Networking system parameter optimization method based on frost ice optimization model predictive control

Abstract

The invention discloses a network construction system parameter optimization method based on frost ice optimization model prediction control. The method comprises the following steps: establishing a first state equation of a power grid region based on a pre-acquired network construction data set; simplifying the first state equation into a second state equation based on a pre-acquired state-space equation; determining a target function according to the network construction data set, an initial weight matrix of key parameters in a preset network construction data set and the second state equation; and iteratively updating the initial weight matrix of the key parameters according to a model prediction control algorithm, a frost ice optimization algorithm and the target function to obtain an optimal weight matrix. According to the technology, under the condition of sudden load change, the stability and the response capability of frequency control among the power grid regions are improved.

Description

Technical Field

[0001] This invention relates to the field of power grid control technology, and in particular to a method for optimizing the parameters of a power grid system based on frost and ice optimization model predictive control. Background Technology

[0002] When there are sudden changes in the load in the power grid area, the circuit is prone to instability. Specifically, the shortcomings of the existing technology are as follows: 1. Most existing studies focus on frequency control or output prediction, without systematically incorporating low-frequency oscillation identification and regulation into the load frequency control framework, resulting in insufficient protection of system stability.

[0003] 2. Traditional load frequency control systems use multi-level PI or distributed controllers, which are complex in structure, difficult to adjust, and have slow response, making it difficult to meet the requirements of high dynamic performance.

[0004] 3. Most load frequency control parameters are set empirically or tuned offline, lacking real-time dynamic adaptive capability, resulting in large prediction errors and an inability to track rapid fluctuations.

[0005] 4. Most existing methods are insufficient in modeling and coupling analysis of GFM converters, and cannot identify or mitigate the oscillation gain path caused by coupling.

[0006] Therefore, ensuring a smooth transition for the power grid region has become an urgent technical problem to be solved. Summary of the Invention

[0007] The purpose of this invention is to provide a method for optimizing the parameters of a grid system based on a frost-ice optimization model predictive control, so as to ensure a smooth transition of the grid area under the condition of sudden load changes in the grid area.

[0008] In a first aspect, embodiments of the present invention provide a method for optimizing parameters of a power grid system based on predictive control using a frost-ice optimization model. The power grid system includes multiple power grid regions. For each power grid region, the method includes: S102: establishing a first state equation for the power grid region based on a pre-acquired power grid data set; S104: simplifying the first state equation into a second state equation based on a pre-acquired state space equation; S106: optimizing the power grid system based on the power grid data set and an initial weight matrix of key parameters in the pre-set power grid data set. The objective function is determined by the second state equation; S108: The initial weight matrix of the key parameters is iteratively updated according to the model predictive control algorithm, the frost optimization algorithm, and the objective function. The optimal weight matrix is ​​obtained. .

[0009] Furthermore, the network data set includes: key parameters and the output power of the diesel generator. and the valve drive amount of the diesel generator The network data set also includes: the output power of the hydrogen fuel cell. Energy storage battery output power One of them; the key parameters include controllable power supply parameters, uncontrollable power supply parameters, and other key parameters.

[0010] Furthermore, the controllable power parameters include: diesel generator output. The controllable power parameters also include the output power of the hydrogen fuel cell. Energy storage battery output One of the uncontrollable power parameters is: fan output power. Photovoltaic output power One of the key parameters mentioned above is the load demand of the power grid area. Frequency change of the power grid area per unit time Frequency variation between power grid regions Power exchange between power grid areas ,in Representing different power grid regions.

[0011] Furthermore, the initial weight matrix of the key parameters include: Where i represents the number of power grid regions and j represents the number of key parameters.

[0012] Furthermore, the second state equation includes: the state variable matrix of the power grid area, the control variable matrix, the disturbance variable matrix, and the controlled output matrix.

[0013] Further, S106 includes: S1062: initial weighting matrix of the key parameters for each power grid region. The system is divided into several weighted matrices; S1064: The objective function is determined based on the second state equation of all power grid regions, the weighted matrices, the pre-acquired prediction time-domain function, and the pre-acquired control time-domain function.

[0014] Further, S108 includes: S1082: an initial weight matrix for the key parameters based on the objective function, a preset upper boundary of model prediction control, and a preset lower boundary of model prediction control. Perform iterative updates to determine the MPC weight matrix. S1084: Based on the objective function and the frost optimization algorithm, adjust the MPC weight matrix. Perform iterative updates to determine the frost weight matrix. S1086: If the current iteration count is greater than the preset iteration threshold, then the frost weight matrix will be adjusted. As the optimal weight matrix .

[0015] Furthermore, S1082 includes: an initial weight matrix for key parameters based on a preset upper boundary of model predictive control and a preset lower boundary of model predictive control. Perform iterative updates to determine the first weight matrix. The initial weight matrix is ​​determined based on S106. objective function value and the first weight matrix objective function value If the preset probability parameters Less than the first weight matrix objective function value And the first weight matrix objective function value Less than or equal to the initial weight matrix objective function value Then the first weight matrix As the MPC weight matrix Otherwise, do not update the weight matrix, and use the initial weight matrix. As the MPC weight matrix .

[0016] Furthermore, S1084 includes: an MPC weight matrix for the key parameters based on preset iterative random parameters, a preset iterative threshold G, and the current iteration number g. Perform iterative updates to determine the second weight matrix. Determine the second weight matrix based on S106. objective function value If the preset probability parameters Less than the second weight matrix objective function value And the second weight matrix objective function value Greater than the MPC weight matrix objective function value Then the second weight matrix As the frost weight matrix Otherwise, do not update the weight matrix, and change the MPC weight matrix. As the frost weight matrix .

[0017] Secondly, embodiments of the present invention provide a network system parameter optimization device based on frost and ice optimization model predictive control. The device includes: a first optimization module, used to establish a first state equation for the power grid region based on a pre-acquired network data set; a second optimization module, used to simplify the first state equation into a second state equation based on a pre-acquired state space equation; and a third optimization module, used to perform optimization based on the network data set and an initial weight matrix of key parameters in the pre-set network data set. The objective function is determined by the second state equation; the fourth optimization module is used to iteratively update the initial weight matrix of the key parameters based on the model predictive control algorithm, the frost optimization algorithm, and the objective function. The optimal weight matrix is ​​obtained. .

[0018] The beneficial effects of the embodiments of the present invention are as follows: This invention discloses a method for optimizing grid system parameters based on frost-free optimization model predictive control, comprising: establishing a first state equation for the power grid region based on pre-acquired grid data sets; simplifying the first state equation into a second state equation based on pre-acquired state-space equations; determining an objective function based on the grid data sets, the initial weight matrices of key parameters in the pre-set grid data sets, and the second state equation; and iteratively updating the initial weight matrices of the key parameters according to the model predictive control algorithm, the frost-free optimization algorithm, and the objective function to obtain the optimal weight matrix. This technology improves the stability and response capability of frequency control between different power grid regions under load abrupt changes. Attached Figure Description

[0019] Figure 1 A flowchart of a method for optimizing network system parameters based on frost-ice optimization model predictive control is provided for the implementation of this invention. Figure 2 A schematic diagram of a network construction system provided for the implementation of this invention; Figure 3 A flowchart of another method for optimizing network system parameters based on frost-ice optimization model predictive control, provided for the implementation of this invention; Figure 4 A schematic diagram of frequency variation in the first power grid region (i.e., region one) provided for the implementation of the present invention; Figure 5 A schematic diagram of frequency variation in the second power grid region (i.e., region two) provided for the implementation of this invention; Figure 6 A schematic diagram of the change in tie-line switching power provided for the implementation of this invention; Figure 7 A schematic diagram of a network system parameter optimization device based on frost and ice optimization model predictive control provided for the implementation of this invention. Detailed Implementation

[0020] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0021] This technology is applicable to advanced power system dispatching and control scenarios, including grid-connected energy storage systems, grid-connected inverter control systems, and distributed energy regulation platforms, and has broad engineering application value.

[0022] This application aims to improve the stability and responsiveness of frequency control in power grid systems, particularly addressing the low-frequency oscillation problem caused by weak grid conditions and grid-forming (GFM) converter coupling. By employing Model Predictive Control (MPC) and the Rime Optimization Algorithm (RIME), a low-complexity, highly responsive, and adaptable load frequency control strategy is proposed. The key technical issues to be addressed are as follows: 1. Under weak grid conditions, the mutual coupling between GFM converters can easily induce low-frequency oscillations, affecting frequency dynamic performance and system stability margin. It is necessary to design a control strategy with predictive and feedforward capabilities to suppress these oscillations.

[0023] 2. Current multi-level control architectures are complex and cumbersome to tune, with a large number of controllers and strong coupling, which limits the system's response speed and practicality.

[0024] 3. Distributed energy sources such as wind and solar power are volatile and uncertain, causing frequent disturbances to the system. A smart control mechanism that is more sensitive to power fluctuations and can adjust more quickly is needed.

[0025] 4. The performance of the load frequency control system is highly dependent on the controller parameter settings. Traditional parameter adjustment methods rely on manual experience or static optimization, which is difficult to adapt to dynamic operating environments. RIME is introduced to adjust the MPC weight parameters online, thereby enhancing robustness and adaptability.

[0026] Example 1 like Figures 1-6 As shown in the figure, this embodiment illustrates a method for optimizing network system parameters based on frost and ice optimization model predictive control.

[0027] The grid system of this application includes multiple power grid areas, such as Figure 2 As shown, in this embodiment, the power grid area includes two parts: the upper part is the first power grid area and the lower part is the second power grid area. Each power grid area contains both controllable power sources, such as diesel generators, batteries and fuel cells, and uncontrollable power sources, such as wind turbines and photovoltaics.

[0028] For each power grid region, this method includes: S102: Establish the first state equation of the power grid area based on the pre-acquired network data set.

[0029] The network data set includes: key parameters and the output power of the diesel generator. and the valve drive amount of the diesel generator ; The network data set also includes: the output power of the hydrogen fuel cell. Energy storage battery output power one; The key parameters include controllable power supply parameters, uncontrollable power supply parameters, and other key parameters.

[0030] The controllable power parameters include: diesel generator output. ①; The controllable power parameters also include the output of the hydrogen fuel cell. Energy storage battery output One of the two power grid areas (one type is set up in each of the two power grid areas); The uncontrollable power parameters include: fan output power. Photovoltaic output power One of the three (one type is set up in each of the two power grid areas); Other key parameters include: ④ Load demand in the power grid area. ⑤ Frequency change of the power grid area per unit time ⑥ Frequency variation between power grid areas ⑦ Power exchange between power grid areas ,in Representing different power grid regions.

[0031] In summary, the key parameters in this embodiment include seven types of data.

[0032] S1022: The process of deriving the first state equation for the first power grid region is as follows: The state-space equations in region 1 are as follows: Formula 1; in, For the output power of hydrogen fuel cells, This is the operating voltage of the hydrogen fuel cell. The time constant of a hydrogen fuel cell. To increase its benefit.

[0033] Formula 2; in, For the valve actuation of diesel generator 1, The time constant of the valve actuation quantity of diesel generator 1. R1 represents the frequency shift in region one, and R1 is the droop coefficient in region one. This is the operating voltage of the diesel generator.

[0034] Formula 3; in, This refers to the output power of diesel generator 1. TD1 and TD2 are the gain and time constant of diesel generator 1, respectively.

[0035] Formula 4; Where ΔPt12 is the increment of the exchange power between region one and region two, Δf12 is the frequency offset between region one and region two, and T12 is the synchronization coefficient between region one and region two.

[0036] Formula 5; Where kp1 and Tp1 are the gain and time constant of the rotor oscillation in region one, respectively, ΔPM is the output power of the fan, and ΔPL1 is the load in region one.

[0037] Based on the above equation, establish the region-state space equation (i.e., the first state equation of the first power grid region):

[0038] Formula 6.

[0039] S1024: Similarly, the process of deriving the second state equation for the second power grid region is as follows: Formula 7; in, For the output power of the energy storage battery, This refers to the operating voltage of the energy storage battery. The time constant of the energy storage battery. To increase its benefit.

[0040] Formula 8; in, For the actuation of valve 2 of the diesel generator, The time constant of the valve actuation quantity of diesel generator 2. R2 represents the frequency shift in region two, and R2 is the droop coefficient in region two. This is the operating voltage for diesel generator 2.

[0041] Formula 9; in, For the output power of diesel generator 2, TD2 and TD2 are the gain and time constant of diesel generator 2, respectively.

[0042] Formula 10; Where ΔPt21 is the increment of the exchange power between region 2 and region 1, ΔF21 is the frequency offset between region 2 and region 1, and T21 is the synchronization coefficient between region 2 and region 1.

[0043] Formula 11; Where kp2 and Tp2 are the gain and time constant of the rotor oscillation in region two, respectively, ΔPPV is the photovoltaic output power, and ΔPL2 is the load in region two.

[0044] Based on the above equation, the two-state-space equations for the region are established as follows: Formula 12.

[0045] S104: Based on the pre-acquired state space equation, simplify the first state equation into a second state equation.

[0046] In this application, power sources are divided into two categories: (a) controllable power sources, such as diesel generators, energy storage batteries, and hydrogen fuel cells; and (b) uncontrollable power sources, such as wind turbines and photovoltaics. When designing the MPC, changes in uncontrollable power sources are considered predictable disturbances, while load changes are considered unpredictable disturbances.

[0047] Based on the frequency load regulation model of the interconnected system constructed in S1, the state-space equations of Region 1 and Region 2 are as follows: Formula 13; Specifically, Equation 13 is referred to as the "pre-acquired state-space equation".

[0048] i represents different power grid regions, which can be equal to 1 or 2. A1 is the system matrix of region one, B1 is the control matrix of region one, D1 is the disturbance matrix of region one, A2 is the system matrix of region two, B2 is the control matrix of region two, and D2 is the disturbance matrix of region two.

[0049] Formula 14; Formula 14 is the simplified second state equation for the first power grid region.

[0050] Formula 15; Formula 15 is the simplified second state equation for the second power grid region.

[0051] Referring to Formulas 14 and 15, the second state equation includes four sets of matrices: the state variable matrix p of the power grid area, the control variable matrix u, the disturbance variable matrix w, and the controlled output matrix C.

[0052] In Formula 14, All seven parameters (i.e., the seven key parameters) have built-in weight coefficients. This application adjusts the stability of the network system by adjusting the weights x of these seven parameters. Formula 15 is similar and will not be elaborated further.

[0053] For Equation 14, p1 is the state variable matrix of region 1, u1 is the control variable matrix of region 1, w1 is the disturbance variable matrix of region 1, and C1 is the controlled output matrix of region 1.

[0054] For Equation 15, x2 represents the state variable matrix of region two, u2 represents the control variable matrix of region two, w2 represents the disturbance variable matrix of region two, and C2 represents the controlled output matrix of region two.

[0055] like Figure 2 As shown, , The values ​​in formulas 14 and 15, The frequency offset constant for the region. R1 is the frequency offset constant for region two; R2 is the droop coefficient in region one and the droop coefficient in region two; ΔVW represents the change in wind speed and ΔG represents the change in illumination amplitude; Load 1 represents ΔPL1 in formula 5; Load 2 represents ΔPL2 in formula 11; Rotating pendulum 1 represents formula 5 and rotating pendulum 2 represents formula 11; T12 represents the synchronization coefficient of region one to region two in formula 4; s represents the Laplace operator; T21 represents the synchronization coefficient of region two to region one in formula 10. Figure 2 The final output of the network system is △f1, △f2, and also △Pt (△Pt=△Pt12-△Pt21).

[0056] S106: Based on the network data set and the preset initial weight matrix of key parameters in the network data set. The objective function is determined by the second state equation.

[0057] The initial weight matrix of the key parameters include: .

[0058] Where i represents the number of power grid regions and j represents the number of key parameters.

[0059] In this embodiment, two power grid regions and seven key parameters are involved; therefore, The expression is as follows: Formula 16; S106 includes: S1062: Initial weight matrix of the key parameters for each power grid region. The weights of the seven key parameters of the first power grid region are divided into several weight matrices according to Formula 17. In this embodiment, the weights of the seven key parameters of the first power grid region are divided into three weight matrices: Q1, R1, and S1; and the weights of the seven key parameters of the second power grid region are divided into three weight matrices: Q2, R2, and S2.

[0060] Formula 17; The parameters in Formula 17 are explained as follows: Output of the diesel generator in Region 1 Weighting parameter x11, hydrogen fuel cell output Weighting parameter x12, weighting parameter x13 for load demand ΔPL1 in region 1, weighting parameter x14 for turbine output power ΔPM, weighting parameter x15 for frequency change Δf1 in region 1, weighting parameter x16 for frequency change Δf12 between region 1 and region 2, and switching power ΔP between region 1 and region 2. t12 The weighting coefficient is x17. Diesel generator output in Region 2. Weighting parameter x21, energy storage battery output Weighting parameter x22, load demand ΔPL2 in region 2, weighting parameter x23, photovoltaic output power ΔP PV The weighting parameter x24, the weighting parameter x25 of the frequency change Δf2 in region two, the weighting parameter x26 of the frequency change Δf21 in region two to region one, and the exchange power ΔP in region two to region one. t21 The weighting coefficient is x27.

[0061] Q1 is the weighted matrix of output error pair in region 1; Q2 is the weighted matrix of output error pair in region 2; R1 is the weighted matrix of control increment pair in region 1; R2 is the weighted matrix of control increment pair in region 2; S1 is the weighted matrix of disturbance variables in region 1; S2 is the weighted matrix of disturbance variables in region 2.

[0062] S1064: Determine the objective function based on the second state equation for all power grid regions, the weighting matrix, the pre-acquired prediction time-domain function, and the pre-acquired control time-domain function.

[0063] Specifically, the second state equations, namely Formulas 14 and 15, combine the state variable matrix p and the controlled output matrix C to obtain the reference trajectory expressions y1 and y2 for the first power grid region and the second power grid region, respectively.

[0064] Formula 18; Formula 19.

[0065] Determine the objective function expression for frequency load optimization and control of the interconnected system: Formula 20; in, This represents the prediction of time k+m from time k. This represents the prediction of time k+n from time k; Np is the prediction time domain; Nc is the control time domain; u1, u2, w1, and w2 are explained in Equations 14 and 15.

[0066] Specifically, the constraints for the controllable power sources (diesel generator 1 and diesel generator 2, energy storage battery and hydrogen fuel cell) are as follows: ; In the formula, k+q|k represents the prediction of time k+q from time k, and the subscripts "max" and "min" represent the upper and lower limits of the corresponding variables, respectively.

[0067] S108: Iteratively update the initial weight matrix of the key parameters based on the model predictive control algorithm, the frost optimization algorithm, and the objective function. The optimal weight matrix is ​​obtained. .

[0068] S108 includes: S1082: The initial weight matrix of the key parameters is determined based on the objective function, the preset upper boundary of the model predictive control, and the preset lower boundary of the model predictive control. Perform iterative updates to determine the MPC weight matrix. .

[0069] S1082 includes: S10822: Initial weight matrix of key parameters based on the preset upper boundary and lower boundary of model predictive control. Perform iterative updates to determine the first weight matrix. .

[0070] Specifically, this step involves initializing the RIME cluster, which means representing the seven key parameters in the two power grid regions as RIME particles and randomly initializing them within the search space.

[0071] Formula 21; Where i is the number of power grid regions, which is 2, and j is the number of key parameters, which is 7. rand is a random value within the given range [0,1]. UB and LB represent the upper and lower bounds of the weight parameters in MPC (Model Predictive Control), respectively, where the upper bound is 1 and the lower bound is 0. The weight of each key parameter has its own preset upper and lower boundaries.

[0072] Formula 21 That is, the first weight matrix of S10822 The weight values ​​in the data.

[0073] S10824: Determine the initial weight matrix based on S106. objective function value and the first weight matrix objective function value .

[0074] For details, see S106, which will not be repeated here.

[0075] S10826: If the preset probability parameter Less than the first weight matrix objective function value (Satisfies Formula 22, Condition 1), and the first weight matrix objective function value Less than or equal to the initial weight matrix objective function value (Satisfying Formula 23, Condition 2), then the first weight matrix will be... As the MPC weight matrix (Formulas 22 and 23).

[0076] S10828: Otherwise, do not update the weight matrix, and use the initial weight matrix. As the MPC weight matrix .

[0077] Specifically, this step (condition one) is the hard frost perforation judgment mechanism. The hard frost perforation mechanism allows information exchange between MPC weight parameter optimizations, enhances convergence, and prevents local optima. Formula 22; in, This represents a preset probability parameter, ranging from -1 to 1, which determines the exchange process.

[0078] Next, it is determined whether condition two is true, that is, RIME replaces the latter with the better-performing MPC weight parameters by comparing the objective function values ​​of the optimization process with those without updated weight parameters.

[0079] Formula 23; Formulas 22 and 23 refer to the following: if Less than the first weight matrix objective function value And the first weight matrix objective function value Less than or equal to the initial weight matrix objective function value Then the first weight matrix As the MPC weight matrix Otherwise, do not update the weights, and use the initial weight matrix. Assign to .

[0080] S1084: Based on the objective function and the Frost Optimization Algorithm, adjust the MPC weight matrix. Perform iterative updates to determine the frost weight matrix. .

[0081] S1084 includes: S10842: MPC weight matrix of key parameters based on preset iterative random parameters, preset iterative threshold G, and current iteration number g. Perform iterative updates to determine the second weight matrix. .

[0082] Specifically, step S10842 involves optimizing the weight parameters based on the soft frost search strategy and the hard frost puncture mechanism: The soft frost search strategy, also known as the Rime Optimization Algorithm (RIME), helps the weight parameters of the Model Predictive Control (MPC) to be extensively optimized during the iteration process. The update process is as follows: Formula 24; Formula 25; Formula 26; in, MPC weight matrix The weight values ​​in; The second weight matrix The weight values ​​in the equation; G is the preset iteration threshold; g is the current iteration number; Let be a variable that changes with the number of iterations; To characterize environmental factors, it changes with iteration; ω is a default parameter that can be set to 5 to control the number of segments in the step function; E is the condensation probability factor that changes with iteration; h is the preset viscosity that can be set to a random number in the range (0, 1); r1, r2, and r3 are random quantities that can be set to random numbers in the range (0, 1).

[0083] Formula 24 means that if Then, the weight data is updated based on physical quantities such as the number of iterations and random numbers to obtain the second weight matrix. ;if In this case, the data is not actually updated, and the original weights are used as the second weight matrix. Proceed to the next steps.

[0084] S10844: Determine the second weight matrix based on S106 objective function value .

[0085] S10846: If the preset probability parameter Less than the second weight matrix objective function value And the second weight matrix objective function value Greater than the MPC weight matrix objective function value Then the second weight matrix As the frost weight matrix .

[0086] Specifically, this step is the same as S10826, and will not be repeated here.

[0087] S10848: Otherwise, do not update the weight matrix, and change the MPC weight matrix. As the frost weight matrix .

[0088] S1086: If the current iteration count is greater than the preset iteration threshold, then the frost weight matrix will be adjusted. As the optimal weight matrix .

[0089] Specifically, it determines whether the iteration threshold G has been reached. If the iteration threshold G has been reached, the optimal combination of weight parameters for the objective function is output. If the search ends, the optimization process ends; otherwise, the iteration continues.

[0090] In summary, the embodiments of this application are as follows: First, establish the state equations, determine the boundary conditions and the objective function J.

[0091] Under weak grid conditions, when the load-side power demand increases (PDold increases to PDnew, the difference is ΔPD, i.e. after being disturbed), the 14 weights are updated based on the state equation and objective function J.

[0092] Adjust separately: x takes values ​​of 1 or 2 (representing 2 regions), j takes values ​​of 1-7 (representing 7 parameters) (Note: Output power is the input parameter, output power is the external output parameter) Diesel generator output of Area 1 Middle School Weighting parameter x11, hydrogen fuel cell output Weighting parameter x12, weighting parameter x13 for load demand ΔPL1 in region 1, weighting parameter x14 for turbine output power ΔPM, weighting parameter x15 for frequency change Δf1 in region 1, weighting parameter x16 for frequency change Δf12 between region 1 and region 2, and switching power ΔP between region 1 and region 2. t12 The weighting coefficient is x17; Diesel generator output in Area 2 Weighting parameter x21, energy storage battery output Weighting parameter x22, load demand ΔPL2 in region 2, weighting parameter x23, photovoltaic output power ΔP PV The weighting parameter x24, the weighting parameter x25 of the frequency change Δf2 in region two, the weighting parameter x26 of the frequency change Δf21 in region two to region one, and the exchange power ΔP in region two to region one. t21 The weighting coefficient is x27.

[0093] Formula 14 is the state equation for the first power grid region, where All seven parameters (i.e., key parameters) have pre-defined weight coefficients x0. This application adjusts the stability of the grid system by adjusting the weights x of these seven parameters. Formula 15 is the state equation for the second power grid region, and will not be elaborated further.

[0094] This embodiment improves the stability and responsiveness of frequency Δf1 and Δf2 control between different power grid regions by adjusting the weights (i.e., the time it takes for the frequency of different regions to recover to stable operation after being disturbed is reduced).

[0095] pass Figure 4 , Figure 5 It can be seen that the 14 weights (7*2) adjusted by the MPC+RIME method used in this embodiment are better than the Δf1 and Δf2 obtained by the MPC and PID methods alone (with smaller amplitude fluctuations) and can reach a stable state faster after being disturbed.

[0096] The beneficial effects of the embodiments of the present invention are as follows: 1. This patent aims to improve the stability and responsiveness of frequency control in power grid systems, especially when the load-side power demand increases under weak grid conditions. By employing Model Predictive Control (MPC) and the Rime Optimization Algorithm (RIME), it adjusts the weight parameters of controllable power sources (diesel generators, hydrogen fuel cells, energy storage batteries), uncontrollable power sources (wind power, photovoltaic), and the weight parameters of power exchange between regions in different areas. This achieves improved stability (i.e., stability of Δf1 and Δf2) and responsiveness of frequency control between regions (i.e., the time taken for different regions to recover to stable operation after being disturbed).

[0097] 2. In this embodiment, the state equations of the network system were constructed and optimized, and the optimal weight matrix of the network system was obtained based on MPC and the Frost Ice algorithm. This technology can improve the stability (i.e., stability of Δf1 and Δf2) and responsiveness (i.e., reduce the time required for different regions' frequencies f to return to stable operation after a disturbance) in the event of sudden load changes (disturbances) within a power grid region. These effects can be achieved through… Figure 4 and Figure 5 reflect.

[0098] Figure 4 The black solid line is a schematic diagram of the frequency change of the first power grid area of ​​the network system after updating the weights using this method during the process from disturbance to recovery. It can be seen that compared with the red line (MPC algorithm to optimize weights) and the blue line (PID method to optimize weights), the frequency change (assignment) of this method is smaller and it can reach a stable state faster.

[0099] Figure 5 The black solid line is a schematic diagram of the frequency change of the second power grid area of ​​the network system after updating the weights using this method during the process from disturbance to recovery. It can be seen that compared with the red line (MPC algorithm to optimize weights) and the blue line (PID method to optimize weights), this method can reach a stable state faster.

[0100] Figure 6 The black solid line is a schematic diagram of the change in the tie-line switching power of the network system after updating the weights using this method. It can be seen that compared with the red line (MPC algorithm to optimize weights) and the blue line (PID method to optimize weights), this method can reach a stable state faster.

[0101] The role of adjusting weight X: When the weight coefficient is initialized to X0, the overall interconnected system (i.e., Figure 2The tracking of load changes (i.e., the load increases from PDold to PDnew, the difference ΔPD is called the load change) in the region is poor, and it cannot quickly and effectively adjust the frequency to restore a stable state; the RIME algorithm is used to adjust the output change of diesel generators in region 1. Weighting parameter x11, change in hydrogen fuel cell output Weighting parameter x12, weighting parameter x13 for load demand change in region 1 ΔPL1, weighting parameter x14 for wind turbine output power change ΔPM, weighting parameter x15 for frequency change in region 1 Δf1, weighting parameter x16 for frequency change between region 1 and region 2 Δf12, weighting parameter x17 for exchange power change between region 1 and region 2 ΔPt12; diesel generator output change in region 2 Weighting parameter x21, change in energy storage battery output The weighting parameters x22, x23, x24, x25, x26, x27 are: x22, x23, x24, x25, x26, x27, x28, x29, x20, x21, x22, x23, x24, x25, x26, x27, x28, x29, x20, x21; and x28, x29, x20, x21. These parameters enable the interconnected system to fully track load changes, accelerate the frequency adjustment to a stable state, and improve system stability and responsiveness.

[0102] Example 2 like Figure 7 As shown, this embodiment of the invention provides a device for optimizing network system parameters based on frost and ice optimization model predictive control. The device includes: The first optimization module is used to establish a first state equation for the power grid region based on a pre-acquired network data set; the second optimization module is used to simplify the first state equation into a second state equation based on a pre-acquired state space equation; the third optimization module is used to determine the initial weight matrix of key parameters in the network data set based on the network data set and a pre-defined network data set. The objective function is determined by the second state equation; the fourth optimization module is used to iteratively update the initial weight matrix of the key parameters based on the model predictive control algorithm, the frost optimization algorithm, and the objective function. The optimal weight matrix is ​​obtained. .

[0103] The beneficial effects of this device embodiment are the same as those of the method embodiment, and will not be repeated in this embodiment.

[0104] The above embodiments are merely illustrative examples and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A method for optimizing the parameters of a network system based on predictive control using a frost-ice optimization model, characterized in that, The network system includes multiple power grid areas; For each of the power grid regions, the method includes: S102: Establish the first state equation of the power grid area based on the pre-acquired network data set; S104: Based on the pre-acquired state space equation, simplify the first state equation into a second state equation; S106: Based on the network data set and the preset initial weight matrix of key parameters in the network data set. The objective function is determined by the second state equation; S108: Iteratively update the initial weight matrix of the key parameters based on the model predictive control algorithm, the frost optimization algorithm, and the objective function. The optimal weight matrix is ​​obtained. .

2. The method for optimizing network system parameters based on frost-ice optimization model predictive control according to claim 1, characterized in that, The network data set includes: key parameters and the output power of the diesel generator. and the valve drive amount of the diesel generator ; The network data set also includes: the output power of the hydrogen fuel cell. Energy storage battery output power one; The key parameters include controllable power supply parameters, uncontrollable power supply parameters, and other key parameters.

3. The method for optimizing network system parameters based on frost-ice optimization model predictive control according to claim 2, characterized in that, The controllable power parameters include: diesel generator output. ; The controllable power parameters also include the output of the hydrogen fuel cell. Energy storage battery output one; The uncontrollable power parameters include: fan output power. Photovoltaic output power one; Other key parameters include: load demand in the power grid area. Frequency change of the power grid area per unit time Frequency variation between power grid regions Power exchange between power grid areas ,in Representing different power grid regions.

4. The method for optimizing network system parameters based on frost-ice optimization model predictive control according to claim 3, characterized in that, The initial weight matrix of the key parameters include: ； Where i represents the number of power grid regions and j represents the number of key parameters.

5. The method for optimizing network system parameters based on frost-ice optimization model predictive control according to claim 4, characterized in that, The second state equation includes: the state variable matrix, control variable matrix, disturbance variable matrix, and controlled output matrix of the power grid area.

6. The method for optimizing network system parameters based on frost-ice optimization model predictive control according to claim 5, characterized in that, S106 includes: S1062: Initial weight matrix of the key parameters for each power grid region. Divide into several weighted matrices; S1064: Determine the objective function based on the second state equation for all power grid regions, the weighting matrix, the pre-acquired prediction time-domain function, and the pre-acquired control time-domain function.

7. The method for optimizing network system parameters based on frost-ice optimization model predictive control according to claim 1, characterized in that, S108 includes: S1082: The initial weight matrix of the key parameters is determined based on the objective function, the preset upper boundary of the model predictive control, and the preset lower boundary of the model predictive control. Perform iterative updates to determine the MPC weight matrix. ; S1084: Based on the objective function and the Frost Optimization Algorithm, adjust the MPC weight matrix. Perform iterative updates to determine the frost weight matrix. ; S1086: If the current iteration count is greater than the preset iteration threshold, then the frost weight matrix will be adjusted. As the optimal weight matrix .

8. The method for optimizing network system parameters based on frost-ice optimization model predictive control according to claim 7, characterized in that, S1082 includes: The initial weight matrix of key parameters is based on the preset upper boundary and lower boundary of the model predictive control. Perform iterative updates to determine the first weight matrix. ; The initial weight matrix is ​​determined based on S106. objective function value and the first weight matrix objective function value ; If the preset probability parameter Less than the first weight matrix objective function value And the first weight matrix objective function value Less than or equal to the initial weight matrix objective function value Then the first weight matrix As the MPC weight matrix ; Otherwise, do not update the weight matrix, and use the initial weight matrix. As the MPC weight matrix .

9. The method for optimizing network system parameters based on frost-ice optimization model predictive control according to claim 8, characterized in that, S1084 includes: The MPC weight matrix for key parameters is based on preset iterative random parameters, preset iterative threshold G, and the current iteration number g. Perform iterative updates to determine the second weight matrix. ; The second weight matrix is ​​determined based on S106. objective function value ; If the preset probability parameter Less than the second weight matrix objective function value And the second weight matrix objective function value Greater than the MPC weight matrix objective function value Then the second weight matrix As the frost weight matrix ; Otherwise, do not update the weight matrix, and change the MPC weight matrix. As the frost weight matrix .

10. A parameter optimization device for a network system based on frost-ice optimization model predictive control, characterized in that, The device includes: The first optimization module is used to establish the first state equation of the power grid area based on the pre-acquired network data set; The second optimization module is used to simplify the first state equation into a second state equation based on the pre-acquired state space equation. The third optimization module is used to optimize the network data set based on the network data set and the preset initial weight matrix of key parameters in the network data set. The objective function is determined by the second state equation; The fourth optimization module is used to iteratively update the initial weight matrix of the key parameters based on the model predictive control algorithm, the frost optimization algorithm, and the objective function. The optimal weight matrix is ​​obtained. .

Images