CN113489073A

CN113489073A - Multi-time-space layered comprehensive frequency modulation control system based on fan cluster

Info

Publication number: CN113489073A
Application number: CN202110826196.XA
Authority: CN
Inventors: 阳同光; 陈长青; 黎灿兵; 李文芳; 蔡振华; 黄志亮
Original assignee: Hunan City University
Current assignee: Hunan City University
Priority date: 2021-07-21
Filing date: 2021-07-21
Publication date: 2021-10-08

Abstract

According to the multi-time-space layered comprehensive frequency modulation control system based on the fan cluster, firstly, different frequency modulation modes and participation time sequences can be adopted for the fans at different wind speeds according to different running states of the fans at different wind speeds; secondly, aiming at the influence of the wind speed on the frequency modulation control of the fans, a clustering method considering the wake effect of the wind power plant is established, so that the number of the units in each cluster can be distributed in a balanced manner, the integral frequency modulation effect of the wind power plant is improved, and the secondary fluctuation problem of the frequency is avoided; and finally, considering the randomness of the wind speed, optimizing the wind power plant cluster by adopting MPC (multi-control computer) rolling per minute, and ensuring the optimal frequency modulation capability and frequency modulation accuracy of the wind power plant.

Description

Multi-time-space layered comprehensive frequency modulation control system based on fan cluster

Technical Field

The invention relates to the technical field of power grid frequency modulation, in particular to a multi-time-space layered comprehensive frequency modulation control system based on a fan cluster.

Background

With the large-scale wind power plant grid connection, a system side puts forward a clear requirement on the wind power plant frequency modulation. As the western part of the countries with high wind power penetration clearly indicate that the grid-connected wind power plant has the same frequency modulation capability as the conventional generator set. According to the national standard GB/T19963-2011 technical provision for accessing wind power plant to power system, the method comprises the following steps: the wind power plant is in accordance with DL/T1040 standard and has the capability of participating in frequency modulation of the power system.

With the deep research of numerous scholars at home and abroad on the frequency modulation control strategy of the grid-connected wind power plant, the control strategy is continuously perfected. Common stand-alone control strategies are: and the kinetic energy of the fan rotor is used as virtual inertia control of the system inertia. Through Maximum Power Point Tracking (MPPT), active and standby overspeed and pitch angle control, droop control, combined control, wind storage coordination control and the like are reserved. However, at low wind speeds, the virtual inertia support capability is limited due to the lower rotor speed. And after inertia is finished, a large number of fans simultaneously enter a rotating speed recovery state, so that the system frequency is easy to drop for the second time. And overspeed and pitch angle control make the fan skew MPPT, not only increased mechanical wear, reduced wind-powered electricity generation field economic nature moreover. In summary, although the control strategy described above has achieved a good control effect in single-machine control, there are still some problems to be solved in the process of multi-spatio-temporal scale combined control in the wind farm.

In an actual large wind power plant, fans are distributed for several kilometers or even dozens of kilometers, all the fans cannot be located at the same wind speed, and the frequency modulation characteristic of the fans is obviously influenced by the wind speed. Therefore, it is impossible to adopt the same frequency modulation control strategy for all fans, and if single-machine control is adopted, the control complexity is increased. Therefore, the interaction of the wind turbines at different wind speeds in time and space is considered, and the overall frequency modulation effect of the wind power plant is optimized through coordination control. The idea of performing group frequency modulation control on the fans in the wind power plant according to the wind speed of the fan in the prior art is proposed, but the situation that a large number of fans may be gathered at a certain wind speed is not considered, so that the grouping strategy is not perfect. An improved fan optimization grouping strategy is provided by research, and the fans in each group are balanced in number by performing degradation processing on the fans. The problem that a large number of fans are concentrated under a certain wind speed is solved, and secondary frequency fluctuation of a system caused by the fans exiting from frequency modulation is avoided. However, the grouping only considers the wind speed factor, a fixed grouping mode is adopted, and the wake effect and the wind speed randomness of the wind power plant are not considered, so that the grouping accuracy is greatly reduced.

However, in large wind farms, the phenomenon that an upstream wind turbine affects the wind speed of a downstream wind turbine is called wake effect, and the effect is determined by the angle θ between the upstream wind turbine and the wind direction and the geographical position of the wind turbine. The geographical position of the fan is taken as a grouping factor, and the influence of the wake effect of the wind power plant can be fully considered when the fans under the same working condition are clustered and optimized; under the condition of considering the wake effect, an optimization objective function can be provided, and the contribution of the wind power plant to system frequency adjustment is researched. However, the method lacks of comprehensive consideration of information such as the running state of the fan, the wind speed prediction and the like.

In recent years, a Model Predictive Control (MPC) theory is gradually applied to the research of wind power plant frequency modulation control due to the fact that the MPC theory has online rolling optimization, error correction and feedback control links. Some scholars establish a multi-space-time scale coordination control strategy of 3 levels of primary frequency modulation, secondary frequency modulation and tertiary frequency modulation in a wind power plant. And students also optimize the rolling from the station layer of the wind power plant, the grouping of the wind power plant and the single machine on 3 levels layer by layer on the space and time scale, and provide a comprehensive frequency control strategy of the wind power cluster. However, in the current stage of research, the wake effect of the wind power plant and the wind speed prediction information are combined, and the research applied to the wind power plant rolling cluster is not much. Meanwhile, the research of combining the hierarchical control idea and the sequential control idea by adopting various control modes for the wind power plant is still incomplete at present.

Aiming at the problems, the theoretical research of the wind power plant participating in system frequency modulation is further perfected. The method aims to provide a wind power plant clustering method considering wind power plant wake effect and wind speed randomness and a multi-space-time-scale hierarchical control strategy.

Disclosure of Invention

The invention aims to overcome the defects of the prior art and provide a multi-time-space layered comprehensive frequency modulation control system based on a fan cluster, wherein when a large wind farm participates in frequency modulation, the operation states of all fans are different due to the influence of factors such as geographical positions, wake effects and the like, and the frequency modulation presents a time-space distribution characteristic; 2) how to optimize the wind turbine generator cluster and enable the frequency modulation capability of the wind power plant to be optimal; 3) how to improve the accuracy of frequency modulation of a wind power plant.

The technical scheme of the invention is as follows: a multi-time-space layered comprehensive frequency modulation control system based on a fan cluster is characterized in that a virtual inertia control is adopted for a fan at a medium wind speed, a droop control is adopted for the fan at a high wind speed, and therefore time-space coordination control among wind power clusters at different wind speeds is achieved; the comprehensive frequency control framework is shown in figure 1; the coordination relationships on the temporal and spatial scales are shown in fig. 2 and 3, respectively.

The multi-time-space layered comprehensive frequency modulation control system is divided into a station layer, a cluster control layer and a fan frequency modulation power distribution layer; each layer is optimally designed.

S1, station layer

Energy for inertia response of thermal power generating unit mainly comes from rotary kinetic energy E stored in rotor of thermal power generating unit_KIt can be expressed as:

the inertia time constant H is generally used to represent the magnitude of the inertia of the thermal power generating unit, which can be expressed as:

in the formula: s_NIs the rated capacity of the generator;

analogy formula (2) can obtain virtual inertia H of wind power plant_W：

In the formula: n is the number of fans in the wind power plant; delta E_opiIs the rotational energy of the ith fan, S_W-NThe total rated capacity of the wind power plant is obtained; wherein the content of the first and second substances,

in the formula: delta E_opTotal power at the rotor side of the fan, P_ATracking point power for pre-frequency-modulated power; p_e(t) is the output electromagnetic power, P_w(t) inputting mechanical power; delta E_kFor total release of kinetic energy of the rotor, J for total moment of inertia of the unit, Delta E_lossAdditional lost wind energy to the rotor due to reduced rotational speed; t is t_onAnd t_offRespectively, the moment of the start of frequency modulation and the moment of inertia exit.

Substituting equations (4) - (6) into equation (3) can obtain a relation equation of the wind power plant inertia constant and the rotating speed omega as follows:

the formula (7) shows that the kinetic energy release power of the rotor is restricted by the variation range of the rotating speed omega at any wind speed v; therefore, the inertia response capability is different at different wind speeds. In the low wind speed region, ω ≈ ω_minThe variable quantity of the rotating speed tends to 0, the release quantity of the kinetic energy of the rotor is almost zero, and the inertia response capability is not realized; in the middle wind speed area, the kinetic energy of the rotor is gradually increased along with the increase of the wind speed, so that sufficient inertia can be providedAnd measuring the power. In a high wind speed area, the rotating speed of the fan is not suitable to be changed when the fan is in a rated rotating speed state, so that inertia response is not suitable to be participated. But sufficient power can be provided by adjusting the pitch angle to fit the primary stage. In summary, the frequency modulation manner of the wind turbine at different wind speeds is shown in fig. 4.

From the analysis, for a large wind power plant, the frequency modulation mode and capacity of the wind turbine are influenced by the wind speed. In order to better exert the integral frequency modulation effect of the large wind power plant, the method provides a timing sequence coordination control strategy according to the frequency modulation characteristic of the fan. The strategy can realize two control purposes by coordinating and controlling the fan units in the large-scale wind power plant at different wind speeds: 1) providing sufficient inertia supporting capacity of the wind power plant; 2) after the inertia of the fan responds, the rotating speed recovery power is provided at the rotating speed recovery stage of the rotor, and the frequency secondary drop caused by the fact that a large number of fans exit the system at the same time is avoided.

In the high and low wind speed areas, the fan does not participate in inertia response control. Therefore, the inertia response capability of the wind power plant is provided by the wind turbine in the medium wind speed region, and the primary frequency modulation stage and the rotation speed recovery power are mainly provided by the wind turbine in the high wind speed region or the thermal power generating unit. The fan participating in frequency modulation has two working modes of MPPT and frequency modulation. And when the system frequency fluctuation is detected, the fan enters a corresponding frequency modulation model according to the wind speed. The frequency modulation coordination control of the fans in the large wind farm is shown in figure 5. The implementation process mainly considers the following principles:

1) according to the actual wind speed of the wind power plant, the fans are clustered into the same group according to the wind speed, and the fans in the same group are distributed with the same control variable. The control complexity of the wind power plant can be simplified without influencing the frequency modulation characteristic of the fan. Meanwhile, in order to ensure that the fan cannot cause cutter cutting accidents due to too low rotating speed, the lowest rotating speed value of a rotor of the fan is kept above 0.7 pu. Thus, the fans with the wind speed of more than 7m/s are divided into a group at intervals of 1 m/s.

2) The number of fans in different wind speed stages is uniformly distributed as much as possible. If the fans are clustered only according to the wind speed, the number of the fans is too small under partial wind speeds, and a large number of fans are clustered under a certain wind speed. The method is difficult to give full play to the benefits of self-coordination of the large wind power plant on system frequency modulation, and is also difficult to reflect the positive influence of uniform grouping on orderly recovery of the rotating speed when the fan exits frequency modulation. And the uniform control of the number of the fans under different wind speeds can be realized by changing the wake effect of the wind power plant.

3) When the system frequency is detected to fluctuate, the high-wind speed unit and the low-wind speed unit keep an MPPT operation mode, and the medium-wind speed area unit (medium-wind speed fan) quickly responds to provide inertia response. When the inertia response judging link detects a medium-wind-speed fan, | delta ω | <4 × 10-7 or df/dt ═ 0, the inertia link is ended, the fan exits the frequency modulation system, the rotation speed recovery stage is entered, and the system enters the primary frequency modulation stage. At the moment, the high-wind-speed unit enters a frequency modulation mode, and power and primary frequency modulation power required by the exiting of the medium-wind-speed fan are provided through the adjustment of the pitch angle. If the number of the high-wind-speed fans in the wind power plant is insufficient, the partial power is provided by the thermal power generating unit.

S2, cluster control layer

The large wind farm has huge number of fans, generally consists of dozens of fans or even hundreds of fans, and is distributed irregularly. Thus, the downstream wind speed (in the wind speed direction) v₂Area A distributed and shielded by upstream fan_sThe longitudinal and lateral distances x and d affect such that the downstream wind speed tends to be lower than the upstream wind speed. Thereby influencing the frequency modulation effect of the downstream fan; the wake effect is shown in fig. 6.

In FIG. 6, r₀The radius of a rotating area of the fan blade is shown, x is the longitudinal distance between an upstream fan and a downstream fan along the wind direction, and d is the transverse distance. A. the₀The wind speed influence area is the rotating area of the fan blade. A. the_sThe shielding area is an effective area of influence of an upstream fan on a downstream fan. Radius r_iIs the wake expansion effect radius. R as the wake expansion effect angle α and or the longitudinal distance increases_iWill increase, the area of influence increases;

r_i＝r₀+x*tanα,(0°≤α≤45°) (8)

the jth downstream region wind speed considering the wake effect is:

wherein the content of the first and second substances,

in the formula: c_TiThe wind energy utilization coefficient of the ith fan is determined by the tip speed ratio lambda and the pitch angle beta. Affected by the working state of the upstream fan.

The output power of the fan depends mainly on the wind speed. Wind speed is affected by two factors, the first being geographical location, i.e. the distance between the shadow areas of the upstream fans. The other is the upstream fan operating condition, i.e., relative to the blade and wind angle θ. Because the position is fixed, the working state of the fan is selected as a cluster clustering index in the method, a k-means clustering algorithm is adopted for clustering, and the clustering process pays attention to the following principles:

1) the wind power plant is divided into several independent subsystems according to the wind speed, so that each subsystem has the same wind speed. In order to avoid too many groups and increase the computational complexity, the wind speed is divided into 5 wind speed sections according to the extremely poor wind speed ((fan cut-out wind speed-fan cut-in wind speed)/5). The frequency modulation control mode can be independently applied to each subsystem, and no wake effect exists between fans in each subsystem. Therefore, the fans in each subsystem have approximate working states, the fans can be equivalent to a single fan externally, and the frequency modulation power is averagely distributed internally. Therefore, the frequency modulation characteristics of the whole wind power plant can be approximately reflected by the subsystems, so that the optimization problem is greatly simplified.

2) On the premise of meeting the frequency modulation response capability, the total output power of the wind power plant is maximized as much as possible. Under normal operation, all fans work in the MPPT mode, but due to the wake effect of the wind power plant, the operation state of the whole wind power plant is not meant to be optimal. As can be seen from fig. 6 and 7, by changing the wind direction angle θ, the force bearing area and the rotation speed of the upstream fan blade can be reduced, although the output power is reduced. But correspondingly reduces the sheltering area of the upstream fan, thereby increasing the downstream wind speed and increasing the output power of the downstream fan. Thus, the total output power of the wind farm may be increased.

3) In order to avoid that a large number of fans are concentrated in a certain wind speed area to influence the overall frequency modulation performance of the wind power plant. The number of the fans in each area is corrected by adjusting the wind direction angle theta, and the number of the fans in each wind speed area is balanced as much as possible. FIG. 7 shows a schematic diagram of each regional fan optimization grouping. In 7(a), when θ is μ, the fan number distribution is unbalanced, the upstream area fan number is larger, and the downstream fan number distribution is smaller. When the angle θ is changed, the number of fans in the upstream area can be reduced and the number of fans in the downstream area can be increased, as shown in fig. 7 (b).

Based on the above analysis, the cluster grouping method of the present application is: by adjusting the wind direction angle theta, the number of each group of units is distributed in a balanced manner as much as possible, and the aim of optimizing the total output power of the wind power plant is fulfilled. Namely, it is

Constraint conditions are as follows:

fan power constraints

0≤P_i≤P_i,max(12) Fan pitch angle constraint

β_i,min≤β_i≤β_i,max (13)

Fan speed constraint

ω_i,min≤ω_i≤ω_i,max (14)

System inertia constraints

In the formula: and m is the number of fans in the middle wind speed area.

S3. fan frequency modulation power distribution layer

In an electric power system, an inertia response link is mainly used for releasing kinetic energy through rotor speed adjustment, and all units participate in uncontrollable adjustment. Therefore, the power distribution of the application mainly refers to the frequency and the rotating speed recovery stage of the fan rotor.

A thermal power generating unit is used as a stable and lasting active power increasing power supply, and a wind power plant is used as a rapid and transient auxiliary active power increasing power supply. Therefore, the frequency modulation active power increment of the wind power plant is as follows:

at this time, the output power of the wind power plant is as follows:

P_wind＝ΔP_wind+P_wind-0 (17)

in the formula, P_{wind_0}Is the output power of the wind power plant in steady state, delta P_{wind_max}For maximum power increase, delta P, of wind farm_tAnd after the power of the thermal power generating unit is increased, the system is in shortage.

However, the mechanical power P captured by the fan_mCan be expressed as:

P_m＝0.5ρπC_P(λ,β)R²v³ (18)

in the formula: ρ is the air density; r is the radius of the wind wheel; λ is tip speed ratio; beta is the pitch angle of the fan; c_P(λ, β) is a wind energy utilization coefficient.

In the formula: omega is the rotating speed of the wind turbine generator; and n is the gear box transformation ratio.

At present, the fan mainly adopts a load shedding mode to provide frequency modulation power, and when the fan runs at different wind speeds by a load shedding coefficient of d%, the frequency modulation power provided by the fan is as follows:

according to the formula (20): the single machine frequency modulation output power is related to the wind speed. Therefore, after the wind turbine generators are clustered according to the wind speed, the frequency modulation power delta P of the wind power plant can be obtained_windAnd distributing the wind power clusters one by one according to the cluster frequency modulation capability. After entering the wind power cluster control layer, the frequency modulation capability is basically the same as the fans are at the same wind speed. Therefore, to simplify the calculation, the individual machines in each cluster are equally distributed. The specific calculation method is as follows:

assume that the power increment above the ith cluster (inclusive) is:

in the formula: j the number of participating frequency modulations and i the total number of clusters participating in frequency modulations.

Wherein, the power reference value of each cluster is as shown in formula (22):

when the thermal power unit is in shortage power Delta P_tWhen the sum of the wind speed cluster and the maximum power increase is larger than the sum of the wind speed cluster and the maximum power increase, the fan in the cluster maintains the maximum power increase state; and continuously accumulating the increased power of the next cluster fan until the increased power reaches delta P_tIs less than the maximum active power increase of the fan participating in frequency modulation. At this time, the single-machine power gain involved in the frequency modulation is calculated by equation (20).

Further, a wind power plant real-time rolling clustering method considering wind speed randomness is adopted during cluster control layer optimization; because the wind speed has randomness, the maximum wind speed and the wind direction angle of the wind power plant are different at different moments. These all cause variability and instability in the wind farm frequency modulation capability. By comprehensively considering the factors, the wind power plant is dynamically clustered every 1min according to the real-time running state of the fan in the wind power plant and the wind speed prediction of the 1min resolution, so that the clusters sequentially participate in frequency modulation according to a power distribution strategy. The MPC (model predictive control) adopted by the application is an alternating process of continuous local optimization and rolling control, and comprises three processes of prediction control, error correction and feedback, so that the influence of wind speed randomness can be effectively reduced, and the basic principle of the MPC is shown in FIG. 8. the implementation process is as follows:

1) at the time t and the current state x (t), solving an objective function under a constraint condition by predicting the future state quantity of the system to obtain a control instruction at each time (t +1, t + k delta t) in the future;

2) feeding back the instruction at the t +1 moment to the control system;

3) and updating the x (t +1) state quantity and repeating the steps.

The system feeds wind power operation data and a wind speed predicted value back to the frequency modulation controller every 1min through a fan operation state feedback link to update (9), so that the correctness of a wind power plant cluster is ensured, and the control accuracy of the system is improved. The dynamic cluster of fans is shown in fig. 9.

Compared with the prior art, the invention has the beneficial effects that: according to the multi-time-space layered comprehensive frequency modulation control system based on the fan cluster, firstly, different frequency modulation modes and participation time sequences can be adopted for the fans at different wind speeds according to different running states of the fans at different wind speeds; secondly, aiming at the influence of the wind speed on the frequency modulation control of the fans, a clustering method considering the wake effect of the wind power plant is established, so that the number of the units in each cluster can be distributed in a balanced manner, the integral frequency modulation effect of the wind power plant is improved, and the secondary fluctuation problem of the frequency is avoided; and finally, considering the randomness of the wind speed, optimizing the wind power plant cluster by adopting MPC (multi-control computer) rolling per minute, and ensuring the optimal frequency modulation capability and frequency modulation accuracy of the wind power plant.

Drawings

FIG. 1 is a hierarchical integrated frequency control framework diagram of a wind farm;

FIG. 2 is a schematic view of wind farm time scale coordination;

FIG. 3 is a schematic diagram of spatial scale coordination of a wind farm;

FIG. 4 is a schematic diagram of a cluster frequency modulation mode with different wind speeds;

FIG. 5 is a schematic diagram of coordination control of a cluster frequency modulation mode at different wind speeds;

FIG. 6 is a schematic diagram of wake effect;

FIG. 7 is a schematic diagram of fan cluster grouping correction;

FIG. 8 is a MPC basic schematic;

FIG. 9 is a schematic diagram of a dynamic cluster strategy for wind turbines;

FIG. 10 is a wind farm distribution diagram in an embodiment;

FIG. 11 is a schematic diagram of fan grouping under different clustering methods;

FIG. 12 is a graph of system frequency variation at a wind speed of 13m/s and θ of 15 °;

FIG. 13 is a graph of an active power output increment of the wind turbine generator at a wind speed of 13m/s and θ of 15 °;

FIG. 14 is a diagram of the output power of the thermal power generating unit when the wind speed is 13m/s and theta is 15 degrees;

fig. 15 is a graph showing the change in system frequency at a wind speed of 11m/s and θ of 15 °;

FIG. 16 is a graph of the increment of active power output of the wind farm when the wind speed is 11m/s and θ is 15 °;

FIG. 17 is a graph of thermal power unit output power at 11m/s wind speed and θ equal to 15 °;

fig. 18 is a graph showing the influence of the prediction error on the fan frequency modulation, in which the respective panels are, from left to right, from top to bottom, (a) wind speed, (b) power, (c) 10% derated reserve power, (d) 5% derated reserve power, (e) 10% derated frequency deviation, and (f) 5% derated frequency deviation.

Detailed Description

The present invention will be described in further detail with reference to specific embodiments, and methods or algorithms not specifically described in the embodiments are all prior art.

Examples

The validity of the method is verified by adopting 5 x 5 matrix distribution and a wind power plant consisting of 25 1.5MW wind turbines in total, as shown in FIG. 10; the wind farm parameters are shown in table 1.

TABLE 1 wind farm parameters

1. Fan cluster results

In the embodiment, 4 different scenes that the wind direction angle θ is 0 °, 15 °, 30 ° and 45 ° are given, and two clustering results of the conventional clustering (only considering the wind speed, hereinafter referred to as clustering 1) and the optimized clustering (considering the wake effect, hereinafter referred to as clustering 2) are compared, as shown in table 2.

TABLE 2 clustering grouping results at different θ

Note: the minus sign indicates a decrease.

FIG. 11 is a schematic diagram of fan cluster distribution under the condition of 15 degrees and wind speed of 12m/s, wherein FIG. 11(a) is a diagram of fan distribution and wind speeds of different clusters under cluster 1. FIG. 11(b) is a wind velocity diagram of the wind turbine distribution and different clusters under cluster 2. Different colors in the figure list different cluster fans. The table 3 shows the maximum total output power of the wind power plant under different wind speeds and by the two clustering methods. Wherein the percentage increase is defined as follows:

as can be seen from the comparison in fig. 11, with the cluster 1, the number of the units is unbalanced, and the units are distributed in a U shape with the increase of the wind direction angle θ, and are balanced first, unbalanced in the middle, and then balanced. With cluster 2, the cluster is less affected by θ. The reason is that the cluster 2 considers two factors of the maximum output total power of the wind power plant and the balance of the number of the units, and the influence of theta on the unit distribution is weakened. Thus, in each scenario, the crew re-clusters. It is further found from fig. 11(b) that the unit clusters are gradually distributed from the periphery of the wind farm to the inner layer, because the wind direction is diagonal from #1 to #25, which is consistent with the actual situation.

Table 3 lists the total output power of the wind farm in the MPPT mode for all the fans in

clusters

1 and 2 at different wind direction angles θ. When θ is 15 °, 30 °, 45 °, the percentage of maximum power added is 2.8%. However, when θ is 0 °, at a wind speed of 8m/s, the power fluctuation reaches 15.1%. Compared with other three wind directions, at 0 DEG, the wake effect has the largest influence and the total output power is the lowest. And with the increase of theta, the total output power of the wind power plant is increased under the same wind speed. This is because at 0 °, only 5 fans directly receive ambient wind, and other fans are affected by wake flow, and the wind speed is reduced, thereby reducing the maximum output power of the fans. And when theta is 15 degrees or 30 degrees, the number of the fans directly receiving the ambient wind is 13, so that the influence of the wind speed is greatly reduced. When θ is 45 °, although the number of fans directly receiving ambient wind is 8, the number of cluster 2 units reaches 7, which greatly increases the total output power. This is consistent with the above analysis of changes θ, which can adjust the overall output power of the wind farm.

Meanwhile, with the increase of theta, in the two clustering methods, the wind speed required by the wind power plant to reach the maximum output power is reduced. If theta is 0 DEG and the wind speed is 14m/s, the whole wind power plant reaches the rated output. However, when θ is 45 °, the wind speed is 12m/s, and the rated output can be achieved. This is mainly because the larger the θ angle, the smaller the downstream unit is shielded from the upstream unit. When the angle exceeds 45 DEG, the wind farm will show a downward trend. This is because the wind farm direction changes, which directly accept ambient wind. And when theta is equal to 0 degrees and 45 degrees, the two clustering method machine set clusters are the same, and in the cluster 2, the total power of the wind power output is smaller than that of the cluster 1. This is mainly due to two reasons: 1. when the wind direction theta is 45 degrees, the wake effect is greatly reduced, and the influence on grouping is weakened; 2. in cluster 2, wake effects are considered, and although clusters are not affected, the wind speed is reduced. Thus, the total output power is reduced under the same cluster. When θ is 0 °, the wake effect is large, and therefore the total output power is reduced to 15.1% at 8 m/s.

2. Frequency modulation response analysis

For further analysis, the method for clustering the wake effect of the wind power plant is provided for improving the frequency modulation capability of the wind power plant. Time domain simulations were built by Simulink. At t 5s, the load increases dramatically by 15MW, resulting in a drop in the system frequency. The effectiveness of the system of the present application is illustrated by the following several cases.

1) Ambient wind is high wind speed

Fig. 12 shows the ambient wind speed at 13m/s, θ 15 °. Three strategies: 1. the fan does not participate in frequency modulation; 2. fan frequency modulation based on cluster 1; 3. and (4) clustering 2 based on the fan frequency modulation frequency change curve. FIG. 13 shows frequency-modulated output power curves of groups of wind turbines in a wind farm. FIG. 14 is a thermal power unit power output curve.

As can be seen from fig. 12: compared with the method that the frequency of the wind power plant is not modulated, the clustering method of the cluster 1 and the cluster 2 can improve the frequency response of the system. And because the primary frequency modulation power reserves of the cluster 1 and the cluster 2 are enough, the frequency recovery states are almost the same after the inertia stage is finished, and no secondary frequency drop occurs. But in the inertia response stage, the adjustment effect of the cluster 2 is better because the number of the wind turbines in the wind power plant at the medium wind speed is not enough in the cluster 1, and enough inertia response capability cannot be provided. And under the clustering 2, the number of the fans of the wind power plant at the medium wind speed is large, so that enough inertia can be provided.

After the fans in the wind power plant are clustered (the cluster 1 is at a high wind speed, the clusters 2 to 4 are at a medium wind speed, and the cluster 5 is at a low wind speed), different frequency modulation control modes are adopted for the fans in each cluster according to a site layer coordination control strategy. Although the time for each cluster fan to enter the frequency modulation state is different, the moment for the inertia response unit to enter the frequency modulation state is the same. And because the frequency modulation capability of each cluster fan is different, the frequency modulation support time is also different, as shown in fig. 13. Because the inertia response capability gradually weakens from the cluster 2 to the cluster 4, the response time is gradually shortened, and the frequency modulation system is exited in turn. When the cluster 4 exits the frequency modulation system, the cluster 1 enters a load shedding state, active power is increased, and rotating speed recovery power is provided for the cluster 4 to exit frequency modulation. And after the cluster 2 exits, the cluster 1 is restored to the MPPT state through the adjustment of the pitch angle. In fig. 13, the curve shows the increasing power above zero and the return absorbed power at the rotational speed below zero. As can be seen from fig. 14, when the wind farm participates in frequency modulation, the output power of the thermal power generating unit is smaller than that of the wind farm in the frequency modulation state. Compared with the cluster 1, the wind power plant can provide sufficient inertia response and rotating speed recovery power in the cluster 2 mode without increasing the frequency modulation output power of the thermal power generating unit. Therefore, the output power of the fire electric power generation unit in cluster 2 is smaller than that in cluster 1. For cluster 1, the wind power plant can provide partial inertia response, but a thermal power generating unit auxiliary providing part is needed, so that the output power of the thermal power generating unit is greater than that of cluster 2. However, when the fans in the medium wind speed area quit frequency modulation, the fans in the high wind speed area can provide enough power increasing power through the coordination effect of the fans in the wind power plant, and the medium wind speed fans are assisted to recover, so that the thermal power generating unit has no subsequent power fluctuation. The coordination control strategy provided by the application can effectively reduce the frequency modulation active power of the thermal power generating unit and reduce the frequency modulation cost of the system. Meanwhile, the clustering method provided by the application is superior to a simple wind speed clustering method.

2) The ambient wind is wind with moderate wind speed

Fig. 15 to 17 are simulation results at a wind speed of 11m/s and θ of 15 °. As can be seen from fig. 15, at this wind speed, 21 fans (#1 to #21) in cluster 1 are all at the medium wind speed, and 16 fans are included in cluster 1, the grouping effect is weakened, and when the frequency modulation is exited, a short power drop occurs, as shown in fig. 17, and the system frequency drops again. This situation will be more severe as the number of units in a cluster increases. By adopting the clustering method provided by the invention, the number of certain cluster units is weakened, the problem of secondary frequency drop caused by the fact that a large number of fans exit the system at the same time can be avoided, and the frequency stability of the system is improved. Compared with the fan frequency modulation, the cluster 1 and the cluster 2 can improve the lower limit of the system frequency. And the lower limit values are almost the same, because the inertia response capabilities of cluster 1 and cluster 2 are similar, the frequency fluctuation after the power disturbance is almost the same. As can be seen from fig. 17, when the ambient wind is a medium wind speed, no fan in the wind farm is in the high wind speed region. Therefore, the fan rotating speed recovery power needs to be provided by the thermal power generating unit, but the power increase of the thermal power generating unit is almost the same in the cluster 1 and the cluster 2. In the cluster 1, the fans are quitted more at the same time, and the thermal power generating unit increases power suddenly. However, the thermal power generating unit is limited by climbing, and cannot meet the requirement of rotating speed recovery instantly, so that the frequency fluctuation phenomenon occurs. After the clustering grouping method is adopted, the number distribution of the units in the cluster is uniform, the phenomenon that a large number of units exit the system at the same time is avoided, for example, the output power of the hot-electric units in fig. 17 is gradually increased, step disturbance is avoided, and the problem of frequency fluctuation can be avoided.

3. Rolling cluster analysis

The wind power plant fan cluster division is influenced by wind speed and wind direction angles. In this example, the frequency modulation effect of the wind farm in the fixed cluster and the rolling cluster will be compared. The rolling calculation period is 1 min. Fig. 18 and table 4 show the wind farm scheduling control results within 60min using the above 2 clustering methods.

TABLE 4 comparative analysis of two modes

As can be seen from fig. 18(a) to 18(d), the rolling clustering method error value is smaller than the fixed clustering method. Especially after 52min, the wind speed fluctuation is more obvious in a period of time. According to the formula (20), when the load is reduced by 10% and 5%, the deviation of the wind power plant frequency modulation reserve power under the two clustering modes is shown in fig. 18(c) and 18(d), the deviation value of the fixed cluster is obviously larger than that of the rolling cluster according to the corresponding frequency modulation deviation curve shown in fig. 18(e) and 18 (f). Meanwhile, in combination with table 4, it is found that the deviation of 10% of the load shedding is greater than 5% in both modes, because the larger the load shedding coefficient is, the more the standby power participating in the frequency modulation is, however, the larger the deviation of the generated frequency modulation standby capacity is, so that the larger the insufficient amount of the generated frequency modulation power is in the frequency modulation process.

4. Conclusion

The rolling clustering method considering wake effect and wind speed randomness and a wind power plant space-time layering coordination control strategy in a large-scale wind power plant are mainly researched. By way of example analysis, the following conclusions are reached:

1) in a 5 x 5 matrix wind farm case study, the units were grouped into 5 clusters under test wind speed conditions. The wind power plant fan cluster is influenced by a wind direction angle theta, and the influence of the wake effect on the grouping is gradually reduced along with the increase of theta.

2) A cluster optimization method considering the wake effect of a wind power plant. The problem of complex control from one fan to another can be solved, and the problem that a large number of fans are concentrated in a certain wind speed interval can be avoided. And further, the problem that secondary frequency fluctuation is caused by system power fluctuation when a certain cluster fan exits is avoided.

3) When the ambient wind speed is higher than or equal to 12m/s, the independent control of the frequency modulation of the wind power plant can be realized through the coordination control among clusters in the wind power plant, and the frequency modulation pressure of the thermal power generating unit is reduced. When the ambient wind speed is the medium wind speed of less than 12m/s, although the wind power plant does not contain the high wind speed area cluster, the wind power plant can inject enough inertia frequency modulation power in the early stage of frequency modulation. When quitting, the method adopts a strategy of quitting in sequence according to the clusters, provides enough time for the climbing of the thermal power generating unit, and avoids the problem of secondary fluctuation of frequency caused by quitting of a large number of fans.

4) Through the cluster strategy of rolling every minute, the influence of the randomness of the wind speed on the frequency modulation capability and the control precision of the wind power plant can be effectively reduced.

In this embodiment, a simpler regularly distributed wind farm model is used. In practical applications, the complexity of the model increases the complexity of the cluster. At the moment, an advanced clustering algorithm is adopted, so that the influence caused by model errors can be reduced. This will be our work in the future. However, the rolling cluster optimization idea considering the wake effect provided by the invention can be applied to actual engineering so as to improve the accuracy of frequency modulation control of the wind power plant.

It should be understood that the present invention may be embodied in many different forms and should not be construed as limited to the specific embodiments set forth herein, but rather should be construed as limited to those embodiments set forth herein.

Claims

1. A multi-time-space layered comprehensive frequency modulation control system based on a fan cluster is characterized in that: the system comprises a station layer, a cluster control layer and a fan frequency modulation power distribution layer; carrying out optimization design aiming at each layer;

s1, station layer

in the formula: s_NIs the rated capacity of the generator;

analogy formula (2) can obtain virtual inertia H of wind power plant_W：

in the formula: delta E_opTotal power at the rotor side of the fan, P_ATracking point power for pre-frequency-modulated power; p_e(t) is the output electromagnetic power, P_w(t) inputting mechanical power; delta E_kFor total release of kinetic energy of the rotor, J for total moment of inertia of the unit, Delta E_lossAdditional lost wind energy to the rotor due to reduced rotational speed; t is t_onAnd t_offRespectively are the moment of frequency modulation starting and inertia quitting;

1) according to the actual wind speed of the wind power plant, firstly clustering the fans into the same group according to the wind speed, and distributing the same control variable to the fans in the same group; meanwhile, in order to ensure that the fan cannot cause the accident of cutter cutting due to the over-low rotating speed, the lowest rotating speed value of the rotor of the fan is kept above 0.7 pu; dividing fans with the wind speed of more than 7m/s into a group at intervals of 1 m/s;

2) the number of the fans at different wind speed stages is uniformly distributed, and the number of the fans is uniformly controlled at different wind speeds by changing the wake effect of the wind power plant;

3) when the system frequency is detected to fluctuate, the high-wind speed unit and the low-wind speed unit keep an MPPT operation mode, and a fan in a medium-wind speed area quickly responds to provide inertia response; when the inertia response judging link detects a medium wind speed fan, | delta ω | <4 × 10-7 or df/dt ═ 0, the inertia link is ended, the medium wind speed fan exits the frequency modulation system, the rotating speed recovery stage is entered, and the system enters the primary frequency modulation stage; at the moment, the high-wind-speed unit enters a frequency modulation mode, and power and primary frequency modulation power required by the exit of the medium-wind-speed fan are provided through the adjustment of the pitch angle; if the number of the high-wind-speed fans in the wind power plant is insufficient, the partial power is provided by the thermal power generating unit;

s2, cluster control layer

According to the wake effect principle, let r be assumed₀Is the radius of the rotating area of the fan blade, x is the longitudinal distance between the upstream fan and the downstream fan along the wind direction, d is the transverse distance, A₀The wind speed is a fan blade rotating area, namely a wind speed influence area; a. the_sThe shielding area is an effective area which is influenced by an upstream fan on a downstream fan; radius r_iIs the wake expansion effect radius; from equation (8), r increases with the wake expansion effect angle α and/or the longitudinal distance_iWill increase, the area of influence increases;

r_i＝r₀+x*tanα,(0°≤α≤45°) (8)；

the jth downstream region wind speed considering the wake effect is:

wherein the content of the first and second substances,

in the formula: c_TiThe wind energy utilization coefficient of the ith fan is determined by the tip speed ratio lambda and the pitch angle beta; it is influenced by the working state of the upstream fan;

selecting the working state of the fan as a cluster clustering index, and clustering by adopting a k-means clustering algorithm;

s3. fan frequency modulation power distribution layer

The thermal power generating unit is used as a stable and lasting active power increasing power supply, and the wind power plant is used as a rapid and transient auxiliary active power increasing power supply; the wind power plant frequency modulation active power increase is as follows:

at this time, the output power of the wind power plant is as follows:

P_wind＝ΔP_wind+P_wind-0 (17)；

in the formula, P_{wind_0}Is the output power of the wind power plant in steady state, delta P_{wind_max}For maximum power increase, delta P, of wind farm_tThe system is used for increasing the system shortage power of the thermal power generating unit;

mechanical power P captured by a fan_mExpressed as:

P_m＝0.5ρπC_P(λ,β)R²v³ (18)；

in the formula: ρ is the air density; r is the radius of the wind wheel; λ is tip speed ratio; beta is the pitch angle of the fan; c_P(lambda, beta) is the wind energy utilization coefficient;

in the formula: omega is the rotating speed of the wind turbine generator; n is the gear box transformation ratio;

according to the formula (20): the single machine frequency modulation output power is related to the wind speed, and the single machines in each cluster are evenly distributed for simplifying calculation.

2. The multi-space-time layered comprehensive frequency modulation control system based on the fan cluster as claimed in claim 1, wherein: when optimizing the cluster control layer, the implementation process considers the following principles:

1) dividing the wind power plant into several independent subsystems according to the wind speed, so that each subsystem has the same wind speed; in order to avoid too many groups and increase the calculation complexity, the wind speed is divided into 5 wind speed sections according to the extreme difference of the wind speed;

2) on the premise of meeting the frequency modulation response capability, the total output power of the wind power plant is maximized as much as possible;

3) in order to avoid that a large number of fans are concentrated in a certain wind speed area to influence the overall frequency modulation performance of the wind power plant; the number of the fans in each area is corrected by adjusting the wind direction angle theta, and the number of the fans in each wind speed area is balanced as much as possible.

3. The system of claim 1, wherein when a control layer of an S2 cluster is optimized, the number of each group of units is distributed as evenly as possible by adjusting a wind direction angle theta, and the aim of optimizing the total output power of a wind power plant is achieved; namely, it is

Constraint conditions are as follows:

fan power constraints

0≤P_i≤P_i,max (12)；

Fan pitch angle constraint

β_i,min≤β_i≤β_i,max (13)；

Fan speed constraint

ω_i，min≤ω_i≤ω_i，max (14)；

System inertia constraints

In the formula: and m is the number of fans in the middle wind speed area.

4. The multi-space-time layered comprehensive frequency modulation control system based on the fan cluster as claimed in claim 1, wherein: the specific calculation method for performing average distribution on the single machines in each cluster is as follows:

assume that the power increment above the ith cluster (inclusive) is:

Wherein, the power reference value of each cluster is as shown in formula (22):

when the thermal power unit is in shortage power Delta P_tWhen the sum of the wind speed cluster and the maximum power increase is larger than the sum of the wind speed cluster and the maximum power increase, the fan in the cluster maintains the maximum power increase state; and continuously accumulating the increased power of the next cluster fan until the increased power reaches delta P_tThe maximum active power increase of the cluster fan participating in frequency modulation is less than that of the cluster fan participating in frequency modulation; at this time, the single-machine power gain involved in the frequency modulation is calculated by equation (20).

5. The multi-space-time layered comprehensive frequency modulation control system based on the fan cluster as claimed in claim 1, wherein: when the cluster control layer is optimized, a wind power plant real-time rolling cluster method considering wind speed randomness is adopted, namely, according to the real-time running state of a fan in the wind power plant and the wind speed prediction of 1min resolution, the wind power plant is dynamically clustered every 1min, so that the clusters sequentially participate in frequency modulation according to a power distribution strategy.

6. The multi-space-time layered comprehensive frequency modulation control system based on the fan cluster as claimed in claim 5, wherein: the wind power plant real-time rolling clustering method considering the wind speed randomness is continuously optimized by adopting an MPC method, and the MPC method comprises three processes of prediction control, error correction and feedback, so that the influence of the wind speed randomness is effectively reduced.

7. The multi-space-time layered comprehensive frequency modulation control system based on the fan cluster as claimed in claim 6, wherein: MPC is implemented as follows:

2) feeding back the instruction at the t +1 moment to the control system;

3) and updating the x (t +1) state quantity and repeating the steps.