MPC wind power converter power prediction control method based on layered dynamic weight adjustment

Abstract

The invention discloses an MPC wind power converter power prediction control method based on layered dynamic weight adjustment, and relates to the technical field of wind power converter power prediction, and the method comprises the steps: building a current prediction model based on a mathematical model of the output current of a converter under a first preset coordinate system; an MPC control system is introduced into the current prediction model, and a first objective function and a second objective function which are independent of each other are set for two control layers of the MPC control system respectively; setting and adjusting extended weight coefficients of the first objective function and the second objective function according to the power grid working condition requirements; according to the control method, an expansion weight coefficient mapping relation between two control layers is established to realize dynamic fusion between the two control layers, so that the output power of the converter is controlled, the control method can realize flexible adjustment of active power and reactive power, and the power tracking performance and the dynamic response capability of the system are remarkably improved.

Description

Technical Field

[0001] The present application generally relates to the technical field of wind power converter power prediction and control, and in particular to an MPC wind power converter power prediction and control method based on hierarchical dynamic weight adjustment. Background Art

[0002] As a clean, renewable energy source, wind power has become a crucial component of the global energy transition. With the continuous advancement of wind power technology, installed wind power capacity and power generation have increased annually, particularly in Europe, North America, and China, where installed wind power capacity has reached world-leading levels. The widespread adoption of wind power has driven continuous innovation in wind turbine control technology, with full-power converter (FPC) control technology being key to ensuring the stable, economical, and efficient operation of wind power systems.

[0003] The primary function of a full-power wind turbine converter is to convert the AC power generated by the wind turbine generator into stable AC power that meets grid requirements. This ensures that the wind turbine maximizes its power generation efficiency under varying wind speeds and loads while maintaining synchronization with the grid. Due to the randomness and volatility of wind energy, traditional control methods (such as PI control and PID control) often fail to provide ideal control results under complex operating conditions, especially when wind speeds fluctuate dramatically or the grid fluctuates significantly.

[0004] Model Predictive Control (MPC) is an advanced control method based on system dynamics models. It achieves control objectives by predicting the system's future state and optimizing control inputs. However, traditional MPC control methods are generally unable to independently track and regulate active power (P) and reactive power (Q). Due to the strong nonlinearity, high coupling, and complex operating conditions of wind power systems, traditional MPC struggles to achieve optimal control results under varying operating conditions. Summary of the Invention

[0005] In view of the above-mentioned defects or deficiencies in the prior art, it is desired to provide an MPC wind power converter power prediction control method based on hierarchical dynamic weight adjustment.

[0006] The present application provides a power prediction and control method for an MPC wind power converter based on hierarchical dynamic weight adjustment, comprising the following steps: Constructing a current prediction model based on a mathematical model of the output current of the converter in a first preset coordinate system; Introducing an MPC control system into the current prediction model, and setting mutually independent first and second objective functions for two control layers of the MPC control system; Setting and adjusting the extended weight coefficients of the first objective function and the second objective function according to the grid operating conditions; Establishing an extended weight coefficient mapping relationship between the two control layers to achieve dynamic fusion between the two control layers, thereby controlling the output power of the converter; Among them, the first objective function is used to optimize the predicted power reference value output by the current prediction model in combination with the grid operating conditions, and the second objective function is used to track and control the actual current and voltage of the converter based on the optimized predicted power reference value; the predicted power reference value includes: active power reference value and reactive power reference value.

[0007] According to the technical solution provided by this application, a current prediction model is constructed based on a mathematical model of the output current of the converter in a first preset coordinate system, specifically including: Establishing a three-phase voltage and current equation of the converter through inductance according to Kirchhoff's law to determine a mathematical model of the output current of the converter in the first preset coordinate system; Transforming the mathematical model into a second preset coordinate system to obtain transformation components corresponding to each axis of the second preset coordinate system; The conversion components corresponding to each axis are discretely processed within a preset sampling period, and a phase-locked loop is introduced to obtain the current prediction model.

[0008] According to the technical solution provided by the present application, the control layer includes: a high control layer and a low control layer, wherein the high control layer and the low control layer correspond to the first objective function and the second objective function which are independent of each other respectively; The first objective function The second objective function is as follows: The following formula (2); Formula (1); Formula (2); in, and Represents the prediction time domain N within, minimizing the future N moment (from K+1 arrive K+N ) after the k+j Active power and reactive power at the moment; and Represented in k+j Active power reference value and reactive power reference value at the moment; and Represents the prediction time domain N within, minimizing the future N moment (fromK+1 arrive K+N ) after the k+j At this moment, in the second preset coordinate system d Axis and q The current components corresponding to the axes respectively; and Represented in k+j At this moment, d Axis and q The current reference value and current reference value corresponding to the axis respectively; λ is the weight coefficient; Representative is k+j At this moment, the converter controls the input voltage to d The amount of change on the axis.

[0009] According to the technical solution provided by this application, the expansion weight coefficient is a fixed coefficient; According to the grid operating conditions, the expanded weight coefficients of the first objective function and the second objective function are set and adjusted, specifically including: Setting fixed expansion weight coefficients for the first objective function and the second objective function respectively to obtain a first objective expansion function and a second objective expansion function; According to the grid operating condition requirements, the weights of the two fixed extended weight coefficients are adjusted; the grid operating condition requirements include at least: actual energy demand of the grid load, wind speed status and grid voltage status.

[0010] According to the technical solution provided by this application, the first target expansion function The second objective expansion function is as follows: The following formula (IV); Formula (III); Formula (IV); in, and are respectively represented as fixed expansion weight coefficients of active power and reactive power in the first objective expansion function; and are respectively expressed as the second objective expansion function d Axis and q Fixed expansion weight coefficient for shaft current shunting; According to the grid operating condition requirements, the weights of the two fixed extended weight coefficients are adjusted, specifically including: When the actual energy demand increases or the wind speed state fluctuates, a fixed expansion weight coefficient of the active power in the first objective expansion function is set. Fixed expansion weight factor greater than reactive power , and set the second objective expansion function d Fixed expansion weight coefficient for shaft current split Greater than q Fixed expansion weight coefficient for shaft current split ; When the grid voltage state is abnormal, the fixed expansion weight coefficient of the active power in the first objective expansion function is set. Fixed expansion weight factor smaller than reactive power , and set the second objective expansion function d Fixed expansion weight coefficient for shaft current split Less than q Fixed expansion weight coefficient for shaft current split .

[0011] According to the technical solution provided by this application, the expansion weight coefficient is a dynamic coefficient; According to the grid operating conditions, the extended weight coefficients of the third objective function and the fourth objective function are set and adjusted, specifically including: Setting dynamic expansion weight coefficients for the first objective function and the second objective function respectively, and weighting the dynamic expansion weight coefficients in the fourth objective function by proportional distribution to obtain a third objective expansion function and a fourth objective expansion function; According to the grid operating condition requirements, dynamic adjustment ranges of two dynamic expansion weight coefficients are set; the grid operating condition requirements include at least: grid load, wind speed fluctuation and grid voltage deviation.

[0012] According to the technical solution provided by this application, the third objective expansion function The fourth objective expansion function is as follows: The following formula (six); Formula (5); Formula (VI); in, and are respectively expressed as the dynamic expansion weight coefficients of active power and reactive power in the third objective expansion function; and Respectively expressed as the fourth objective expansion function d Axis and q Dynamic expansion weight coefficient of shaft current shunt.

[0013] According to the technical solution provided by this application, an extended weight coefficient mapping relationship is established between the two control layers to achieve dynamic fusion between the two control layers, specifically including: Establishing a mapping function including a fusion coefficient based on the dynamic expansion weight coefficients of the active power and reactive power and the dynamic expansion weight coefficients of the d-axis and q-axis current splits; According to the grid operating condition requirements, different fusion coefficients are set to adjust the control priorities of the high-level control and the low-level control.

[0014] In summary, the present technical solution specifically discloses a power prediction and control method of an MPC wind power converter based on hierarchical dynamic weight adjustment, including: constructing a current prediction model based on a mathematical model of the converter's output current in a first preset coordinate system; introducing an MPC control system into the current prediction model, and setting independent first and second objective functions for the two control layers of the MPC control system; setting and adjusting the extended weight coefficients of the first and second objective functions according to the grid operating conditions; establishing an extended weight coefficient mapping relationship between the two control layers to achieve dynamic fusion between the two control layers, thereby controlling the output power of the converter; wherein the first objective function is used to optimize the predicted power reference value output by the current prediction model in combination with the grid operating conditions, and the second objective function is used to track and control the actual current and voltage of the converter based on the optimized predicted power reference value; the predicted power reference value includes: an active power reference value and a reactive power reference value.

[0015] Traditional MPC control methods are usually unable to track and adjust active power P and reactive power Q independently, but wind power systems usually have strong nonlinearity, high coupling and complex operating conditions. This makes it difficult for traditional MPC to achieve the best control effect under different operating conditions. In this application, by setting the first objective function and the second objective function with extended weight coefficients for the two control layers of the MPC control system, the active power P and reactive power Q can be controlled according to the grid operating conditions by adjusting the extended weight coefficients. At the same time, dynamic fusion is achieved between the two control layers, and the control targets of the high and low control layers are coordinated to effectively avoid conflicts, thereby accurately controlling the output power of the converter. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Other features, objects and advantages of the present application will become more apparent upon reading the detailed description of non-limiting embodiments made with reference to the following drawings: Figure 1 The figure is a flow chart of a power prediction control method for an MPC wind power converter based on hierarchical dynamic weight adjustment.

[0017] Figure 2 Schematic diagram of the expanded flow of step S100 in a power prediction and control method for an MPC wind power converter based on hierarchical dynamic weight adjustment.

[0018] Figure 3 4 is a first expanded flow chart of step S300 in a power prediction and control method for an MPC wind power converter based on hierarchical dynamic weight adjustment.

[0019] Figure 4 4 is a second expanded flow chart of step S300 in an MPC wind power converter power prediction and control method based on hierarchical dynamic weight adjustment.

[0020] Figure 5 Schematic diagram of the hierarchical structure of the MPC control system.

[0021] Figure 6 is the global parameter in the objective function corresponding to the lower layer Schematic diagram of the influence curve of control input smoothness and the value range under different working conditions.

[0022] Figure 7 This is the weight mapping relationship diagram of the MPC high and low control layers.

[0023] Figure 8 It is the relationship curve between the weight of MPC high and low control layers and the change of fusion coefficient.

[0024] Figure 9 The fixed expansion weight coefficient and dynamic expansion weight coefficient of MPC are used in active power P Effect comparison chart in reference value tracking.

[0025] Figure 10 The fixed expansion weight coefficient and dynamic expansion weight coefficient of MPC in reactive power Q Effect comparison chart in reference value tracking. DETAILED DESCRIPTION

[0026] The present application will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the relevant invention and are not intended to limit the invention. It should also be noted that, for ease of description, only the portions relevant to the invention are shown in the accompanying drawings.

[0027] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0028] Example 1 In order to make the technical solutions of the embodiments of the present application clearer and easier to understand, the application background of the embodiments of the present application is introduced below.

[0029] Model predictive control (MPC) is an advanced control method based on system dynamics models. It achieves control objectives by predicting the future state of the system and optimizing the control input. MPC has been widely used in industrial control, particularly in power electronics and energy management. Compared with traditional feedback control methods, MPC can handle constraints in complex systems (such as voltage, current, and power limits) and optimize multi-objective control. For wind turbine converters, MPC can optimize the converter's control strategy in real time based on the dynamic characteristics of the wind turbine and the operating conditions of the power grid. MPC not only performs maximum power point tracking (MPPT) and power regulation, but also effectively improves system stability and reliability under complex conditions such as wind speed fluctuations and grid frequency changes.

[0030] The operating conditions of wind power systems are complex and changeable. For example, under light load, the demand for reactive power may be relatively small. However, under special circumstances such as heavy load or grid failure, it may be necessary to prioritize the stable supply of active power to maintain basic system operation, or to quickly adjust reactive power to stabilize voltage. Traditional MPC control methods are usually unable to track and adjust active power P and reactive power Q independently. Due to the strong nonlinearity, high coupling and complex operating conditions of wind power systems, traditional MPC has difficulty achieving optimal control effects under different operating conditions. In some scenarios, such as when the priority of voltage support and frequency regulation changes dynamically, the controller will not be able to respond quickly to grid demand, resulting in power deviation and reduced dynamic performance.

[0031] In view of this, the present application provides a hierarchical dynamic weight-adjusted MPC wind power converter power prediction and control method, whose target object can be a full-power wind power converter, and the control method includes: constructing a current prediction model based on a mathematical model of the converter's output current in a first preset coordinate system; introducing an MPC control system into the current prediction model, and setting a first objective function and a second objective function for the two control layers of the MPC control system respectively; setting and adjusting the extended weight coefficients of the first objective function and the second objective function according to the grid operating conditions; establishing an extended weight coefficient mapping relationship between the two control layers to achieve dynamic fusion between the two control layers, thereby controlling the output power of the converter; wherein the first objective function is used to optimize the predicted power reference value output by the current prediction model in combination with the grid operating conditions, and the second objective function is used to track and control the actual current and voltage of the converter based on the optimized predicted power reference value; the predicted power reference value includes: an active power reference value and a reactive power reference value.

[0032] Therefore, to overcome the limitations of traditional MPC control systems, this paper proposes a power prediction and control method for wind turbine converters based on hierarchical dynamic weight adjustment. Hierarchical dynamic weight MPC achieves flexible allocation and priority adjustment of power targets (active and reactive) by establishing a dynamic weight adjustment mechanism between the upper and lower layers. One control layer (the upper control layer or the upper layer) in the MPC control system is responsible for optimizing global objectives, such as optimizing the reference values and dynamic weights of the wind turbine converter's active power P and reactive power Q, and dynamically adjusting the priority weights of active and reactive power based on grid demand. Another control layer (the lower control layer or the lower layer) performs real-time control based on the reference values and set expanded weight coefficients provided by the upper layer, in conjunction with the current operating status, to ensure rapid tracking of active and reactive power while ensuring strict compliance with control input smoothness and system constraints. In addition, the present invention introduces a weight fusion mechanism, which combines the global objectives of the high-level and the local adaptive control of the low-level by setting mapping parameters with fusion coefficients in the high-level and low-level layers. Different fusion coefficients are set under different working conditions to dynamically adjust the control priorities of the high-level and low-level layers, thereby balancing the global objectives and local responses.

[0033] Specifically, please refer to Figure 1 The flowchart of the MPC wind power converter power prediction control method based on hierarchical dynamic weight adjustment provided by this embodiment includes the following steps: S100, constructing a current prediction model based on a mathematical model of the output current of the converter in a first preset coordinate system; In order to more accurately describe the dynamic characteristics of the converter output current, promptly detect potential unstable factors in the system, adjust the control strategy in advance, and ensure the stable operation of the power system, a current prediction model is constructed here based on the mathematical relationship presented by the mathematical model of the converter output current in a first preset coordinate system. This current prediction model can provide forward-looking information for the control strategy, helping the system to effectively improve the stability and reliability of the system under complex conditions such as wind speed fluctuations and grid frequency changes.

[0034] Specifically, see Figure 2 , the expansion of step S100 includes the following steps: S101. Establishing a three-phase voltage and current equation of the converter using inductance according to Kirchhoff's law to determine a mathematical model of the converter's output current in a first preset coordinate system. exist abc In the coordinate system, the three-phase winding of the converter is assumed to be A Mutually, B Harmony C Phase, the inductance of each phase winding is L f, the three-phase currents are i a 、 i b 、 i c , the three-phase grid voltages are U ia 、U ib 、U ic , the components of the three-phase grid voltage on each axis in the coordinate system are U Ida 、U Idc 、U Iac ; Then based on Kirchhoff's law, the three-phase voltage and current equation can be obtained as follows (7): Formula (VII); Next, in the output current model of the full power converter, the current dynamics is determined by the output voltage U ia and U iβ Therefore, the output current of the converter is in the first preset coordinate system ( αβ The mathematical model of the coordinate system is as follows: Formula (VIII); In this formula, i α and i β The three-phase bridge arms are αβ Output current axis component in the coordinate system, U Idα 、U Idβ The grid voltage is αβ Coordinate system α and β Axis component, U iα and U iβ It is the output voltage axis component of the three-phase bridge arm in the coordinate system.

[0035] S102, transforming the mathematical model into a second preset coordinate system to obtain transformation components corresponding to each axis of the second preset coordinate system; Furthermore, due to αβ In the coordinate system, the current cannot be decoupled and controlled. PQ Decoupling control, to achieve independent control of active power and reactive power, requires the use of Park transformation to transform the mathematical model into the second preset coordinate system ( dq coordinate system), converted into dq In the coordinate system d, q Axis component, please refer to the following formula (9): (Nine); S103 , performing discrete processing on the conversion components corresponding to each axis within a preset sampling period, and introducing a phase-locked loop to obtain a current prediction model.

[0036] By discretizing formula (9), and setting the sampling period to T s , we can get the following formula (X): Formula (10); In this formula, 、 yes k+1 time d Axis and q The predicted value of the shaft current, 、 yes k time d Axis and q The sampling value of the shaft current, is the grid angular frequency, 、 The grid voltage is dq Components in the coordinate system, 、 is the output voltage of the converter at the current moment dq Components in the coordinate system used to control and , and then control the power to meet the power demand of the system. From formula (10), we can see that the bridge arm inductor current d Axis component Not only affected by the grid connection point d Axis component , converter output voltage d Axis component is also affected by the bridge arm inductor current q Axis component The impact of grid current d 、 q In addition to being affected by the voltage across the inductor, the components are also affected by the coupling between each other. P Mainly depends on but also affected by The impact of × ); reactive power Q To depend on but also affected by The impact of × ). This coupling relationship means that adjusting or Will affect P and QTherefore, the goal of decoupling is to achieve and Independent control is provided to meet the requirements of wind power grid connection for independent regulation of active and reactive power.

[0037] In view of this, a phase-locked loop is introduced in the embodiment of the present application, which tracks the phase of the grid voltage in real time. θ and frequency , in power decoupling, the phase-locked loop ensures dq The coordinate system is synchronized with the grid voltage, so that the grid voltage is dq The coordinate system will satisfy the following characteristics: (1) =0, that is, the grid voltage q The axis component is 0; (2) , that is, the grid voltage d The axis component is close to the grid voltage amplitude.

[0038] exist dq In the coordinate system, the instantaneous active power P and reactive power Q It can be expressed as the following formula (11): Formula (XI): After the phase-locked loop is introduced, formula (11) is simplified to the following formula (12): Formula (12); It can be seen from formula (12) that the active power P Only with Related to irrelevant; reactive power Q Only with Related to This shows that in theory, after synchronization through the phase-locked loop P and Q Decoupling has been achieved mathematically, so in the end, the discretized dynamic current prediction model after the phase-locked loop is introduced, which is the final current prediction model. For details, see the following formula (13): Formula (XIII); Among them, and The current prediction at subsequent moments depends on the prediction value at the previous moment, and is calculated recursively step by step. The prediction model will and With control input and Dynamic association is performed.

[0039] It should be noted that although the phase-locked loop can theoretically achieve decoupling in the power formula, in the discretized current dynamic equation shown in formula (13), and The coupling term still exists. This is because the second preset coordinate system ( dq Coordinate system) is a synchronous rotating reference system. The rotation of the reference system will introduce cross terms, that is, and , these items lead to and Therefore, the embodiment of the present application introduces an MPC control system based on this, and compensates for the cross terms in the subsequent first objective function and second objective function to achieve dynamic decoupling.

[0040] So after step S100, there are the following steps: S200, introducing an MPC control system into the current prediction model, and setting a mutually independent first objective function and a second objective function for two control layers of the MPC control system; After the introduction of the MPC control system, the objective functions of each MPC layer will be designed according to the functions of the hierarchical structure. The basic idea of hierarchical control is to divide the control task into two parts: a high control layer (slow global optimization) and a low control layer (fast local tracking). Although each layer has an independent first objective function and a second objective function, they will cooperate with each other to achieve the overall optimization of the system. The first objective function is used to optimize the predicted power reference value output by the current prediction model in combination with the grid operating conditions. The second objective function is used to track and control the actual current and voltage of the converter based on the optimized predicted power reference value. The predicted power reference value includes: active power reference value and reactive power reference value.

[0041] The hierarchical control structure is as follows Figure 5 As shown, it can be clearly seen that the high-level control layer is responsible for global optimization and the low-level control layer is responsible for real-time tracking. The high-level power mapping module inputs the predicted power reference value into the high-level objective function, and then optimizes the active power reference value and the reactive power reference value to provide a reference for the low-level control layer. Therefore, the high-level control layer is responsible for global optimization, taking into account the system's operating status (such as wind speed fluctuations, grid demand, energy storage system) and the dynamics of external disturbances, and calculating the converter output current reference value. and , whose optimization goal is global power management, that is, active power P and reactive power Q The low-level control layer is responsible for quickly responding to the reference value calculated by the high-level control layer to achieve converter current tracking and output voltage control. Its main task is to accurately track the reference value to ensure the dynamic performance and stability of the system.

[0042] Specifically, considering that high-level control mainly focuses on the global optimization goal of the system, the first objective function corresponding to the high-level is designed The following formula (1): Formula (1); The first is to optimize maximum power point tracking (MPPT). Wind turbines typically need to operate in MPPT mode to maximize wind energy utilization. The second is to track the reactive power reference value, taking into account the grid's requirements for reactive power and voltage regulation.

[0043] At the same time, according to 、 And the current voltage sampling value, calculate dq Axis reference current, calculated and The following formula (14): Formula (XIV); Considering that the main task of the low-level control is to quickly track the predicted power reference value provided by the high-level control and ensure the dynamic response and stability of the system, the second objective function corresponding to the low-level control is designed. The following formula (2): Formula (2); in, and Represents the prediction time domain N within, minimizing the future N moment (from K+1 arrive K+N ) after the k+j Active power and reactive power at the moment; and Represented in k+j Active power reference value and reactive power reference value at the moment; and Represents the prediction time domain N within, minimizing the future N moment (from K+1 arrive K+N ) after the k+j At this moment, in the second preset coordinate system d Axis and q The current components corresponding to the axes respectively; and Represented in k+j At this moment, d Axis and q The current reference value and current reference value corresponding to the axis respectively; λis the weight coefficient used to balance the priority between the first term (current tracking error) and the second term (control input smoothness) of the second objective function; Representative is k+j At this moment, the converter controls the input voltage to d The amount of change on the axis.

[0044] Furthermore, in order to ensure that the objective function can be implemented in the actual system, the second objective function is Set the following constraints: Constraint 1:

[0045] Constraint 2:

[0046] Constraint 3:

[0047] Among them, constraint 1 ensures that the optimally calculated control input and Do not exceed the voltage output capacity of the inverter, and avoid the control signal exceeding the hardware limit, which may cause the system to fail. is the maximum output voltage of the converter; Constraint 2 is to avoid excessive input fluctuations, protect system stability and hardware equipment, and is the maximum rate of change of the control input; Constraint 3 is to ensure that the current does not exceed the rated value of the converter, avoid overcurrent and cause system damage, and meet the grid's regulatory requirements for current amplitude.

[0048] The first item in formula (2) is the main control target, which means that in the forecast time domain N within, minimizing the future N moment (from K+1 arrive K+N ) current and Relative to the reference value and The error makes and Accurately track reference values and ;because and is and Calculated, then track and Essentially controlling P and Q， Moreover, the objective function is to optimize the current tracking error. When calculating the current prediction value, the model will naturally include cross terms.

[0049] The second term in formula (2) is used to limit the change of the control input by penalizing the change of the control input. and In converter control, the control input is and , is the output voltage of the converter dq These voltage signals are applied to the filter or load through their modulation strategy. If the control input changes too quickly, it may cause active power P and reactive power Q Unnecessary fluctuations occur, affecting the stability of the power grid. and Represents the converter control input and The change in , that is, the following formula (15): Formula (XV); It needs to be explained that, Figure 6 As shown in the figure, if λ is large, the optimization process will pay more attention to the smoothness of the control input change; if λ is small, the optimization process will pay more attention to the current tracking accuracy, which may lead to more drastic changes in the control input. Figure 6 The value range of λ in three cases is given in the figure: the red area in the figure is the priority of fast response, within this range , the control input responds to the reference value more quickly but with lower smoothness, and is suitable for scenarios that require fast dynamic adjustment; Figure 6 The middle green area is the priority for stability. , the control input changes are smoother and the stability is higher, but a certain response speed may be sacrificed. It is suitable for scenarios with high requirements for smooth control; Figure 6 The middle yellow area is the equilibrium condition. , taking into account both response speed and smoothness, and is suitable for working conditions that need to balance the needs of both. Figure 6 , you can select a suitable value range of λ according to specific needs to find the best balance between fast response and stability.

[0050] At the same time, the second objective function It is a comprehensive performance indicator of the entire prediction time domain, taking into account the current error and control input changes at the current moment and in the future. The goal is to find a set of control input sequences and , so that the second objective function Minimize, and consider the system constraints during the optimization process; after the optimization is completed, extract the first value in the control input sequence and , which is applied to the converter; however, it should be noted that for future control input It is only a predicted value and will be recalculated when actually used. Finally, the above process is repeated in the next sampling period through rolling updates. Based on this, the above mechanism ensures that the control input at the current moment is the global optimal.

[0051] In summary, through the MPC control system, power control is finally achieved through the mutual cooperation of the two objective functions. From the above, the PQ control principle diagram based on hierarchical model prediction can be obtained as follows: Figure 7 As shown in the figure, after a series of online calculations and optimizations, the switching quantity is directly output to act on the converter.

[0052] However, due to the is the global weight, which is used to limit the control input and The rate of change is not targeted separately and Therefore, by adjusting , can affect the overall dynamic performance of the system, but cannot adjust the active power alone P and reactive power Q Therefore, in order to solve the above problems, in the embodiment of the present application, the first objective function and the second objective function of the high-level and low-level layers are respectively expanded, and different expansion weight coefficients are introduced to weight them respectively. In this way, by introducing the expansion weight coefficients at both the high-level and low-level layers, the system is comprehensively optimized from global power management to local dynamic control. The high-level layer is responsible for setting the global priority, and the low-level layer can accurately implement specific current control on this basis. That is, there is step S300: setting and adjusting the expansion weight coefficients of the first objective function and the second objective function according to the grid working condition requirements; Depending on the working conditions, for example, in scenarios where dynamic response speed and accuracy are not required, or where balanced resource allocation is required, the expansion weight coefficient can be set as a fixed coefficient, see Figure 3 At this time, step S300 specifically includes the following steps: S301a, setting fixed extension weight coefficients for the first objective function and the second objective function respectively, to obtain the first objective extension function and the second objective extension function; For high-rise buildings, considering the active power P and reactive power Q For different needs, weight coefficients can be added to these two items respectively. The first objective expansion function The following formula (3): Formula (III); in, and are respectively expressed as fixed expansion weight coefficients of active power and reactive power in the first objective expansion function. In other words, is the importance weight of active power tracking, The importance weight of reactive power tracking is adjusted by and , you can control the high-level optimization priority of active or reactive power.

[0053] For the lower layers, since the lower layers are responsible for current tracking, the weights can be directly applied to and The tracking error term is and Corresponding to P and Q , the second extended objective function The following formula (IV): Formula (IV); in, and are respectively expressed as the second objective expansion function d Axis and q Fixed expansion weight coefficient of shaft current shunt; in the first term, fixed expansion weight coefficient Adjustment Importance of tracking error, fixed expansion weight coefficient Adjustment The importance of tracking error.

[0054] S302a. Adjust the weights of the two fixed extended weight coefficients according to the grid operating condition requirements; the grid operating condition requirements include at least: the actual energy demand of the grid load, the wind speed status, and the grid voltage status.

[0055] Specifically, based on the above content, different terms in the two target expansion functions are involved in the adjustment 、 and 、 , which means that under different grid operating conditions, each fixed expansion weight coefficient also has a different setting method, so as to make the optimal control decision under each operating condition. Therefore, step S302 also includes the following steps: Step 1: When the actual energy demand increases or the wind speed fluctuates, set the fixed expansion weight coefficient of the active power in the first objective expansion function. Fixed expansion weight factor greater than reactive power , and set the second objective expansion function d Fixed expansion weight coefficient for shaft current split Greater thanq Fixed expansion weight coefficient for shaft current split ; In this case, since the main function of active power is to meet the actual energy demand of the grid load, when the load demand increases and the wind speed fluctuates, the adjustment goal of the fixed expansion weight coefficient is to prioritize smoothing the active power output, ensure the stability of the grid power transmission, and appropriately weaken the priority of reactive power.

[0056] So the following setting method: (1) Set the high-level objective function weight to > ; (2) Set the weight of the low-level objective function to > .

[0057] Step 2: When the grid voltage is abnormal, set the fixed expansion weight coefficient of the active power in the first objective expansion function. Fixed expansion weight factor smaller than reactive power , and set the second objective expansion function d Fixed expansion weight coefficient for shaft current split Less than q Fixed expansion weight coefficient for shaft current split .

[0058] In such a case, since the main function of reactive power is to regulate the voltage level, support the stable operation of the power grid and provide electromagnetic energy for inductive or capacitive loads, when the grid voltage is abnormal, the system needs to give priority to regulating reactive power.

[0059] So the following setting method: (1) Set the high-level objective function weight to > ; (2) Set the weight of the low-level objective function to > .

[0060] In addition, since the method of setting the expansion weight coefficient as a fixed coefficient is also applicable to the working conditions that require balanced allocation of resources, under the normal operation of the power grid, it is required P and Q The response dynamics are consistent, and the balanced control is consistent with and ,ensure P and Q Synchronous response, you can also set = , = .

[0061] In addition, although the above-mentioned design of fixed expansion weight coefficient is suitable for scenarios with low requirements for dynamic response speed and accuracy, or working conditions that require balanced resource allocation, due to the randomness and volatility of wind energy in actual scenarios, the MPC control system with fixed expansion weight coefficient cannot flexibly respond to changes in working conditions during operation. Therefore, in the embodiment of the present application, an MPC control system with dynamic weight adjustment is also introduced, that is, the expansion weight coefficient is a dynamic coefficient. Its advantage is that it can flexibly switch priorities and adapt to the needs of multiple working conditions, thereby giving priority to allocating resources to key targets, making resource utilization more efficient, and also improving response speed and accuracy, and giving priority to adjusting key needs; see Figure 4 At this time, step S300 specifically includes the following steps: S301b, setting dynamic expansion weight coefficients for the first objective function and the second objective function respectively, and weighting the dynamic expansion weight coefficients in the fourth objective function by proportional distribution to obtain a third objective expansion function and a fourth objective expansion function; Specifically, for the first objective function corresponding to the high level, the third objective expansion function of the following formula (5) can be designed: : Formula (5); in, and They are respectively expressed as the dynamic expansion weight coefficients of active power and reactive power in the third objective expansion function.

[0062] For the second objective function corresponding to the lower layer, the fourth objective expansion function can be designed as follows: : Formula (VI); in, and Respectively expressed as the fourth objective expansion function d Axis and q Dynamic expansion weight coefficient of shaft current shunt.

[0063] S302b. According to the grid operating condition requirements, set the dynamic adjustment ranges of the two dynamic expansion weight coefficients; the grid operating condition requirements include at least: grid load, wind speed fluctuation, and grid voltage deviation.

[0064] Since the third objective expansion function and the fourth objective expansion function It is used to flexibly respond to changes in operating conditions, so for the third objective expansion function in and The following update rules can be followed: (1) Wind speed fluctuation Wind speed fluctuations will cause the active power of the wind power system to P Output fluctuations require an increased weight to smooth the active power output, and the update rule is as follows: Formula (XVI); Among them, in formula (16), is the basic weight of active power, is the wind speed fluctuation adjustment coefficient, is the standard deviation of wind speed fluctuation measured by the wind speed sensor.

[0065] (2) Increased grid load When the grid load increases, the active power demand When the power increases, it is necessary to output more active power first, increase The weight of , its update rule is as follows formula (17): Formula (XVII); Among them, in formula (17), is the load demand adjustment coefficient when the grid load increases, Add the amount for load demand.

[0066] (3) Grid voltage deviation When the grid voltage Deviation from reference value When the reactive power Q Adjust the voltage to increase The weight of , its update rule is: Formula (18) Among them, in formula (18), is the basic weight of reactive power, is the adjustment coefficient when the grid voltage deviates, is the voltage deviation.

[0067] Combining the above update rules, we can get the comprehensive update rule shown in the following formula (19): Formula (19).

[0068] Under such an update rule, the formula (19) and The dynamic adjustment range should follow the following setting principles: (1) Update rule parameter settings It should be set according to the overall goal of the system. If the system is based on power generation economy, that is, giving priority to ensuring active power output, then Set the range to If the system needs to take reactive power into account, that is, grid voltage stability, then Set the range to .

[0069] Should fluctuate according to wind speed Set the degree of influence on system operation. If the wind speed fluctuation has a significant impact on the active power output, then Set the range to ; If the system is not sensitive to wind speed fluctuations, then Set the range to .

[0070] Should change according to load The sensitivity setting for the system active power demand. If the load demand has a greater pressure on the active power, then Set the range to ; If the load change has little pressure on the system, then Set the range to .

[0071] (2) Update rule parameter settings According to the system operation goals, Should be .

[0072] It should be set according to the impact of grid voltage fluctuation on the system. If the grid voltage fluctuation is large or the system is connected to a weak grid, Set the range to If the grid voltage is relatively stable or not sensitive to voltage fluctuations, then Set the range to .

[0073] After introducing the third target expansion function After setting the principle, the fourth objective expansion function is The setting principles are as follows: First, in the embodiment of the present application, in the fourth target expansion function The objective function designed here is Do not participate in dynamic expansion weight coefficient adjustment because the core goal of dynamic weight adjustment is to dynamically change according to the operating conditions. and , thereby giving priority to regulating active power P or reactive powerQ . is the weight parameter of the control input smoothing term, which affects the converter output voltage u d and u q The magnitude of the change (smoothness), but it does not directly affect i d and i q Therefore, in the design of dynamic weight adjustment, the focus is on and Dynamic adjustment without involving dynamic adjustment.

[0074] Secondly, dynamic weight adjustment requires the use of running state perception and dynamic weight calculation function to dynamically allocate dynamic expansion weight coefficients in real time. and , in order to achieve the active power P and reactive power Q Specifically, the following is a detailed implementation of the allocation of low-level dynamic expansion weight coefficients: (1) Operating status perception and acquisition of main parameters: Perceiving the operating status is the core of the dynamic weight adjustment method, which relies on real-time monitoring and analysis of key parameters in the wind turbine converter and grid operating environment.

[0075] A. Grid frequency deviation

[0076] The grid frequency deviation is a direct reflection of the active power balance in the grid: an increase in frequency indicates an excess of active power in the grid, while a decrease in frequency indicates an insufficient active power in the grid. Therefore, the grid frequency is measured in real time through a phase-locked loop, and the grid frequency deviation is calculated using the following formula (20): Formula (XX); in, f n It is the rated frequency of the power grid, usually 50Hz.

[0077] B. Grid voltage deviation

[0078] The grid voltage deviation reflects the reactive power balance in the grid: voltage increase usually indicates excess reactive power, while voltage decrease usually indicates insufficient reactive power. The grid voltage is measured in real time by the voltage sensor of the converter. , calculate the voltage deviation using formula (XXI): Formula (XXI); C. Active power reference value deviation

[0079] Active power reference value deviation Indicates the active power of the wind farm P Whether the real-time dispatching instructions of the power grid are met: If P < , active power needs to be increased if P > , need to limit active power, active power reference value deviation Calculated according to the following formula (XXII): Formula (XXII); in, is the active power reference value, Calculated in real time by the converter output current and grid voltage.

[0080] D. Active power reference value deviation

[0081] Active power reference value deviation Represents the reactive power of the wind farm Q Whether the real-time dispatching instructions of the power grid are met: If Q < , reactive power needs to be increased if Q > , it is necessary to limit the reactive power and active power reference value deviation Calculated according to the following formula (XXIII): Formula (XXIII); in, is the reactive power reference value, Calculated in real time by the converter output current and grid voltage.

[0082] After that, the proportional distribution method is also used in the embodiment of the present application to dynamically allocate weights. This method is simple and easy to implement. The weight distribution is determined by the size of the deviation, which is intuitive and easy to understand. At the same time, the weight is dynamically adjusted according to the deviation to adapt to different operating conditions. The amount of calculation is small, and the calculation or optimization of the weight coefficient is avoided, reducing the burden of online calculation. Specifically, the present invention here uses weighted fusion to guide the dynamic adjustment of high-level weights to low-level targets, while retaining the adaptability of low-level weights. This fusion requires the introduction of high-level information in weight distribution and ensures the consistency of high-level and low-level targets.

[0083] It can be seen that the dynamic weight of the lower layer is proportionally distributed as follows (24): Formula (XXIV); and It is based on the weights distributed proportionally in the lower layers. It should be noted that if these deviations are used directly for weight calculation, the variables with large deviation values (such as P or Q ) will play a dominant role in weight distribution, while variables with smaller deviations (such as or ) may be ignored, and the dimensional differences between different deviations lead to the calculation results being unable to reflect the actual regulation requirements. Therefore, the deviations in the above formula are normalized to obtain the following formula (XXV): Formula (XXV); Substituting the normalized deviation into the weight distribution formula, we get the following formula (XXVI): Formula (XXVI); The above formula (XXVI) shows the weight and Will follow the deviation 、 、 and Adjust according to the dynamic changes of the system.

[0084] In order to enable the two high and low control layers to achieve global and local balance, and to coordinate the control objectives of the high and low layers to avoid conflicts between the two, the embodiment of the present application also introduces a fusion coefficient to make the combination of the high-level weight and the low-level weight more flexible, that is, after step S300, the following steps are further included: S400, establishing an extended weight coefficient mapping relationship between the two control layers to achieve dynamic fusion between the two control layers, thereby controlling the output power of the converter; Specifically, see Figure 7 ,Should Figure 7 Shows the mapping relationship between high-level weights and low-level weights and the fusion coefficient The role of high-level weight and It is the output of optimization, used to guide the low-level weight distribution; combined with the fusion coefficient through the mapping function , to achieve dynamic combination of high-level weights and low-level weights; fusion coefficient Adjust the influence ratio of high-level and low-level weights, low-level weights and , combining the global optimization of high-level and its own local needs, Figure 7 With fusion coefficient The mapping function is shown in the following formula (XXVII): Formula (XXVII); Among them, in formula (XXVII) As the fusion coefficient of the high-level and low-level weights, it is used to adjust the relative influence of the high-level and low-level layers. Its setting range is .

[0085] like Figure 8 The figure shows the relationship curve of high-level and low-level weight mapping, which shows the mapping relationship between high-level weight and low-level weight, and reflects the influence of fusion coefficient on the relationship between the two. and , reflecting the high-level target allocation priority; the red solid line represents the low-level weight and , reflecting the low-level priority of the fusion result based on the high-level weight. At the same time, the fusion coefficient Control the degree of integration of high-level and low-level weights. When completely dependent on the top, The decision is entirely made by the lower-level weights. Figure 8 The green points in the middle show different fusion coefficients The value of and the corresponding low-level weights are adjusted by The mapping relationship between high and low layer weights can be flexibly changed.

[0086] Specifically, due to and Determine the objective function i d and i q The importance of the error term is adjusted by and , we can pay more attention to minimizing a certain error, thereby making the active power P and reactive power Q The adjustments can be made independently.

[0087] Through the above steps, the weight of the high-level and Directly participate in the allocation of low-level weights to ensure that high-level goals guide low-level control. , you can set different The control priorities of the high-level and low-level layers are changed, and the low-level weights retain the characteristics of deviation-based distribution, which is still real-time and can quickly respond to dynamic changes in the system without completely losing the guiding role of the high-level layers.

[0088] Based on the above and Figure 9 and Figure 10It can be seen that the MPC wind power converter power prediction control method based on hierarchical dynamic weight adjustment proposed in this application has important significance in the field of power prediction control of full-power wind power converters. Specifically, Figure 9 and Figure 10 As shown, the fixed expansion weight coefficient and the dynamic expansion weight coefficient are shown in P and Q Performance differences in tracking, where Figure 9 For active power tracking comparison, the black dotted line indicates P ref , the red curve represents the tracking result under fixed weight P , we can see that there is lag and deviation; the blue curve represents the tracking result under dynamic weight P , can follow the reference value more accurately, reducing hysteresis and deviation; Figure 10 For reactive power tracking comparison, the black dotted line indicates Q ref , the red curve represents the tracking result under fixed weight Q , the tracking error is large; the blue curve represents the tracking result under dynamic weight Q , which in contrast can better adapt to changes in reference values.

[0089] By dynamically adjusting weight distribution, this control method enables flexible regulation of active and reactive power, significantly improving power tracking performance and the system's dynamic response capabilities. Its hierarchical optimization architecture provides technical support for wind power control under complex operating conditions and provides an important guarantee for the development of smart grids with a high proportion of wind power integration. In the future, as wind power continues to expand, the hierarchical dynamic weight MPC method will play an increasingly important role, providing a solid technical foundation for the stable operation of wind farms and the security and reliability of power grids.

[0090] The above description is merely a preferred embodiment of the present application and an illustration of the technical principles employed. Those skilled in the art should understand that the scope of the invention herein is not limited to the technical solutions formed by the specific combination of the above-mentioned technical features, but also encompasses other technical solutions formed by any combination of the above-mentioned technical features or their equivalents without departing from the inventive concept. For example, a technical solution formed by replacing the above-mentioned features with (but not limited to) technical features having similar functions disclosed in this application.

Claims

1. A power prediction and control method for an MPC wind power converter based on hierarchical dynamic weight adjustment, characterized in that: The steps include: Constructing a current prediction model based on a mathematical model of the output current of the converter in a first preset coordinate system; Introducing an MPC control system into the current prediction model, and setting mutually independent first and second objective functions for two control layers of the MPC control system; Setting and adjusting the extended weight coefficients of the first objective function and the second objective function according to the grid operating conditions; Establishing an extended weight coefficient mapping relationship between the two control layers to achieve dynamic fusion between the two control layers, thereby controlling the output power of the converter; Among them, the first objective function is used to optimize the predicted power reference value output by the current prediction model in combination with the grid operating conditions, and the second objective function is used to track and control the actual current and voltage of the converter based on the optimized predicted power reference value; the predicted power reference value includes: active power reference value and reactive power reference value.

2. The MPC wind power converter power prediction and control method based on hierarchical dynamic weight adjustment according to claim 1 is characterized in that: Based on the mathematical model of the output current of the converter in the first preset coordinate system, a current prediction model is constructed, specifically including: Establishing a three-phase voltage and current equation of the converter through inductance according to Kirchhoff's law to determine a mathematical model of the output current of the converter in the first preset coordinate system; Transforming the mathematical model into a second preset coordinate system to obtain transformation components corresponding to each axis of the second preset coordinate system; The conversion components corresponding to each axis are discretely processed within a preset sampling period, and a phase-locked loop is introduced to obtain the current prediction model.

3. The MPC wind power converter power prediction and control method based on hierarchical dynamic weight adjustment according to claim 1 is characterized in that: The control layer includes: a high control layer and a low control layer, wherein the high control layer and the low control layer correspond to the first objective function and the second objective function which are independent of each other respectively; The first objective function As shown in formula (1), the second objective function The following formula (2); Formula (1); Formula (2); in, and Represents the prediction time domain N within, minimizing the future N moment (from K+1 arrive K+ N ) after the k+j Active power and reactive power at the moment; and Represented in k+j Active power reference value and reactive power reference value at the moment; and Represents the prediction time domain N within, minimizing the future N moment (from K+1 arrive K+N ) after the k+j At this moment, in the second preset coordinate system d Axis and q The current components corresponding to the axes respectively; and Represented in k+ j At this moment, d Axis and q The current reference value and current reference value corresponding to the axis respectively; λ is the weight coefficient; Representative is k+j At this moment, the converter controls the input voltage to d The amount of change on the axis.

4. The MPC wind power converter power prediction and control method based on hierarchical dynamic weight adjustment according to claim 3 is characterized in that: The expansion weight coefficient is a fixed coefficient; According to the grid operating conditions, the expanded weight coefficients of the first objective function and the second objective function are set and adjusted, specifically including: Setting fixed expansion weight coefficients for the first objective function and the second objective function respectively to obtain a first objective expansion function and a second objective expansion function; According to the grid operating condition requirements, the weights of the two fixed extended weight coefficients are adjusted; the grid operating condition requirements include at least: actual energy demand of the grid load, wind speed status and grid voltage status.

5. The MPC wind power converter power prediction and control method based on hierarchical dynamic weight adjustment according to claim 4 is characterized in that: The first target expansion function The second objective expansion function is as follows: The following formula (IV); Formula (III); Formula (IV); in, and are respectively represented as fixed expansion weight coefficients of active power and reactive power in the first objective expansion function; and are respectively expressed as the second objective expansion function d Axis and q Fixed expansion weight coefficient for shaft current shunting; According to the grid operating condition requirements, the weights of the two fixed extended weight coefficients are adjusted, specifically including: When the actual energy demand increases or the wind speed state fluctuates, a fixed expansion weight coefficient of the active power in the first objective expansion function is set. Fixed expansion weight factor greater than reactive power , and set the second objective expansion function d Fixed expansion weight coefficient for shaft current split Greater than q Fixed expansion weight coefficient for shaft current split ; When the grid voltage state is abnormal, the fixed expansion weight coefficient of the active power in the first objective expansion function is set. Fixed expansion weight factor smaller than reactive power , and set the second objective expansion function d Fixed expansion weight coefficient for shaft current split Less than q Fixed expansion weight coefficient for shaft current split .

6. The MPC wind power converter power prediction and control method based on hierarchical dynamic weight adjustment according to claim 3 is characterized in that: The expansion weight coefficient is a dynamic coefficient; According to the grid operating conditions, the extended weight coefficients of the third objective function and the fourth objective function are set and adjusted, specifically including: Setting dynamic expansion weight coefficients for the first objective function and the second objective function respectively, and weighting the dynamic expansion weight coefficients in the fourth objective function by proportional distribution to obtain a third objective expansion function and a fourth objective expansion function; According to the grid operating condition requirements, dynamic adjustment ranges of two dynamic expansion weight coefficients are set; the grid operating condition requirements include at least: grid load, wind speed fluctuation and grid voltage deviation.

7. The MPC wind power converter power prediction and control method based on hierarchical dynamic weight adjustment according to claim 6 is characterized in that: The third objective expansion function The fourth objective expansion function is as follows: The following formula (six); Formula (5); Formula (VI); in, and are respectively expressed as the dynamic expansion weight coefficients of active power and reactive power in the third objective expansion function; and Respectively expressed as the fourth objective expansion function d Axis and q Dynamic expansion weight coefficient of shaft current shunt.

8. The MPC wind power converter power prediction and control method based on hierarchical dynamic weight adjustment according to claim 6 is characterized in that: Establishing an extended weight coefficient mapping relationship between the two control layers to achieve dynamic fusion between the two control layers specifically includes: Establishing a mapping function including a fusion coefficient based on the dynamic expansion weight coefficients of the active power and reactive power and the dynamic expansion weight coefficients of the d-axis and q-axis current splits; According to the grid operating condition requirements, different fusion coefficients are set to adjust the control priorities of the high-level control and the low-level control.

Images