Automatic power generation control method for AC / DC interconnected power grid

Abstract

The present invention provides an automatic power generation control method for an AC / DC interconnected power grid based on a two-level hierarchical model predictive control. The method uses a decentralized control method at the lower level to implement model predictive control for multiple regional power grids; and uses a dynamic coordination mode at the upper level to monitor the decentralized model predictive controllers at the lower level. Therefore, the method organically combines the centralized control method with the decentralized control method, switches the control method through the progressive relationship between the upper and lower levels, and has both the global optimality of the centralized control and the scalability of the decentralized control method.

Description

Technical Field

[0001] The present invention relates to the field of automatic power generation control of an electric power system, and in particular to an automatic power generation control method of an AC / DC interconnected power grid based on two-level hierarchical model predictive control. Background Art

[0002] Automatic Generation Control (AGC) is an important function in the energy management system (EMS). It adjusts the output of the generator set in real time according to a certain regulation rate to meet the requirements of power system frequency and tie line power control. With the large-scale access of wind power to the AC / DC interconnected power grid, the traditional AGC control method is difficult to suppress the regional frequency stability and tie line oscillation caused by the volatility of wind power.

[0003] Model Predictive Control (MPC) has been widely used in the field of power system control in recent years. Model Predictive Control is a type of feedback control method. Its mechanism is: at each sampling moment, based on the current measurement information obtained, a finite time domain open-loop optimization problem is solved online, and the first element of the obtained control sequence is applied to the controlled object; at the next sampling moment, the above process is repeated: the optimization problem is refreshed with the new measurement value and solved again. Therefore, the model predictive control method has the advantages of low system model requirements, good control performance, convenient online optimization, and the ability to handle multi-variable constraints. It has been successfully applied to the design of AGC system controllers.

[0004] At present, the controller structures used in model predictive control are mainly centralized model predictive control (CMPC) and decentralized model predictive control (DMPC). Centralized model predictive control is to feed back all the output or state information of the whole system to a central controller, and the controller calculates the control input of the whole system. The advantage of this control method is that the system structure is simple, but the ability in terms of calculation amount and reliability is very limited. Decentralized model predictive control is that the MPC controllers distributed in various regions calculate the corresponding control quantity according to the system state information and ACE signal, and distribute it to the AGC units in the region to realize the decentralized control of the interconnected power grid AGC system. The decentralized model predictive control method can improve the scalability and fault tolerance of the system through multiple controllers, and solve the dispersion of the geographical distribution of subsystems. When the coupling between the subsystems is relatively weak, the controller design is relatively simple; when the coupling between the subsystems is relatively strong, the decentralized control system may not achieve satisfactory control performance, and even the problem of system instability may occur. Therefore, both centralized model predictive control and decentralized model predictive control have certain defects and cannot be applied to the design of AGC system controller for AC / DC interconnected power grids containing wind power. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide an automatic power generation control method for an AC / DC interconnected power grid based on a two-level hierarchical model predictive control. The method uses a decentralized control method at the lower level to implement model predictive control for multiple regional power grids; and uses a dynamic coordination mode at the upper level to monitor the decentralized model predictive controllers at the lower level. Therefore, the method organically combines the centralized control method with the decentralized control method, switches the control method through the progressive relationship between the upper and lower levels, and has both the global optimality of the centralized control and the scalability of the decentralized control method.

[0006] To solve the above technical problems, the present invention adopts the following technical solutions:

[0007] An automatic power generation control method for an AC / DC interconnected power grid based on a two-level hierarchical model predictive control is characterized in that: the lower level decomposes the entire AGC system into a number of independent but interrelated subsystems, and decentralized MPC controllers implement autonomous control to achieve the optimization goals of the subsystems themselves; the upper level implements dynamic coordinated control of the decentralized MPC controllers to achieve centralized global control.

[0008] The specific steps include:

[0009] 1) Establish the state space model of the AGC system in each sub-area;

[0010] 2) Form a two-level hierarchical AGC model predictive control: Based on the state space model of the AGC system in each sub-region, a decentralized AGC model predictive control is established at the upper layer to achieve local autonomy; based on the decentralized AGC model predictive controller established at the lower layer, a centralized AGC model predictive control is established at the upper layer to coordinate the overall situation;

[0011] 3) Control parameter correction and simulation.

[0012] Through repeated simulation, the prediction time domain, control time domain, weighting factor and other related parameters of the upper and lower MPC controllers are corrected; and verified through simulation of scenarios such as wind power fluctuations and load steps.

[0013] The state space model of the sub-region AGC system established in step 1) is the basis of the present invention, wherein the transfer function model of the sub-region AGC system used can be Figure 1 Description, converted into a state space model is

[0014]

[0015] In formula (1), x i is the state variable, u i is the control variable, d i is the disturbance variable; A i , B i 、E i , C i is the corresponding system matrix parameter, and the specific elements are as follows:

[0016] x i =[ΔX ti ,ΔP Rti ,ΔP Gti ,ΔX RHi ,ΔX hi ,ΔP Ghi ,ΔX gi ,ΔX VPi ,ΔX Rgi ,ΔP Ggi ,ΔP tieAC,i ,ΔP tieDC,i ,Δf i ] T (2)

[0017] u i =ΔP ci ; (3)

[0018]

[0019] yi =ACE i ; (5)

[0020]

[0021] in

[0022]

[0023]

[0024] C i =[0,0,0,0,0,0,0,0,0,0,1,1,β i ] T (8)

[0025]

[0026] The step 2) is the core content of the present invention, and the implementation framework is as follows Figure 2 As shown in the figure: at the lower level, decentralized model predictive control is adopted for each sub-region, multiple MPC controllers are dispersed in multiple regional power grids, WAMS transmits the ACE signals of each regional power grid to the decentralized MPC controllers, and all ACE signals are transmitted to the upper-level controller; the centralized MPC controller at the upper level determines whether it needs to act based on all ACE signals: when it needs to act, the control signal ACE is input to the MPC controllers dispersed in each region. ref , control signal ACE when no action is needed ref In the off state.

[0027] The specific steps are as follows:

[0028] Step 1: Form a decentralized AGC model predictive control model: The MPC controllers of each sub-area are based on x i is the state variable, u i is the control variable, d i is the disturbance variable;

[0029] Step 2: The distributed AGC model predictive controller is implemented in three steps: model prediction - rolling optimization - feedback correction;

[0030] First, the discrete state space model is obtained by discretizing equation (1):

[0031]

[0032] In the formula, k is the starting time; T s is the sampling period of the control system.

[0033] Then, based on the discrete state space model (10), it can be predicted that the system starts at y i (k) Model output y in the future i (k+jk), j∈(1,p), p is the prediction time domain. In the control time domain m(m≤p), a quadratic performance index function is constructed:

[0034]

[0035] In the formula, q j 、r j are the output weighting factor and the control weighting factor respectively; r i is the reference trajectory value. When the AGC system is controlled by the ACE signal, its value is all 0.

[0036] Finally, the future control action sequence u is determined by solving the quadratic performance index function i (k+j-1k), j∈(1,m), and then the first control sequence u in the control action sequence i (k) Act on the system and solve the refreshed optimization problem, and the cycle continues.

[0037] Step 3: Determine whether the upper centralized MPC controller needs to act. If necessary, input the control signal ACE to the MPC controllers distributed in each area. ref,i , according to the three steps of model predictive control: model prediction - rolling optimization - feedback correction, the control variable here is ACE ref,i , and the reference trajectory is also the area error signal ACE i , to achieve global coordinated control. If no action is required, the control signal ACE ref In the off state.

[0038] From the above solution steps, it can be seen that: ① The model predictive control algorithm has the advantages of good control performance, strong robustness and low requirements on model accuracy due to the use of control strategies such as multi-step prediction, rolling optimization and feedback correction; ② The two-level hierarchical model predictive control organically combines the centralized control method and the decentralized control method, and switches the control method through the progressive relationship between the upper and lower layers. Therefore, the two-level hierarchical model predictive control has the advantages of both centralized control and decentralized control, and is an advanced control method with excellent control performance and high reliability.

[0039] In summary, the automatic generation control method of AC / DC interconnected power grid based on two-level hierarchical model predictive control can ensure the rapidity and effectiveness of local control because the calculation amount of each decentralized MPC controller in the lower layer of the two-level hierarchical control is not large. In addition, the decomposition of the entire AGC system allows fault separation and risk dispersion, thereby improving the reliability of the entire AGC control system. The dynamic coordination of the centralized MPC controller at the upper layer can eliminate the adjustment action conflicts between decentralized controllers and achieve global monitoring.

[0040] The advantages and positive effects of the present invention are: the automatic power generation control method of the AC / DC interconnected power grid based on the two-level hierarchical model predictive control can not only maintain the frequency and AC / DC interconnection line exchange power fluctuations within a smaller range, but also has a control effect that is significantly better than the CMPC and DMPC control methods; decentralized control and centralized control are achieved by switching between upper and lower layer controllers, and the reliability is very high. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 : Transfer function block diagram of AGC model for AC / DC interconnected power grid area i;

[0042] Figure 2 :Block diagram of two-level hierarchical control of AC / DC interconnected power grid based on model predictive control;

[0043] Figure 3 : Schematic diagram of two-region AC / DC interconnected power grid with multiple power sources;

[0044] Figure 4 : Deviation curve between actual and predicted wind power in region 1;

[0045] Figure 5 : Frequency deviation of regional power grid 1;

[0046] Figure 6 : Frequency deviation of regional power grid 2;

[0047] Figure 7 : AC tie line power deviation;

[0048] Figure 8 : DC tie line power deviation;

[0049] Fig. 9 : Frequency deviation of regional power grid 1;

[0050] Figure: 10: Frequency deviation of regional power grid 2;

[0051] Fig.11 : Frequency deviation of regional power grid 1;

[0052] Fig.12 : Frequency deviation of regional power grid 2;

[0053] in Figure 5-8 It is the dynamic time domain response curve of the AGC system when the wind power fluctuates; Figure 9-10 It is the comparison curve of the frequency deviation of two regions (CMPC and HMPC) when the communication fails; Figure 11-12 It is the comparison curve of the frequency deviation of the two regions (DMPC and HMPC) when the communication fails. DETAILED DESCRIPTION

[0054] The embodiments of the present invention are briefly described below with reference to the accompanying drawings.

[0055] As Figure 3 Taking the simulation study of the AGC system of two-region AC / DC interconnected power grid with multiple power sources as an example, the specific implementation steps of the present invention can be summarized as follows:

[0056] 1) First, the transfer function model of the AGC system is converted into a state space model and discretized to establish the corresponding state space model of the discrete time system;

[0057] 2) Relevant parameter settings of upper and lower MPC controllers:

[0058] Set the discrete point sampling period T s = 0.1s, prediction time domain p = 20, control time domain m = 10, output weighting factor is 1,

[0059] The control weighting factor is 0.1.

[0060] 3) Based on the short-term wind power forecast value, a wind power output deviation curve is given: Here, the embodiment of the present invention assumes that the deviation curve between the actual wind power value and the forecast value of region 1 has been obtained, see the attached Figure 4 ;

[0061] 4) The lower-level distributed MPC controller implements the steps of "model prediction, rolling optimization, and feedback correction", and the upper-level centralized MPC controller integrates all ACE signals to determine whether to act. If necessary, the control signal is input, and if not, the control signal is disconnected;

[0062] 5) According to the simulation scenario, output the simulation results: Dynamic time domain response curve of the AGC system when wind power fluctuates ( Figure 5 ).

[0063] In order to further illustrate the superiority of the method of the present invention, namely high reliability, load step disturbances of 0.1 pu and 0.2 pu are applied to regional power grids 1 and 2 at 20 s and 50 s respectively, and simulation analysis is performed under the assumption that a communication failure occurs in the WAMS system.

[0064] Figure 9-10 It is assumed that the centralized MPC controller cannot control the regional power grid due to the WAMS communication failure, that is, when ΔPc1 = 0, the control of regional power grid 1 is lost, that is, when ΔP c1 =0,ΔP c2 = 0, the control of the two regional power grids is lost. The frequency deviation curves of the two regional power grids in these two cases are shown in Appendix 6. It can be seen from the figure that when the centralized MPC controller fails partially or completely, the two-level hierarchical control method of the present invention can still maintain the frequency stability of the system due to the effect of the lower-level decentralized MPC controller.

[0065] same, Figure 11-12 It is assumed that the WAMS communication failure causes the decentralized MPC controller to be unable to control the regional power grid, that is, when ΔP c1 = 0, the control of regional power grid 1 is lost, that is, when ΔP c1 =0,ΔP c2 = 0, the control of the two regional power grids is lost. The frequency deviation curves of the two regional power grids in these two cases are shown in the appendix. Figure 7 As shown in the figure, it can be seen that when the decentralized MPC controller fails partially or completely, the two-level hierarchical control strategy proposed in this paper is still able to maintain the frequency stability of the system due to the action of the upper centralized MPC controller.

[0066] The above comparison results fully demonstrate that the automatic power generation control method based on the two-level hierarchical model predictive control realizes upper-level centralized control and lower-level decentralized control, and can switch the control mode under different communication failures to maintain the frequency stability of the system, with high reliability.

[0067] The above are only preferred specific implementation modes of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical solutions and inventive concepts of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for automatic power generation control of an AC / DC interconnected power grid, characterized in that: This method is an automatic power generation control method for AC / DC interconnected power grids based on two-level hierarchical model predictive control. In the lower layer, a decentralized control method is used to implement model predictive control for multiple regional power grids. In the upper layer, a dynamic coordination mode is used to monitor the decentralized model predictive controllers in the lower layer. The lower layer decomposes the entire AGC system into several independent but interrelated subsystems, and decentralized MPC controllers implement autonomous control to achieve the optimization goals of the subsystems themselves. The upper layer implements dynamic coordination control on the decentralized MPC controllers to achieve centralized global control. The specific steps include: 1) Establish the state space model of the AGC system in each sub-area; 2) Form a two-level hierarchical AGC model predictive control: Based on the state space model of the AGC system in each sub-region, a decentralized AGC model predictive control is established at the upper layer to achieve local autonomy; based on the decentralized AGC model predictive controller established at the lower layer, a centralized AGC model predictive control is established at the upper layer to coordinate the overall situation; 3) Control parameter correction and simulation: The prediction time domain, control time domain, and weighting factor related parameters of the upper and lower MPC controllers are calibrated through repeated simulations; verified through simulations of wind power fluctuations and load step scenarios; The state space model of the sub-region AGC system used in step 1) is In formula (1), x i is the state variable, u i is the control variable, d i is the disturbance variable; A i , B i 、E i , C i is the corresponding system matrix parameter, and the specific elements are as follows: x i =[ΔX ti ,ΔP Rti ,ΔP Gti ,ΔX RHi ,ΔX hi ,ΔP Ghi ,ΔX gi ,ΔX VPi ,ΔX Rgi ,ΔP Ggi ,ΔP tieAC,i ,ΔP tieDC,i ,Δf i ] T (2) u i =ΔP ci ; (3) and i =ACE i ; (5) in C i ＝[0,0,0,0,0,0,0,0,0,0,1,1,β i ] T (8) 2. The method for automatic power generation control of an AC / DC interconnected power grid according to claim 1, characterized in that: The step 2): at the lower level, a decentralized model predictive control is adopted for each sub-region, multiple MPC controllers are dispersed in multiple regional power grids, WAMS transmits the ACE signals of each regional power grid to the decentralized MPC controllers, and transmits all ACE signals to the upper level controller; the centralized MPC controller at the upper level determines whether to act based on all ACE signals: when it is necessary to act, the control signal ACE is input to the MPC controllers dispersed in each region. ref , control signal ACE when no action is needed ref In the off state.

3. The method for automatic power generation control of an AC / DC interconnected power grid according to claim 2, characterized in that: The specific steps of step 2) are as follows: Step 1: Form a decentralized AGC model predictive control model: The MPC controllers of each sub-area are based on x i is the state variable, u i is the control variable, d i is the disturbance variable; Step 2: The distributed AGC model predictive controller is implemented in three steps: model prediction - rolling optimization - feedback correction; First, the discrete state space model is obtained by discretizing equation (1): In the formula, k is the starting time; T s is the sampling period of the control system; Then, based on the discrete state space model, it can be predicted that the system starts at y i (k) Model output y in the future i (k+j|k), j∈(1,p), p is the prediction time domain; in the control time domain m, by constructing a quadratic performance index function: In the formula, q j 、r j are the output weighting factor and the control weighting factor respectively; r i is the reference trajectory value, and its value is all 0 when the AGC system is the control target of the ACE signal; Finally, the future control action sequence u is determined by solving the quadratic performance index function i (k+j-1|k), j∈(1,m), and then the first control sequence u in the control action sequence i (k) Act on the system and solve the refreshed optimization problem, and the cycle continues.

Images