A double-layer control method and device for wind power station participating in primary frequency modulation of power grid

Abstract

The present invention discloses a double-layer control method and device for a wind-storage power station to participate in primary frequency regulation of a power grid, and relates to the technical field of power system operation and control. The method comprises obtaining the frequency of a grid-connected point of a wind-storage power station, and performing frequency regulation according to a control moment when the frequency of the grid-connected point of the wind-storage power station is outside a normal range. When the control moment is a model predictive control (MPC) moment, a centralized model predictive control (MPC) algorithm is used to perform primary frequency regulation control. When the control moment is a distributed adaptive control moment, a station-level model predictive control algorithm is used to perform distributed adaptive control. The device comprises a wind-storage power station controller, a wind turbine controller, and an energy storage controller, as well as three program modules, namely a frequency regulation moment determination module, a centralized primary frequency regulation control module, and a distributed adaptive control module. The method achieves high efficiency and good effect in the wind-storage power station's participation in primary frequency regulation of a power grid through the double-layer control steps of centralized model predictive control and distributed adaptive control.

Description

Technical Field

[0001] The present invention relates to the technical field of power system operation and control, and in particular to a double-layer control method and device for a wind power station to participate in primary frequency modulation of a power grid. Background Art

[0002] Faced with the dual pressures of energy and the environment, the development and utilization of renewable energy remains a crucial path to addressing current challenges. Among renewable energy generation technologies, wind power, due to its inexhaustible and environmentally friendly nature, significantly contributes to alleviating environmental and energy pressures. Consequently, the rapid growth of wind power in recent years has significantly altered the structure and operation of traditional power systems. An increasing number of regions will face challenges associated with large-scale wind power integration. The inherent uncertainty of wind resources can negatively impact overall system stability. Furthermore, to maximize economic efficiency, most wind turbines operate in Maximum Power Point Tracking (MPPT) mode. In this mode, fluctuations in wind resources can significantly impact the system and even compromise safe operation. Furthermore, wind turbines operate as asynchronous motors, connected to the grid via rectifiers and inverters. This decouples rotor speed from system frequency, resulting in relatively low system inertia and potentially leading to frequency control issues.

[0003] At present, most wind farms use doubly fed wind turbines for power generation and permanent magnet direct drive wind turbines. The research object of this application is permanent magnet direct drive wind turbines. The kinetic energy stored in the blades of permanent magnet direct drive wind turbines and the flexibility of converter PQ control make it possible for wind power to participate in frequency regulation. By adjusting the power electronic devices on the rotor side of the permanent magnet direct drive wind turbine, its active power output can be quickly adjusted, and the kinetic energy stored in the wind turbine itself can be used to provide fast primary frequency regulation services. At the same time, in the power system, energy storage equipment can temporarily store excess electrical energy in the power grid, and convert this stored energy into electrical energy when necessary and release it back to the power grid, thereby achieving flexible regulation of the power grid. Due to the flexible charging and discharging characteristics of energy storage equipment, energy storage equipment has been built on a large scale in the power grid. In terms of assisting new energy stations to participate in the primary frequency regulation of the power system, energy storage equipment has a large application prospect and will become an important part of the new generation of smart grids.

[0004] Currently, most control methods for wind-storage power plants participating in power system primary frequency regulation focus solely on the "standalone" state of wind turbines and energy storage batteries. Specifically, individual wind turbines and energy storage batteries independently provide simulated inertia and droop characteristics based on their measured local frequencies. However, in a power system, wind-storage power plants should respond to system frequency fluctuations as a whole. Especially in the context of power market reform, the overall output power of wind-storage power plants is a criterion for their performance and settlement as market participants. However, if wind-storage power plants are considered to respond to frequency fluctuations as a whole, power allocation between wind and energy storage within the power plant becomes a challenge. Centralized optimization control methods are generally employed to address this issue, and existing methods both domestically and internationally fall into this category. However, due to the large number of devices within large-scale wind-storage power plants and their widespread geographical distribution, centralized control requires long communication and computation times, resulting in significant latency. Furthermore, conventional centralized control methods fail to account for electromechanical transients during wind turbine frequency regulation. Model predictive control (MPC) approaches rely heavily on the wind turbine model and prediction accuracy. Large errors in predicting wind turbine status within a station can lead to irrational power distribution between turbines, and in severe cases, cause turbines to disconnect from the grid. This can also lead to poor frequency characteristics of wind-storage power plants, large system frequency fluctuations, high requirements for energy storage battery power capacity, and high frequency regulation costs. Furthermore, centralized model maintenance and optimization calculations are time-consuming and unsuitable for primary frequency regulation, which requires fast response times.

[0005] The authorization announcement number is CN106712058B, and its title is "Coordinated Control Method for Doubly Fed Wind Turbine Wind Farm Participating in Power System Primary Frequency Regulation," hereinafter referred to as Reference Document 1. The present application differs from Reference Document 1 in that Reference Document 1 only adopts a distributed control method, which is a single-layer device-level wind farm frequency regulation control architecture, while the present application combines centralized model predictive control with distributed adaptive control, adopting a two-layer control architecture that combines station-level control and device-level control.

[0006] Authorization announcement number CN106532746B, entitled "A Control System and Implementation Method for Wind Farm Participating in Primary Frequency Regulation," hereinafter referred to as Reference Document 2. The present application differs from Reference Document 2 in that Reference Document 2 only addresses wind farm participation in primary frequency regulation, employing a site-level control strategy based on wind turbine speed evaluation, without controlling power distribution between wind turbines. This application, however, does control power distribution between wind turbines.

[0007] The authorization announcement number is CN110048440B, and its name is "A control method and model for wind turbines participating in the primary frequency regulation of the power grid", hereinafter referred to as Comparative Document 3. Compared with Comparative Document 3, the difference between this application and Comparative Document 3 is that Comparative Document 3 only adopts a distributed control method for wind farms participating in the primary frequency regulation of the power system, which is a single-layer wind farm frequency regulation control architecture at the device level. It only considers the operating status of a single wind turbine in the station and does not consider the frequency characteristics of the entire wind farm. However, this application combines centralized model predictive control with distributed adaptive control, adopts a two-layer control architecture that combines station-level control and device-level control, and considers the frequency characteristics of the entire wind farm.

[0008] Authorization announcement number is CN112636374B, entitled "Primary Frequency Regulation and Virtual Inertia Response Control Method and Apparatus for Wind Farms," ​​hereinafter referred to as Reference 4. This application differs from Reference 4 in that Reference 4 addresses a site-level control strategy for wind farms with flywheel energy storage participating in grid primary frequency regulation, and does not control power distribution between wind turbines, whereas this application does control power distribution between wind turbines.

[0009] The authorization announcement number is CN108599234B, and its name is "Virtual Inertia and Primary Frequency Regulation Control Method for Wind Turbine Generators", hereinafter referred to as Comparative Document 5. The difference between this application and Comparative Document 5 is that Comparative Document 5 adopts a distributed control strategy based on virtual inertia control for wind farm participation in frequency regulation. It is a device-level single-layer wind farm frequency regulation control architecture, which only considers the operating status of a single wind turbine in the station and does not consider the frequency characteristics of the entire wind farm. The present application combines centralized model predictive control with distributed adaptive control, adopts a two-layer control architecture that combines station-level control and device-level control, and considers the frequency characteristics of the entire wind farm.

[0010] Existing technical problems and considerations:

[0011] At present, among the existing methods for controlling the primary frequency regulation of wind-storage power stations participating in the power grid, the centralized control method is more dependent on the wind-storage power station's prediction accuracy and model accuracy for wind turbines. Once the prediction error and model error are large, the control result will be unreasonable, and may even cause problems such as wind turbines being disconnected from the grid. Moreover, the centralized method has a long communication time, and the wind-storage power station responds slowly to frequency changes at the grid connection point. In addition, in the traditional distributed control method for frequency regulation of wind-storage power stations, although its response speed is fast, it cannot guarantee the overall frequency response of the wind-storage power station, and cannot fully tap the frequency regulation capability of the wind-storage power station, resulting in poor frequency stability of the power grid. In response to the defects of the above-mentioned existing control technologies, this application proposes a two-layer control method to fully tap the frequency regulation capability of the wind-storage power station and ensure the frequency stability of the power system. Summary of the Invention

[0012] The technical problem to be solved by the present invention is to provide a double-layer control method and device for a wind power station to participate in the primary frequency modulation of the power grid, so as to solve the technical problem that the wind power station has low efficiency and poor effect in participating in the primary frequency modulation of the power grid.

[0013] In order to solve the above technical problems, the technical solution adopted by the present invention is: a two-layer control method for a wind-storage power station to participate in the primary frequency regulation of the power grid, obtaining the grid-connected point frequency of the wind-storage power station, and performing frequency regulation according to the control time when the grid-connected point frequency of the wind-storage power station is outside the normal range. The normal range is 49.9Hz~50.1Hz. When the control time is the model predictive control MPC time, a centralized model predictive control MPC algorithm is used for primary frequency regulation control. When the control time is the distributed adaptive control time, a station-level model predictive control algorithm is used for distributed adaptive control.

[0014] A further technical solution is: specifically including the following steps, S1 determines the frequency regulation time, obtains the frequency of the grid connection point of the wind-storage power station, and performs frequency regulation according to the control time when the frequency of the grid connection point of the wind-storage power station is not within the normal range. When the control time is the model predictive control MPC time, execute step S2, and when the control time is the distributed adaptive control time, execute step S3; S2 centralized primary frequency regulation control, obtains the virtual droop coefficient and virtual inertia coefficient of the wind turbine and controls the output power of the wind turbine during the primary frequency regulation period, obtains the virtual droop coefficient and virtual inertia coefficient of the energy storage battery and controls the output power of the energy storage battery during the primary frequency regulation period; S3 distributed adaptive control, the distributed adaptive control time is between the times when the two site-level model predictive control MPCs issue instructions, and the site-level model predictive control algorithm is used to perform distributed adaptive control, including the steps of wind turbine error adaptive control and energy storage battery adaptive control.

[0015] A further technical solution is that the step of the fan error adaptive control includes the following steps: obtaining the fan speed prediction value and the fan mechanical power prediction value, calculating the prediction error coefficient η e,i , calculate the power of each fan at time n under error adaptive control Then the error adaptive control of each fan is completed;

[0016]

[0017] In formula (28), η e,i is the prediction error coefficient of wind turbine i; ω mea,i is the actual measured value of the mechanical power of the i-th fan, in rad / s; i is the i-th fan; ω min is the minimum speed of the fan, in rad / s; ω mpc,iis the predicted speed of the i-th fan, in rad / s;

[0018]

[0019] In formula (29), is the power of the i-th wind turbine at time n under error adaptive control, in MW; is the virtual droop coefficient of the i-th wind turbine, in MW / Hz; Δf(n) is the frequency deviation at time n, in Hz; is the virtual inertia coefficient of the i-th wind turbine, in MW*s / Hz; f(n) is the system frequency at time n, in Hz; f(n-1) is the system frequency at time n-1, in Hz; t0 is the starting time of the frequency modulation phase, that is, the initial time when the frequency is in the abnormal range; P e,i (t0) is the power of the i-th wind turbine at time t0, in MW; P m,mpc,i is the predicted value of the mechanical power of the i-th wind turbine, in MW; P m,mea,i is the actual measured value of the mechanical power of the i-th wind turbine, in MW.

[0020] A further technical solution is to obtain the overall output coefficient η of the wind power station p , combined with the overall output coefficient η of the wind storage power station p and the prediction error coefficient η of each wind turbine e,i , calculate the active power of each wind turbine at time n under error adaptive control Then the error adaptive control of each fan is completed;

[0021]

[0022] In formula (30), η p is the overall output coefficient of the wind-storage power station; K f is the virtual droop coefficient of the power frequency of the wind power station, in MW / Hz; K in is the virtual inertia coefficient of the wind power station power frequency, in MW*s / Hz; f(n-2) is the system frequency at time n-2, in Hz; T is the MPC sampling time, set to 1s; P e,i (t0) is the initial power of the i-th wind turbine, in MW; is the sum of the initial power of the wind turbines, in MW; is the power of the tie line at the wind turbine of the wind storage power station at time n-1, in MW;

[0023]

[0024] In formula (31), is the active power of the i-th wind turbine at time n under error adaptive control, in MW; P m,i is the mechanical power of the i-th wind turbine in the wind power station, in MW.

[0025] A further technical solution is that the distributed adaptive control using the station-level model predictive control algorithm also includes a fan speed adaptive control step located after the fan error adaptive control step, and the fan speed adaptive control step includes the following steps: calculating the fan speed adaptive coefficient η ω,i , combined with each fan speed adaptive coefficient η ω,i and the active power of the corresponding wind turbine at time n under error adaptive control Calculate the output power of each wind turbine at time n Then the adaptive control of the speed of each fan is completed;

[0026]

[0027] In formula (32), η ω,i is the fan speed adaptive coefficient; ω i represents the rotation speed of the i-th wind turbine blade in the wind power station, in rad / s; ω' min is the minimum warning value of the fan rotor speed, in rad / s; ω max is the maximum speed of the fan, in rad / s; ω' max is the maximum warning value of the fan rotor speed, in rad / s; f is the system frequency, in Hz; f n is the rated frequency of the system frequency, in Hz;

[0028]

[0029] In formula (33), is the output power of the i-th wind turbine at time n, in MW.

[0030] A further technical solution is that the step of adaptive control of the energy storage battery includes the following steps: obtaining the virtual droop coefficient and virtual inertia coefficient of the energy storage battery and the grid connection point frequency of the wind storage power station, and calculating the reference power P of the wind storage power station. ref , calculate the energy storage battery power P ESS , and then complete the adaptive control of energy storage batteries;

[0031]

[0032] In formula (34), P PCC The power of the interconnecting line at the wind turbine of the wind storage power station, in MW; is the frequency change rate.

[0033] A double-layer control device for a wind-storage power station participating in the primary frequency modulation of a power grid includes a wind-storage power station controller, a wind turbine controller, an energy storage controller, and three program modules: a frequency modulation moment determination module, a centralized primary frequency modulation control module, and a distributed adaptive control module. The wind-storage power station controller is connected to and communicates with the wind turbine controller and the energy storage controller respectively. The frequency modulation moment determination module is used for the wind-storage power station controller to obtain the grid-connected frequency of the wind-storage power station. When the grid-connected frequency of the wind-storage power station is outside a normal range, frequency modulation is performed according to the control moment. When the control moment is a model predictive control (MPC) moment, the centralized primary frequency modulation control module is executed, and the wind-storage power station controller uses a centralized model predictive control (MPC) algorithm to perform primary frequency modulation control through the wind turbine controller and the wind-storage power station controller. When the control moment is a distributed adaptive control moment, the distributed adaptive control module is executed, and the wind-storage power station controller performs distributed adaptive control through a site-level model predictive control algorithm.

[0034] A further technical solution is: a centralized primary frequency regulation control module is used for the wind power station controller to obtain the virtual droop coefficient and virtual inertia coefficient of the wind turbine, which are sent to the wind turbine controller and control the output power of the wind turbine during the primary frequency regulation period; the wind power station controller obtains the virtual droop coefficient and virtual inertia coefficient of the energy storage battery, which are sent to the energy storage controller and control the output power of the energy storage battery during the primary frequency regulation period; a distributed adaptive control module is used for each wind turbine controller to obtain the wind turbine speed prediction value and wind turbine mechanical power prediction value sent by the wind power station controller, and calculate the prediction error coefficient η of each wind turbine e,i , calculate the power of each fan at time n under error adaptive control Each wind turbine controller obtains the overall output coefficient η of the wind storage power station sent by the wind storage power station controller p , combined with the overall output coefficient η of the wind storage power station p and the prediction error coefficient η of each wind turbine e,i , calculate the active power of each wind turbine at time n under error adaptive control Then the error adaptive control of each fan is completed.

[0035] A further technical solution is that the distributed adaptive control module is also used for each fan controller to calculate the corresponding fan speed adaptive coefficient η ω,i , combined with each fan speed adaptive coefficient η ω,i and the active power of the corresponding wind turbine at time n under error adaptive control Calculate the output power of each wind turbine at time n Then, the adaptive control of the speed of each wind turbine is completed; the energy storage controller obtains the virtual droop coefficient and virtual inertia coefficient of the energy storage battery sent by the wind storage power station and the grid connection point frequency of the wind storage power station, and calculates the reference power P of the wind storage power station. ref , calculate the energy storage battery power P ESS , and then complete the adaptive control of the energy storage battery.

[0036] A two-layer control device for a wind-storage power station participating in the primary frequency regulation of a power grid includes a computer-readable storage medium, the computer-readable storage medium stores a computer program, the computer program includes the above-mentioned frequency regulation moment determination module, the centralized primary frequency regulation control module and the distributed adaptive control module, and when the computer program is executed by a processor, the corresponding steps in the above-mentioned method are implemented.

[0037] The beneficial effects of adopting the above technical solution are:

[0038] A two-tier control method for wind-storage power plants participating in primary frequency regulation of the power grid is described. The method obtains the grid-connection frequency of the wind-storage power plant and performs frequency regulation based on the control moment when the grid-connection frequency is outside the normal range of 49.9Hz to 50.1Hz. When the control moment is a model predictive control (MPC) moment, a centralized MPC algorithm is used for primary frequency regulation. When the control moment is a distributed adaptive control (DAC) moment, a station-level MPC algorithm is used for distributed adaptive control. This technical solution, through the dual-tier control steps of centralized MPC and distributed adaptive control, achieves high efficiency and effective results in wind-storage power plants participating in primary frequency regulation of the power grid.

[0039] A two-layer control device for a wind-storage power station participating in grid primary frequency regulation includes a wind-storage power station controller, a wind turbine controller, and an energy storage controller, as well as three program modules: a frequency regulation time determination module, a centralized primary frequency regulation control module, and a distributed adaptive control module. The wind-storage power station controller is connected to and communicates with the wind turbine controller and the energy storage controller, respectively. The frequency regulation time determination module is used by the wind-storage power station controller to obtain the grid connection frequency of the wind-storage power station. When the grid connection frequency of the wind-storage power station is outside the normal range, frequency regulation is performed according to the control time. When the control time is a model predictive control (MPC) time, the centralized primary frequency regulation control module is executed. The wind-storage power station controller uses a centralized model predictive control (MPC) algorithm to perform primary frequency regulation through the wind turbine controller and the wind-storage power station controller. When the control time is a distributed adaptive control time, the distributed adaptive control module is executed. The wind-storage power station controller performs distributed adaptive control using a site-level model predictive control algorithm. By utilizing the wind-storage power station controller, the wind turbine controller, the energy storage controller, the frequency regulation time determination module, the centralized primary frequency regulation control module, and the distributed adaptive control module, the device achieves high efficiency and good results in the wind-storage power station's participation in grid primary frequency regulation.

[0040] Please refer to the detailed description of the specific implementation method. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a principle block diagram of the present invention;

[0042] Figure 2 is a data flow diagram of the present invention;

[0043] Figure 3 It is a time diagram of the relationship between the upper and lower layer control time scales in the present invention;

[0044] Figure 4 This is a flow chart of the wind power station's coordinated control of the power system's primary frequency regulation;

[0045] Figure 5 It is the flow chart of model predictive control MPC;

[0046] Figure 6 This is a flow chart for calculating the reference output of a wind-storage power station;

[0047] Figure 7 This is a flow chart for calculating the reference output of a fan. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. The following description of at least one exemplary embodiment is actually only illustrative and is in no way intended to limit the present application and its application or use. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0049] In the following description, many specific details are set forth to facilitate a full understanding of the present application. However, the present application may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present application. Therefore, the present application is not limited to the specific embodiments disclosed below.

[0050] Example 1:

[0051] The present invention discloses a double-layer control method for a wind power station to participate in primary frequency modulation of a power grid, comprising the following steps:

[0052] S1 determines the frequency modulation time

[0053] The wind-storage power station controller obtains the grid-connected point frequency of the wind-storage power station. When the grid-connected point frequency of the wind-storage power station is within the normal range, steps S2 and S3 are skipped and the frequency is not adjusted. When the grid-connected point frequency of the wind-storage power station is not within the normal range, the frequency is adjusted according to the control time. The normal range is 49.9Hz~50.1Hz.

[0054] When the control moment is the model predictive control MPC moment, step S2 is executed, and the wind power station controller uses a centralized model predictive control MPC algorithm to perform a frequency modulation control through the wind turbine controller and the wind power station controller.

[0055] When the control moment is the distributed adaptive control moment, step S3 is executed, and the wind-storage power station controller performs distributed adaptive control through the site-level model predictive control algorithm.

[0056] S2 centralized primary frequency modulation control

[0057] The wind-storage power station controller adopts a centralized model predictive control MPC algorithm for control, which means that the wind-storage power station controller obtains the virtual droop coefficient and virtual inertia coefficient of the wind turbine and sends them to the wind turbine controller to control the output power of the wind turbine during a frequency modulation period. The wind-storage power station controller obtains the virtual droop coefficient and virtual inertia coefficient of the energy storage battery and sends them to the energy storage controller to control the output power of the energy storage battery during a frequency modulation period.

[0058] S3 Distributed Adaptive Control

[0059] The distributed adaptive control moment is between the moments when the two site-level model predictive control MPCs issue instructions. The wind-storage power station controller performs distributed adaptive control through the site-level model predictive control algorithm, including three steps: wind turbine error adaptive control, wind turbine speed adaptive control, and energy storage battery adaptive control.

[0060] The step of wind turbine error adaptive control includes the following steps: each wind turbine controller obtains the wind turbine speed prediction value and wind turbine mechanical power prediction value sent by the wind power station controller, and calculates the prediction error coefficient η of each wind turbine e,i , calculate the power of each fan at time n under error adaptive control Then the error adaptive control of each fan is completed.

[0061]

[0062] In formula (28), η e,i is the prediction error coefficient of wind turbine i; ω mea,i is the actual measured value of the mechanical power of the i-th fan, in rad / s, measured by the fan controller; i is the i-th fan; ω min is the minimum speed of the fan, in rad / s; ω mpc,i is the predicted speed of the i-th wind turbine, in rad / s, which is sent from the wind power station controller to the wind turbine controller.

[0063]

[0064] In formula (29), is the power of the i-th wind turbine at time n under error adaptive control, in MW; is the virtual droop coefficient of the i-th wind turbine, in MW / Hz; Δf(n) is the frequency deviation at time n, in Hz; is the virtual inertia coefficient of the i-th wind turbine, in MW*s / Hz; f(n) is the system frequency at time n, in Hz; f(n-1) is the system frequency at time n-1, in Hz; t0 is the starting time of the frequency modulation phase, that is, the initial time when the frequency is in the abnormal range; P e,i (t0) is the power of the i-th wind turbine at time t0, in MW; P m,mpc,i is the predicted value of the mechanical power of the i-th wind turbine, in MW, which is sent from the wind power station controller to the wind turbine controller; P m,mea,i is the actual measured value of the mechanical power of the i-th wind turbine, in MW, measured by the wind turbine controller.

[0065] Each wind turbine controller obtains the overall output coefficient η of the wind storage power station sent by the wind storage power station controller p, combined with the overall output coefficient η of the wind storage power station p and the prediction error coefficient η of each wind turbine e,i , calculate the active power of each wind turbine at time n under error adaptive control Then the error adaptive control of each fan is completed.

[0066]

[0067] In formula (30), η p is the overall output coefficient of the wind-storage power station; K f is the virtual droop coefficient of the power frequency of the wind power station, in MW / Hz; K in is the virtual inertia coefficient of the power frequency of the wind-storage power station, in MW*s / Hz, which is generally adjusted according to the scale of the wind-storage power station; f(n-2) is the system frequency at time n-2, in Hz; T is the MPC sampling time of the model predictive control, which is set to 1s; P e,i (t0) is the initial power of the i-th wind turbine, in MW; is the sum of the initial power of the wind turbines, in MW; The power of the interconnection line at the wind turbine of the wind-storage power station at time n-1 is measured by the wind-storage power station controller, in MW, and then unidirectionally broadcast to all wind turbine controllers.

[0068]

[0069] In formula (31), is the active power of the i-th wind turbine at time n under the error adaptive control of the wind turbine controller, in MW; P m,i is the mechanical power of the i-th wind turbine in the wind power station, in MW.

[0070] The fan speed adaptive control step includes the following steps: each fan controller calculates and obtains the corresponding fan speed adaptive coefficient η ω,i , combined with each fan speed adaptive coefficient η ω,i and the active power of the corresponding wind turbine at time n under error adaptive control Calculate the output power of each wind turbine at time n Then the adaptive control of the speed of each fan is completed.

[0071]

[0072] In formula (32), η ω,i is the fan speed adaptive coefficient; ω i represents the rotational speed of the i-th wind turbine blade in the wind-storage power station, in rad / s, which is measured by the wind turbine controller and uploaded to the wind-storage power station controller; ω' minis the minimum warning value of the fan rotor speed, in rad / s; ω max is the maximum speed of the fan, in rad / s; ω' max is the maximum warning value of the wind turbine rotor speed, in rad / s; f is the system frequency, i.e. the grid frequency, in Hz; f n The rated frequency of the system, in Hz.

[0073]

[0074] In formula (33), is the output power of the i-th wind turbine at time n, in MW.

[0075] The energy storage battery adaptive control step includes the following steps: the energy storage controller obtains the virtual droop coefficient and virtual inertia coefficient of the energy storage battery sent by the wind storage power station and the grid connection point frequency of the wind storage power station, and calculates the reference power P of the wind storage power station. ref , calculate the energy storage battery power P ESS , and then complete the adaptive control of the energy storage battery.

[0076]

[0077] In formula (34), P PCC The power of the interconnecting line at the wind turbine of the wind storage power station, in MW; is the frequency change rate.

[0078] Example 2:

[0079] The present invention discloses a double-layer control device for a wind-storage power station to participate in the primary frequency regulation of a power grid, comprising a wind-storage power station controller, a wind turbine controller and an energy storage controller, as well as three program modules: a frequency regulation time determination module, a centralized primary frequency regulation control module and a distributed adaptive control module. The wind-storage power station controller is electrically connected to and communicates with the wind turbine controller, and the wind-storage power station controller is electrically connected to and communicates with the energy storage controller.

[0080] The frequency regulation moment judgment module is used for the wind storage power station controller to obtain the frequency of the wind storage power station grid connection point. When the frequency of the wind storage power station grid connection point is outside the normal range, the frequency is adjusted according to the control moment. When the control moment is the model predictive control MPC moment, the centralized primary frequency regulation control module is executed, and the wind storage power station controller adopts the centralized model predictive control MPC algorithm to perform primary frequency regulation control through the wind turbine controller and the wind storage power station controller. When the control moment is the distributed adaptive control moment, the distributed adaptive control module is executed, and the wind storage power station controller performs distributed adaptive control through the site layer model predictive control algorithm.

[0081] The centralized primary frequency regulation control module is used for the wind power station controller to obtain the virtual droop coefficient and virtual inertia coefficient of the wind turbine, send them to the wind turbine controller, and control the output power of the wind turbine during the primary frequency regulation period. The wind power station controller obtains the virtual droop coefficient and virtual inertia coefficient of the energy storage battery, sends them to the energy storage controller, and controls the output power of the energy storage battery during the primary frequency regulation period.

[0082] Distributed adaptive control module, used for each wind turbine controller to obtain the wind turbine speed prediction value and wind turbine mechanical power prediction value sent by the wind power station controller, and calculate the prediction error coefficient η of each wind turbine e,i , calculate the power of each fan at time n under error adaptive control Each wind turbine controller obtains the overall output coefficient η of the wind storage power station sent by the wind storage power station controller p , combined with the overall output coefficient η of the wind storage power station p and the prediction error coefficient η of each wind turbine e,i , calculate the active power of each wind turbine at time n under error adaptive control Then complete the adaptive control of each fan error; each fan controller calculates the corresponding fan speed adaptive coefficient η ω,i , combined with each fan speed adaptive coefficient η ω,i and the active power of the corresponding wind turbine at time n under error adaptive control Calculate the output power of each wind turbine at time n Then, the adaptive control of the speed of each wind turbine is completed; the energy storage controller obtains the virtual droop coefficient and virtual inertia coefficient of the energy storage battery sent by the wind storage power station and the grid connection point frequency of the wind storage power station, and calculates the reference power P of the wind storage power station. ref , calculate the energy storage battery power P ESS , and then complete the adaptive control of the energy storage battery.

[0083] Example 3:

[0084] The present invention discloses a double-layer control device for a wind-storage power station to participate in primary frequency modulation of a power grid, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of embodiment 1 are implemented.

[0085] Example 4:

[0086] The present invention discloses a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in embodiment 1 are implemented.

[0087] The greatest technical contribution of this application lies in the two-layer control method: the wind storage power station controller obtains the frequency of the wind storage power station grid connection point, and does not adjust the frequency when the frequency of the wind storage power station grid connection point is within the normal range. When the frequency of the wind storage power station grid connection point is not within the normal range, it adjusts the frequency according to the control time. The normal range is 49.9Hz~50.1Hz. When the control time is the model predictive control MPC time, the wind storage power station controller adopts the centralized model predictive control MPC algorithm to perform a frequency adjustment control through the wind turbine controller and the wind storage power station controller. When the control time is the distributed adaptive control time, the wind storage power station controller performs distributed adaptive control through the site layer model predictive control algorithm.

[0088] Among them, the model predictive control MPC moment, the centralized model predictive control MPC algorithm, the distributed adaptive control moment and the station-level model predictive control algorithm themselves are existing technologies and will not be described in detail here.

[0089] A further technical contribution is to establish the prediction error coefficient η for each wind turbine e,i , calculate the power of each fan at time n under error adaptive control Each wind turbine controller obtains the overall output coefficient η of the wind storage power station sent by the wind storage power station controller p , combined with the overall output coefficient η of the wind storage power station p and the prediction error coefficient η of each wind turbine e,i , calculate the active power of each wind turbine at time n under error adaptive control Then complete the adaptive control of each fan error; establish each fan controller to calculate the corresponding fan speed adaptive coefficient η ω,i , combined with each fan speed adaptive coefficient η ω,i and the active power of the corresponding wind turbine at time n under error adaptive control Calculate the output power of each wind turbine at time n Then the adaptive control of the speed of each fan is completed.

[0090] The concept of this application:

[0091] Among the traditional control methods for wind-storage power stations to participate in the primary frequency regulation of the power system, although the centralized control method can better consider the overall characteristics of the wind-storage power station, it cannot have the response speed that meets the primary frequency regulation due to the problem of communication delay. In addition, the prediction accuracy requirements for the equipment status in the wind-storage power station are relatively high, and the equipment in the wind-storage power station may operate in an unsafe state. Although the distributed control method can respond to frequency changes at the grid connection point of the wind-storage power station more quickly, it cannot guarantee the primary frequency regulation characteristics of the wind-storage power station.

[0092] In addition, since the control equipment of the wind-storage power station includes a wind-storage power station control server and a device-level controller, and the device-level controller includes a wind turbine controller and an energy storage battery controller, the primary frequency regulation control method in this patent comprehensively utilizes a centralized control method and a distributed control method, combining the advantages of both control methods. The server at the station level adopts a centralized control method with a longer time scale to ensure better overall frequency characteristics of the wind-storage power station. At the same time, the controller at the device level adopts a distributed adaptive control method with a shorter time scale to ensure a faster primary frequency regulation response speed of the wind-storage power station and safe operation of the equipment within the wind-storage power station.

[0093] Technical contributions of this application:

[0094] The method of the present invention comprehensively utilizes the site-level controller of the wind-storage power station, namely the wind-storage power station controller, and the equipment-level controller including the wind turbine controller and the energy storage controller, as well as distributed and centralized communication technologies, to meet the goal of the wind-storage power station as a whole participating in the primary frequency regulation of the system, so that it exhibits frequency response characteristics similar to those of a synchronous generator, and can allocate power according to the frequency response capabilities of different wind turbines and energy storage batteries, thereby ensuring the safe operation of the wind turbines and energy storage batteries and improving the dynamic performance of the primary frequency regulation of the wind-storage power station. Under the control of the control method of the present invention, the wind-storage power station controller uses centralized communication to receive the operating status of the wind turbine and energy storage, and predicts the future status of the wind turbine and energy storage. It adopts a centralized model predictive control (MPC) method with a longer rolling cycle to meet the actual communication physical requirements, and sends the optimized droop coefficient and inertia coefficient to the device-layer controller of the wind turbine and energy storage. Then, the measured output power and grid connection point frequency of the wind-storage power station are sent to the device-layer controller via broadcast communication. To address the problem of inaccurate prediction of the wind turbine operating status by the site-layer model predictive control (MPC), the device controller of the wind turbine and energy storage measures the operating status of the equipment itself and adopts a distributed adaptive control method. This method can increase the output of wind turbines and energy storage batteries with more frequency regulation margin, i.e., a larger power adjustment range, and reduce the output of wind turbines and energy storage batteries with less frequency regulation margin, i.e., a smaller power adjustment range, so that the power distribution between the wind turbine and energy storage is reasonable, thereby ensuring that the overall frequency response of the wind-storage power station is relatively ideal and the operation of the wind turbine and energy storage is safe.

[0095] Description of beneficial effects:

[0096] The present invention comprehensively utilizes local communication technology and centralized communication technology to design a two-layer coordinated feedback control method for a wind-storage power station based on permanent magnet direct-drive wind turbines and energy storage batteries to participate in primary frequency modulation as a whole. The method is divided into a station control layer and a device control layer. In the present invention, the wind-storage power station controller uses the predicted data of the controller wind speed and uses a centralized model predictive control (MPC) method to obtain the virtual droop coefficient and virtual inertia coefficient sequence of the wind turbine and the virtual droop coefficient and virtual inertia coefficient sequence of the energy storage device, and sends them to the device layer controller including the wind turbine controller and the energy storage controller to control the output power of the wind turbine and the energy storage during the primary frequency modulation period, so that the overall wind-storage power station presents a relatively ideal frequency response to the outside world; the device layer controller uses an adaptive control algorithm to measure the operating status of the wind turbine and the energy storage, and corrects and controls the active power of the wind turbine and the energy storage in the case of large prediction errors of the station layer model predictive control (MPC), thereby achieving a rapid primary frequency modulation response of the wind-storage power station. At the same time, the method proposed in the present invention fully considers the operating limits of wind turbines and energy storage batteries. Therefore, while participating in frequency regulation, it can also ensure the reasonable distribution of power between wind turbines and energy storage batteries, ensure the safe operation of wind storage batteries, and reduce the cost of primary frequency regulation.

[0097] The specific features of this method are as follows:

[0098] 1. Compared with traditional centralized control methods, this method reduces the computational communication time and prediction accuracy requirements for the centralized model predictive control (MPC) used by the wind-storage power station controller. Compared with traditional distributed control methods, it achieves a more ideal frequency response for the wind-storage power station, resolving the conflict between a good frequency response for the wind-storage power station and reasonable power distribution between the wind and the storage. This method also ensures that the wind-storage power station exhibits synchronized droop characteristics with the system during a single frequency modulation period, ensuring the engineering feasibility of the method.

[0099] 2. The concept of an energy status index is proposed, which can conveniently measure the rotor kinetic energy released by a permanent magnet direct-drive wind turbine during the frequency modulation phase. Compared to the traditional method of using wind speed to measure the wind turbine's frequency modulation capability, this method directly calculates the wind turbine's rotor kinetic energy, which is related to the wind turbine's frequency modulation capability. This method can rationally allocate power between wind turbines while ensuring safe operation, fully utilizing the wind turbine's rotor kinetic energy and reducing wind energy loss. See steps S2021-S2022.

[0100] 3. For site-level control, a centralized model predictive control (MPC) method based on optimizing the virtual droop coefficients and virtual inertia coefficients of wind turbines and energy storage systems is proposed. Compared to control methods that optimize wind turbine power and energy storage battery power, this method allows wind turbine and energy storage battery power to respond to changes in the frequency at the wind power station's grid connection point, even when there are certain prediction errors in the frequency prediction at the wind power station's grid connection point during the primary frequency regulation phase. This results in a wind power station with excellent frequency characteristics. See step S204.

[0101] 4. To address the impact of wind energy prediction errors and inaccurate dynamic models, we innovatively proposed a prediction error coefficient and a speed adaptation coefficient. This method, based on wind turbine state prediction errors and wind turbine rotor speed, is a speed adaptive control method. Compared to other distributed control algorithms, this method can directly perform corrections based on wind turbine prediction errors and wind turbine speed conditions. This ensures reasonable wind turbine and energy storage battery output power, meeting real-time correction requirements. See steps S301-S302.

[0102] Technical solution description:

[0103] The method includes two control processes, namely, the wind-storage power station coordination controller executes a large-time-scale centralized model predictive control process, and the equipment controller of the permanent magnet direct-drive wind turbine and energy storage battery executes a small-time-scale distributed adaptive control process. Both processes are executed cyclically according to a certain period.

[0104] like Figure 1 As shown, the relationship between the control devices of the double-layer frequency regulation control method of the wind-storage power station.

[0105] like Figure 2 As shown, the control architecture of the two-layer control strategy.

[0106] like Figure 3 As shown, the time scale of centralized control and distributed control.

[0107] like Figure 4 As shown in the figure, the single frequency modulation double-layer control process, the specific steps are as follows:

[0108] The steps of this application include S1 determining the frequency modulation time, S2 the wind power station controller obtaining the droop coefficient of the wind turbine, the droop coefficient and inertia coefficient of the energy storage battery and sending them to the equipment controller, and S3 the distributed adaptive control of the equipment layer controller.

[0109] S1 determines the frequency modulation time

[0110] like Figure 4 As shown, the frequency of the wind-storage power station's grid connection point must first be measured to determine whether it is within the normal range (49.9Hz to 50.1Hz). If so, the wind-storage power station does not participate in frequency regulation. If not, the frequency regulation control steps are executed based on the control time during the period when the frequency is not within the normal range. If it is a model predictive control (MPC) time, the wind-storage power station controller is called and step S2 is executed. If it is a distributed adaptive control time, the device layer controller is called and distributed adaptive control is executed.

[0111] See Table 1 for the wind-storage power station site-level control data interaction table:

[0112] Table 1: Wind power station site level control data interaction table

[0113]

[0114] The S2 wind power station controller obtains the droop coefficient of the wind turbine, the droop coefficient of the energy storage battery and the inertia coefficient and sends them to the device controller

[0115] The equipment controller includes a fan controller and an energy storage controller.

[0116] like Figure 5 As shown, this step gives the model predictive control MPC process of the wind power station controller.

[0117] like Figure 3 As shown in the time scale relationship, the control moments in the frequency modulation stage can be divided into model predictive control (MPC) moments and distributed adaptive control moments. Entering the two-layer control process, the station layer model predictive control (MPC) process is as follows:

[0118] S201 predicts the frequency change of the grid connection point and obtains the reference power of the wind power station

[0119] In the model predictive control MPC prediction time period T P Inside, T P It is generally about 8s, which is used to predict the frequency change of the wind-storage power station grid connection point and calculate the reference power of the wind-storage power station.

[0120] like Figure 6 As shown, the specific steps are as follows:

[0121] S2011 wind power station controller obtains wind power station reference power

[0122] First, the wind power station controller uses a single model predictive control MPC rolling period T f Since the centralized communication and calculation time is at least 2 to 3 seconds, the recommended value is 5 seconds. Measure the frequency f of the wind-storage power station grid connection point in Hz, and calculate the reference power P of the wind-storage power station according to formula (1): ref .

[0123] P ref =P ref * +P f * +P in * (1)

[0124] In formula (1), P ref is the reference power of the wind power station, in MW; P ref *It represents the output power command value of the wind-storage power station in steady state, in MW, which is given by the upper-level power grid dispatching center. Its value is related to the wind resources, and the maximum power does not exceed the rated power of the wind-storage power station; P f * is the droop response power of the primary frequency regulation of the wind power station, in MW; P in * is the inertial response power of the primary frequency regulation of the wind-storage power station, in MW, and the values ​​are as follows:

[0125] P f * =K f (f * -f) (2)

[0126]

[0127] In formula (2), K f is the virtual droop coefficient of the power frequency of the wind power station, in MW / Hz, f * is the rated frequency of the system, usually 50Hz.

[0128] In formula (3), K in is the virtual inertia coefficient of the power frequency of the wind-storage power station, with the unit of MW*s / Hz, which is generally adjusted according to the scale of the wind-storage power station; t is the time point in the nonlinear sequence, with the unit of second.

[0129] S2012 Subsequently, the first-order inertia equation (4) was used to approximate the dynamic process of power system frequency oscillation

[0130]

[0131] In formula (4), H is the system inertia coefficient, the unit is MW*s / Hz, which is given by the power grid dispatching center, P G It is the power generation of the conventional units in the system, which is sent from the grid dispatching center to the wind power station controller in MW, P L The system load is sent from the grid dispatching center to the wind power station controller in MW, P WF It is the power generation capacity of the wind power station, in MW.

[0132] The wind power station controller will record the frequency changes of the grid connection point and estimate the unbalanced power of the power system through formula (5):

[0133] P UB =P G -P L +P WF =2H·(f(t)-f(t-1)) / (f * ·T) (5)

[0134] In formula (5), T is the MPC sampling time. Since MPC needs to consider the electromechanical transient process of the wind turbine, it is generally set to 1s, and t is the time point in the nonlinear sequence.

[0135] Then, combined with Equation (5), Equation (4) is discretized and linearized with sampling time T to obtain Equation (6).

[0136] f(t+1)=f(t)+K f ·Tf n (f(t)-f n ) / (2H)+K in ·f n (f(t)-f(t-1)) / (2H)+f(t)-f(t-1) (6)

[0137] In formula (6), (t) represents the value of a certain data at time t, for example, f(t) represents the frequency of the wind power station grid connection point at time t.

[0138] S2013 uses the virtual inertia coefficient K preset by the wind power station in and virtual droop coefficient K f The predicted or measured grid connection point frequency of the wind-storage power station is combined with formula (1) to calculate the reference power of the wind-storage power station.

[0139] S2014 uses the discretization formula (6) of the power system frequency dynamics and the estimated system unbalanced power P UB , calculate the grid connection point frequency f(t+1) at time t+1.

[0140] S2015 repeats steps S2013 and S2014 until the entire station-level model predictive control MPC prediction period T P The reference power of wind-storage power station of all model predictive control MPC points in the system is calculated until the reference power of wind-storage power station of all model predictive control MPC points P is obtained. ref .

[0141] After obtaining the reference power of the wind-storage power station based on the grid connection point frequency and the wind-storage power station frequency modulation coefficient, S202 needs to allocate power based on the predicted status of the wind turbine and energy storage battery. To reasonably allocate power, before executing a frequency modulation coordination control, first establish relevant evaluation indicators and set relevant parameters:

[0142] S2021 establishes the following energy state evaluation indicators for the stored kinetic energy of permanent magnet direct drive wind turbine blades:

[0143]

[0144] In formula (7), SOEi It represents the energy status evaluation index of the i-th permanent magnet direct-drive wind turbine in the wind-storage power station, without unit, ω i It represents the rotation speed of the blade of the i-th permanent magnet direct-drive wind turbine in the wind-storage power station, in rad / s, which is measured by the wind turbine controller and uploaded to the wind-storage power station controller. max,i It represents the upper limit of the blade speed of the i-th permanent magnet direct-drive wind turbine in the wind-storage power station to ensure normal power generation, in rad / s, ω min,i The lower limit of the blade speed of the i-th permanent magnet direct-drive wind turbine in the wind-storage power station to ensure normal power generation is expressed in rad / s. This value is obtained from the parameters of the wind turbine itself. i is an integer between 1 and l, and l is the number of permanent magnet direct-drive wind turbines in the wind-storage power station.

[0145] S2022 establishes the following evaluation indicators for the power generation level of permanent magnet direct-drive wind turbines:

[0146]

[0147] In formula (8), x i It represents the power generation level evaluation index of the i-th permanent magnet direct-drive wind turbine in the wind-storage power station, without unit, P e,i It represents the electromagnetic power output by the i-th permanent magnet direct-drive wind turbine in the wind power station, in MW, P m,i It represents the mechanical power captured by the i-th permanent magnet direct-drive wind turbine in the wind-storage power station, in MW, which is measured by the wind turbine controller and uploaded to the power station controller.

[0148] S203 After obtaining the reference power sequence of the wind power station through the prediction method and having the index for evaluating the frequency regulation capability of the wind turbine, this method will use the frequency regulation capability index of the wind turbine to calculate the frequency regulation capability of the wind turbine in the model predictive control MPC prediction period T P The reference power of the wind turbine is calculated and the state change of the wind turbine is predicted.

[0149] like Figure 7 As shown, the specific steps are as follows:

[0150] S2031 uses the first-order electromechanical transient model of the fan to predict the state change of the fan frequency regulation stage. The differential equation of its rotation dynamic process can be shown in formula (9):

[0151]

[0152] In formula (9), J C is the moment of inertia, which is obtained from the fan parameters itself, and the unit is kg*m2; ω i is the mechanical speed of the i-th wind turbine in the wind power station, in rad / s, P e,i is the electromagnetic power of the i-th wind turbine in the wind power station, in MW, Pm,i is the mechanical power of the i-th wind turbine in the wind-storage power station, in MW, measured by the wind turbine controller and uploaded to the wind-storage power station controller; T m,i is the mechanical torque of the i-th wind turbine in the wind power station, T e,i is the electromagnetic torque of the i-th wind turbine in the wind power station, which can be calculated by formula (9). It can be seen from the above formula that the electromagnetic torque of the wind turbine is directly related to the electromagnetic power. The wind turbine electromechanical transient equation is discretely linearized with a sampling time T of generally 1s, and the discrete model at time t is:

[0153] ω i (t)=(P m,i (t-1)-P e,i (t-1))T / (J C ω i (t-1))+ω i (t-1) (10)

[0154] Formula (10) is the discrete model of the wind turbine. In formula (10), ω i (t) is the mechanical speed of the i-th fan at time t, in rad / s.

[0155] S2032 Prediction Model Predictive Control MPC prediction cycle T P The wind speed changes of each wind turbine in the wind storage power station.

[0156] S2033 uses the wind turbine rotor speed obtained by measurement or the speed of each wind turbine in the wind power station predicted by formula (10) i , combined with formula (11) to calculate the mechanical power P of the fan m,i ,for:

[0157]

[0158]

[0159] In formula (11), R i is the blade radius of the i-th permanent magnet direct drive wind turbine, in m, C p,i is the wind energy capture coefficient of the i-th wind turbine, which is obtained from the wind turbine parameters and has no unit; ρ is the air density in kg / m3, v s,i is the wind speed of the i-th wind turbine, in m / s, β i is the pitch angle of the i-th permanent magnet direct drive wind turbine, in rad, λ i is the tip speed ratio of the i-th permanent magnet direct-drive wind turbine, unitless, and measured by the wind turbine controller.

[0160] S2034 uses formula (7) to calculate the SOE of each wind turbine in the wind storage power station, and uses formula (8) to calculate the power generation level index x of each wind turbine i , and calculate the reference power of the fan according to formula (13)

[0161]

[0162] In formula (13), is the reference power of the i-th permanent magnet direct-drive wind turbine in the wind-storage power station, in MW; is the reference power of the jth permanent magnet direct drive wind turbine in the wind storage power station, in MW; P m,j is the mechanical power of the jth wind turbine in the wind power station, in MW; SOE j Energy status evaluation index of the jth wind turbine in a wind power station.

[0163] The S2035 wind power station controller calculates the wind turbine electromagnetic power calculated by formula (13) and the mechanical power calculated by formula (11), and combines formula (10) to calculate the wind turbine speed ω at time t+1 i (t+1).

[0164] S2036 repeats steps S2033, S2034 and S2035 until the entire station-level model predictive control MPC prediction cycle T P The fan reference power of all control points within and speed ω i All are calculated.

[0165] S204 first uses steps S201, S202, and S203 to obtain the reference power of the wind power station and the reference power of the wind turbine. Then, the wind power station controller will establish a model predictive control (MPC) model based on the reasonable distribution of wind turbine power and the minimum output of energy storage power in the wind power station. The objective function is:

[0166]

[0167] In formula (14), {u * (k)} is the objective function of the model predictive control model, n is the model predictive control MPC point in the control process, k is the initial moment of the model predictive control MPC, Δx(n) is the penalty function of the energy storage device power output; Δz(n) is the penalty function of the wind turbine power tracking error.

[0168]

[0169]

[0170] In formula (15), C1 is the weight coefficient of the penalty function of the battery output of the energy storage device, is the droop coefficient of the energy storage battery, in MW / Hz; Δf(n) is the frequency deviation at time n, in Hz; is the inertia coefficient of the energy storage battery, in MW*s / Hz; f(n) is the system frequency at time n, in Hz; f(n-1) is the system frequency at time n-1, in Hz; t0 is the starting time of the frequency modulation phase, that is, the initial time when the frequency is in the abnormal range; P ESS (t0) is the energy storage power at time t0, in MW.

[0171] In formula (16), the weight coefficient of the penalty function of the C2 wind turbine power tracking error is, is the droop coefficient of the i-th wind turbine, in MW / Hz; is the inertia coefficient of the i-th wind turbine, in MW*s / Hz; P e,i (t0) is the power of the i-th wind turbine at time t0, in MW; is the reference power of the i-th wind turbine at time n, in MW.

[0172] The site-level model predictive control (MPC) proposed in the present invention needs to consider wind turbine power ramping constraints, wind turbine output power constraints, wind turbine rotor speed constraints, and frequency response constraints of the wind power station.

[0173] -ΔP max ≤P e,i (n)-P e,i (n-1)≤ΔP max (17)

[0174] In formula (17), ΔP max is the ramp constraint between two model predictive control MPC points, in MW; P e,i (n) is the wind turbine power of the i-th wind turbine at time n, in MW; P e,i (n-1) is the wind turbine power of the i-th wind turbine at time n-1, in MW.

[0175] P i,min <P e,i (n)<P i,max (18)

[0176] In formula (18), P i,min is the lower limit of wind turbine power constraint, in MW; P i,max is the upper limit of wind turbine power constraint, in MW.

[0177] ω min <ω i (n)<ω max(19)

[0178] In formula (19), ω min is the minimum speed of the fan, in rad / s; ω i (n) is the speed of the i-th fan at time n, in rad / s; ω max The maximum speed of the fan, in rad / s.

[0179] P ESS,min ≤P ESS ≤P ESS,max (20)

[0180] In formula (20), P ESS,min is the minimum power of the energy storage battery, in MW; P ESS is the energy storage battery power, unit MW; P ESS,max It is the maximum and minimum power of the energy storage battery, in MW.

[0181]

[0182]

[0183]

[0184]

[0185] In formula (24), P ESS (n) is the energy storage power at time n, in MW; P ESS (t0) is the energy storage battery power at time t0, in MW.

[0186] The complete model of the model predictive control (MPC) includes the objective function (14), the transient model (6) and (10), the SOE constraint (2), and the operation constraint (17) to (24). The interior point method is used to solve the model, and the optimization variables are the droop coefficient and inertia coefficient of the wind turbine and the droop coefficient and inertia coefficient of the energy storage battery.

[0187] In addition, during the MPC process, the difference between the MPC sequence results and the reference value must be monitored. If the difference is greater than a given threshold, the reference value is replaced with the optimized result. Subsequent reference values ​​are updated according to steps S201 and S202, and the process proceeds to step S203 to continue MPC optimization. The optimized sequence generated by the MPC is sent to the local wind turbine controller until the difference for all control points meets the requirements.

[0188] The S206 wind-storage power station controller sends the calculated droop coefficient and inertia coefficient of the wind turbine and energy storage battery to the equipment controller.

[0189] Distributed Adaptive Control of S3 Device Layer Controller

[0190] In the frequency response phase of the wind-storage power station, at the time of distributed adaptive control at the device layer, the distributed adaptive control process of the device layer controller of the primary frequency modulation dual-layer control is as follows:

[0191] The S301 distributed adaptive control point is located between the moments when the two station-level model predictive control MPCs issue instructions.

[0192] like Figure 3 As shown, the device layer controller will first perform error adaptive control of the fan. The error adaptive control process of the fan is as follows:

[0193] S3011 First, the wind turbine controller receives the predicted values ​​of wind turbine speed and wind turbine mechanical power sent by the wind power station controller, and determines the prediction error of wind turbine speed and mechanical power at the distributed adaptive control time point.

[0194] P m,mea,i -ΔP≤P m,mpc,i ≤P m,mea,i +ΔP (25)

[0195] In formula (25), P m,mea,i is the actual measured value of the mechanical power of the i-th wind turbine, in MW, measured by the wind turbine controller; ΔP is the mechanical power deviation range, in MW, set by the wind turbine controller, generally around 5% of the rated value; P m,mpc,i is the predicted mechanical power of the i-th wind turbine, in MW, which is sent from the wind power station controller to the wind turbine controller.

[0196] ω mea,i -Δω≤ω mpc,i ≤ω mea,i +Δω (26)

[0197] In formula (26), ω mea,i is the actual measured value of the mechanical power of the i-th fan, in rad / s, measured by the fan controller; Δω is the speed deviation range, in rad / s, set by the fan controller, generally around 5% of the rated value; ω mpc,i is the predicted speed value of the i-th wind turbine, in rad / s, which is sent from the wind power station controller to the wind turbine controller. If the error is within the allowable range, the error adaptive control of the wind turbine is directly terminated, and the power P of the i-th wind turbine at time n is calculated by formula (27): e,i (n), jump to step S302, and perform adaptive control of the fan speed.

[0198]

[0199] S3012 For the prediction error of fan speed and fan mechanical power, the equipment controller uses the ratio of the actual rotor kinetic energy of the fan to the predicted rotor kinetic energy to establish a prediction error coefficient η for fan i e,i

[0200]

[0201] In formula (28), η e,i is the prediction error coefficient of wind turbine i.

[0202] S3013 combines the prediction error coefficient η of wind turbine i e,i The fan controller relies on the prediction error coefficient to calculate the power of the i-th fan at time n under error adaptive control The unit is MW, which corrects the unreasonable control results caused by prediction errors.

[0203]

[0204] In formula (29), is the power of the i-th wind turbine at time n under error adaptive control, in MW.

[0205] In addition to the wind turbine speed prediction error coefficient, in order to ensure a good overall frequency response of the wind-storage power station, the equipment controller uses the ratio of the measured actual wind-storage power station output power to the wind-storage power station prediction power to propose an overall wind-storage power station output coefficient η. p , ensuring the frequency response of wind-storage power stations:

[0206]

[0207] In formula (30), η p is the overall output coefficient of the wind storage power station, The power of the interconnection line at the wind turbine of the wind-storage power station at time n-1 is measured by the wind-storage power station controller, in MW, and then unidirectionally broadcast to all wind turbine controllers.

[0208] S3015 combined with the overall output coefficient η of the wind-storage power station p and the prediction error coefficient η of wind turbine i e,i , the active power of the i-th wind turbine at time n under the error adaptive control of the wind turbine controller (unit: MW) is:

[0209]

[0210] In formula (31), is the active power of the i-th wind turbine at time n under the error adaptive control of the wind turbine controller, in MW.

[0211] After the fan error adaptive control is executed in S302, in order to prevent the fan speed from exceeding the limit, if the fan speed reaches the warning range, the speed adaptive coefficient will be set. According to the fan speed, the fan output power will be reduced. The fan speed adaptive control needs to be executed. The control process is as follows:

[0212] S3021 fan controller determines whether the fan speed is within the speed warning range. If not, the speed adaptive control is not started and the calculated fan power P of the i-th fan at that moment is directly used. e,i The command is sent down and the process goes directly to step S303. Otherwise, the fan will adaptively adjust the power according to the speed.

[0213] In order to prevent the fan speed from exceeding the limit, S3022 uses the per-unit value of the fan's rotor kinetic energy that can be used for primary frequency modulation, that is, the ratio of the fan's rotor kinetic energy that can be used for frequency modulation to the fan's rotor kinetic energy within the warning range, to define the fan speed adaptive coefficient η. ω,i for:

[0214]

[0215] In formula (32), η ω,i is the fan speed adaptive coefficient, ω' min is the minimum warning value of the fan rotor speed, in rad / s, ω' max It is the maximum warning value of the fan rotor speed, in rad / s.

[0216] S3023 combined with the fan speed adaptive coefficient η ω,i Active power of wind turbine under error adaptive control After the speed adaptive control, the output power of the i-th fan at time n is for:

[0217]

[0218] In formula (33), is the output power of the i-th wind turbine at time n, in MW.

[0219] S303 addresses the adaptive control of the energy storage batteries within the power station. Since the primary function of the energy storage batteries in frequency regulation at a wind power station is to adjust the output power to ensure an ideal frequency response, the energy storage controller first calculates the wind power station's output power reference value (i.e., the ideal value) based on the droop coefficient and inertia coefficient set for the wind power station and the measured grid connection point frequency. The controller then measures the sum of the wind turbine powers via the tie line. The difference between the ideal wind turbine output and the wind turbine tie line power is then calculated as the output of the energy storage battery under distributed adaptive control.

[0220]

[0221] In formula (34), P PCC It is the interconnection line power at the wind turbine of the wind storage power station, in MW.

[0222] S304 The device controller sends the calculated wind turbine and energy storage power to the wind turbine and energy storage battery.

[0223] After this application was run confidentially for a period of time, the on-site technicians reported that the following benefits were achieved:

[0224] The method includes: establishing energy state evaluation indicators and power generation level evaluation indicators for wind turbines; designing a two-layer control architecture for wind-storage power stations to participate in frequency regulation; the central control layer of the wind-storage power station measures the frequency, estimates the power imbalance, and uses the model predictive control (MPC) method to determine the reference value of the power of each wind turbine and the centralized energy storage power station on a large time scale; the local control layer of the wind turbine uses a distributed adaptive control method, relying on the local operating data of the wind turbine and the energy storage battery and the reference value obtained by the upper-level model predictive control (MPC) method to further correct and control the wind turbine and the energy storage battery on a smaller time scale, so as to achieve reasonable power distribution between the wind turbine and the energy storage battery and better frequency response characteristics of the wind-storage power station. The present invention utilizes the kinetic energy stored in the permanent magnet direct-drive wind turbine rotor and the electricity stored in the centralized energy storage battery, fully tapping the primary frequency regulation capability of the wind storage power station. At the same time, the designed two-layer architecture reduces the dependence on the prediction accuracy, calculation and communication speed of the central control layer. After the model predictive control calculation is performed at the station level, the equipment level is corrected. According to the status of the wind turbine, the distributed adaptive control method is used to correct the unreasonable power output of the wind turbine and energy storage equipment, thereby realizing the overall rapid primary frequency regulation response of the wind storage power station and the reasonable distribution of power between the wind and storage, and ensuring the safe operation of the wind storage.

[0225] At present, the technical solution of the present invention has been put into pilot production, that is, a smaller-scale test of the product before large-scale mass production; after the pilot production was completed, a user usage survey was carried out on a small scale, and the survey results showed that user satisfaction was high; now preparations have begun for the formal production and industrialization of the product (including intellectual property risk warning surveys).

Claims

1. A two-tier control method for a wind power station to participate in primary frequency modulation of a power grid, characterized by: Obtain the grid connection point frequency of the wind-storage power station. When the grid connection point frequency of the wind-storage power station is outside the normal range, frequency modulation is performed according to the control time. The normal range is 49.9 Hz to 50.1 Hz. When the control time is the model predictive control (MPC) time, a centralized model predictive control (MPC) algorithm is used for primary frequency modulation control. When the control time is the distributed adaptive control time, a station-level model predictive control algorithm is used for distributed adaptive control. Specifically, the following steps are included: S1 determines the frequency modulation moment, obtains the grid-connected point frequency of the wind-storage power station, and performs frequency modulation according to the control moment when the grid-connected point frequency of the wind-storage power station is not within the normal range. When the control moment is the model predictive control MPC moment, step S2 is executed; when the control moment is the distributed adaptive control moment, step S3 is executed; S2 is centralized primary frequency modulation control, obtains the virtual droop coefficient and virtual inertia coefficient of the wind turbine and controls the output power of the wind turbine during the primary frequency modulation period, obtains the virtual droop coefficient and virtual inertia coefficient of the energy storage battery and controls the output power of the energy storage battery during the primary frequency modulation period; S3 is distributed adaptive control, the distributed adaptive control moment is between the moments when the two site-level model predictive control MPCs issue instructions, and the site-level model predictive control algorithm is used to perform distributed adaptive control, including the steps of wind turbine error adaptive control and energy storage battery adaptive control; The step of the fan error adaptive control comprises the following steps: obtaining the fan speed prediction value and the fan mechanical power prediction value, and calculating the prediction error coefficient η. e,i , calculate the power of each fan at time n under error adaptive control Then the error adaptive control of each fan is completed; In formula (28), η e,i is the prediction error coefficient of wind turbine i; ω mea,i is the actual measured value of the mechanical power of the i-th fan, in rad / s; i is the i-th fan; ω min is the minimum speed of the fan, in rad / s; ω mpc,i is the predicted speed of the i-th fan, in rad / s; In formula (29), is the power of the i-th wind turbine at time n under error adaptive control, in MW; is the virtual droop coefficient of the i-th wind turbine, in MW / Hz; Δf(n) is the frequency deviation at time n, in Hz; is the virtual inertia coefficient of the i-th wind turbine, in MW*s / Hz; f(n) is the system frequency at time n, in Hz; f(n-1) is the system frequency at time n-1, in Hz; t0 is the starting time of the frequency modulation phase, that is, the initial time when the frequency is in the abnormal range; P e,i (t0) is the power of the i-th wind turbine at time t0, in MW; P m,mpc,i is the predicted value of the mechanical power of the i-th wind turbine, in MW; P m,mea,i is the actual measured value of the mechanical power of the i-th wind turbine, in MW; Obtain the overall output coefficient η of the wind storage power station p , combined with the overall output coefficient η of the wind storage power station p and the prediction error coefficient η of each wind turbine e,i , calculate the active power of each wind turbine at time n under error adaptive control Then the error adaptive control of each fan is completed; In formula (30), η p is the overall output coefficient of the wind-storage power station; K f is the virtual droop coefficient of the power frequency of the wind power station, in MW / Hz; K in is the virtual inertia coefficient of the wind power station power frequency, in MW*s / Hz; f(n-2) is the system frequency at time n-2, in Hz; T is the MPC sampling time, set to 1s; P e,i (t0) is the initial power of the i-th wind turbine, in MW; is the sum of the initial power of the wind turbines, in MW; is the power of the tie line at the wind turbine of the wind storage power station at time n-1, in MW; In formula (31), is the active power of the i-th wind turbine at time n under error adaptive control, in MW; P m,i is the mechanical power of the i-th wind turbine in the wind-storage power station, in MW; The distributed adaptive control using the station-level model predictive control algorithm also includes a fan speed adaptive control step located after the fan error adaptive control step. The fan speed adaptive control step includes the following steps: calculating the fan speed adaptive coefficient η ω,i , combined with each fan speed adaptive coefficient η ω,i and the active power of the corresponding wind turbine at time n under error adaptive control Calculate the output power of each wind turbine at time n Then the adaptive control of the speed of each fan is completed; In formula (32), η ω,i is the fan speed adaptive coefficient; ω i represents the rotation speed of the i-th wind turbine blade in the wind power station, in rad / s; ω' min is the minimum warning value of the fan rotor speed, in rad / s; ω max is the maximum speed of the fan, in rad / s; ω' max is the maximum warning value of the fan rotor speed, in rad / s; f is the system frequency, in Hz; f n is the rated frequency of the system frequency, in Hz; In formula (33), is the output power of the i-th wind turbine at time n, in MW; The steps of the energy storage battery adaptive control include the following steps: obtaining the virtual droop coefficient and virtual inertia coefficient of the energy storage battery and the grid connection point frequency of the wind storage power station, and calculating the reference power P of the wind storage power station. ref , calculate the energy storage battery power P ESS , and then complete the adaptive control of energy storage batteries; In formula (34), P PCC The power of the interconnecting line at the wind turbine of the wind storage power station, in MW; is the frequency change rate.

2. A double-layer control device for a wind-storage power station participating in primary frequency modulation of a power grid, used in the double-layer control method for a wind-storage power station participating in primary frequency modulation of a power grid as claimed in claim 1, characterized in that: It includes three program modules: a wind power station controller, a wind turbine controller, an energy storage controller, a frequency modulation time determination module, a centralized primary frequency modulation control module, and a distributed adaptive control module. The wind power station controller is connected to and communicates with the wind turbine controller and the energy storage controller respectively. The frequency modulation time determination module is used for the wind power station controller to obtain the grid-connected frequency of the wind power station. When the grid-connected frequency of the wind power station is outside the normal range, the frequency is modulated according to the control time. When the control time is the model predictive control MPC time, the centralized primary frequency modulation control module is executed. The wind power station controller adopts the centralized model predictive control MPC algorithm to perform primary frequency modulation control through the wind turbine controller and the wind power station controller. When the control time is the distributed adaptive control time, the distributed adaptive control module is executed. The wind power station controller performs distributed adaptive control through the site-level model predictive control algorithm. The centralized primary frequency regulation control module is used for the wind power station controller to obtain the virtual droop coefficient and virtual inertia coefficient of the wind turbine and send them to the wind turbine controller to control the output power of the wind turbine during the primary frequency regulation period. The wind power station controller obtains the virtual droop coefficient and virtual inertia coefficient of the energy storage battery and sends them to the energy storage controller to control the output power of the energy storage battery during the primary frequency regulation period. The distributed adaptive control module is used for each wind turbine controller to obtain the wind turbine speed prediction value and wind turbine mechanical power prediction value sent by the wind power station controller, and calculate the prediction error coefficient η of each wind turbine. e,i , calculate the power of each fan at time n under error adaptive control Each wind turbine controller obtains the overall output coefficient η of the wind storage power station sent by the wind storage power station controller p , combined with the overall output coefficient η of the wind storage power station p and the prediction error coefficient η of each wind turbine e,i , calculate the active power of each wind turbine at time n under error adaptive control Then the error adaptive control of each fan is completed; The distributed adaptive control module is also used for each fan controller to calculate the corresponding fan speed adaptive coefficient η ω,i , combined with each fan speed adaptive coefficient η ω,i and the active power of the corresponding wind turbine at time n under error adaptive control Calculate the output power of each wind turbine at time n Then, the adaptive control of the speed of each wind turbine is completed; the energy storage controller obtains the virtual droop coefficient and virtual inertia coefficient of the energy storage battery sent by the wind storage power station and the grid connection point frequency of the wind storage power station, and calculates the reference power P of the wind storage power station. ref , calculate the energy storage battery power P ESS , and then complete the adaptive control of the energy storage battery.

3. A two-tier control device for a wind-storage power station participating in primary frequency modulation of a power grid, comprising a computer-readable storage medium storing a computer program, characterized in that: The computer program includes the frequency modulation moment determination module, the centralized primary frequency modulation control module and the distributed adaptive control module in claim 2, and implements the corresponding steps in claim 1 when the computer program is executed by the processor.

Images