Photovoltaic participation in grid frequency regulation based on droop and model prediction

Abstract

The application discloses a photovoltaic participation in grid frequency modulation method based on droop and model prediction. The method is: on the basis of photovoltaic active reserve, the output increment of droop control is changed by frequency difference, and the output increment of model prediction control is changed by frequency deviation rate, so that the output power of photovoltaic power generation is changed, and the photovoltaic power generation participates in grid frequency regulation. The application starts from the connection between active power and frequency, establishes a prediction model by using the rotor motion equation of a synchronous generator on the basis of photovoltaic active reserve, combines with active droop control, changes the output increment of droop control by frequency difference, changes the output increment of model prediction control by frequency deviation rate, so that the output power of photovoltaic power generation is changed, the photovoltaic power generation participates in grid frequency modulation, and the method not only improves transient and dynamic effects, but also adjusts the steady state.

Description

Technical Field

[0001] The present invention relates to the field of photovoltaic power generation system grid-connected frequency regulation, and in particular to a method for photovoltaic participation in grid frequency regulation based on droop and model prediction. Background Art

[0002] To achieve sustainable development of human society, the application of renewable energy such as wind and solar energy has been continuously expanded. According to statistics from the National Energy Administration, by the end of 2020, China's installed photovoltaic power generation capacity reached 223 million kW, an increase of 18.7 million kW over the previous year. It is estimated that by 2030, the installed photovoltaic power generation capacity will reach 1,050 GW. The frequency of traditional power systems is determined by the rotor speed of synchronous machines. The rotor has a large rotational inertia, which can provide power support, suppress grid frequency fluctuations, and have good frequency response characteristics. However, with the gradual increase in photovoltaic power generation, the proportion of the total capacity of conventional generator sets in the system has gradually decreased, reducing the reserve capacity of the grid's primary frequency regulation resources, resulting in a gradual decrease in the rotational inertia of the traditional power system and weakening the grid's frequency regulation capabilities.

[0003] Furthermore, because photovoltaic power generation systems lack inherent inertia, their power output does not respond to changes in system frequency and does not participate in the grid's primary frequency regulation. Furthermore, large-scale photovoltaic power generation grid integration squeezes the space available for synchronous generators with rotational inertia, leading to reduced grid inertia and insufficient frequency regulation capabilities. Therefore, there is an urgent need for photovoltaic power generation to actively participate in grid frequency regulation. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for photovoltaic participation in grid frequency regulation based on droop and model prediction. This method can better follow the changes in frequency to change the output power of photovoltaic power generation, so that it can participate in the frequency regulation of the grid, which not only improves the transient and dynamic effects, but also regulates the steady state.

[0005] The technical solution of the present invention is to use frequency deviation and frequency deviation rate to perform power regulation on the basis of active standby, and the process includes the following steps:

[0006] Step S1, phase-locked loop and coordinate transformation: sampling the grid-side current and voltage to obtain the three-phase inverter-side current I a , I b , I c With voltage U ga 、U gb 、U gc , for the grid voltage U ga 、U gb 、U gcPhase-locked processing is performed to obtain the grid electrical angle θ, grid frequency f and grid angular velocity ω, and the grid current and voltage are transformed by abc / dq to obtain I d , I q With E d 、E q , and calculate the active power P at the same time;

[0007] Step S2, DC voltage control: DC side voltage reference value U dcref with U dc The error signal is passed through the PI controller to generate a duty cycle d, which in turn generates a drive signal to drive the Boost converter to work;

[0008] Step S3, frequency modulation control: This part consists of a control strategy unit and a power tracking unit. The control strategy unit can compare the f obtained in step S1 with the reference frequency f d The output increment ΔP2 of the droop control and the output increment ΔP1 of the model predictive control are calculated, and the power P after the photovoltaic active reserve is combined o , get the FM power P * ; The power tracking unit is to obtain the frequency modulation power P * Based on the sampling of photovoltaic output voltage U PV , output current I PV , and then the photovoltaic output reference voltage U is obtained by the power tracking algorithm PVref , then with the photovoltaic output voltage U PV After comparison, the reference value of active power P is obtained through the PI controller. ref ;

[0009] Step S4, dual-loop control: P ref and P are the reference value and actual value of active power respectively; Q ref and Q are the reference value and actual value of reactive power respectively, where the reactive reference power Q ref =0; the power outer loop control generates d-axis and q-axis current command values ​​according to the requirements of active and reactive power respectively; the grid-side current is used as the feedback quantity, and after the PI controller and voltage feedforward, the voltage U is obtained. d , U q Then, the control signal voltage U is obtained by inverse transformation from the dq coordinate system to the abc coordinate system. a , U b , U c Finally, a drive signal is generated to control the operation of the inverter switch tube:

[0010] Step S3: FM power P * The specific formula is as follows:

[0011] P * =P o+ ΔP1+ ΔP2

[0012] where P o The formula is as follows:

[0013] P o = P max (1-δ)

[0014] where P max is the maximum power of photovoltaic power generation output, and δ is a photovoltaic load shedding coefficient.

[0015] Step S3 is to express the output increment ΔP1 of the model predictive control, and the expression is as follows:

[0016]

[0017] wherein:

[0018]

[0019]

[0020]

[0021] ω′=ω-ω0

[0022] ω′ p (k+1|k)=[ω′(k+1) ω′(k+2) ω′(k+3)] T

[0023] ΔU(k)=[ΔT m (k) ΔT m (k+1) ΔT m (k+2)] T

[0024]

[0025] I=[1 1 1] T

[0026] S d =[B cd B cd +B cd 2 B cd +B cd 2 +B cd 3 ] T

[0027]

[0028] Q=diag(q,q,q),(q>0)

[0029] R = diag(r, r, r), (r > 0)

[0030] Where: J is the moment of inertia; D is the damping coefficient; T s is the system sampling time; ω is the grid synchronization angular velocity; ω0 is the mechanical angular velocity of the synchronous generator; Δω′ is the grid angular frequency change increment; ΔT e is the electromagnetic torque increment; ΔT m is the mechanical torque increment; q is the angular frequency difference weight coefficient; r is the mechanical torque increment weight coefficient.

[0031] Compared with the prior art, the present invention has the following significant advantages: (1) considering that the inverter structure of the photovoltaic power generation system can be equivalent to the synchronous generator structure, the frequency prediction model is established using the synchronous generator rotor motion equation, and a control method combining model predictive control and droop control is proposed; (2) compared with the traditional droop control strategy, the method proposed in the present invention has a significant improvement on the transient effect, a faster dynamic adjustment speed, and does not affect the steady-state value of the system frequency; (3) considering the change of light intensity, the method can still achieve good results. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 This is a control strategy diagram for photovoltaic power generation to actively participate in grid frequency regulation.

[0033] Figure 2 This is a flow chart of the photovoltaic frequency regulation power tracking algorithm based on active reserve.

[0034] Figure 3 It is the regional power grid simulation topology diagram.

[0035] Figure 4 This is the simulation waveform when the load suddenly increases by 100MW.

[0036] Figure 5 This is the simulation waveform when the load suddenly decreases by 100MW.

[0037] Figure 6 (a) is the light intensity change diagram; (b) is the frequency waveform diagram when the light intensity changes. DETAILED DESCRIPTION

[0038] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0039] Combine Figure 1 The present invention is suitable for photovoltaic power generation systems to participate in the frequency regulation of the power grid. In the active loop control link, the photovoltaic output reference power P ref, the output increment ΔP2 of the droop control and the output increment ΔP1 of the model predictive control, and the power P after the photovoltaic active reserve o The FM power P is obtained by superposition * , and then after the power tracking algorithm, the reference value of active power P can be obtained ref .

[0040] The present invention will be further described in detail below with reference to the embodiments.

[0041] The technical solution of the present invention is: a method for photovoltaic participation in grid frequency regulation based on droop and model prediction, characterized in that, on the basis of active reserve, frequency deviation and frequency deviation rate are used to perform power regulation, and the process includes the following steps:

[0042] Step S1, phase-locked loop and coordinate transformation: sampling the grid-side current and voltage to obtain the three-phase inverter-side current I a , I b , I c With voltage U ga 、U gb 、U gc , for the grid voltage U ga 、U gb 、U gc Phase-locked processing is performed to obtain the grid electrical angle θ, grid frequency f and grid angular velocity ω, and the grid current and voltage are transformed by abc / dq to obtain I d , I q With E d 、E q , and calculate the active power P at the same time.

[0043] Step S2, DC voltage control: DC side voltage reference value U dcref with U dc The error signal is passed through the PI controller to generate a duty cycle d, which in turn generates a drive signal to drive the Boost converter to work;

[0044] Step S3, frequency modulation control: This part consists of a control strategy unit and a power tracking unit. The control strategy unit can compare the f obtained in step S1 with the reference frequency f d The output increment ΔP2 of the droop control and the output increment ΔP1 of the model predictive control are calculated, and the power P after the photovoltaic active reserve is combined o , get the FM power P * ; The power tracking unit is to obtain the frequency modulation power P * Based on the sampling of photovoltaic output voltage U PV , output current I PV , and then the photovoltaic output reference voltage U is obtained by the power tracking algorithm PVref, then with the photovoltaic output voltage U PV After comparison, the reference value of active power P is obtained through the PI controller. ref .

[0045] Step S4, dual-loop control: P ref and P are the reference value and actual value of active power respectively; Q ref and Q are the reference value and actual value of reactive power respectively, where the reactive reference power Q ref =0; the power outer loop control generates d-axis and q-axis current command values ​​according to the requirements of active and reactive power respectively; the grid-side current is used as the feedback quantity, and after the PI controller and voltage feedforward, the voltage U is obtained. d , U q Then, the control signal voltage U is obtained by inverse transformation from the dq coordinate system to the abc coordinate system. a , U b , U c , and finally further generates a driving signal to control the operation of the inverter switch tube.

[0046] Further, in step S3, the frequency modulation power P * The specific formula is as follows:

[0047] P * =P o +ΔP1+ΔP2 (1)

[0048] Among them, P o The formula is as follows:

[0049] P o =P max (1-δ) (2)

[0050] Where P max is the maximum power output of photovoltaic power generation, and δ is the photovoltaic load reduction coefficient.

[0051] The detailed steps of the output increment ΔP2 of the droop control are as follows:

[0052] According to step S1, the grid frequency f can be obtained. According to the droop control strategy, the output increment ΔP2 can be obtained, which is expressed as follows:

[0053]

[0054] Where P n is the rated active power of the photovoltaic array; K is the frequency modulation control gain, which is divided into the upper value (underfrequency) K1 and the lower value (overfrequency) K2; f is the current measurement frequency, f L1 The system under-frequency action dead zone threshold is 49.96, f H1 The system over-frequency action dead zone threshold is 50.03.

[0055] The detailed steps of the output increment ΔP1 of the model predictive control are as follows:

[0056] According to step S1, the grid angular velocity ω and active power P can be obtained. From the rotor motion equation of the synchronous generator, it can be known that:

[0057]

[0058] Where: P e is the electromagnetic power; J is the moment of inertia; D is the damping coefficient; ω is the synchronous angular velocity of the grid; ω0 is the mechanical angular velocity of the synchronous generator.

[0059] Define variables ω′, T m , T e , ω′ is as follows:

[0060]

[0061] According to formula (4) and formula (5), formula (6) can be obtained:

[0062]

[0063] Since all states of the system can be measured, in order to reduce static errors, the control model under the continuous state is discretized and changed to an incremental model, and the formula is as follows:

[0064]

[0065] Where: Δω′(k) is the controlled output increment, and its physical meaning is the rate of change of the system angular frequency; T s is the system sampling time.

[0066] For the convenience of calculation, formula (8) is defined as follows:

[0067]

[0068] The system output ω′(k) can be expressed as:

[0069] ω′(k)=Δω′(k)+ω′(k-1) (9)

[0070] Define the prediction time domain as p, the control time domain as m, and m≤p.

[0071] Because outside the control time domain, the controlled electromagnetic torque ΔT m unchanged, that is:

[0072] ΔT m (k+i)=0 i=m,m+1…p-1 (10)

[0073] Measurable interference quantity: electromagnetic torque ΔT e After time k, it remains unchanged, that is:

[0074] ΔT e (k+i)=0i=1,2…p-1 (11)

[0075] The present invention assumes that the prediction time domain is equal to the control time domain and is 3, that is, m=p=3. According to formulas (7), (8), and (9), the output prediction equation can be obtained as follows:

[0076] ω′ p (k+1|k)=S ω Δω′(k)+Iω′(k)+S d ΔT e (k)+S u ΔU(k) ​​(12)

[0077] Where:

[0078]

[0079] The control goal is to obtain the output increment ΔP1 of the model predictive control, so the objective function can be defined as formula (14):

[0080]

[0081] The weight coefficients Q and R matrices are expressed as follows:

[0082]

[0083] Wherein, the output weight coefficient q and the control weight coefficient r.

[0084] Finally, according to the objective function, the optimal control sequence solution at time k can be obtained, that is, the optimal control amount ΔT m The sequence solution is:

[0085]

[0086] According to the basic principle of model predictive control, the first element obtained by the optimization solution acts on the control system, and the output increment ΔP1 of the model predictive control is transformed as follows:

[0087] ΔP1=

[100] ΔU(k)*314 (17)

[0088] Furthermore, the photovoltaic output reference power P in step S3 is ref , is obtained in the control strategy unit P * Based on the sampling of photovoltaic output voltage U PV , output current I PV, and then the photovoltaic output reference voltage U is obtained by the power tracking algorithm PVref , then with the photovoltaic output voltage U PV After comparison, the reference value of active power P is obtained through the PI controller. ref , the specific algorithm flow chart is as follows Figure 2 shown.

[0089] Example 1

[0090] This embodiment is based on MATLAB / Simulink to build Figure 3 The regional power grid simulation model shown in Figure 1 shows a PV power station with a rated capacity of 125MW, an active reserve of δ=20%, a capacity of 198MW for each synchronous generator set, and an active power consumption of 480MW for the load. Figure 3 The overall control of a single photovoltaic power generation unit in a photovoltaic power station is as follows: Figure 1 As shown in Figure 1, the control method is a photovoltaic participation in grid frequency regulation method based on droop and model prediction. The simulation parameters are shown in Table 1.

[0091] Table 1 Simulation parameters

[0092]

[0093] Condition 1: The output power of the photovoltaic power plant remains unchanged for a short period of time and is in a power-limited operation state. When the simulation time is t=8s, the load suddenly increases by 100MW. The simulation results are shown in the figure below. Figure 4 shown.

[0094] Depend on Figure 4 It can be seen that: 1) When photovoltaic power generation adopts a droop control strategy based on active reserve, the frequency setting will control the photovoltaic power generation to increase the active power generated according to the set active power-frequency curve to participate in the frequency regulation of the power grid. The frequency fluctuates downward, and the system minimum frequency is lower than 49.77Hz. After 10 seconds, the frequency stabilizes again, and the stable frequency is approximately 49.9Hz. 2) When the control strategy proposed by this invention is adopted, the system minimum frequency is lower than 49.8Hz, and the transient effect is better. After 9 seconds, the frequency stabilizes again, and the stable frequency remains consistent with the droop control.

[0095] Condition 2: The output power of the photovoltaic system remains unchanged for a short period of time and is in a power-limited operation state. When the simulation time is t=8s, the load suddenly decreases by 100MW. The simulation results are shown in the figure below. Figure 5 shown.

[0096] Depend on Figure 5The results show that: 1) When the photovoltaic power generation adopts the active reserve droop control strategy, the load suddenly drops by 100MW, and the frequency fluctuates upward. The transient effect is better than that under constant power control, and the frequency stabilizes after 15 seconds, reaching approximately 50.1Hz. 2) It is clearly seen that compared with droop control, the control strategy proposed in this invention achieves better transient effect, and the frequency stabilizes after 12 seconds, remaining consistent with the droop control.

[0097] Condition 3: Since the output power of photovoltaic cells varies greatly with light intensity but is less affected by temperature, the present invention only analyzes the scenario of sudden changes in light intensity when considering the characteristics of the photovoltaic source end.

[0098] Depend on Figure 6 It can be seen that when the light intensity changes, the frequency waveform obtained by using the control strategy proposed in this invention improves the transient effect of the frequency within the frequency dead zone threshold (49.96Hz-50.03Hz), and the dynamic adjustment speed is faster, while the steady-state value remains unchanged. Outside the frequency dead zone threshold (49.96Hz-50.03Hz), not only the transient and dynamic adjustment effects are improved, but the steady-state frequency is also regulated. In summary, the control strategy proposed in this invention remains effective when considering changes in photovoltaic intensity.

[0099] In summary, the present invention proposes a method for photovoltaic participation in grid frequency regulation based on droop and model prediction. This method can better follow frequency changes to change the output power of photovoltaic power generation, allowing it to participate in grid frequency regulation. Compared with traditional droop control, this method improves transient and dynamic effects and regulates steady state. At the same time, considering the impact of changes in light intensity, the method of the present invention remains effective, and can regulate transient and dynamic effects of frequency within the frequency threshold, and can also improve steady state outside the frequency threshold.

Claims

1. A method for photovoltaic participation in grid frequency regulation based on droop and model prediction, characterized in that: On the basis of active standby, frequency deviation and frequency deviation rate are used to perform power regulation. The process includes the following steps: Step S1, phase-locked loop and coordinate transformation: sampling the AC side grid current I a , I b , I c , AC side grid voltage U ga 、U gb 、U gc , for the grid voltage U ga 、U gb 、U gc Perform phase-locking processing to obtain the grid electrical angle θ, grid frequency f and grid angular velocity ω, and perform abc / dq transformation on the grid current to obtain I d , I q , perform abc / dq transformation on the grid voltage to obtain E d 、E q , according to I d , I q 、E d 、E q Calculate the active power P; Step S2, DC voltage control: DC side voltage reference value U dcref with U dc The error signal is passed through the PI controller to generate a duty cycle d, which in turn generates a drive signal to drive the Boost converter to work; Step S3, frequency modulation control: This part consists of a control strategy unit and a power tracking unit. The control strategy unit can compare the f obtained in step S1 with the reference frequency f d The output increment ΔP2 of the droop control and the output increment ΔP1 of the model predictive control are calculated, and the power P after the photovoltaic active reserve is combined o , get the FM power P * ; The power tracking unit is to obtain the frequency modulation power P * Based on the sampling of photovoltaic output voltage U PV , output current I PV , the photovoltaic output reference voltage U is obtained by the power tracking algorithm PVref , then with the photovoltaic output voltage U PV After comparison, the reference value of active power P is obtained through the PI controller. ref ; Step S4, dual-loop control: P ref and P are the reference value and actual value of active power respectively; Q ref and Q are the reference value and actual value of reactive power respectively, where the reactive reference power Q ref =0; the power outer loop control generates d-axis and q-axis current command values ​​according to the requirements of active and reactive power respectively; the grid-side current is used as the feedback quantity, and after the PI controller and voltage feedforward, the voltage U is obtained. d , U q Then, the control signal voltage U is obtained by inverse transformation from the dq coordinate system to the abc coordinate system. a , U b , U c , and finally further generates a driving signal to control the operation of the inverter switch tube; The FM power P in step S3 * The specific formula is as follows: * =P o +ΔP1+ΔP2; Among them, P o The formula is as follows: o =P max (1-δ); Where P max is the maximum power output of photovoltaic power generation, δ is the photovoltaic load reduction coefficient; Step S3 is the output increment ΔP1 of the model predictive control, which is expressed as follows: in: ω'=ω-ω0; oh p (k+1|k)=S ω Dω'(k)+Iω'(k)+S d ΔT e (k)+S u ΔU(k); ΔU(k)=[ΔT m (k) ΔT m (k+1) ΔT m (k+2)] T ; I＝[1 1 1] T ； S d =[B cd B cd +B cd 2 B cd +B cd 2 +B cd 3 ] T ; Q = diag(q,q,q), (q>0); R = diag(r, r, r), (r > 0); Where: J is the moment of inertia; D is the damping coefficient; T s is the system sampling time; ω is the grid synchronization angular velocity; ω0 is the mechanical angular velocity of the synchronous generator; Δω′ is the grid angular frequency change increment; ΔT e is the electromagnetic torque increment; ΔT m is the mechanical torque increment; q is the angular frequency difference weight coefficient; r is the mechanical torque increment weight coefficient.

Images