CN113890019A - Stability-constrained intermittent power system self-adaptive CGPC-PI control method - Google Patents
Stability-constrained intermittent power system self-adaptive CGPC-PI control method Download PDFInfo
- Publication number
- CN113890019A CN113890019A CN202111138149.2A CN202111138149A CN113890019A CN 113890019 A CN113890019 A CN 113890019A CN 202111138149 A CN202111138149 A CN 202111138149A CN 113890019 A CN113890019 A CN 113890019A
- Authority
- CN
- China
- Prior art keywords
- cgpc
- control
- equation
- stability
- output
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 39
- 230000008859 change Effects 0.000 claims abstract description 9
- 238000012545 processing Methods 0.000 claims abstract description 9
- 230000008569 process Effects 0.000 claims description 14
- 239000011159 matrix material Substances 0.000 claims description 9
- 238000013528 artificial neural network Methods 0.000 claims description 6
- 238000009499 grossing Methods 0.000 claims description 6
- 230000007704 transition Effects 0.000 claims description 6
- 238000004422 calculation algorithm Methods 0.000 claims description 4
- 238000011217 control strategy Methods 0.000 claims description 4
- 230000010354 integration Effects 0.000 claims description 4
- 238000005457 optimization Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 239000005367 kimax Substances 0.000 claims description 3
- 238000012886 linear function Methods 0.000 claims description 3
- 238000004088 simulation Methods 0.000 claims description 3
- 238000012549 training Methods 0.000 claims description 3
- 230000009466 transformation Effects 0.000 claims description 3
- 230000003044 adaptive effect Effects 0.000 claims 9
- 238000005096 rolling process Methods 0.000 claims 1
- 230000033228 biological regulation Effects 0.000 abstract description 3
- 230000035515 penetration Effects 0.000 abstract description 3
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000002803 fossil fuel Substances 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
Landscapes
- Engineering & Computer Science (AREA)
- Power Engineering (AREA)
- Feedback Control In General (AREA)
Abstract
The invention discloses a stability-constrained intermittent power system self-adaptive CGPC-PI control method, which aims to calculate the track of PI control parameters and optimize the future output y, firstly obtains the stable region of a load frequency control LFC system, and then converts the independent variable of a constrained generalized predictive control CGPC into a K in an upper-layer control systemPAnd KiFinally, a shorter time lag and a smaller overshoot are obtained by adding an anti-saturation scheme in a lower control level, thereby further optimizing KpAnd KiThe method has the advantages that the CGPC-PI has better control performance than PI and CGPC, the CGPC-PI has more obvious superiority when processing change disturbance which is large enough to trigger an anti-saturation scheme, and the CGPC-PI can improve the frequency regulation performance of an intermittent power penetration system.
Description
Technical Field
The invention relates to the field of power systems, in particular to a stability-constrained intermittent power system self-adaptive CGPC-PI control method.
Background
The increasing demand for reduction of fossil fuel consumption and development of environmentally friendly energy sources requires a more powerful power system to cope with the attendant intermittent problems. The ever-increasing popularity of renewable energy sources will present the following challenges: (1) inertia will be reduced because more and more electronic inverters separate the power supply from the load and output variations will be more active due to atmospheric volatility. Therefore, maintaining frequency, power angle, and voltage stability is more challenging.
In consideration of the decoupling characteristic of the system, the frequency stability of the system can be researched by using a linear frequency modulation method under the condition of small interference. The load frequency control LFC mainly aims to maintain stability of local frequency and tie line switching power, and may be divided into a centralized load frequency control LFC, a distributed load frequency control LFC, and a distributed load frequency control LFC. The centralized load frequency control LFC measures all system outputs using a single controller and calculates the control variables for all actuators in the system, which is computationally and geographically difficult to implement. Therefore, the distributed load frequency control LFC and the distributed load frequency control LFC are more widely used in large-scale power grids.
With the increasing popularity of renewable energy sources, modern power systems face greater challenges in maintaining frequency stability, which can be studied by load frequency control LFC, but traditional Proportional Integral (PI) controllers cannot meet the increasing robustness requirement.
Disclosure of Invention
In order to overcome the defects and shortcomings in the prior art, the invention provides a stability-constrained intermittent power system self-adaptive CGPC-PI control method.
The method adopts the technical scheme that the method aims at calculating the track of PI control parameters and optimizing future output y, firstly, a stable region of a load frequency control LFC system is obtained, and then, the independent variable of a constraint generalized predictive control CGPC is converted into K in an upper-layer control systemPAnd KiFinally, a shorter time lag and a smaller overshoot are obtained by adding an anti-saturation scheme in a lower control level, thereby further optimizing KpAnd Ki。
In the present invention, the stability region and the stability constraint of the system are determined by the simple structure of the actual implementation device PI, the stability boundary of the system is determined by the time series obtained from historical or experimental data, and three steps are required to achieve the goal:
step 1: given a scalar time seriesThe system can use xt=f(xt-L,xt-2L,...,xt-mL) By a form fit of the expressionThe feedforward neural network approximates any non-linear function with a suitably chosen set of parameters, which is expressed as equation (1) below:
wherein α 0, α 1,. alpha.j, α j and β 0, β 1, 1,. beta.m,qTwo sets of parameters, [ m, q, L ], representing the need for training]The complexity of the fitting process and the accuracy of the results are determined and set to [5, 6, 5 ]];
Step 2: then, the jacobian matrix is calculated according to the formula in step 1, as shown in the following formula (2):
step 2: the maximum lyapunov coefficient is obtained by the following formula (3):
wherein M ═ (length (X) -M L)2/3,And v1Max (eig (Tm' × Tm)), if the maximum lyapunov coefficient is not greater than 0, the system is proved to be stable;
calculating the maximum Lyapunov coefficient of the linear frequency modulation system under different PI control parameters through an exhaustive algorithm to obtain the boundary of a stable domain, wherein the calculation logic is based on equations (1) - (2), starting from Kp ═ 0 and Ki ═ 0, and a time sequence { x } t ═ 1 can be collected through simulationTBy applying a feed-forward neural network to { x } t ═ 1TEquation (1) can be derived, then, from equations (2) and (3), the maximum lyapunov coefficient can be calculated, then Kp and Ki can be iteratively increased until λ is greater than 0, and finally, by separating the regions where (Kp, Ki) is consistent or inconsistent with the stability criterion, a stable region is foundThe boundary of (2).
In the present invention, the upper control structure of the CGPC-PI controller restricts the argument of the CGPC controller for the generalized predictive control to be modified to the control parameter of the PI controller, adds the stability restriction using the result of claim 2, and for the following formula, "(k)" indicates that the value of a certain state variable is obtained at the k-th step. "| k" denotes the kth step in the prediction horizon;
the prediction model, GPC, was designed based on the CARIMA model and is described by the formula:
A(z-1)y(k)=z-dB(z-1)u(k)+C(z-1)ξ(k)/Δ (4),
where u (k) is a manipulated variable, the input variable, y (k) is a process output, ζ (k) represents a white noise sequence, d represents a time delay, Δ is a differential operator, and is expressed as Δ ═ 1-z-1,A(z-1)、B(z-1) And C (z)-1) Expressed by equation (5).
When d is equal to 1, we can perform a difference operation on both sides, and finally obtain equation (6),
Formula (6) shows a fitting formula of the input and output sequences, and on the basis of the fitting formula, a prediction model in an optimization window is established;
and (3) performing multi-step prediction for GPC by using a charpy equation, wherein the expression is formula (7):
wherein Ej, Gj, Fj should satisfy formula (8):
in the invention, the CARIMA model is introduced by using a loss map equation, and the output prediction is obtained through a formula (9):
Y=F1ΔU+F2ΔU(k)+G′Yk+E′ξ (9)
in equation (9), the parameters may be interpreted as follows:
wherein N is1And N respectively represents the predicted view NpInitial and final values of (2), NuRepresents a control range;
the output prediction Y is divided into two parts, F1Δ U represents the variable influence from the future control increment, F2ΔU(k)+G’Yk+ E' ζ represents a fixed contribution from the past, and the increment of the manipulated value directly determines the variables (9) and (10) of the output predictions in the equation.
In the present invention, the scrolling is optimized by controlling the FOV NuThe optimal delta U sequence is found, the cost function can be minimized, and compared with the CGPC controlled by the constraint generalized prediction, two special processes are carried out: first, the cost function and constraints are rewritten to meet continuity and stability requirements, and second, Δ u is converted to PI (K)pAnd Ki) The control parameter of (2);
smoothing of said first reference output trajectory, which should be set to a first order process, as shown in equation (11), ensures a smooth transition of the output y (k) from the original value to the setpoint ω,
in the formula (11), j represents the prediction horizon NpA is a smoothing factor.
In the present invention, the second objective function treats;
from the definition of the reference output trajectory, the cost function can be written as equation (12):
J=min{[Y-Yr]T[Y-Yr]+ΔUTΓ1ΔU+[K(k)-K(k-1)]Tr2[K(k)-K(k-1)]} (12),
wherein Y isr=[yr(k+N1|k);yr(k+N1+1|k);...;yr(k+Np|k)],Γ1And Γ2Are respectively made of gamma1=γ1INu×Nu,Γ2=γ2INu×NuIs represented by two diagonal matrices, where γ1And gamma2Is an adjustable weighting parameter, K is a matrix of the form of equation (13):
The cost function is to track the set output track, reduce the control cost and ensure the continuity of the control parameters;
since practical operating constraints can reduce performance, a set of inequalities is taken as a feasible solution range of CGPC-PI as shown in formula (14):
wherein Δ uminAnd Δ umaxIs a constraint on the incremental change of the control variable, uminAnd umaxIs a constraint controlling the amplitude of the variable, likewise, yminAnd ymaxLimiting the amplitude of the output variable, KminAnd KmaxAre approximate limits obtained by the stable region, respectively denoted as [ Kpmin,Kimin]And [ K ]pmax,Kimax]Avoiding over-compressing the feasible solution range, application e1、e2And e3To ensure that the constraints only affect the control variables in the current step, representing [1, 0]1×Nu、[1,0,...,0]1×NpAnd [1, 1; 0, 0; ...; 0,0]Nu×2;
Subsequently, the previous adjustable variable needs to be converted into k (k) (12) and (14) in the equation.
In the invention, the characteristics of the discrete time PI control strategy can be used for deducing the CGPC-PI controller, and the increment of the PI manipulation value can be expanded as shown in a formula (15):
in the formula (15), i is within [1, 2.. N ]u-1]At represents the time step, e represents the error between the setpoint and the predicted output,
equation (15) needs to be expanded to matrix form to describe the relationship between Δ U (k) and k (k), to resolve the contradiction between predicting output y (k) requiring Δ U (k) and calculating Δ U (k) requiring y (k), a trade-off is being made, and by reviewing the final objective of this step, i.e., finding equivalent transitions for Δ U (k) and k (k), it is assumed that if γ 2 is sufficiently small, Δ U '(k) calculated by GPC is almost equal to Δ U (k) calculated by CGPC-PI, and Δ U' (k) and the output prediction can be calculated in equations (16) and (17), respectively.
Y′(k|k)=F1ΔU′+F2dU(k)+GYk (17),
In the present invention, the equations (15) to (17) are equivalent transformation equations as shown in the following equation (18):
ΔU(k)=E(k)S(k)K(k) (18),
where e (k), s (k), k (k) are matrices corresponding to the form given in equation (19).
In the present invention, J is expressed as a standard quadratic programming form by applying equations (12) to (18), and k (k) is a decision variable.
In equation (20), H and F yield from the following equations:
constraint processing, the constraint quantity in equation (14) may be rewritten into a linear form, the argument is k (k), the incremental change of the control variable may be directly replaced by Δ U (k) ═ e (k) s (k) k (k), and the control variable U (k) may be replaced by U (k) ═ U (k-1) + e (k) s (k) k (k), and by combining equations (9) and (18), the relationship between the output variables y (k) and k (k) is obtained, and the constraint on the control parameter does not need further processing, and in conclusion, the modified constraint is as follows:
the output of the upper control hierarchy, K (k), is computed (29) - (31) by solving a quadratic programming problem composed of equations and only the first column of K (k), namely [ Kp (k | k), Ki (k | k) ], is implemented.
In the invention, the lower control level of the CGPC-PI controller adopts an anti-saturation scheme to shorten time lag and reduce overshoot, as shown in the following formula (22):
as shown in equation (22), Δ u (k) is described using a piecewise function. If u (k-1) exceeds the preset limit um, only a negative error can be added in the integration of the k-th step. Vice versa, if u (k-1) < -um, only positive errors can be added.
The method has the advantages that the CGPC-PI has better control performance than PI and CGPC, the CGPC-PI has more obvious superiority when processing the change disturbance which is large enough to trigger the anti-saturation scheme, and the CGPC-PI can improve the frequency regulation performance of the intermittent power penetration system.
Drawings
FIG. 1 is a flow chart of a stabilization zone determination process of the present invention;
FIG. 2 is a flow chart of the CGPC-PI method of the present invention.
Detailed Description
It should be noted that the embodiments and features of the embodiments can be combined with each other without conflict, and the present application will be further described in detail with reference to the drawings and specific embodiments.
As shown in figure 1, the self-adaptive CGPC-PI control method of the intermittent power system with stability constraint has the overall aim of calculating the track of PI control parameters and optimizing future output y, firstly, a stable region of a load frequency control LFC system is obtained, and then, the independent variable of the constraint generalized predictive control CGPC is converted into K in an upper-layer control systemPAnd KiFinally, a shorter time lag and a smaller overshoot are obtained by adding an anti-saturation scheme in a lower control level, thereby further optimizing KpAnd Ki。
In the present invention, the stability region and the stability constraint of the system are determined by the simple structure of the actual implementation device PI, the stability boundary of the system is determined by the time series obtained from historical or experimental data, and three steps are required to achieve the goal:
step 1: given a scalar time seriesThe system can use xt=f(xt-L,xt-2L,...,xt-mL) Approximating any non-linear function by a feed-forward neural network with a set of suitably selected parameters, as shown in equation (1) below:
wherein α 0, α 1,. alpha.j, α j and β 0, β 1, 1,. beta.m,qTwo sets of parameters, [ m, q, L ], representing the need for training]The complexity of the fitting process and the accuracy of the results are determined and set to [5, 6, 5 ]];
Step 2: then, the jacobian matrix is calculated according to the formula in step 1, as shown in the following formula (2):
step 2: the maximum lyapunov coefficient is obtained by the following formula (3):
wherein M ═ (length (X) -M L)2/3,And v1Max (eig (Tm' × Tm)), if the maximum lyapunov coefficient is not greater than 0, the system proves to be stable:
calculating the maximum Lyapunov coefficient of the linear frequency modulation system under different PI control parameters through an exhaustive algorithm to obtain the boundary of a stable domain, wherein the calculation logic is based on equations (1) - (2), starting from Kp ═ 0 and Ki ═ 0, and a time sequence { x } t ═ 1 can be collected through simulationTBy applying a feed-forward neural network to { x } t ═ 1TEquation (1) can be derived, then, from equations (2) and (3), the maximum lyapunov coefficient can be calculated, then Kp and Ki can be iteratively increased until λ is greater than 0, and finally, the boundary of the stable region is found by separating the regions where (Kp, Ki) is consistent or inconsistent with the stability criterion.
As can be seen from the flow chart in fig. 1, the proposed stability criterion is easy to implement and is highly dependent on the operating parameters of the system compared to the conventional method of adding equal or equal internal constraints, since the stability region changes when the operating conditions are switched, and therefore, in engineering practice, it is preferable to use the conservative stability region result when determining the search boundary of the PI parameter.
In the present invention, the upper control structure of the CGPC-PI controller restricts the argument of the CGPC controller for the generalized predictive control to be modified to the control parameter of the PI controller, adds the stability restriction using the result of claim 2, and for the following formula, "(k)" indicates that the value of a certain state variable is obtained at the k-th step. "| k" denotes the kth step in the prediction horizon;
the prediction model, GPC, was designed based on the CARIMA model and is described by the formula:
A(z-1)y(k)=z-dB(z-1)u(k)+C(z-1)ξ(k)/Δ (4),
where u (k) is a manipulated variable, the input variable, y (k) is a process output, ζ (k) represents a white noise sequence, d represents a time delay, Δ is a differential operator, and is expressed as Δ ═ 1-z-1,A(z-1)、B(z-1) And C (z)-1) Expressed by equation (5).
When d is equal to 1, we can perform a difference operation on both sides, and finally obtain equation (6),
Formula (6) shows a fitting formula of the input and output sequences, and on the basis of the fitting formula, a prediction model in an optimization window is established;
and (3) performing multi-step prediction for GPC by using a charpy equation, wherein the expression is formula (7):
wherein Ej, Gj, Fj should satisfy formula (8):
in the invention, the CARIMA model is introduced by using a loss map equation, and the output prediction is obtained through a formula (9):
Y=F1ΔU+F2ΔU(k)+G′Yk+E′ξ (9)
in equation (9), the parameters may be interpreted as follows:
wherein N is1And N respectively represents the predicted view NpInitial and final values of (2), NuRepresents a control range;
the output prediction Y is divided into two parts, F1Δ U represents the variable influence from the future control increment, F2ΔU(k)+G’Yk+ E' zeta represents a fixed influence from the past, and the increment of the manipulated value directly determines the variables (9) and (1) of the output prediction in the equation0)。
In the present invention, the scrolling is optimized by controlling the FOV NuThe optimal delta U sequence is found, the cost function can be minimized, and compared with the CGPC controlled by the constraint generalized prediction, two special processes are carried out: first, the cost function and constraints are rewritten to meet continuity and stability requirements, and second, Δ u is converted to PI (K)pAnd Ki) The control parameter of (2);
smoothing of said first reference output trajectory, which should be set to a first order process, as shown in equation (11), ensures a smooth transition of the output y (k) from the original value to the setpoint ω,
in the formula (11), j represents the prediction horizon NpA is a smoothing factor.
In the present invention, the second objective function treats;
from the definition of the reference output trajectory, the cost function can be written as equation (12):
J=min{[Y-Yr]T[Y-Yr]+ΔUTΓ1ΔU+[K(k)-K(k-1)]Tr2[K(k)-K(k-1)]} (12),
wherein Y isr=[yr(k+N1|k);yr(k+N1+1|k);...;yr(k+Np|k)],Γ1And Γ2Are respectively made of gamma1=γ1INu×Nu,Γ2=γ2INu×NuIs represented by two diagonal matrices, where γ1And gamma2Is an adjustable weighting parameter, K is a matrix of the form of equation (13):
the cost function is to track the set output track, reduce the control cost and ensure the continuity of the control parameters;
since practical operating constraints can reduce performance, a set of inequalities is taken as a feasible solution range of CGPC-PI as shown in formula (14):
wherein Δ uminAnd Δ umaxIs a constraint on the incremental change of the control variable, uminAnd umaxIs a constraint controlling the amplitude of the variable, likewise, yminAnd ymaxLimiting the amplitude of the output variable, KminAnd KmaxAre approximate limits obtained by the stable region, respectively denoted as [ Kpmin,Kimin]And [ K ]pmax,Kimax]Avoiding over-compressing the feasible solution range, application e1、e2And e3To ensure that the constraints only affect the control variables in the current step, representing [1, 0]1×Nu、[1,0,...,0]1×NpAnd [1, 1; 0, 0; ...; 0,0]Nu×2;
Subsequently, the previous adjustable variable needs to be converted into k (k) (12) and (14) in the equation.
In the invention, the characteristics of the discrete time PI control strategy can be used for deducing the CGPC-PI controller, and the increment of the PI manipulation value can be expanded as shown in a formula (15):
in the formula (15), i is within [1, 2.. N ]u-1]At represents the time step, e represents the error between the setpoint and the predicted output,
equation (15) needs to be expanded to matrix form to describe the relationship between Δ U (k) and k (k), to resolve the contradiction between predicting output y (k) requiring Δ U (k) and calculating Δ U (k) requiring y (k), a trade-off is being made, and by reviewing the final objective of this step, i.e., finding equivalent transitions for Δ U (k) and k (k), it is assumed that if γ 2 is sufficiently small, Δ U '(k) calculated by GPC is almost equal to Δ U (k) calculated by CGPC-PI, and Δ U' (k) and the output prediction can be calculated in equations (16) and (17), respectively.
Y′(k|k)=F1ΔU′+F2dU(k)+GYk (17),
In the present invention, the equations (15) to (17) are equivalent transformation equations as shown in the following equation (18):
ΔU(k)=E(k)S(k)K(k) (18),
where e (k), s (k), k (k) are matrices corresponding to the form given in equation (19).
In the present invention, J is expressed as a standard quadratic programming form by applying equations (12) to (18), and k (k) is a decision variable.
In equation (20), H and F yield from the following equations:
constraint processing, the constraint quantity in equation (14) may be rewritten into a linear form, the argument is k (k), the incremental change of the control variable may be directly replaced by Δ U (k) ═ e (k) s (k) k (k), and the control variable U (k) may be replaced by U (k) ═ U (k-1) + e (k) s (k) k (k), and by combining equations (9) and (18), the relationship between the output variables y (k) and k (k) is obtained, and the constraint on the control parameter does not need further processing, and in conclusion, the modified constraint is as follows:
the output of the upper control hierarchy, K (k), is computed (29) - (31) by solving a quadratic programming problem composed of equations and only the first column of K (k), namely [ Kp (k | k), Ki (k | k) ], is implemented.
Feedback correction, common knowledge, is impossible to establish an absolutely accurate model for a practical system, the uncertainty of the prediction can be caused by the introduction of nonlinearity, time-varying characteristics and interference, and online identification can process a relatively inaccurate model, so that the superiority of the CGPC is ensured.
The PI controller introduces an integrating element into the system, providing a pole-zero that contributes to stability. However, due to integration, the control outputs may also add when faced with errors of the same sign, even into the saturation region, when an inverse error is introduced, a slower response will be obtained since the system must first exit the saturation region, and therefore an anti-saturation scheme is employed to reduce skew and overshoot, as shown in equation (22):
as shown in equation (22), Δ u (k) is described using a piecewise function. If u (k-1) exceeds the preset limit um, only a negative error can be added in the integration of the k-th step. Vice versa, if u (k-1) < -um, only positive errors can be added.
To apply equation (22), two methods may be employed. First, the increment of the GPC control variable may be rewritten in the form of equation (22). Then, the cost function and the constraint condition under different conditions are respectively deduced. Or we can simply modify the lower hierarchy of CGPC-PI to the form of equation (22). This demonstrates one of the advantages of CGPC-PI, which can be used directly with advanced PI control theory without further modification of the GPC algorithm, and thus can achieve control performance by a simpler approach.
In summary, the flow chart of the CGPC-PI control strategy for each sampling step is shown in fig. 2.
The method has the advantages that the CGPC-PI has better control performance than PI and CGPC, the CGPC-PI has more obvious superiority when processing the change disturbance which is large enough to trigger the anti-saturation scheme, and the CGPC-PI can improve the frequency regulation performance of the intermittent power penetration system.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various equivalent changes, modifications, substitutions and alterations can be made herein without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims (10)
1. A stability-constrained intermittent power system self-adaptive CGPC-PI control method is characterized in that the method is used for calculating the track of PI control parameters and optimizing future output y, firstly, a stable region of a load frequency control LFC system is obtained, and then, the independent variable of the constrained generalized predictive control CGPC is converted into K in an upper-layer control systemPAnd KiFinally, K is further optimized by adding an anti-saturation scheme in a lower control level to obtain shorter time lag and smaller overshootpAnd Ki。
2. The intermittent power system adaptive CGPC-PI control method of the stability constraint is characterized in that a stable area of the system, the stability constraint utilizes a simple structure of an actual implementation device PI, a stable boundary of the system is determined by a time sequence obtained by historical or experimental data, and three steps are needed to achieve the aim:
step 1: given a scalar time seriesThe system can use xt=f(xt-L,xt-2L,...,xt-mL) Approximating any non-linear function by a feed-forward neural network with a set of suitably selected parameters, as shown in equation (1) below:
wherein α 0, α 1,. alpha.j, α j and β 0, β 1, 1,. beta.m,qTwo sets of parameters, [ m, q, L ], representing the need for training]The complexity of the fitting process and the accuracy of the results are determined and set to [5, 6, 5 ]];
Step 2: then, the jacobian matrix is calculated according to the formula in step 1, as shown in the following formula (2):
step 2: the maximum lyapunov coefficient is obtained by the following formula (3):
wherein M ═ (length (X) -M L)2/3,And v1Max (eig (Tm' × Tm)), if the maximum lyapunov coefficient is not greater than 0, the system is proved to be stable;
calculating the maximum Lyapunov coefficient of the linear frequency modulation system under different PI control parameters through an exhaustive algorithm to obtain the boundary of a stable domain, wherein the calculation logic is based on equations (1) - (2), starting from Kp ═ 0 and Ki ═ 0, and a time sequence { x } t ═ 1 can be collected through simulationTBy applying a feed-forward neural network to { x } t ═ 1TEquation (1) can be derived, then, from equations (2) and (3), the maximum lyapunov coefficient can be calculated, then Kp and Ki can be iteratively increased until λ is greater than 0, and finally, the boundary of the stable region is found by separating the regions where (Kp, Ki) is consistent or inconsistent with the stability criterion.
3. The stability-constrained intermittent power system adaptive CGPC-PI control method of claim 2, wherein an upper-layer control structure of the CGPC-PI controller restricts an argument of a generalized predictive control CGPC controller to be modified into a control parameter of the PI controller, adds stability constraint using a result of claim 2, and for the following formula, "(k)" indicates that a value of a certain state variable is obtained at a kth step, and "| k" indicates the kth step in a predictive view;
the prediction model, GPC, was designed based on the CARIMA model and is described by the formula:
A(z-1)y(k)=z-dB(z-1)u(k)+C(z-1)ξ(k)/Δ (4),
where u (k) is a manipulated variable, the input variable, y (k) is a process output, ζ (k) represents a white noise sequence, d represents a time delay, Δ is a differential operator, and is expressed as Δ ═ 1-z-1,A(z-1)、B(z-1) And C (z)-1) Expressed by equation (5).
When d is equal to 1, we can perform a difference operation on both sides, and finally obtain equation (6),
Formula (6) shows a fitting formula of the input and output sequences, and on the basis of the fitting formula, a prediction model in an optimization window is established;
and (3) performing multi-step prediction for GPC by using a charpy equation, wherein the expression is formula (7):
wherein Ej, Gj, Fj should satisfy formula (8):
4. the stability-constrained intermittent power system adaptive CGPC-PI control method of claim 3, wherein the CARIMA model is introduced using a charpy equation to obtain an output prediction by equation (9):
Y=F1ΔU+F2ΔU(k)+G′Yk+E′ξ (9)
in equation (9), the parameters may be interpreted as follows:
wherein N is1And N respectively represents the predicted view NpInitial and final values of (2), NuRepresents a control range;
the output prediction Y is divided into two parts, F1Δ U represents the variable influence from the future control increment, F2ΔU(k)+G’Yk+ E' ζ represents a fixed contribution from the past, and the increment of the manipulated value directly determines the variables (9) and (10) of the output predictions in the equation.
5. The stability-constrained intermittent power system adaptive CGPC-PI control method of claim 4, wherein said rolling optimization is performed by controlling FOV NuThe optimal delta U sequence is found, the cost function can be minimized, and compared with the CGPC controlled by the constraint generalized prediction, two special processes are carried out: first, the cost function and constraints are rewritten to meet continuity and stability requirements, and second, Δ u is converted to PI (K)pAnd Ki) The control parameter of (2);
smoothing of said first reference output trajectory, which should be set to a first order process, as shown in equation (11), ensures a smooth transition of the output y (k) from the original value to the setpoint ω,
in the formula (11), j represents the prediction horizon NpA is a smoothing factor.
6. The stability-constrained intermittent power system adaptive CGPC-PI control method of claim 5, wherein the second objective function treats;
from the definition of the reference output trajectory, the cost function can be written as equation (12):
J=min{[Y-Yr]T[Y-Yr]+ΔUTF1ΔU+[K(k)-K(k-1)]TΓ2[K(k)-K(k-1)]} (12),
wherein Y isr=[yr(k+N1|k);yr(k+N1+1|k);...;yr(k+Np|k)],Γ1And Γ2Are respectively made of gamma1=γ1INu×Nu,Γ2=γ2INu×NuIs represented by two diagonal matrices, where γ1And gamma2Is an adjustable weighting parameter, K is a matrix of the form of equation (13):
the cost function is to track the set output track, reduce the control cost and ensure the continuity of the control parameters;
since practical operating constraints can reduce performance, a set of inequalities is taken as a feasible solution range of CGPC-PI as shown in formula (14):
wherein Δ uminAnd Δ umaxIs a constraint on the incremental change of the control variable, uminAnd umaxIs a constraint controlling the amplitude of the variable, likewise, yminAnd ymaxLimiting the amplitude of the output variable, KminAnd KmaxAre approximate limits obtained by the stable region, respectively denoted as [ Kpmin,Kimin]And [ K ]pmax,Kimax]Avoiding over-compressing the feasible solution range, application e1、e2And e3To ensure thatThe constraints only affect the control variables in the current step, representing [1, 0., 0, respectively]1×Nu、[1,0,...,0]1×NpAnd [1, 1; 0, 0; ...; 0,0]Nu×2;
Subsequently, the previous adjustable variable needs to be converted into k (k) (12) and (14) in the equation.
7. A stability-constrained intermittent power system adaptive CGPC-PI control method as claimed in claim 1, wherein the characteristics of the discrete-time PI control strategy can be used to derive a CGPC-PI controller, and the increment of PI manipulated value can be extended as shown in equation (15):
in the formula (15), i ∈ [1, 2 … Nu-1]At represents the time step, e represents the error between the setpoint and the predicted output,
equation (15) needs to be expanded to matrix form to describe the relationship between Δ U (k) and k (k), to resolve the contradiction between predicting output y (k) requiring Δ U (k) and calculating Δ U (k) requiring y (k), a trade-off is being made, and by reviewing the final objective of this step, i.e., finding equivalent transitions for Δ U (k) and k (k), it is assumed that if γ 2 is sufficiently small, Δ U '(k) calculated by GPC is almost equal to Δ U (k) calculated by CGPC-PI, and Δ U' (k) and the output prediction can be calculated in equations (16) and (17), respectively.
Y′(k|k)=F1ΔU′+F2dU(k)+GYk (17)。
8. A stability-constrained intermittent power system adaptive CGPC-PI control method as claimed in claim 7 wherein said equations (15) - (17), equivalent transformation equation is shown as equation (18):
ΔU(k)=E(k)s(k)K(k) (18),
wherein E (k), S (k), k (k) are matrices corresponding to the forms given in equation (19),
9. the stability-constrained intermittent power system adaptive CGPC-PI control method of claim 8, characterized by: said J is represented as a standard quadratic programming form by applying equations (12) - (18), k (k) is a decision variable,
in equation (20), H and F yield from the following equations:
constraint processing, the constraint quantity in equation (14) may be rewritten into a linear form, the argument is k (k), the incremental change of the control variable may be directly replaced by Δ U (k) ═ e (k) s (k) k (k), and the control variable U (k) may be replaced by U (k) ═ U (k-1) + e (k) s (k) k (k), and by combining equations (9) and (18), the relationship between the output variables y (k) and k (k) is obtained, and the constraint on the control parameter does not need further processing, and in conclusion, the modified constraint is as follows:
the output of the upper control hierarchy, K (k), is computed (29) - (31) by solving a quadratic programming problem composed of equations and only the first column of K (k), namely [ Kp (k | k), Ki (k | k) ], is implemented.
10. The stability-constrained intermittent power system adaptive CGPC-PI control method of claim 1, characterized by: the lower control level of the CGPC-PI controller adopts an anti-saturation scheme to shorten the time lag and reduce overshoot, as shown in the following formula (22):
as shown in equation (22), Δ u (k) is described using a piecewise function, which can only add negative errors in the integration at step k if u (k-1) exceeds a preset limit, um, and vice versa, which can only add positive errors if u (k-1) < -um.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111138149.2A CN113890019A (en) | 2021-09-27 | 2021-09-27 | Stability-constrained intermittent power system self-adaptive CGPC-PI control method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111138149.2A CN113890019A (en) | 2021-09-27 | 2021-09-27 | Stability-constrained intermittent power system self-adaptive CGPC-PI control method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113890019A true CN113890019A (en) | 2022-01-04 |
Family
ID=79007051
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111138149.2A Pending CN113890019A (en) | 2021-09-27 | 2021-09-27 | Stability-constrained intermittent power system self-adaptive CGPC-PI control method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113890019A (en) |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111897211A (en) * | 2020-05-31 | 2020-11-06 | 吉林大学 | Piezoelectric ceramic micro-positioning platform trajectory tracking control method considering constraint conditions |
-
2021
- 2021-09-27 CN CN202111138149.2A patent/CN113890019A/en active Pending
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111897211A (en) * | 2020-05-31 | 2020-11-06 | 吉林大学 | Piezoelectric ceramic micro-positioning platform trajectory tracking control method considering constraint conditions |
Non-Patent Citations (2)
Title |
---|
WANYING LIU 等: "An adaptive CGPC based anti-windup PI controller with stability constraints for the intermittent power penetrated system", 《INTERNATIONAL JOURNAL OF ELECTRICAL POWER AND ENERGY SYSTEMS 》, vol. 130, pages 1 - 18 * |
张守艮;: "电力系统供电负荷稳定性优化控制研究", 数字通信世界, no. 11, pages 64 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111459051B (en) | Discrete terminal sliding mode model-free control method with disturbance observer | |
Ellis et al. | Integrating dynamic economic optimization and model predictive control for optimal operation of nonlinear process systems | |
CN110048606B (en) | DC-DC boost converter dynamic sliding mode voltage control method based on interval two-type self-adaptive fuzzy neural network | |
Sun | A Barbalat-like lemma with its application to learning control | |
Wang et al. | Adaptive T–S fuzzy-neural modeling and control for general MIMO unknown nonaffine nonlinear systems using projection update laws | |
Wang et al. | Cascade optimal control for tracking and synchronization of a multimotor driving system | |
Chi et al. | Active disturbance rejection control for nonaffined globally Lipschitz nonlinear discrete-time systems | |
Jiang et al. | Robust adaptive dynamic programming | |
Khettab et al. | Enhanced fractional order indirect fuzzy adaptive synchronization of uncertain fractional chaotic systems based on the variable structure control: robust H∞ design approach | |
Ye et al. | Backstepping design embedded with time-varying command filters | |
Kögel et al. | Cooperative distributed MPC using the alternating direction multiplier method | |
CN113625573B (en) | Fractional order system backstepping sliding mode control method influenced by asymmetric dead zone input | |
Zandi et al. | Voltage control of DC–DC converters through direct control of power switches using reinforcement learning | |
Zandi et al. | Voltage control of a quasi z-source converter under constant power load condition using reinforcement learning | |
CN114879508A (en) | Grinding robot path tracking control method based on model prediction control | |
Chang et al. | Variance and passivity constrained sliding mode fuzzy control for continuous stochastic non-linear systems | |
Zhao et al. | Stabilization of linear systems with time delay and disturbance: A new state prediction approach | |
Chen et al. | Optimal tracking control for unknown nonlinear systems with uncertain input saturation: A dynamic event-triggered ADP algorithm | |
CN111240201B (en) | Disturbance suppression control method | |
CN113890019A (en) | Stability-constrained intermittent power system self-adaptive CGPC-PI control method | |
CN110908286A (en) | Design method of indirect self-adaptive fuzzy optimal synchronous controller of uncertain chaotic system | |
Zhu et al. | Stabilization and optimization of discrete-time Markovian jump linear systems via mode feedback control | |
Cui et al. | Adaptive Horizon Seeking for Generalized Predictive Control via Deep Reinforcement Learning With Application to DC/DC Converters | |
Van Kien et al. | Adaptive MIMO fuzzy controller for double coupled tank system optimizing by jaya algorithm | |
CN110442028B (en) | Fuzzy prediction based anti-bifurcation control method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220104 |
|
RJ01 | Rejection of invention patent application after publication |