 matrix with N rows and n p columns. Optionally, N=45 and n p =4.

[0103] Step S212: at the current k moment, updating support vector (d in,up ,d out,up ) with perturbation observation value

[0104]

[0105] Step S213: updating training sample: deleting the first row of D and adding the updated support vector (d in,up ,d out,up ) to the last row of D, i.e.

[0106]

[0107] Step S22: fitting the least square parameters b and a i of the perturbation predictor with the updated training sample;

[0108]

[0109] wherein E=[1,1,…,1] T and H is the kernel function matrix;

[0110] Step S23: inputting as test sample, the perturbation predictor predicts the predicted value of aggregate perturbation at the next moment k+1 as

[0111]

[0112] wherein K a is the kernel function;

[0113] Step S24: continuously updating based on the predicted value of aggregate perturbation at the moment k+1 until the future dynamics of aggregate perturbation in the prediction time domain are obtained

[0114] ​​

[0115] where n a is the prediction length.

[0116] Specifically, after obtaining , update According to formula (9), predict In this way, the future dynamics of the lumped disturbance As an option, n a = 1, that is, the future 1-step dynamics of the lumped disturbance is considered.

[0117] Step S3: fuse the observed value of the lumped disturbance and the future dynamics into the state space model of the wind heat unit to solve the steady-state target sequence (x ss|k+i , u ss|k+i ), and construct an infinite prediction time domain constraint optimization problem based on the steady-state target sequence and the state of the wind heat unit.

[0118] Step S31: at the current time k, fuse the observed value of the lumped disturbance and the future dynamics into the state space model of the wind heat unit containing the steady-state target sequence (x ss|k+i , u ss|k+i );

[0119]

[0120] where x ss|k+i and u ss|k+i are the steady-state target values of the state variable and the control variable at k+i considering the disturbance; is the predicted value of the lumped disturbance at k+i, A, B, B d , and C are model parameters of the discrete state space model of the wind heat unit; y a|k+i is the steady-state output that the system can reach under the action of the control variable constraint and the steady-state target sequence; the constraint of the control variable is:

[0121]

[0122] where and are the lower limit and the upper limit of the control variable pitch angle increment, u and are the lower limit and the upper limit of the control variable pitch angle, u k+i is the control variable at k+i, and Δu k+i is the increment of the control variable at k+i; in this embodiment, u=0, (normalized), Δu=-0.2,

[0123] Step S32: Minimize the steady-state output y that the system can achieve a|k+i With the set value (y r ,u r ) to dynamically solve the steady-state target sequence (x ss|k+i ,u ss|k+i );

[0124]

[0125] in, and are the steady-state target sequences of state variables and input variables, Q a and R a are two weight matrices respectively; in this embodiment, the weight matrix Q a =50 and R a =1.

[0126] Step S33: constructing an infinite prediction time domain constrained optimization problem based on the steady-state target sequence and the state of the wind-heat unit;

[0127]

[0128] in, x k+i+1 is the state variable at time k+i+1, u k+i is the control variable at time k+i, Q and R are and In this embodiment, the weight matrix Q=0.1 and R=30.

[0129] Step S4: Combining the steady-state target sequence and the bimodal control strategy, a maximum controlled admissible set with disturbance prediction compensation is constructed.

[0130] Step S41: Divide the infinite prediction time domain into two modes: a transient adjustment region and a gradual convergence region, and adopt a dual-mode control law;

[0131]

[0132] Where K is the feedback gain, u k+i is the control variable of the wind-heating unit at time k+i, c k+i is the control degree of freedom at time k+i, n c is the step size of the transient regulation area, x ss|k+i and u ss|k+i are the steady-state target values ​​of the state variables and the control variables at time k+i when considering disturbances;

[0133] Step S42: Solve the one-step dynamic characteristic of the wind heat unit in deviation form based on the dual-mode control law:

[0134]

[0135] Step S43: Define the variable The autonomous expression of the wind heat unit control model about the variable s k is:

[0136]

[0137] Wherein,

[0138] Wherein, is the control freedom degree sequence starting at time k in the prediction time domain, is the steady-state target sequence of the state variable; A and B are both model parameters after the wind heat unit state space model is discretized; Ω and Φ are both parameters for simplifying the formula;

[0139] Step S44: Convert the constraint of the control variable into an expression about the autonomous variable s k ;

[0140]

[0141] Wherein,

[0142] Step S45: According to the constraint expression converted into the autonomous variable s k , construct the maximum controlled allowable set with disturbance estimation compensation.

[0143] Step S45 further includes the following sub-steps:

[0144] Step a: Define n = 0,

[0145] Step b: Solve the linear programming problem:

[0146]

[0147] Step c: Judge whether the s k obtained in step b satisfies the constraint If not, n = n + 1, and return to step b; if yes, n is large enough, the algorithm stops, and the maximum controlled allowable set is F·s k ≤ t, convert the maximum controlled allowable set into the form about the control freedom degree :

[0148] That is, convert to

[0149] The maximum controlled allowable set is obtained as:

[0150]

[0151] wherein, N s , M s and V s are the corresponding weights.

[0152] Step S5: Based on the maximum controlled allowable set and the steady-state target sequence, the infinite prediction horizon constrained optimization problem is converted into an equivalent quadratic programming type constrained optimization problem for solving, and the anti-interference control law of the wind heat unit is obtained.

[0153] Step S51: The state model of the wind heat unit and the maximum controlled allowable set with disturbance estimation compensation are solved, and the infinite prediction horizon constrained optimization problem is converted into an equivalent quadratic programming type constrained optimization problem:

[0154]

[0155] wherein, is the control freedom sequence at time k, S cs , S xs , S ss and S us are the corresponding weights, x k , and x k is the state variable at time k,

[0156] Step S52: The quadratic programming solver is used to solve the equivalent quadratic programming type constrained optimization problem, and the optimal control freedom sequence

[0157] Step S53: The first element c k of the optimal control freedom sequence is combined with the dual-mode control law to obtain the anti-interference control law u k at the current time k.

[0158] u k = -K(x k -x ss|k )+u ss|k-1 +c k (22)

[0159] wherein, x k is the state variable at the current time k, K is the feedback gain, and x ss|k ​​is a steady-state target value of the state variable at the current time k, u ss|k-1 is a steady-state target value of the control variable at the time k-1.

[0160] In the embodiment, the external disturbance added is a wind speed disturbance: in the simulation 0-1000s, the wind speed is 12m / s, 1001s-2000s, the wind speed is stepped down to 8m / s, 2001s-3000s, the wind speed is stepped up to 12m / s. The results are shown in the attached Figure 2 .

[0161] It can be seen from the attached Figure 2 that after the feedforward compensation considering the future dynamics of the disturbance, the output power of the wind turbine can still be quickly stabilized at the set value after a small fluctuation, which shows that the wind turbine unit can be effectively controlled by the application.

[0162] Embodiment 2

[0163] A wind turbine unit constraint optimal model predictive anti-interference control system, comprising:

[0164] A lumped disturbance observation module for using the established extended state observer to estimate the observation value of the lumped disturbance of the wind turbine unit in real time

[0165] A lumped disturbance prediction module for predicting the future dynamics of the lumped disturbance in the prediction time domain according to the observation value of the lumped disturbance using the established disturbance predictor

[0166] A constraint optimization problem construction module for fusing the observation value of the lumped disturbance and the future dynamics into the wind turbine unit state space model to solve the steady-state target sequence (x ss|k+i , u ss|k+i ) and construct an infinite prediction time domain constraint optimization problem based on the steady-state target sequence and the state of the wind turbine unit;

[0167] A maximum controlled allowable set construction module for constructing a maximum controlled allowable set with disturbance prediction compensation in combination with the steady-state target sequence and the dual-mode control strategy;

[0168] An anti-interference control law acquisition module for converting the infinite prediction time domain constraint optimization problem into an equivalent quadratic programming type constraint optimization problem based on the maximum controlled allowable set and the steady-state target sequence to solve the wind turbine unit anti-interference control law.

[0169] The more specific process of the above method can refer to the corresponding content disclosed in the foregoing embodiments, which will not be repeated here.

[0170] Embodiment 3

[0171] The application provides a computer device, comprising a processor and a memory; wherein the processor implements the steps of the wind heat unit constraint optimal model predictive anti-interference control method when executing the computer program stored in the memory.

[0172] The more specific process of the method can refer to the corresponding content disclosed in the foregoing embodiments, and will not be described here.

[0173] Embodiment 4

[0174] The application provides a computer readable storage medium for storing a computer program; the computer program is executed by a processor to implement the steps of the wind heat unit constraint optimal model predictive anti-interference control method.

[0175] The more specific process of the method can refer to the corresponding content disclosed in the foregoing embodiments, and will not be described here.

[0176] The embodiments in the specification are described in a progressive manner, and each embodiment focuses on the difference from other embodiments. The same or similar parts of each embodiment can be referred to each other. For the system, device and storage medium disclosed by the embodiments, since it corresponds to the method disclosed by the embodiments, the description is relatively simple, and the related part can be referred to the method part.

[0177] Those skilled in the art can clearly understand that the technology in the embodiments of the application can be realized by means of software and necessary general hardware platforms. Based on such understanding, the technical solutions in the embodiments of the application can be embodied in the form of a software product, which can be stored in a storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the method described in the embodiments of the application or some parts of the embodiments.

[0178] The above is only the preferred embodiment of the application, and the protection scope of the application is not limited to the above-mentioned embodiments. Any technical solution falling within the idea of the application shall fall within the protection scope of the application. It should be noted that, for ordinary skilled in the art, some improvements and refinements without departing from the principle of the application shall be considered as the protection scope of the application.

Claims

1. A method for predicting anti-interference control of a wind-heat unit with a constrained optimal model, characterized in that: include: The established extended state observer is used to estimate the observed value of the lumped disturbance of the wind-heat unit in real time According to the observation value of the lumped disturbance The future dynamics of the lumped disturbance in the prediction domain are predicted using the established disturbance predictor Specifically, the disturbance predictor predicts the predicted value of the aggregate disturbance at the next time k+1 for: Among them, K a is the kernel function; is the test sample, n p is the dimension of the input variable in the training sample; k is the current moment; N is the sample size, b and α i are the two least square parameters of the disturbance predictor; d in,i is the aggregate disturbance sequence observed in the past; The predicted value based on the aggregate disturbance at time k+1 is continuously updated Until the future dynamics of the lumped disturbance in the prediction time domain are obtained Among them, n a is the predicted length; The observation value of the lumped disturbance and future developments Fusion into the state space model of the wind thermal unit to conduct steady-state target sequence (x ss|k+i ,u ss|k+i ) and construct an infinite prediction time domain constrained optimization problem based on the steady-state target sequence and the state of the wind-heat unit. The specific constrained optimization problem is: in, x k+i+1 is the state variable at time k+i+1, u k+i is the control variable at time k+i, Q and R are and The weight matrix of Δu and are the lower and upper limits of the control variable pitch angle increment, u and are the lower and upper limits of the variable pitch angle, u k+i is the control variable at time k+i, Δu k+i is the increment of the control variable at time k+i; x ss|k+i and u ss|k+i are the steady-state target values ​​of the state variables and the control variables at time k+i when considering disturbances; Combining the steady-state target sequence and the dual-mode control strategy, the maximum controlled admissible set with disturbance prediction compensation is constructed; the maximum controlled admissible set is about the control degrees of freedom The form is: in, N s 、M s and V s They are and The corresponding weight; is the control degree of freedom sequence at time k; Based on the maximum controlled admissible set and the steady-state target sequence, the infinite prediction time domain constrained optimization problem is transformed into an equivalent quadratic programming constrained optimization problem and solved to obtain the anti-interference control law of the wind-heat unit; specifically: By combining the state model of the wind-thermal unit and the maximum controllable admissible set with disturbance estimation compensation, the infinite prediction horizon constrained optimization problem is transformed into an equivalent quadratic programming constrained optimization problem: Among them, S cs 、S xs 、S ss and S us They are x k 、 and The corresponding weight, x k is the state variable at time k, The quadratic programming solver is used to solve the equivalent quadratic programming constrained optimization problem to obtain the optimal control degree of freedom sequence. The first element c of the optimal control degree of freedom sequence is k Combine the dual-mode control law to obtain the anti-interference control law u at the current time k k ; u k =-K(x k -x ss|k )+u ss|k-1 +c k Among them, x k is the state variable at the current moment k, K is the feedback gain, x ss|k is the steady-state target value of the state variable at the current moment k, u ss|k-1 is the steady-state target value of the control variable at time k-1.

2. The wind-heat unit constraint optimal model prediction anti-interference control method according to claim 1 is characterized in that: The expression of the extended state observer is: in, and are the derivatives of z1 and z2 respectively; z1 and z2 are the estimated values ​​of wind turbine output power y and lumped disturbance d respectively, u is the pitch angle of wind turbine, A m and B m are all model parameters of the state space model of the wind thermal unit; L = [β 01 ,β 02 ] T is the gain of the extended state observer.

3. The wind-heat unit constraint optimal model prediction anti-interference control method according to claim 1 is characterized in that: The least squares parameters b and α of the disturbance predictor are i Obtained through the following methods: Specifically: According to the current time k and the aggregate disturbance sequence observed in the past, the training sample D = (d in ,d out ); Where N is the sample size, n p The dimension of the input variables in the training sample; Using the updated training samples, fit the least squares parameters b and α of the disturbance predictor i ; Where E = [1, 1, …, 1] T , H is the kernel function matrix.

4. The wind-heat unit constraint optimal model prediction anti-interference control method according to claim 1 is characterized in that: The observed value of the lumped disturbance and future developments Fusion into the state space model of the wind thermal unit to conduct steady-state target sequence (x ss|k+i ,u ss|k+i ), the specific process is: At the current time k, the lumped disturbance observation value and future developments Fused to contain the steady-state target sequence (x ss|k+i ,u ss|k+i ) in the state space model of the wind-thermal unit; in, is the predicted value of the lumped disturbance at time k+i, A, B, B d , C are the model parameters of the discretized state space model of the wind thermal unit; y a|k+i It is the steady-state output that the system can achieve under the constraints of the control variables and the steady-state target sequence; By minimizing the steady-state output y that the system can achieve a|k+i With the set value (y r ,u r ) to dynamically solve the steady-state target sequence (x ss|k+i ,u ss|k+i ); in, and are the steady-state target sequences of state variables and input variables, Q a and R a are two weight matrices respectively.

5. The wind-heat unit constraint optimal model prediction anti-interference control method according to claim 1 is characterized in that: The above-mentioned combination of steady-state target sequence and dual-mode control strategy constructs the maximum controlled admissible set with disturbance prediction compensation. The specific process is as follows: The infinite prediction time domain is divided into two modes: transient regulation region and gradual convergence region, and a dual-mode control law is adopted; Where K is the feedback gain, u k+i is the control variable of the wind-heating unit at time k+i, c k+i is the control degree of freedom at time k+i, n c is the step size of the transient regulation area, x ss|k+i and u ss|k+i are the steady-state target values ​​of the state variables and the control variables at time k+i when considering disturbances; Based on the dual-mode control law, the deviation between the state variables of the wind-heating unit and the steady-state target value at time k+1 is obtained; Defining variables Control model of wind-heat unit with respect to variable s k The autonomous expression of is: in, is the control degree of freedom sequence starting at time k in the prediction domain, is the steady-state target sequence of state variables; A and B are the model parameters after the discretization of the state space model of the wind-heat unit; Ω and Φ are parameters used to simplify the formula; Transform the constraints on the control variables into constraints on the autonomous variables s k Expressions of in, K s =[K,[-1,0,...,0],0], According to the transformation into the autonomous variable s k The constraint expression of is used to construct the maximum controlled admissible set with disturbance prediction compensation.

6. The wind-heat unit constraint optimal model prediction anti-interference control method according to claim 5, characterized in that: The basis is transformed into the autonomous variable s k The constraint expression of the maximum controlled admissible set with disturbance prediction compensation is constructed. The specific process is: Step a: define n=0, Step b: Solve the linear programming problem: Step c: Determine the s obtained in step b k Whether the constraints are met If not, then n=n+1, and return to step b; if so, n is large enough, the algorithm stops, and the maximum controlled admissible set is F·s k ≤t, transform the maximum controlled admissible set into the control degrees of freedom form.

7. A wind-heating unit constraint optimal model prediction anti-interference control system, based on the wind-heating unit constraint optimal model prediction anti-interference control method according to any one of claims 1 to 6, characterized in that: include: The lumped disturbance observation module is used to estimate the observed value of the lumped disturbance of the wind-heat unit in real time using the established extended state observer The lumped disturbance prediction module is used to predict the observed value of the lumped disturbance The future dynamics of the lumped disturbance in the prediction domain are predicted using the established disturbance predictor Constrained optimization problem building block for transforming the observations of lumped perturbations into and future developments Fusion into the state space model of the wind thermal unit to conduct steady-state target sequence (x ss|k+i ,u ss|k+i ) and construct an infinite prediction time domain constrained optimization problem based on the steady-state target sequence and the state of the wind-thermal unit; The maximum controlled admissible set construction module is used to combine the steady-state target sequence and the bimodal control strategy to construct the maximum controlled admissible set with disturbance estimation compensation; The anti-interference control law acquisition module is used to transform the infinite prediction time domain constrained optimization problem into an equivalent quadratic programming constrained optimization problem based on the maximum controlled admissible set and the steady-state target sequence, and obtain the anti-interference control law of the wind-heat unit.

8. A computer device, characterized in that: It comprises a processor and a memory; wherein, when the processor executes the computer program stored in the memory, the steps of the wind-heat unit constraint optimal model prediction anti-interference control method described in any one of claims 1-6 are implemented.

9. A computer-readable storage medium, characterized in that Used to store computer programs; when the computer programs are executed by the processor, the steps of the wind-heat unit constraint optimal model prediction anti-interference control method described in any one of claims 1-6 are implemented.

Images