Network construction type converter cluster system collaborative prediction control method and system
By constructing a two-layer control strategy for a grid-type converter cluster system, the frequency and voltage regulation of the converters are optimized, solving the problems of slow regulation speed and poor adaptability of the system under high communication latency, minimizing total network loss, and improving the stability and efficiency of the power system.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-17
- Publication Date
- 2026-04-07
AI Technical Summary
Existing grid-type converter cluster systems have slow adjustment speed and poor adaptability under high communication latency, resulting in nonlinear growth of system network losses and affecting power system stability and energy utilization efficiency.
A two-layer control strategy is adopted. First, a second-layer control is performed through a distributed control architecture of a sparse communication network to obtain the secondary adjustment of angular frequency and voltage. Then, a first-layer control is performed through a droop control strategy to optimize the output frequency and voltage of the converter in order to minimize the total network loss.
It improves the system's dynamic response speed and adaptability under high communication latency, reduces network loss, and enhances the system's economy and stability.
Smart Images
Figure CN121813432A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of control technology for grid-type converter cluster systems, and particularly to a collaborative predictive control method and system for grid-type converter cluster systems. Background Technology
[0002] As the penetration rate of new energy power generation in the power system continues to rise, grid-connected converters, as core equipment supporting the stable operation of new power systems, significantly improve the stability of distributed power generation grid connection by actively constructing grid voltage and frequency. However, in scenarios such as power plants or industrial parks with a high proportion of new energy integration, traditional grid-connected converter control strategies often neglect the global optimization of reactive power flow, leading to a non-linear increase in system network losses as the scale of new energy power generation expands. This not only wastes energy but also directly deteriorates power supply quality due to excessive network losses, threatening the stable operation of the power system. Against this backdrop, optimizing the reactive power output of grid-connected converters through cluster collaborative control technology and constructing a distributed collaborative control scheme for grid-connected converter systems that considers network loss optimization can effectively reduce network losses in the converter system, which is of great significance for promoting the construction of new power systems and the low-carbon energy transition.
[0003] To minimize total network loss, existing grid-type converter cluster systems typically use PI (proportional-integral) regulation to synchronously optimize the allocation of reactive power at each node and implement secondary frequency / voltage regulation control of the system. This approach aims to minimize overall network loss while maintaining grid frequency and voltage stability. However, this strategy is slow to adjust and has poor adaptability to communication delays. Summary of the Invention
[0004] The technical problem to be solved by the present invention is to provide a collaborative predictive control method and system for a grid-type converter cluster system, which can improve the dynamic response speed of the system and enhance the system's adaptability under high communication latency.
[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A collaborative predictive control method for a grid-type converter cluster system includes the following steps: Constructing a grid-type converter cluster system; With the goal of minimizing total network loss, each grid-type converter in the grid-type converter cluster system is subjected to two-level control according to the first control cycle to obtain the secondary regulation of angular frequency and the secondary regulation of voltage. The two-level control adopts a distributed control architecture based on sparse communication network. Based on the secondary adjustment of the angular frequency and the secondary adjustment of the voltage, a layer of control is performed on the angular frequency and voltage of each grid-type converter according to the second control cycle to obtain the output angular frequency and output voltage of each grid-type converter. The layer of control adopts a droop control strategy.
[0006] To solve the above-mentioned technical problems, another technical solution adopted by the present invention is as follows: A collaborative predictive control system for a grid-type converter cluster system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the various steps in the aforementioned collaborative predictive control method for a grid-type converter cluster system.
[0007] The beneficial effects of this invention are as follows: A grid-type converter cluster system is constructed with the goal of minimizing total network loss. Each grid-type converter in the system is subjected to two-level control according to a first control cycle to obtain secondary angular frequency and voltage regulation values. This two-level control adopts a distributed control architecture based on a sparse communication network. Based on the secondary angular frequency and voltage regulation values, a first-level control is performed on the angular frequency and voltage of each grid-type converter according to a second control cycle to obtain the output angular frequency and output voltage of each grid-type converter. This first-level control employs a droop control strategy, thereby incorporating the goal of minimizing total network loss into the two-level distributed cooperative predictive control of the grid-type converter cluster system. Furthermore, the use of dual-level control, compared to PI regulation, improves the regulation speed, thereby increasing the system's dynamic response speed and enhancing its adaptability under high communication latency. Attached Figure Description
[0008] Figure 1 This is a flowchart of a collaborative predictive control method for a grid-type converter cluster system according to an embodiment of the present invention; Figure 2 This is a schematic diagram of a collaborative predictive control system for a grid-type converter cluster system according to an embodiment of the present invention; Figure 3 This is a schematic diagram of a network converter cluster system in a collaborative predictive control method for a network converter cluster system according to an embodiment of the present invention. Figure 4 This is a schematic diagram of the control structure of the i-th grid converter in a collaborative predictive control method for a grid converter cluster system according to an embodiment of the present invention. Figure 5 This is a control schematic diagram of a collaborative predictive control method for a grid-type converter cluster system according to an embodiment of the present invention. Detailed Implementation
[0009] To explain in detail the technical content, objectives, and effects of the present invention, the following description is provided in conjunction with the embodiments and accompanying drawings.
[0010] Before detailing the embodiments of this application, some related concepts will first be explained: Droop control strategy: By simulating the external characteristics of a synchronous generator, the output of distributed power sources is adjusted to maintain the frequency and voltage stability of the power system.
[0011] In existing technologies, to achieve the goal of minimizing total network loss, PI regulation is usually used to synchronously optimize the allocation of reactive power at each node and control the secondary frequency / voltage regulation of the system. While maintaining the stability of the grid frequency and voltage, the overall network loss of the system is minimized. However, this strategy has a slow adjustment speed and poor adaptability to communication delays.
[0012] To at least solve the above problems, please refer to Figure 1 This invention provides a cooperative predictive control method for a grid-type converter cluster system, comprising the following steps: Constructing a grid-type converter cluster system; With the goal of minimizing total network loss, each grid-type converter in the grid-type converter cluster system is subjected to two-level control according to the first control cycle to obtain the secondary regulation of angular frequency and the secondary regulation of voltage. The two-level control adopts a distributed control architecture based on sparse communication network. Based on the secondary adjustment of the angular frequency and the secondary adjustment of the voltage, a layer of control is performed on the angular frequency and voltage of each grid-type converter according to the second control cycle to obtain the output angular frequency and output voltage of each grid-type converter. The layer of control adopts a droop control strategy.
[0013] As can be seen from the above description, the beneficial effects of the present invention are as follows: A grid-type converter cluster system is constructed, aiming to minimize total network loss. Each grid-type converter in the system is subjected to two-level control according to a first control cycle to obtain secondary angular frequency and voltage regulation values. This two-level control adopts a distributed control architecture based on a sparse communication network. Based on the secondary angular frequency and voltage regulation values, a first-level control is performed on the angular frequency and voltage of each grid-type converter according to a second control cycle to obtain the output angular frequency and output voltage of each grid-type converter. This first-level control adopts a droop control strategy, thereby incorporating the goal of minimizing total network loss into the two-level distributed cooperative predictive control of the grid-type converter cluster system. Furthermore, the use of dual-level control, compared to PI regulation, improves the regulation speed, thereby increasing the system's dynamic response speed and enhancing its adaptability under high communication latency.
[0014] Furthermore, minimizing total network loss includes: Based on the active power loss generated by the reactive power of the grid converters in the grid converter cluster system, an objective function to minimize the total network loss is constructed, and constraints corresponding to the objective function are established. Based on the objective function and the constraints, Lagrange multipliers are introduced to construct the Lagrange function; The objective expression for minimizing total network loss is obtained by taking the partial derivative of the Lagrange function and rearranging it. When the objective expression is satisfied, the total network loss of the grid-type converter cluster system is minimized.
[0015] Furthermore, based on the active power loss generated by the reactive power of the grid-type converters in the grid-type converter cluster system, an objective function to minimize the total network loss is constructed, specifically as follows: ; In the formula, This represents the total active power loss generated by reactive power in a grid-type converter cluster system. This represents the active power loss generated by the reactive power of the i-th grid-type converter. , where n represents the total number of grid-type converters in the grid-type converter cluster system; Establish the constraints corresponding to the objective function, specifically as follows: ; In the formula, Q i This represents the reactive power of the i-th grid-type converter. Q i This represents the reactive power loss of the i-th grid-type converter. Q Load This represents the total reactive power load of a grid-type converter cluster system. The objective formula for minimizing total network loss is as follows: ; In the formula, This represents the incremental rate of network loss for the i-th grid-type converter. .
[0016] As described above, on the distribution network side, short lines and low voltage levels result in a high resistance-to-reactance ratio in the line impedance. This makes the line network loss mainly dominated by active power loss on the line resistance. Furthermore, due to the strict constraints of load demand on active power, the adjustment space for active power output is extremely limited, making it difficult to significantly reduce losses by adjusting the active power distribution. However, distributed power sources based on power electronic converters have flexible reactive power control capabilities and can dynamically inject or absorb reactive power to change the reactive power flow of the system. Therefore, the minimum control of total network loss in a large-scale grid-type converter cluster system is mainly achieved by optimizing the reactive power output of each converter. Based on the active power loss generated by the reactive power of the grid-type converters in the system, an objective function to minimize the total network loss is constructed, and corresponding constraints are established. Based on the objective function and constraints, a Lagrange function is constructed, and after taking its partial derivative and rearranging, the objective expression for minimizing the total network loss is obtained. When the partial derivative of the Lagrange function with respect to the reactive power output of each distributed power source is the same, the incremental rate of network loss of each grid-type converter is consistent. Therefore, when the grid-type converter cluster system satisfies the principle of equal incremental rate of network loss, the global network loss can be minimized.
[0017] Furthermore, according to the first control cycle, each grid-type converter in the grid-type converter cluster system is subjected to two-level control to obtain the secondary regulation of angular frequency and the secondary regulation of voltage, including: According to the first control cycle, active power-frequency distributed cooperative predictive control is performed on each grid-type converter in the grid-type converter cluster system to obtain the secondary angular frequency regulation. At the same time, according to the first control cycle, reactive power-voltage distributed cooperative predictive control is performed on each grid-type converter in the grid-type converter cluster system to obtain the secondary voltage regulation.
[0018] As described above, since there is a coupling relationship between frequency and active power, and voltage and reactive power in the grid-type converter cluster system, the secondary control of the grid-type converter cluster system includes active power-frequency distributed cooperative predictive control and reactive power-voltage distributed cooperative predictive control, which respectively realize the secondary regulation of angular frequency and voltage.
[0019] Furthermore, according to the first control cycle, active power-frequency distributed cooperative predictive control is performed on each grid-type converter in the grid-type converter cluster system to obtain the secondary angular frequency regulation, which includes: Determine the first self-penalty term for the second-order adjustment of angular frequency; The output active power of each grid converter in the grid converter cluster system is obtained based on instantaneous power theory. An active power prediction equation is established based on the output active power. Establish the constraint equations corresponding to the active power prediction equation; The active power penalty term is obtained based on the active power prediction equation and the constraint equation. An angular frequency prediction equation is established based on the output active power; The angular frequency penalty term is obtained based on the angular frequency prediction equation; A first cost function for active-frequency distributed cooperative predictive control is constructed based on the first self-penalty term, the active power penalty term, and the angular frequency penalty term; Solving the first cost function yields the optimal control quantity for the first system; The optimal control quantity of the first system is integrated to obtain the secondary adjustment quantity of angular frequency.
[0020] Furthermore, the first self-penalty term of the second-order angular frequency adjustment is determined as follows: ; In the formula, This represents the first self-penalty term for the second-order adjustment of angular frequency. This represents the weight coefficient of the first self-penalty term. [ k ] indicates the secondary adjustment amount of angular frequency; Based on the active power prediction equation and the constraint equation, the active power penalty term is obtained, specifically as follows: ; In the formula, This indicates the active power penalty term. This represents the weighting coefficient of the active power penalty term. k p,i This represents the active power droop factor of the locally grid-type converter i. P i [ k +1] indicates that the grid-type converter i is in k The active power output at time +1 k p,j This represents the active power droop factor of converter j in a neighbor-connected grid configuration. P j [ k [This indicates that the neighbor-networked converter j is in...] k The active power output at any given time, where m represents the total number of neighboring grid-type converters of grid-type converter i; The angular frequency penalty term is obtained based on the aforementioned angular frequency prediction equation, specifically as follows: ; In the formula, This represents the angular frequency penalty term. This represents the weighting coefficient of the angular frequency penalty term. [ k+1] indicates that the local grid-type converter i is in k Angular frequency at time +1 Indicates the rated angular frequency; The first cost function for active-frequency distributed cooperative predictive control is constructed based on the first self-penalty term, the active power penalty term, and the angular frequency penalty term, specifically as follows: ; In the formula, J 1 represents the first cost function of active-frequency distributed cooperative predictive control.
[0021] As described above, by determining the first self-penalty term, the active power penalty term, and the angular frequency penalty term, a first cost function for active-frequency distributed cooperative predictive control is constructed. Solving the first cost function yields the first optimal control quantity for the system. Then, by performing integration, a secondary angular frequency adjustment quantity is obtained. This enables the active power of each converter to be accurately allocated according to its respective capacity, and the system frequency to track the rated value without error. Thus, the angular frequency correction of the grid-type converter cluster system and the accurate allocation of active power output according to capacity are realized.
[0022] Furthermore, reactive power-voltage distributed cooperative predictive control is performed on each grid-type converter in the grid-type converter cluster system according to the first control cycle, resulting in the following voltage secondary regulation quantities: Determine the second self-penalty term for the secondary voltage regulation; The output reactive power of each grid converter in the grid converter cluster system is obtained based on instantaneous power theory. A reactive power prediction equation is established based on the output reactive power. By combining the reactive power prediction equation with the objective of minimizing total network loss, the network loss incremental rate prediction model of each grid-type converter is obtained. The reactive power penalty term is determined based on the network loss incremental rate prediction model. Establish the expression for the dynamic consistency algorithm of bus voltage; Based on the bus voltage dynamic consistency algorithm expression, the bus voltage prediction equation for each grid-type converter is obtained. The bus voltage penalty term is obtained based on the bus voltage prediction equation. A second cost function for reactive-voltage distributed collaborative predictive control is established based on the second self-penalty term, the reactive power penalty term, and the bus voltage penalty term; Solving the second cost function yields the optimal control quantity for the second system. The optimal control quantity of the second system is integrated to obtain the secondary voltage regulation quantity.
[0023] Furthermore, the second self-penalty term of the voltage secondary regulation is determined as follows: ; In the formula, This represents the second self-penalty term for the secondary voltage regulation. This represents the weighting coefficient of the second self-penalty term. [ k [] indicates the secondary voltage regulation amount; The reactive power penalty term is determined based on the aforementioned network loss incremental rate prediction model, specifically as follows: ; In the formula, This indicates a reactive power penalty term. This represents the weighting coefficient for the reactive power penalty term. [ k +1] indicates that the grid-type converter i is in k The incremental rate of network loss at time +1 [ k +1] indicates that the neighbor-networked converter j is in k The incremental rate of network loss at time +1, where m represents the total number of neighboring network converters of network converter i; The bus voltage penalty term is obtained based on the bus voltage prediction equation, specifically as follows: ; In the formula, This indicates the bus voltage penalty term. This represents the weighting coefficient of the bus voltage penalty term. [ k +1] indicates that the local grid-type converter i is in k Bus voltage observation at time +1 E n Indicates the rated voltage; A second cost function for reactive-voltage distributed collaborative predictive control is established based on the second self-penalty term, the reactive power penalty term, and the bus voltage penalty term, specifically as follows: ; In the formula, J 2 represents the second cost function of reactive-voltage distributed collaborative predictive control.
[0024] As described above, a second cost function for reactive-voltage distributed collaborative predictive control is established based on the second self-penalty term, reactive power penalty term, and bus voltage penalty term. Solving the second cost function yields the second optimal control quantity of the system. Integrating this quantity yields the secondary voltage regulation quantity, enabling precise allocation of reactive power in each converter according to the consistent incremental rate of network loss, and ensuring that the AC bus voltage tracks the rated value without error. This achieves correction of the system bus voltage and minimizes system network loss.
[0025] Furthermore, based on the secondary adjustment of the angular frequency and the secondary adjustment of the voltage, a single layer of control is performed on the angular frequency and voltage of each grid-type converter according to the second control cycle, resulting in the output angular frequency and output voltage of each grid-type converter, including: The rated angular frequency, rated voltage, active droop coefficient, reactive droop coefficient, output active power, and output reactive power of each grid-type converter are obtained according to the second control cycle. The output angular frequency and output voltage of each grid-type converter are calculated based on the rated angular frequency, the rated voltage, the active droop coefficient, the reactive droop coefficient, the output active power, the output reactive power, the secondary regulation of the angular frequency, and the secondary regulation of the voltage.
[0026] As described above, the output angular frequency and output voltage of each grid-type converter are calculated based on the rated angular frequency, rated voltage, active power droop coefficient, reactive power droop coefficient, output active power, output reactive power, secondary regulation of angular frequency, and secondary regulation of voltage. By adopting a droop control strategy, the converter simulates the operating characteristics of the governor and exciter of a synchronous generator, which can significantly improve the frequency and voltage regulation capabilities of the converter equipment. In addition, since the first-level control is essentially differential control, a second-level control is introduced to compensate for the system output frequency and voltage, restoring them to their rated values. This significantly reduces the network loss of the grid-type converter cluster system, and even under high system communication latency, distributed power supply collaborative output can still be achieved, and the system frequency and voltage remain stable.
[0027] Please refer to Figure 2 Another embodiment of the present invention provides a cooperative predictive control system for a grid-type converter cluster system, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the various steps of the above-described cooperative predictive control method for a grid-type converter cluster system.
[0028] The cooperative predictive control method and system for a grid-type converter cluster system described above are applicable to large-scale grid-type converter cluster systems. The following detailed implementation methods illustrate these methods: Please refer to Figure 1One embodiment of the present invention is as follows: A collaborative predictive control method for a grid-type converter cluster system includes the following steps: S1. Construct a grid-type converter cluster system.
[0029] like Figure 3 As shown, the grid-type converter cluster system adopts a radial network, including n distributed generation sources (DGs) and n grid-type converters (DC / AC), as well as the line equivalent resistance. R ( R 1. R i … R n , R i (representing the equivalent resistance of the i-th line) and the equivalent inductance of the line. L ( L 1. L i … L n , L i (representing the equivalent inductance of the i-th line), all distributed power sources supply power to the load on the common bus, and each distributed power source shares information through a sparse communication network. Due to its simple and reliable topology, it is widely used, especially in remote areas or small power supply systems far from the main grid.
[0030] like Figure 4 As shown, the control structure of the i-th grid-type converter is illustrated. The grid-type converter cluster system consists of a DC side, a three-phase voltage source converter, and line impedance. Figure 4 middle U dc This represents the DC-side port voltage. The DC side is typically connected to energy storage equipment or photovoltaic devices with active power backup. The inductance and capacitance of an LC filter are respectively... L f and C f express, R g and L g These represent the resistance and inductance values of the line impedance, respectively. i 1 and i o These represent the converter port output current and the filtered current fed into the AC bus, respectively. u c Indicates capacitor voltage. E ref Indicates the reference value of voltage amplitude. Indicates the output power angle. S kIndicates a switch quantity.
[0031] S2. With the goal of minimizing total network loss, each grid-type converter in the grid-type converter cluster system is subjected to two-level control according to the first control cycle to obtain the secondary regulation of angular frequency and voltage. The two-level control adopts a distributed control architecture based on a sparse communication network, such as... Figure 5 As shown, by collecting information from neighboring grid-type converters, a secondary regulation quantity is generated using a distributed model predictive control module, specifically including S21-S24: S21. Based on the active power loss generated by the reactive power of the grid converter in the grid converter cluster system, construct an objective function to minimize the total network loss, and establish the constraints corresponding to the objective function.
[0032] The total power loss of each grid-type converter is: ; In the formula, i This represents the total power loss of the i-th grid-type converter. Q i This represents the reactive power loss of the i-th grid-type converter. P i This represents the active power loss of the i-th grid-type converter. R i Let represent the equivalent resistance of the i-th line. X i Let i represent the equivalent reactance of the i-th line.
[0033] Since this invention primarily focuses on the impact of reactive power on active power loss, an objective function to minimize total network loss is constructed based on the active power loss generated by the reactive power of the grid-type converters in the grid-type converter cluster system. Specifically: ; In the formula, This represents the total active power loss generated by reactive power in a grid-type converter cluster system. This represents the active power loss generated by the reactive power of the i-th grid-type converter. , where n represents the total number of grid-type converters in the grid-type converter cluster system.
[0034] Establish the constraints corresponding to the objective function, specifically as follows: ; In the formula, Q i This represents the reactive power of the i-th grid-type converter. QLoad This represents the total reactive power load of a grid-type converter cluster system.
[0035] S22. Based on the objective function and the constraints, introduce Lagrange multipliers to construct the Lagrange function, specifically as follows: ; In the formula, L Represent the Lagrange function, It represents the Lagrange multiplier.
[0036] S23. After taking the partial derivative of the Lagrange function and rearranging, the objective expression for minimizing the total network loss is obtained. When the objective expression is satisfied, the total network loss of the grid-type converter cluster system is minimized.
[0037] Taking the partial derivative of the Lagrange function yields: ; … ; The objective formula for minimizing total network loss is as follows: ; In the formula, This represents the incremental rate of network loss for the i-th grid-type converter. .
[0038] S24. Perform active power-frequency distributed cooperative predictive control on each grid-type converter in the grid-type converter cluster system according to the first control cycle to obtain the secondary angular frequency regulation. Simultaneously, perform reactive power-voltage distributed cooperative predictive control on each grid-type converter in the grid-type converter cluster system according to the first control cycle to obtain the secondary voltage regulation. Specifically, this includes S241-S2421: In one alternative implementation, the first control period is on the order of milliseconds or seconds.
[0039] S241. Determine the first self-penalty term for the second-order adjustment of angular frequency, specifically: ; In the formula, This represents the first self-penalty term for the second-order adjustment of angular frequency. This represents the weight coefficient of the first self-penalty term. [ k ] indicates the secondary adjustment amount of angular frequency.
[0040] The expression for the second-order adjustment of the angular frequency is: ; In the formula, [ k +1] indicates that the i-th grid-type converter... k The output angular frequency at time +1 [ k ] indicates that the i-th grid-type converter is in k The output angular frequency at a given time.
[0041] To ensure that the active power of each converter is accurately allocated according to its own capacity, the active power of each unit needs to be controlled. First, an active power prediction model is established, as follows.
[0042] S242. Based on instantaneous power theory, the output active power of each grid-type converter in the grid-type converter cluster system is obtained as follows: ; In the formula, E di This represents the d-axis component of the port output voltage of the i-th grid-type converter. i di This represents the d-axis component of the port output current of the i-th grid-connected converter. E qi This represents the q-axis component of the port output voltage of the i-th grid-connected converter. i qi Let represent the q-axis component of the port output current of the i-th grid-connected converter. Assuming the system output voltages are all on the d-axis, then... E qi =0. Since the port output current is the inductor current, it is assumed that... k +1 time and k The current remains almost constant at any given time, that is... i di [ k +1] i di [ k ]、 i qi [ k +1] i qi [ k ].
[0043] S243. Establish an active power prediction equation based on the output active power.
[0044] Specifically, discretizing the above equation yields the active power prediction equation, which is as follows: ; In the formula, P i [ k] indicates that the grid-type converter i is in k The active power output at all times. E di [ k ] indicates that the i-th grid-type converter is in k The d-axis component of the port output voltage at time t. [ k ] indicates in k Voltage power angle at time , i di [ k ] indicates that the i-th grid-type converter is in k The d-axis component of the port output current at any given time; because k Time and k Voltage power angle at time +1 The change is very small, therefore sin [ k ] Therefore: ; In the formula, T s This represents the first control cycle; thus, the final active power prediction equation is obtained, specifically: .
[0045] S244. Establish the constraint equations corresponding to the active power prediction equation, specifically as follows: .
[0046] S245. Based on the active power prediction equation and the constraint equation, the active power penalty term is obtained, specifically as follows: ; In the formula, This indicates the active power penalty term. This represents the weighting coefficient of the active power penalty term. k p,i This represents the active power droop factor of the locally grid-type converter i. P i [ k +1] indicates that the grid-type converter i is in k The active power output at time +1 k p,j This represents the active power droop factor of converter j in a neighbor-connected grid configuration. P j [ k [This indicates that the neighbor-networked converter j is in...] k The active power output at any given time, where m represents the total number of neighboring grid-type converters of grid-type converter i.
[0047] S246. Establish an angular frequency prediction equation based on the output active power.
[0048] Specifically, to achieve the system's frequency error-free tracking rating, by We can obtain: ; Combining the above equation with the formula for the output active power of a grid-type converter, we obtain the angular frequency prediction equation as follows: .
[0049] S247. Based on the angular frequency prediction equation, the angular frequency penalty term is obtained, specifically as follows: ; In the formula, This represents the angular frequency penalty term. This represents the weighting coefficient of the angular frequency penalty term. [ k +1] indicates that the local grid-type converter i is in k Angular frequency at time +1 Indicates the rated angular frequency.
[0050] S248. Construct a first cost function for active-frequency distributed collaborative predictive control based on the first self-penalty term, the active power penalty term, and the angular frequency penalty term, so as to achieve accurate allocation of active power of each converter according to its respective capacity, and the system frequency zero-failure tracking rated value, specifically: ; In the formula, J 1 represents the first cost function of active-frequency distributed cooperative predictive control. The first cost function is a quadratic function.
[0051] S249. Solve the first cost function to obtain the optimal control quantity of the first system. .
[0052] S2410. Perform integral calculation on the optimal control quantity of the first system to obtain the secondary adjustment quantity of angular frequency.
[0053] S2411. Determine the second self-penalty term for the secondary voltage regulation to prevent instability caused by over-regulation of the control system. Specifically: ; In the formula, This represents the second self-penalty term for the secondary voltage regulation. This represents the weighting coefficient of the second self-penalty term. [ k [] indicates the secondary voltage regulation amount; The expression for the secondary voltage regulation is as follows: ; In the formula, [ k +1] indicates that the i-th grid-type converter... k Output voltage at +1 time [ k ] indicates that the i-th grid-type converter is in k The output voltage at any given time.
[0054] To ensure accurate allocation of reactive power across all converters according to the consistent incremental rate of network loss, a reactive power prediction model is established, as follows.
[0055] S2412. Based on instantaneous power theory, the output reactive power of each grid-type converter in the grid-type converter cluster system is obtained as follows: .
[0056] S2413. Establish a reactive power prediction equation based on the output reactive power.
[0057] Specifically, discretizing the above equation yields the reactive power prediction equation as follows: ; In the formula, Q i [ k +1] indicates that the grid-type converter i is in k The reactive power output at time +1 Q i [ k ] indicates that the grid-type converter i is in k The reactive power output at all times.
[0058] S2414. Combine the reactive power prediction equation with the objective of minimizing total network loss to obtain the network loss incremental rate prediction model for each grid-type converter, specifically: .
[0059] S2415. Based on the aforementioned network loss incremental rate prediction model, determine the reactive power penalty term to achieve consistent network loss incremental rates for each converter, specifically as follows: ; In the formula, This indicates a reactive power penalty term. This represents the weighting coefficient for the reactive power penalty term. [ k +1] indicates that the grid-type converter i is in kThe incremental rate of network loss at time +1 [ k +1] indicates that the neighbor-networked converter j is in k The incremental rate of network loss at time +1.
[0060] Since voltage is a local variable and there is line impedance between the converter and the AC bus, an average consistent voltage observer is introduced to achieve error-free tracking of the AC bus voltage rating in order to observe the AC bus voltage, as detailed below.
[0061] S2416. Establish the expression for the dynamic consistency algorithm of bus voltage, specifically as follows: ; In the formula, This represents the observed bus voltage value of local grid-type converter i. This represents the observed bus voltage value of the neighboring grid-connected converter j. a ij This represents the communication weighting coefficient between local grid converter i and its neighboring grid converter j. This represents the measured output voltage value of the local grid-type converter i.
[0062] S2417. Based on the bus voltage dynamic consistency algorithm expression, the bus voltage prediction equation for each grid-type converter is obtained.
[0063] Specifically, the expression for the dynamic consensus algorithm of the bus voltage is discretized to obtain the initial bus voltage prediction equation for each grid-type converter, as follows: ; In the formula, [ k ] indicates that the local grid-type converter i is in k The observed value of the bus voltage at time 10:00; Substituting the expression for the secondary voltage regulation into the initial bus voltage prediction equation, we obtain the final bus voltage prediction equation, which is as follows: .
[0064] S2418. Based on the bus voltage prediction equation, the bus voltage penalty term is obtained, specifically: ; In the formula, This indicates the bus voltage penalty term. This represents the weighting coefficient of the bus voltage penalty term. [ k +1] indicates that the local grid-type converter i is in k Bus voltage observation at time +1E n This indicates the rated voltage.
[0065] S2419. Based on the second self-penalty term, the reactive power penalty term, and the bus voltage penalty term, establish a second cost function for reactive-voltage distributed collaborative predictive control to achieve accurate allocation of reactive power of each converter according to the consistent incremental rate of network loss, and to achieve zero-error tracking of the rated value of AC bus voltage, specifically: ; In the formula, J 2 represents the second cost function of reactive-voltage distributed collaborative predictive control.
[0066] S2420. Solve the second cost function to obtain the optimal control quantity of the second system. .
[0067] S2421. Perform integral calculation on the optimal control quantity of the second system to obtain the secondary voltage regulation quantity.
[0068] S3. Based on the secondary adjustment of the angular frequency and the secondary adjustment of the voltage, perform a single-level control on the angular frequency and voltage of each grid-type converter according to the second control cycle to obtain the output angular frequency and output voltage of each grid-type converter. This single-level control employs a droop control strategy. Figure 5 As shown, specifically including S31-S32: S31. According to the second control cycle, obtain the rated angular frequency, rated voltage, active droop coefficient, reactive droop coefficient, output active power and output reactive power of each grid-type converter.
[0069] In one alternative implementation, the second control cycle is on the order of microseconds.
[0070] S32. Calculate the output angular frequency and output voltage of each grid-type converter based on the rated angular frequency, the rated voltage, the active droop coefficient, the reactive droop coefficient, the output active power, the output reactive power, the secondary adjustment of the angular frequency, and the secondary adjustment of the voltage.
[0071] For each grid-type converter, the original single-layer control can be described as follows: ; In the formula, This represents the output angular frequency of the i-th grid-type converter. E i This represents the output voltage of the i-th grid-type converter. k q,i Let represent the reactive power droop coefficient of the i-th grid-type converter. Pi This represents the active power of the i-th grid-type converter; Since the first-level control is essentially differential control, a second-level control is needed to compensate for the system output frequency and voltage, restoring them to their rated values. After introducing the second-level control, the output angular frequency and output voltage of each grid-type converter are calculated based on the rated angular frequency, rated voltage, active power droop coefficient, reactive power droop coefficient, output active power, output reactive power, the secondary regulation of the angular frequency, and the secondary regulation of the voltage. Specifically: ; In the formula, This represents the secondary angular frequency regulation of the i-th grid-type converter. This represents the secondary voltage regulation of the i-th grid-type converter.
[0072] In summary, the above-described collaborative predictive control method for a grid-type converter cluster system of the present invention constructs a grid-type converter cluster system with the goal of minimizing total network loss. It performs two-level control on each grid-type converter in the system according to a first control cycle to obtain secondary angular frequency and voltage regulation values. This two-level control adopts a distributed control architecture based on a sparse communication network. Based on the secondary angular frequency and voltage regulation values, it performs one-level control on the angular frequency and voltage of each grid-type converter according to a second control cycle to obtain the secondary regulation values of each grid-type converter. The output angular frequency and output voltage are controlled by a droop control strategy in this layer-two distributed cooperative predictive control of the grid-type converter cluster system. This strategy incorporates the goal of minimizing total network loss into the system's operational objective. The use of dual-layer control improves the adjustment speed compared to PI regulation, thereby enhancing the system's dynamic response speed and adaptability under high communication latency. Furthermore, the incremental rate of network loss is incorporated into the cost function of the layer-two distributed cooperative predictive control, enabling each converter to optimize reactive power output according to the same incremental rate of network loss, thus minimizing the total network loss and improving the economic efficiency of the converter system operation.
[0073] According to another aspect of the invention, Figure 2 This is a schematic diagram illustrating a cooperative predictive control system for a grid-type converter cluster system according to an embodiment of the present invention. The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the various steps of the cooperative predictive control method for a grid-type converter cluster system as described above.
[0074] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent modifications made based on the content of the present invention specification and drawings, or direct or indirect applications in related technical fields, are similarly included within the patent protection scope of the present invention.
Claims
1. A cooperative predictive control method for a grid-type converter cluster system, characterized in that, Including the following steps: Constructing a grid-type converter cluster system; With the goal of minimizing total network loss, each grid-type converter in the grid-type converter cluster system is subjected to two-level control according to the first control cycle to obtain the secondary regulation of angular frequency and the secondary regulation of voltage. The two-level control adopts a distributed control architecture based on sparse communication network. Based on the secondary adjustment of the angular frequency and the secondary adjustment of the voltage, a layer of control is performed on the angular frequency and voltage of each grid-type converter according to the second control cycle to obtain the output angular frequency and output voltage of each grid-type converter. The layer of control adopts a droop control strategy.
2. The collaborative predictive control method for a grid-type converter cluster system according to claim 1, characterized in that, Minimizing total network loss includes: Based on the active power loss generated by the reactive power of the grid converters in the grid converter cluster system, an objective function to minimize the total network loss is constructed, and constraints corresponding to the objective function are established. Based on the objective function and the constraints, Lagrange multipliers are introduced to construct the Lagrange function; The objective expression for minimizing total network loss is obtained by taking the partial derivative of the Lagrange function and rearranging it. When the objective expression is satisfied, the total network loss of the grid-type converter cluster system is minimized.
3. The collaborative predictive control method for a grid-type converter cluster system according to claim 2, characterized in that, Based on the active power loss generated by the reactive power of the grid-type converters in the aforementioned grid-type converter cluster system, an objective function to minimize the total network loss is constructed, specifically: ； In the formula, This represents the total active power loss generated by reactive power in a grid-type converter cluster system. This represents the active power loss generated by the reactive power of the i-th grid-type converter. , where n represents the total number of grid-type converters in the grid-type converter cluster system; Establish the constraints corresponding to the objective function, specifically as follows: ； In the formula, Q i This represents the reactive power of the i-th grid-type converter. Q i This represents the reactive power loss of the i-th grid-type converter. Q Load This represents the total reactive power load of a grid-type converter cluster system. The objective formula for minimizing total network loss is as follows: ； In the formula, This represents the incremental rate of network loss for the i-th grid-type converter. .
4. The collaborative predictive control method for a grid-type converter cluster system according to claim 3, characterized in that, According to the first control cycle, each grid-type converter in the grid-type converter cluster system is subjected to two-level control to obtain the secondary regulation of angular frequency and the secondary regulation of voltage, including: According to the first control cycle, active power-frequency distributed cooperative predictive control is performed on each grid-type converter in the grid-type converter cluster system to obtain the secondary angular frequency regulation. At the same time, according to the first control cycle, reactive power-voltage distributed cooperative predictive control is performed on each grid-type converter in the grid-type converter cluster system to obtain the secondary voltage regulation.
5. The collaborative predictive control method for a grid-type converter cluster system according to claim 4, characterized in that, According to the first control cycle, active power-frequency distributed cooperative predictive control is performed on each grid-type converter in the grid-type converter cluster system to obtain the secondary angular frequency adjustment, which includes: Determine the first self-penalty term for the second-order adjustment of angular frequency; The output active power of each grid converter in the grid converter cluster system is obtained based on instantaneous power theory. An active power prediction equation is established based on the output active power. Establish the constraint equations corresponding to the active power prediction equation; The active power penalty term is obtained based on the active power prediction equation and the constraint equation. An angular frequency prediction equation is established based on the output active power; The angular frequency penalty term is obtained based on the angular frequency prediction equation; A first cost function for active-frequency distributed cooperative predictive control is constructed based on the first self-penalty term, the active power penalty term, and the angular frequency penalty term; Solving the first cost function yields the optimal control quantity for the first system; The optimal control quantity of the first system is integrated to obtain the secondary adjustment quantity of angular frequency.
6. The collaborative predictive control method for a grid-type converter cluster system according to claim 5, characterized in that, The first self-penalty term for determining the second-order adjustment of angular frequency is as follows: ； In the formula, This represents the first self-penalty term for the second-order adjustment of angular frequency. This represents the weight coefficient of the first self-penalty term. [ k ] indicates the secondary adjustment amount of angular frequency; Based on the active power prediction equation and the constraint equation, the active power penalty term is obtained, specifically as follows: ； In the formula, This indicates the active power penalty term. This represents the weighting coefficient of the active power penalty term. k p,i This represents the active power droop factor of the locally grid-type converter i. P i [ k +1] indicates that the grid-type converter i is in k The active power output at time +1 k p,j This represents the active power droop factor of converter j in a neighbor-connected grid configuration. P j [ k ] indicates that the neighbor-networked converter j is in k The active power output at any given time, where m represents the total number of neighboring grid-type converters of grid-type converter i; The angular frequency penalty term is obtained based on the aforementioned angular frequency prediction equation, specifically as follows: ； In the formula, This represents the angular frequency penalty term. This represents the weighting coefficient of the angular frequency penalty term. [ k +1] indicates that the local grid-type converter i is in k Angular frequency at time +1 Indicates the rated angular frequency; The first cost function for active-frequency distributed cooperative predictive control is constructed based on the first self-penalty term, the active power penalty term, and the angular frequency penalty term, specifically as follows: ； In the formula, J 1 represents the first cost function of active-frequency distributed cooperative predictive control.
7. The collaborative predictive control method for a grid-type converter cluster system according to claim 4, characterized in that, According to the first control cycle, reactive power-voltage distributed collaborative predictive control is performed on each grid converter in the grid converter cluster system to obtain the voltage secondary regulation quantity, which includes: Determine the second self-penalty term for the secondary voltage regulation; The output reactive power of each grid converter in the grid converter cluster system is obtained based on instantaneous power theory. A reactive power prediction equation is established based on the output reactive power. By combining the reactive power prediction equation with the objective of minimizing total network loss, the network loss incremental rate prediction model of each grid-type converter is obtained. The reactive power penalty term is determined based on the network loss incremental rate prediction model. Establish the expression for the dynamic consistency algorithm of bus voltage; Based on the bus voltage dynamic consistency algorithm expression, the bus voltage prediction equation for each grid-type converter is obtained. The bus voltage penalty term is obtained based on the bus voltage prediction equation. A second cost function for reactive-voltage distributed collaborative predictive control is established based on the second self-penalty term, the reactive power penalty term, and the bus voltage penalty term; Solving the second cost function yields the optimal control quantity for the second system. The optimal control quantity of the second system is integrated to obtain the secondary voltage regulation quantity.
8. The collaborative predictive control method for a grid-type converter cluster system according to claim 7, characterized in that, The second self-penalty term for determining the secondary voltage regulation is as follows: ； In the formula, This represents the second self-penalty term for the secondary voltage regulation. This represents the weighting coefficient of the second self-penalty term. [ k [] indicates the secondary voltage regulation amount; The reactive power penalty term is determined based on the aforementioned network loss incremental rate prediction model, specifically as follows: ； In the formula, This indicates a reactive power penalty term. This represents the weighting coefficient for the reactive power penalty term. [ k +1] indicates that the grid-type converter i is in k The incremental rate of network loss at time +1 [ k +1] indicates that the neighbor-networked converter j is in k The incremental rate of network loss at time +1, where m represents the total number of neighboring network converters of network converter i; The bus voltage penalty term is obtained based on the bus voltage prediction equation, specifically as follows: ； In the formula, This indicates the bus voltage penalty term. This represents the weighting coefficient of the bus voltage penalty term. [ k +1] indicates that the local grid-type converter i is in k Bus voltage observation at time +1 E n Indicates the rated voltage; A second cost function for reactive-voltage distributed collaborative predictive control is established based on the second self-penalty term, the reactive power penalty term, and the bus voltage penalty term, specifically as follows: ； In the formula, J 2 represents the second cost function of reactive-voltage distributed collaborative predictive control.
9. The collaborative predictive control method for a grid-type converter cluster system according to claim 1, characterized in that, Based on the secondary regulation of angular frequency and the secondary regulation of voltage, a single layer of control is performed on the angular frequency and voltage of each grid-type converter according to the second control cycle, resulting in the output angular frequency and output voltage of each grid-type converter, including: The rated angular frequency, rated voltage, active droop coefficient, reactive droop coefficient, output active power, and output reactive power of each grid-type converter are obtained according to the second control cycle. The output angular frequency and output voltage of each grid-type converter are calculated based on the rated angular frequency, the rated voltage, the active droop coefficient, the reactive droop coefficient, the output active power, the output reactive power, the secondary regulation of the angular frequency, and the secondary regulation of the voltage.
10. A cooperative predictive control system for a grid-type converter cluster system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements each step of the collaborative predictive control method for a grid-type converter cluster system according to any one of claims 1 to 9.