A method for dynamically setting active power droop coefficient of distributed power supply controller

Abstract

The present invention discloses a method for dynamically setting the active power droop coefficient of a distributed power supply controller, which comprises: determining the distributed power supplies in a region, collecting the rated active power P of all distributed power supplies, i，e ; Establish a historical data set; perform data cleaning on the historical data set and construct an optimized data set; use support vector machine to establish an active power droop coefficient calculation model and perform training regression; determine whether power oscillation occurs. If power oscillation occurs, the active power droop coefficient m i Set to 0, the distributed energy power maintains the original power generation power, and no droop characteristic adjustment is performed. This method applies a variety of existing simulation technologies on site, collects and analyzes actual operating parameters, selects data with good results, and uses support vector machines to train and iterate the active power droop coefficient calculation model, and considers the power fluctuation problem that may be caused by the dynamic change of the droop coefficient to meet the requirements of grid frequency regulation and stable operation.

Description

Technical Field

[0001] The present invention relates to the technical field of distributed power sources, and in particular to a method for dynamically setting an active power droop coefficient of a distributed power source controller. Background Art

[0002] Today, with the increasing shortage of conventional energy and the increasingly serious environmental problems, distributed power sources based on renewable energy such as solar energy and wind energy have been widely valued and applied. Distributed power sources are connected to the low-voltage distribution network through controllers to achieve local consumption, with the advantages of small grid losses and adjustable distribution network voltage.

[0003] At present, the controllers of distributed power sources mostly adopt droop control, which is a method in which multiple controllers in the region jointly assume frequency control. By setting the linear function relationship between power and frequency given by the droop coefficient, all controllers can balance the power fluctuations generated by the load. Since the actual power generation of distributed energy varies with the change of meteorological conditions during operation, it needs to be considered in the dynamic adjustment technology of the active power droop coefficient. The existing technology proposes a method of adaptively adjusting the value of the droop coefficient by introducing it into an odd function with the change of the difference between the rated load power and the actual load power, and a method of introducing the power deviation and the voltage deviation into the series correction link of the traditional droop control strategy, and sending the output active power information to the central controller, automatically adjusting the respective PV droop coefficients through the PI regulator, and proposing that as the output power moves away from the reference value, the droop coefficient will decrease with the sine function, and proposing that the droop coefficient will automatically adjust with the change of line impedance and load. The above technology takes into account the change of the actual power generation of distributed power sources, but it is a laboratory simulation result, and the actual operation effect on site has not been verified and iteratively optimized, and the power fluctuations caused by the dynamic change of the droop coefficient have not been considered, which leads to the problem of not being able to meet the grid regulation requirements and stable operation. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0005] In view of the problems existing in the above-mentioned existing method for dynamically setting the active power droop coefficient of the distributed power supply controller, the present invention is proposed.

[0006] Therefore, an object of the present invention is to provide a method for dynamically setting the active power droop coefficient of a distributed power supply controller.

[0007] To solve the above technical problems, the present invention provides the following technical solutions: determine the distributed power sources in the area, collect the rated active power P of all distributed power sources i，e ;

[0008] Building historical data sets;

[0009] Performing data cleaning on the historical data set to construct an optimized data set;

[0010] Use support vector machine to establish active power droop coefficient calculation model and perform training regression;

[0011] Determine whether power oscillation occurs. If power oscillation occurs, the active power droop coefficient m i If it is set to 0, the distributed energy power maintains the original power generation power and does not adjust the droop characteristics.

[0012] As a preferred solution of the method for dynamically setting the active power droop coefficient of the distributed power supply controller of the present invention, wherein: the distributed power supply in the area is a distributed energy source connected to the power grid in the area and participating in the power grid regulation, including photovoltaic power generation and wind power generation;

[0013] The division range is the substation under a transformer in the 35kV power grid.

[0014] As a preferred solution of the method for dynamically setting the active power droop coefficient of the distributed power supply controller of the present invention, the establishment of the historical data set includes:

[0015] Establish historical data set θ={P i，e , P i，α , U i，α , Hz i，α , m i，α}, including rated power generation P i，e , real-time power generation P i，α , real-time voltage U i，α , real-time frequency Hz i，α , real-time active power droop coefficient m i，α， α is the absolute time tag when the data is collected;

[0016] The active power droop coefficient m is dynamically set by introducing the droop coefficient into an odd function according to the change of the difference between the rated load power and the actual load power and adaptively adjusting the value. i (i=1.2......n);

[0017] Among them, the active power P i , voltage U i , grid frequency Hz i Collected regularly from power grid operation.

[0018] As a preferred solution of the method for dynamically setting the active power droop coefficient of the distributed power supply controller of the present invention, wherein: the period τ of periodic data collection during power grid operation is determined by the distributed power supply control system and complies with τ <min(δ,γ);

[0019] That is, the period τ of periodic data collection is less than the minimum value of the time lengths δ and γ.

[0020] As a preferred solution of the method for dynamically setting the active power droop coefficient of the distributed power supply controller described in the present invention, the method of introducing the value of the droop coefficient into the odd function and adaptively adjusting it along with the change of the difference between the rated load power and the actual load power is replaced by a method of introducing the power deviation and the voltage deviation into the series correction link of the traditional droop control strategy to dynamically set the active power droop coefficient m. i ;

[0021] Among them, the active power P is regularly collected during the operation of the power grid. i , voltage U i , grid frequency Hz i , enter the collected data into the historical data set θ.

[0022] As a preferred solution of the method for dynamically setting the active power droop coefficient of the distributed power supply controller of the present invention, wherein: the historical data set θ is cleaned to construct an optimized data set ζ={P i，e , P i，α , U i，α , Hz i，α , m i，α};

[0023] The data cleaning standard is to select data whose grid frequency is within the qualified range at time (α+δ) and whose voltage is within the qualified range at time (α+γ);

[0024] δ is the time from when the distributed power supply controller receives the adjustment command to when the active power actually changes. It is a relative time. δ is an integer multiple of the data collection period τ. If it is not, it is rounded off.

[0025] Grid frequency Hz i(α+δ) The conditions for judging whether the data is within the qualified frequency range are:

[0026] Hz lim，min <Hz i(α+δ) <Hz lim，max

[0027] Hz lim，min is the minimum limit value of the power grid frequency, Hz lim，max It is the maximum limit value of the power grid frequency;

[0028] γ is the time from when the distributed power supply controller receives the adjustment command to when the voltage actually changes, which is a relative time;

[0029] Voltage U i(α+δ) The conditions for judging whether the data is within the qualified range are:

[0030] U im，min <U i(α+δ) <U lim，max

[0031] U lim，min is the minimum voltage limit value, U lim，max This is the maximum voltage limit.

[0032] As a preferred solution of the method for dynamically setting the active power droop coefficient of the distributed power supply controller of the present invention, wherein: a sampling data set ε of a support vector machine SVM is established, ε={(x i，j， y i，j ), i=1, 2...n, j=1, 2,...z};

[0033] Among them, x i，j ∈R N , is a four-dimensional vector, composed of the rated power generation power P of the distributed generation i，e , real-time power generation P i，α , real-time voltage U i，α , real-time frequency Hz i，α Composition, y i，j ∈R is the active power droop coefficient m at the corresponding moment i，α ;

[0034] Use support vector machine SVM to establish a mathematical calculation model:

[0035] m i =f[(P i，e , P i，j , U i，j， Hz i，j ), a, b].

[0036] As a preferred solution of the method for dynamically setting the active power droop coefficient of the distributed power controller of the present invention, the real-time data of the distributed power sources in the acquisition area includes the real-time power generation power P of the i-th distributed power source. i，r , voltage U i，r and grid frequency Hz i，r (i=1.2......n), use the trained support vector machine SVM model to calculate the active power droop coefficient m in real time i，r , the result m i，r Send it to the corresponding distributed power controller to collect the real-time power generation power P i，α, real-time voltage U i，α , real-time frequency Hzi, α and real-time active power droop coefficient m i，α And add it to the historical data set θ.

[0037] As a preferred solution of the method for dynamically setting the active power droop coefficient of the distributed power supply controller of the present invention, the power oscillation judgment method is:

[0038] Set the judgment cycle time Δt and power upper limit value P lim，max , power lower limit P lim，min ;

[0039] When the actual power P of the unit reaches the upper power limit P at the same time within a judgment cycle time Δt lim，max And the power lower limit P lim，min , determine the occurrence of power oscillation;

[0040] Among them, the power upper limit value P lim，max , power lower limit P lim，min It can be set as an absolute value or a relative value.

[0041] As a preferred solution of the method for dynamically setting the active power droop coefficient of the distributed power supply controller of the present invention, wherein: if power oscillation occurs, the controller stabilizes the active power output, and if the power oscillation is not eliminated within a period of time length S, the active power output is reduced in proportion β until the power oscillation is eliminated or the active power of the unit is reduced to zero;

[0042] Among them, the time length S and the ratio β are set in advance by the grid dispatching operation mode.

[0043] The beneficial effects of the present invention are as follows: a method for dynamically setting the active power droop coefficient of a distributed power supply controller is constructed, a variety of currently available simulation technologies are applied on-site, actual operating parameters are collected and analyzed, data with good effects are selected, and the active power droop coefficient calculation model is trained and iteratively optimized through a support vector machine, and the power fluctuation problem that may be caused by the dynamic change of the droop coefficient is considered to meet the requirements of grid frequency regulation and stable operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0045] Figure 1The present invention is a flow chart of a method for dynamically setting the active power droop coefficient of a distributed power supply controller. DETAILED DESCRIPTION

[0046] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0047] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0048] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0049] Secondly, the present invention is described in detail with reference to the schematic diagram. When describing the embodiments of the present invention in detail, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0050] Example 1

[0051] Reference Figure 1 , a flow chart of a method for dynamically setting an active power droop coefficient of a distributed power supply controller is provided, and a method for dynamically setting an active power droop coefficient of a distributed power supply controller includes:

[0052] S1: Determine the distributed power sources in the area and collect the rated available power P of all distributed power sources i，e It should be noted that:

[0053] Determine that there are n distributed power sources in the area and collect the rated active power P of all distributed power sources i，e (i=1.2......n).

[0054] The distributed power sources in the region are distributed energy sources that are connected to the regional power grid and participate in grid regulation, including photovoltaic power generation and wind power generation. The scope of division is the substation under a transformer in the 35kV power grid.

[0055] S2: Establish a historical data set. It should be noted that:

[0056] Establish historical data set θ={P i，e , P i，α , U i，α , Hz i，α , m i，α}.

[0057] Among them, P i，e is the rated power generation, P i，α is the real-time power generation, U i，α is the real-time voltage, Hz i，α is the real-time frequency, m i，α is the real-time active power droop coefficient, and α is the absolute time tag when collecting data.

[0058] The active power droop coefficient m is dynamically set by introducing the value of the droop coefficient into the odd function according to the change of the difference between the rated load power and the actual load power and adaptively adjusting the value. i (i=1.2......n).

[0059] Among them, the active power P i , voltage U i , grid frequency Hz i Collected regularly from power grid operation.

[0060] The period τ of regular data collection during grid operation is determined by the distributed power control system and complies with τ <min(δ,γ);

[0061] That is, the period τ of periodic data collection is less than the minimum value of the time lengths δ and γ.

[0062] S3: Replace the experimental simulation technology to dynamically set the active power droop coefficient m i , repeat S2 and enter the data into the historical data set θ. It should be noted that:

[0063] The method of introducing the droop coefficient value into the odd function as the difference between the rated load power and the actual load power changes is replaced by the method of introducing the power deviation and the voltage deviation into the series correction link of the traditional droop control strategy to dynamically set the active power droop coefficient m. i .

[0064] Among them, the active power P is regularly collected during the operation of the power grid. i , voltage U i , and grid frequency Hz i , enter the collected data into the historical data set θ.

[0065] S4: Clean the historical data set to construct an optimized data set.

[0066] The standard for data cleaning is to select data whose grid frequency is within the qualified range at time (α+δ) and whose voltage is within the qualified range at time (α+γ); the qualified range of grid frequency and the qualified range of voltage shall be implemented in accordance with the national standard "Frequency Deviation of Power Quality Power System (GB / T 15945)".

[0067] δ is the time from when the distributed power supply controller receives the adjustment instruction to when the active power actually changes, which is a relative time. It is determined according to the control parameters of the same type of power supply unit, or after the distributed power supply unit is tested. δ must be an integer multiple of the data collection period τ. If not, it is rounded off.

[0068] Grid frequency Hz i(α+δ) The conditions for judging whether the data is within the qualified frequency range are:

[0069] Hz lim，min <Hz i(α+δ) <Hz lim，max

[0070] Hz lim，min is the minimum limit value of the power grid frequency, Hz lim，max It is the maximum limit value of the power grid frequency.

[0071] γ is the time from when the distributed power supply controller receives the adjustment instruction to when the voltage actually changes, which is a relative time; it is determined according to the control parameters of the same type of power supply unit, or after the distributed power supply unit is tested. δ must be an integer multiple of the data collection period τ. If not, it is rounded off.

[0072] Voltage U i(α+δ) The conditions for judging whether the data is within the qualified range are:

[0073] U im，min <U i(α+δ) <U lim，max

[0074] U lim，min is the minimum voltage limit value, U lim，max This is the maximum voltage limit.

[0075] The optimized data set after data cleaning ζ={P i，e , P i，α , U i，α , Hz i，α , m i，α}, where P i，e is the rated power generation, P i，α is the real-time power generation, U i，α is the real-time voltage, Hz i，α is the real-time frequency, m i，αis the real-time active power droop coefficient, and α is the absolute time tag when collecting data.

[0076] S5: Use support vector machine to establish active power droop coefficient calculation model and perform training regression. It should be noted that:

[0077] Establish the sampling data set of support vector machine SVM ε={(x i，j ,y i，j ), i = 1, 2, ..., n, j = 1, 2, ..., z}, where x i，j ∈R N , is a four-dimensional vector, composed of the rated power generation power P of the distributed generation i，e , real-time power generation P i，α , real-time voltage U i，α and real-time frequency Hz i，α Composition; i，j ∈R is the active power droop coefficient m at the corresponding moment i，α .

[0078] The mathematical calculation model established by using support vector machine SVM is:

[0079] m i =f[(P i，e , P i，j , U i，j , Hz i，j ), a, b]

[0080] The real-time data of distributed power sources in the acquisition area includes the real-time power generation power P of the i-th distributed power source. i，r , voltage U i，r and grid frequency Hz i，r (i=1.2......n), use the trained support vector machine SVM model to calculate the active power droop coefficient m in real time i，r , the result m i，r Send it to the corresponding distributed power controller to collect the real-time power generation power P i，α , real-time voltage U i，α , Real-time frequency Hz i，α and real-time active power droop coefficient m i，α , adding these to the historical dataset θ.

[0081] S6: Regularly clean the historical data set θ and update the optimized data set ζ. Repeat steps S4 and S5. The time for cleaning the historical data set θ and updating the optimized data set ζ is determined according to the implementation effect. If the implementation effect is good, the update time is extended, and if the implementation effect is poor, the update time is shortened.

[0082] S7: Determine whether power oscillation occurs. If power oscillation occurs, the active power droop coefficient mi If it is set to 0, the distributed energy power maintains the original power generation power and does not adjust the droop characteristics. It should be noted that:

[0083] The method for judging power oscillation during grid operation is as follows:

[0084] Set the judgment cycle time Δt and the power upper limit value P lim，max , power lower limit P lim，min , when the actual power P of the unit reaches the upper power limit P at the same time within a judgment cycle time Δt lim，max And the power lower limit P lim，min , it is determined that power oscillation occurs.

[0085] Power upper limit P lim，max , power lower limit P lim，min It can be set as an absolute value or a relative value.

[0086] When power oscillation occurs, the active power droop coefficient is set to 0, the distributed energy power maintains the original power generation power, and the droop characteristic is not adjusted. The controller stabilizes the active power output. If the power oscillation is not eliminated within the period of time length S, the active power output is reduced in proportion β until the power oscillation is eliminated or the active power of the unit drops to zero.

[0087] The time length S and the ratio β are set in advance by the grid dispatching operation mode.

[0088] Example 2

[0089] This embodiment is the second embodiment of the present invention. This embodiment is different from the first embodiment in that it provides a verification test of a method for dynamically setting the active power droop coefficient of a distributed power supply controller, and verifies and explains the technical effects used in this method.

[0090] Traditional technical solutions: No actual on-site operation effect verification and iterative optimization were carried out, and the power fluctuations that may be caused by the dynamic changes of the droop coefficient were not considered, resulting in the inability to meet the grid regulation requirements and stable operation.

[0091] In view of the fact that the power generation of distributed power sources varies greatly due to meteorological factors during the power generation process, a variety of active power droop coefficient setting methods are verified on site, the actual operating parameters are collected and analyzed to establish an optimal data set, and the active power droop coefficient calculation model is trained and regressed using support vector machine (SVM) to improve the accuracy of parameters and ensure the actual regulation effect of the power grid. Considering the power oscillation problem of power grid operation that may be caused by the change of droop coefficient, monitoring methods and treatment measures are formulated to ensure the safe and stable operation of the power grid.

[0092] In this embodiment, the traditional method and the present method are used to measure and compare the dynamic adjustment of the droop coefficient of the distributed power source during the power generation process.

[0093] A1: Take the distributed energy of a transformer substation in a 35kV power grid, including two 10kW distributed power sources.

[0094] A2: The period τ of regular data collection during power grid operation is 15ms.

[0095] The following table is obtained by adopting a method of adaptively adjusting the droop coefficient value by introducing it into an odd function along with the change of the difference between the rated load power and the actual load power;

[0096]

[0097] The calculation model is а=(kp+ki / s)(Pi-Pref), is the corrected active power droop coefficient, m i is the original active power droop coefficient, kp and ki are the proportional coefficient and integral coefficient of the proportional-integral controller, Pi is the output active power, and Pref is calculated by the central controller after collecting the actual output active power of each inverter to establish the historical data set θ.

[0098] A3: The power deviation and voltage deviation are introduced into the series correction link of the traditional droop control strategy to obtain the following table;

[0099]

[0100] The calculation model is:

[0101]

[0102] Among them, m1 and m2 are the corrected active power droop coefficients of distributed power sources No. 1 and No. 2 respectively, and m 01 、m 02 are the original active power droop coefficients of No. 1 and No. 2 distributed power sources, ΔP1 and ΔP1 are the power deviations of No. 1 and No. 2 distributed power sources, ki1 and ki2 are the integral gains of No. 1 and No. 2 distributed power sources, and s is the time constant.

[0103] A4: Clean the historical data set θ and construct an optimized data set ζ = {P i，e , P i，α , U i，α , Hz i，α , m i，α}, including rated power generation P i，e , real-time power generation P i，α, real-time voltage U i，α , real-time frequency Hz i，α , real-time active power droop coefficient m i，α， α is the absolute time tag when the data is collected.

[0104] The standard for data cleaning is to select data whose grid frequency is within the qualified range at (α+0.5min) and whose voltage is within the qualified range at (α+1min). The qualified range of grid frequency and voltage shall be implemented in accordance with the national standard "Frequency Deviation of Power Quality Power System (GB / T 15945)".

[0105] Grid frequency Hz i(α+0.5min) The conditions for judging whether the data is within the qualified frequency range are:

[0106] 49.8Hz<Hz i(α+δ) <50.2Hz

[0107] Voltage U i(α+δ) The conditions for judging whether the data is within the qualified range are:

[0108] 31.5kV<U i(α+δ) <38.5kV

[0109] The optimized data set after data cleaning ζ={P i，e , P i，α , U i，α , Hz i，α , m i，α}, including rated power generation P i，e , real-time power generation P i，α , real-time voltage U i，α , real-time frequency Hz i，α , real-time active power droop coefficient m i，α , α is the absolute time tag when collecting data.

[0110] A5: Create a sampling data set for support vector machine SVM ε = {(x i，j ,y i，j ), i = 1, 2, ..., n, j = 1, 2, ..., z}, where x i，j ∈R N , is a four-dimensional vector, composed of the rated power generation power P of the distributed generation i，e , real-time power generation P i，α , real-time voltage U i，α , real-time frequency Hz i，α Composition. i，j ∈R is the active power droop coefficient m at the corresponding moment i，α .

[0111] The mathematical calculation model established by using support vector machine SVM is:

[0112] m i =f[(P i，e , P i，j , U i，j， Hz i，j ), a, b].

[0113] A6: Collect real-time data of distributed power sources in the area, including the real-time power generation power P of the i-th distributed power source i，r , voltage U i，r , grid frequency Hz i，r (i=1.2......n), use the trained support vector machine SVM model to calculate the active power droop coefficient m in real time i，r , the result m i，r Send it to the corresponding distributed power controller to collect the real-time power generation power P i，α , real-time voltage U i，α , real-time frequency Hz i，α , real-time active power droop coefficient m i，α Added to the historical dataset θ.

[0114] A7: Clean the historical data set θ every month, update and optimize the data set ζ, and repeat steps A4, A5, and A6.

[0115] A8: Methods for judging power oscillation during grid operation:

[0116] Set the judgment cycle time to 1 minute, the power upper limit to 11kW, and the power lower limit to 9kW. When the actual power P of the unit reaches both the power upper limit of 11kW and the power lower limit of 9kW within 1 minute, it is determined that power oscillation has occurred.

[0117] When power oscillation occurs, the active power droop coefficient is set to 0, the distributed energy power maintains the original power generation power, and the droop characteristic is not adjusted. The controller stabilizes the active power output. If the power oscillation is not eliminated within 10 minutes, the active power output is reduced by 10% until the power oscillation is eliminated or the active power of the unit drops to zero.

[0118] The comparison of the results of this method with the original technology is shown in the following table:

[0119]

[0120] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for dynamically setting the active power droop coefficient of a distributed power supply controller, characterized in that: include, Determine the distributed power sources in the area and collect the rated active power of all distributed power sources P i，e ; Building historical data sets; Establishing the historical data set includes: Building a historical dataset θ ={ P i，e , P i，α , U i，α , Hz i，α , m i，α }; in, P i，e is the rated power generation, P i，α For real-time power generation, U i，α is the real-time voltage, Hz i，α is the real-time frequency, m i，α is the real-time active power droop coefficient, α It is the absolute time tag when the data is collected; The active power droop coefficient is dynamically set by introducing the value of the droop coefficient into the odd function according to the change of the difference between the rated load power and the actual load power and adaptively adjusting the value. m i (i=1.2......n); Among them, the power generation P i ,Voltage U i , Grid frequency Hz i Regularly collected from power grid operation; Performing data cleaning on the historical data set to construct an optimized data set; For the historical data set θ Perform data cleaning and build optimized data sets ζ ={ P i，e , P i，α , U i，α , Hz i，α , m i，α }; Among them, the standard of data cleaning is to select ( α+δ ) when the grid frequency is within the qualified range, and ( α+γ ) Data when the voltage is within the qualified range; δ is the time from when the distributed power controller receives the adjustment command to when the active power actually changes, which is the relative time. δ It is an integer multiple of the data collection period τ. If not, it is rounded off. Grid frequency Hz i（α+δ） The conditions for judging whether the data is within the qualified frequency range are: Hz lim，min ＜ Hz i（α+δ） ＜ Hz lim，max Hz lim，min is the minimum limit value of the power grid frequency, Hz lim，max It is the maximum limit value of the power grid frequency; γ It is the time from when the distributed power supply controller receives the adjustment instruction to when the voltage actually changes, which is a relative time; Voltage U i（α+δ） The conditions for judging whether the data is within the qualified range are: U im，min ＜ U i（α+δ） ＜ U lim，max U lim，min is the minimum voltage limit value, U lim，max is the maximum voltage limit; Use support vector machine to establish active power droop coefficient calculation model and perform training regression; Build a sampling data set for support vector machine SVM ε ={( x i，j， y i，j ), i=1, 2...n, j=1, 2,...z}; in, x i，j ∈ R N , is a four-dimensional vector, which is composed of the rated power generation of distributed generation P i，e , real-time power generation P i，α , real-time voltage U i，α , real-time frequency Hz i，α composition, y i，j ∈ R is the active power droop coefficient at the corresponding moment m i，α ; Use support vector machine SVM to establish a mathematical calculation model: m i = f [( P i，e ,P i，j , U i，j， Hz i，j ),a,b] The real-time data of distributed power sources in the acquisition area includes the real-time power generation of the i-th distributed power source P i，r ,Voltage U i，r and grid frequency Hz i，r (i=1.2......n), use the trained support vector machine SVM model to calculate the active power droop coefficient in real time m i，r , the result m i，r Send to the corresponding i A distributed power controller to collect real-time power generation P i，α , real-time voltage U i，α , Real-time frequency Hz i，α and real-time active power droop coefficient m i，α And add to the historical dataset θ; Determine whether power oscillation occurs. If power oscillation occurs, the active power droop coefficient m i If it is set to 0, the distributed energy power maintains the original power generation power and does not adjust the droop characteristics.

2. The method for dynamically setting the active power droop coefficient of a distributed power supply controller according to claim 1, characterized in that: The distributed power sources in the region are distributed energy sources that are connected to the regional power grid and participate in grid regulation, including photovoltaic power generation and wind power generation; The division range is the substation under a transformer in the 35kV power grid.

3. The method for dynamically setting the active power droop coefficient of a distributed power supply controller according to claim 2, characterized in that: The period τ of regular data collection during grid operation is determined by the distributed power control system and complies with τ <min( δ, γ ); That is, the period of regular data collection τ is less than δ and γ Minimum value for duration.

4. The method for dynamically setting the active power droop coefficient of a distributed power supply controller according to claim 3, characterized in that: The method of introducing the droop coefficient value into the odd function as the difference between the rated load power and the actual load power changes and adaptively adjusting it is replaced by a method of introducing the power deviation and the voltage deviation into the series correction link of the traditional droop control strategy to dynamically set the active power droop coefficient. m i ; Among them, the power generation power is regularly collected during the operation of the power grid. P i ,Voltage U i , grid frequency Hz i , enter the collected data into the historical data set θ .

5. The method for dynamically setting the active power droop coefficient of a distributed power supply controller according to claim 1, characterized in that: The power oscillation judgment method is: Set the judgment cycle time Δt , Power upper limit P lim，max , power lower limit P lim，min ; When a judgment cycle time Δt The actual power of the unit P At the same time, the power upper limit is reached P lim，max and power lower limit P lim，min , determine that power oscillation occurs; Among them, the power upper limit P lim，max , power lower limit P lim，min Set to an absolute value.

6. The method for dynamically setting the active power droop coefficient of a distributed power supply controller according to claim 1 or 5, characterized in that: If power oscillation occurs, the controller stabilizes the active power output. If the time length S If the power oscillation is not eliminated within the period of β Reduce active power output until power oscillation is eliminated or the unit active power drops to zero; Among them, the length of time S and proportion β It is set in advance by the grid dispatching operation mode.

Images