CN114256865A

CN114256865A - Wind power installed capacity calculation method considering load increase direction randomness

Info

Publication number: CN114256865A
Application number: CN202011046650.1A
Authority: CN
Inventors: 郑焕坤; 韩超超; 吕婷婷; 孙耀斌
Original assignee: North China Electric Power University
Current assignee: North China Electric Power University
Priority date: 2020-09-23
Filing date: 2020-09-23
Publication date: 2022-03-29

Abstract

A wind power installed capacity calculation method considering randomness of a load growth direction includes the steps of firstly, establishing a continuous power flow model of an alternating current-direct current system containing wind power, calculating unbalanced region load growth coefficients considering the randomness, establishing a wind power installed capacity optimization model, then utilizing a hybrid intelligent algorithm formed by combining an extreme learning machine and a few-parameter multi-target backbone particle swarm algorithm to solve the model, and comparing and calculating wind power installed capacities under different wind power access points and different output line impedances. The method considers the static voltage stability of the system, considers the randomness of the load increasing direction in the process of calculating the load margin of the system, provides an objective function of a wind installed capacity optimization model, utilizes a hybrid intelligent algorithm to solve, obtains the wind installed capacity considering the randomness of the load increasing direction under the AC/DC background, and provides a certain reference for the planning and design of the wind power plant.

Description

Wind power installed capacity calculation method considering load increase direction randomness

Technical Field

The invention relates to a wind power installed capacity calculation method considering randomness of a load increasing direction, and belongs to the field of wind power plant planning.

Background

In recent years, wind power generation and direct current transmission technologies are rapidly developed, alternating current and direct current hybrid systems are more and more appeared, however, due to randomness and intermittency of wind energy, the change of system tide is caused by the access of wind power, and many problems in the aspects of safety and stability are brought to the current power grid, so that the problem that the wind power installed capacity is determined to be necessary to be considered in the planning and design of the wind power plant on the premise of ensuring the safe and stable operation of the alternating current and direct current hybrid systems is solved.

The large wind power plant is generally far away from a load center, and when the active power output of the wind power plant is increased, the reactive loss of a booster transformer and a transmission line of the wind power plant is increased, so that the static voltage stability of the system can be reduced, and therefore the static voltage stability of the system needs to be considered when the installed capacity of wind power is calculated.

Continuous power flow is a common method for calculating the static voltage stability of a system, in the calculation process, most of the current researches consider that the load increases in an equal proportion mode, actually, the load characteristics of all regions are different, and the load increasing directions are also different. Therefore, the difference of the load increasing directions and the randomness thereof are considered in the static voltage stability calculation process, and the wind power installed capacity which is more practical and takes the static voltage stability of the system into consideration can be obtained.

Disclosure of Invention

The invention provides a wind power installed capacity calculation method considering randomness of a load increase direction on the premise of taking an alternating current-direct current hybrid system as a background and considering system static voltage stability.

The problems of the invention are solved by the following technical scheme:

a wind power installed capacity calculation method considering randomness of a load growth direction includes the steps of firstly, establishing a continuous power flow model of an alternating current-direct current system containing wind power, calculating a load growth coefficient of an unbalanced area considering randomness, establishing a wind power installed capacity optimization model, then utilizing a hybrid intelligent algorithm formed by combining an Extreme Learning Machine (ELM) and a less-parameter multi-target backbone particle swarm algorithm to solve the model, and calculating and comparing wind power installed capacities under different wind power access points and different wind power output line impedances.

The wind power installed capacity calculation method considering the randomness of the load growth direction comprises the following specific steps of firstly establishing a continuous power flow model of the wind power-containing alternating current-direct current system and calculating the load growth coefficient of the unbalanced area considering the randomness:

the continuous power flow equation comprising the wind power system is as follows:

(1+k_Gi)P_Gi0+P_Wi-(1+k_PLi)P_Li0＝P_i(x)

Q_Gi0+Q_Wi-(1+k_QLi)Q_Li0＝Q_i(x)

in the formula, P_i(x)、Q_i(x)Representing the active and reactive power injected into node i; k is a radical of_GiRepresenting a growth coefficient representing the generator; k is a radical of_PLiAnd k_QLiRespectively representing the active and reactive growth coefficients of the load; p_Gi0And P_Li0Respectively representing the active power and the active load of the ground state generator of the node i; p_wiAnd Q_wiRespectively representing active power and reactive power of the wind power plant accessed to an alternating current and direct current system; q_Gi0And Q_Li0And respectively representing the reactive power and the reactive load of the ground state generator corresponding to the node i.

The unbalanced area load increase growth coefficient definition considering randomness is defined as follows:

in the calculation of the continuous power flow, each load increase coefficient k_PLiAnd k_QLiThe load growth direction is formed, and the load growth direction has randomness because the growth coefficient has randomness.

The wind power installed capacity calculation method considering the randomness of the load growth direction establishes a wind power installed capacity optimization model, and specifically comprises the following steps:

1) objective function of model

a. Maximum installed wind power capacity

In the formula, j represents the serial number of the wind power plant accessed by the system; n is a radical of_wjThe number of the fans in the wind power plant j is; p is a radical of_rThe rated capacity of a single fan.

b. Maximum expected value of system load margin

max f₂＝E(λ)

In the formula, E (×) represents the expectation function.

c. Maximum system load margin standard deviation inverse value

max f₃＝1/Std(λ)

In the formula, Std (×) represents a function for calculating the standard deviation.

2) Constraint conditions of model

The constraints considered are equality constraints and inequality constraints. The equality constraints are the system power flow equations. The inequality constraint conditions comprise active power output constraint of a conventional generator, line power flow constraint and system up-down rotation standby constraint.

The wind power installed capacity calculation method considering the randomness of the load growth direction solves the model by using a hybrid intelligent algorithm formed by combining an extreme learning machine and a less-parameter multi-target backbone particle swarm algorithm, and comprises the following specific steps of:

a. randomly generated 1 particle, the control variable of the system:

(N_W1，...，N_Wm，C_W1，...C_Wm，P_G1，...P_Gi)

wherein N is_W1，...，N_WmThe number of the fans of each wind power plant is counted; c_W1，...，C_WmThe number of parallel capacitor groups at a collection bus of the wind power plant is shown; p_G1，...，P_GiThe power output of the conventional unit participating in optimization is large.

And carrying out random simulation check on the particles, if the particles meet the constraint condition, retaining the particles, otherwise, regenerating the particles until N particles which can meet the constraint condition are generated, and taking the N particles as an initial population.

b. And calculating a fitness function of each particle by using the trained ELM, calculating an individual leader of each particle, and finding out an initial non-inferior solution by comparing function values, thereby forming an external reserve set and further calculating a global leader of the particle.

c. Updating the position of a particle by a Gaussian sampling formula, performing certain variation treatment on the particle according to the variation probability to form a new particle position, performing random simulation verification on the newly formed particle, repeatedly updating the particle which does not meet the random verification for n times, and keeping the position of the particle unchanged if the particle does not meet the constraint condition after the n times of updating.

d. Estimating the fitness function value f of the new particle by ELM₂And f₃. And continuously updating the individual leader of the particle, and updating the external reserve set and the global leader of the particle until a certain iteration number is met to obtain an optimized calculation result.

In the background of an alternating current-direct current system, the system static voltage stability after the wind power plant is connected is considered, the randomness of the load increasing direction in the process of calculating the load margin is considered, a wind power installed capacity optimization model taking the maximum installed capacity of the wind power plant, the maximum load margin expected value and the maximum load margin standard deviation inverse value as objective functions is established, the model is solved through a hybrid intelligent algorithm, the wind power installed capacities under different wind power access points and the impedance of a transmission line are calculated and compared, and the reference is provided for the planning of the wind power plant.

Drawings

The present invention will be described in further detail with reference to the accompanying drawings.

FIG. 1 is a schematic diagram of an improved IEEE30 node system;

FIG. 2 is a flow chart of a wind power installed capacity calculation method considering load growth direction randomness;

FIG. 3 is a schematic diagram of an Extreme Learning Machine (ELM) estimation objective function;

FIG. 4 is a flow chart of a hybrid intelligent algorithm based on stochastic simulation techniques;

FIG. 5 is a modified IEEE30 node system Pareto frontier (wind farm access system from node 7);

FIG. 6 is f₁-f₂Relationship diagram (wind farm access to system from node 7);

FIG. 7 is f₁-f₃Relationship diagram (wind farm access to system from node 7);

FIG. 8 is f₂-f₃Relationship diagram (wind farm slave node 7)An access system);

FIG. 9 is a modified IEEE30 node system Pareto frontier (wind farm access system from node 28);

FIG. 10 is f₁-f₂Relationship diagram (wind farm access to system from node 28);

FIG. 11 is f₁-f₃Relationship diagram (wind farm access to system from node 28);

FIG. 12 is f₂-f₃Relationship diagram (wind farm access to system from node 28);

FIG. 13 is a Pareto front edge comparison diagram of a system when the impedances of wind power transmission lines are different;

FIG. 14 shows the impedance of the wind power transmission line at different times₁-f₂Comparing the images;

FIG. 15 shows the impedance of the wind power transmission line at different times f₁-f₃Compare the figures.

The symbols used in the figures or text are: f. of₁For installed capacity of wind power generation, f₂For load margin expectation, f₃The standard deviation is the inverse value of the load margin.

Detailed Description

The invention provides a wind power installed capacity calculation method considering randomness of a load increase direction on the premise of taking an alternating current-direct current hybrid system as a background and considering system static voltage stability. The method considers the randomness of the load increasing direction in the process of calculating the load margin of the system, takes the maximum wind power installed capacity, the maximum load margin expected value and the maximum load margin standard deviation inverse value as target functions, adopts a hybrid intelligent algorithm to solve a model, and calculates and compares the wind power installed capacities under different wind power plant access places and sending line impedances, thereby providing a certain reference for the planning of the wind power plant under an alternating current and direct current power grid.

The invention is realized by adopting the following steps:

1. and establishing a continuous power flow model of the wind power-containing alternating current and direct current system and calculating an unbalanced regional load increase coefficient considering randomness. A Weibull (Weibull) distribution model of wind speed is adopted to describe the randomness of the wind speed, the system load is divided into a plurality of regions, the load growth direction in each region is assumed to be the same, and the load growth fluctuation meets normal distribution.

2. And establishing a wind power installed capacity optimization model. Considering the randomness of the load growth direction in the process of calculating the load margin of the system, taking the maximum installed capacity of the wind power plant, the maximum expected value of the load margin and the minimum standard deviation of the load margin as target functions, and calculating the expected value and the reciprocal of the standard deviation of an output random variable by using a random simulation method based on Latin hypercube sampling. The equality constraints are the system power flow equations. The inequality constraint conditions comprise active power output constraint of a conventional generator, line power flow constraint and system up-down rotation standby constraint.

3. And solving the model by using a hybrid intelligent algorithm formed by combining an extreme learning machine and a less-parameter multi-target backbone particle swarm algorithm. Considering that the wind speed has randomness, an opportunity constraint planning theory is introduced, inequality constraint is established under a certain confidence level, so that the uncertainty of the wind speed is processed, and then whether a certain control variable meets a constraint condition is verified by utilizing a random simulation technology based on Latin hypercube. And wind power installed capacities under different wind power transmission line impedances and different place access conditions are calculated by comparison.

The technical solution of the present invention will be described in detail below with reference to the accompanying drawings.

The method is based on an alternating current-direct current hybrid system, and a wind power installed capacity optimization calculation model considering randomness of a load increasing direction is established. The model takes the randomness of the wind speed into consideration, considers the constraint conditions of system tide, active power output of the generator and the like, introduces an opportunity constraint planning theory, and adopts a random simulation technology based on the Latin hypercube to carry out random check on the control variable of the optimization model. Meanwhile, the randomness of the load increasing direction in the process of calculating the load margin of the system is considered, the maximum wind power installed capacity, the maximum load margin expected value and the maximum load margin standard deviation inverse value are used as target functions, the target function values are estimated based on the extreme learning machine, and the target function values and the multi-target backbone particle swarm algorithm jointly form a hybrid intelligent algorithm for solving the optimization model. And finally, taking an improved IEEE30 node alternating current-direct current system as an example, obtaining the installed capacity of the wind power under a certain load margin index, and verifying the effectiveness and the correctness of the model and the algorithm. And wind power installed capacities under different wind power plant access points and output line impedances are calculated and compared, so that certain reference is provided for planning of the wind power plant under an alternating current and direct current power grid.

1. And establishing an alternating current-direct current continuous power flow model containing wind power and calculating an unbalanced regional load increase coefficient considering randomness.

1) Wind power plant output model

The wind speed has randomness, so that the output of the wind turbine generator has uncertainty, and the output of the wind turbine generator is described by adopting the following piecewise function:

in the formula, v_in、v_r、v_outThe cut-in wind speed, the rated wind speed and the cut-out wind speed of the wind turbine generator are respectively. p is a radical of_rThe rated power of the wind turbine generator is obtained.

A Weibull (Weibull) distribution model of wind speed is used to describe the randomness of wind speed:

in the formula, k is a shape parameter and has a value range of 1.8-2.3. c is a scale parameter, i.e. the average wind speed in the area over a certain time period.

A lumped model is used for simulating the whole wind power plant, the same type of fan is installed in the same wind power plant, and the wind speed is the same, so that the output of the wind power plant is equal to the sum of the output of each fan:

P_W＝N_WP_r (3)

in the formula, N_WNumber of fans, P_rThe rated output of a single fan.

2) The continuous power flow equation comprising the wind power system is as follows:

In the continuous load flow calculation process, the load flow of the system needs to be calculated, for the alternating current-direct current hybrid system, the alternating iteration method is used for calculating the load flow, and the converter station is equivalent to a variable load on a bus of the alternating current system, namely P_L＝-P_S，Q_L＝-Q_SWherein P is_LFor active power, Q, of converter stations injected into the system via an AC bus_LReactive power is injected into the system for the converter station via the ac bus.

3) In the calculation of the continuous power flow, each load increase coefficient k_PLiAnd k_QLiConstituting the direction of load growth.

Dividing the system load into m regions, and assuming that the load increasing direction in each region is the same, the load increasing coefficient of the jth region is as follows:

wherein K_AjFor the expected value of the load increase coefficient of the jth area, assuming that the load increase coefficients in the same area are the same; p_AjPredicting the total load P of all nodes after planning year for the region j_Aj，0The current total load for zone j.

For the load nodes after the partition, the fluctuation of the load increment is consideredThe active growth coefficient of the nodes in each region is listed as a random variable, and the load growth fluctuation meets the normal distribution, namely K_A，j～N(μ_j，σ_j)。

In order to realize the balance between the load increment and the power generation increment, each generator distributes active power increase power according to the active reserve capacity proportion, and the active power output increase coefficient of the generator is as follows:

wherein m is the number of area divisions, P_res，iFor the active reserve capacity of generator i, P_G，i，0Is the initial active output of the generator.

2. And establishing a wind power installed capacity optimization model.

2.1 objective function

1) Maximum installed wind power capacity

2) Maximum expected value of system load margin

max f₂＝E(λ) (8)

In the formula, E (×) represents the expectation function.

3) Maximum system load margin standard deviation inverse value

In order to reduce the risk caused by the fluctuation of the load margin caused by the fluctuation of the growing direction, the standard deviation of the load margin is minimum as an objective function, which is equivalent to the maximum reciprocal thereof, and the objective function can provide a certain reference of the fluctuation range of the expected value of the load margin for a decision maker:

max f₃＝1/Std(λ) (9)

2.2 constraint Condition

The constraint conditions include equality constraint conditions and inequality constraint conditions. The equality constraints are the system power flow equations. The inequality constraint conditions comprise active power output constraint of a conventional generator, line power flow constraint and system up-down rotation standby constraint. Considering that wind speed has randomness, state variables of a power system containing wind power obtained through load flow calculation are also random variables, an opportunity constraint planning theory is introduced, inequality constraints are established under a certain confidence level, uncertainty of the wind speed is processed, and then whether a certain control variable meets constraint conditions or not is tested by using a random simulation technology based on Latin hypercube.

A system power flow constraint equation:

in the formula P_Gi、Q_GiThe active power and reactive power output of the generator at the node i are obtained. P_Wi、Q_WiAnd the active and reactive power output of the wind power plant at the node i is obtained. P_Li、Q_LiThe active and reactive magnitudes of the load at node i. U shape_i、U_j、θ_ijThe magnitude of the voltage amplitude at nodes i, j and the phase angle difference between them. P_Gimin、P_GimaxThe upper and lower limits of the active output of the generator i. S_liIs the power flow on line i. Eta is the system rotation standby coefficient. Alpha is alpha₁、a₂Is the confidence of the corresponding inequality constraint.

3. Solving models using hybrid intelligent algorithms

Although the continuous power flow can accurately reflect the static voltage stability margin of the power grid as a margin index, the continuous power flow has the defect of long calculation time. Considering the randomness of the load growth direction, the expected value and the inverse standard deviation value of the output random variable are calculated by using a random simulation method based on Latin hypercube sampling. If each sample needs to be subjected to one continuous power flow calculation, the total calculation amount is very large, in order to reduce the calculation amount and the calculation time, a trained Extreme Learning Machine (ELM) with the characteristics of high learning speed, good generalization performance and the like is adopted to estimate the expected value and the inverse standard deviation value of the load margin, and the flow is shown in fig. 3. And embedding the model into a subsequent optimization solution method to form a hybrid intelligent algorithm for solving the model.

3.1 training and testing sample formation procedure was as follows:

1) randomly generating a control variable x_iDetecting whether the control variable meets the constraint condition in the model by using a random simulation technology based on Latin hypercube, and if not, generating again; if so, assuming the wind power plant as rated output, calculating the system load flow corresponding to the control variable, and recording the control variable value and the voltage amplitude v of each load node_iAnd phase angle theta_iThis is used as the input of the training sample.

2) And calculating the distribution characteristics of output variables by using a Monte Carlo method based on a Latin hypercube sampling technology in consideration of the randomness of the load growth direction in the continuous power flow calculation to obtain an expected value E (lambda) and a reciprocal value 1/Std (lambda) of the standard deviation of the load margin. The result is used as the output of the training sample, thereby obtaining a complete sample (x)_i，v_i，θ_i，E(λ)，1/Std(λ))。

And repeating the steps 1) and 2) until a certain number of samples are generated, and then generating a training sample set and a testing sample set of the ELM.

3) The ELM is trained by a training set, and a simulation test is performed by a test set. The test results are described in terms of mean square error:

wherein y is_iOutput value size, y, for predicted test set samples i_i，0Is its actual value.

3.2 model solution flow

The optimization model is solved by using a hybrid intelligent algorithm, the algorithm flow is shown in fig. 4, and the specific steps are as follows:

1) randomly generating 1 particle, and randomly generating 1 particle, namely a control variable of the system:

(N_W1，...，N_Wm，C_W1，...C_Wm，P_G1，...P_Gi) (12)

2) And calculating a fitness function of each particle by using the trained ELM, calculating an individual leader of each particle, and finding out an initial non-inferior solution by comparing function values, thereby forming an external reserve set and further calculating a global leader of the particle.

3) Updating the position of a particle by a Gaussian sampling formula, performing certain variation treatment on the particle according to the variation probability to form a new particle position, performing random simulation verification on the newly formed particle, repeatedly updating the particle which does not meet the random verification for n times, and keeping the position of the particle unchanged if the particle does not meet the constraint condition after the n times of updating.

4) Estimating the fitness function value f of the new particle by ELM₂And f₃. And continuously updating the individual leader of the particle, and updating the external reserve set and the global leader of the particle until a certain iteration number is met to obtain an optimized calculation result.

4. Simulation example

The simulation is carried out by using an improved IEEE30 node system shown in FIG. 1, the total load of the system is 283.4MW, the reference value of the system power is 100MW, the rated power of a single fan is 1MW, and the IEEE30 node system is divided into 4 areas with unbalanced load increase modes, as shown in Table 1:

TABLE 1 System load node zone partitioning

Load growth direction expectation K of 4 load regions_A＝[0.638 0.708 0.912 1.567]. The number of random analog samples based on latin hypercube is 600.

4.1 wind farm Slave node 7 Access

4.1.1 analysis of relationships between objective functions

1) Solving the model results in a more uniform front edge of the Pareto solution as shown in fig. 5. Projecting Pareto solution to f₁-f₂On the other hand, as shown in fig. 6, it can be seen that the installed wind power capacity of the system and the expected load margin value are mutually exclusive, and when the installed wind power capacity gradually increases from 30MW to 70MW, the expected load margin value of the system considering the randomness of the load increasing direction gradually decreases.

The reason for this is that: with the increase of installed capacity of the wind power plant, when rated output of the wind power plant is increased, reactive power consumption on a main transformer and a sending line of the wind power plant is increased, and the voltage stability margin of the system is reduced, so that the expected value of the load margin considering the randomness of the load increasing direction is reduced.

2) Projecting Pareto solution to f₁-f₃Counting the planes, a graph of the relationship shown in FIG. 7 is obtained. As can be seen from the figure, a synergistic relationship is presented between the two objective functions. Meanwhile, as can be seen from fig. 6, with the increase of the installed capacity of the wind power, although the expected value of the load margin is gradually reduced, the inverse value of the standard deviation of the load margin is gradually increased, that is, the fluctuation of the expected value of the load margin caused by the randomness of the load increasing direction is reduced, so that a certain fluctuation risk reference of the expected value of the load margin is provided for a decision maker.

3) Projecting Pareto solution to f₂-f₃And (4) flattening to obtain a relation graph shown in the figure 8. As can be seen from the figure, the two objective functions are mutually exclusive, that is, as the expected value of the load margin of the system increases, the inverse value of the standard deviation of the load margin increasesThe standard deviation of the load margin is gradually reduced, namely the fluctuation risk of the expected value of the load margin after considering the randomness of the load increasing direction is increased.

4.1.2 determination of installed wind Capacity

As shown in fig. 6, when a lower limit value of the load margin is given, the wind installed capacity corresponding to the lower limit value can be determined, and the obtained standard difference value of the load margin can provide a fluctuation risk reference of the expected value of the load margin. The load margin is generally in the range of 0.15-0.3 of the load multiple, and here the lower limit of the load margin is taken to be 0.2, so that when the wind farm is connected to the grid from the node 7, it can be determined that the installed wind power capacity is 63MW, the load margin expected value is 0.2002, the load margin standard deviation inverse value is 62.4474, and the load margin standard deviation is 0.0160, and the corresponding control variables of the solution are:

X＝[63 16 30.116 19.923 29.511]

4.2 wind farm Slave node 28 Access

To further verify the correctness of the analysis and calculation herein, the wind farm is connected from the node 28 to the grid, its Pareto solution set is as in fig. 9 and its projections on the planes are as in fig. 10, 11, 12.

Comparing fig. 6-8 and 10-12, it can be seen that the resulting curves from node 28 access and node 7 access for the wind farm are similar. The static voltage stability of the system is really influenced by the access of the wind power plant, namely, the static voltage stability of the system is reduced along with the increase of the installed capacity of the wind power. And after the randomness of load increase in different areas is considered, the fluctuation risk of the system load margin is reduced along with the increase of the installed capacity of the wind power, so that a certain risk reference can be provided for planning personnel. This result also further demonstrates the correctness and validity of the analysis herein and the established optimization model and its computational method.

And still taking the lower limit of the load margin as 0.2, obtaining that the installed wind power capacity is 68MW when the wind power plant is connected to the power grid from the node 28, the expected value of the load margin is 0.2058, the inverse value of the standard deviation of the load margin is 60.9248, the standard deviation of the load margin is 0.0164, and the standard deviation of the load margin can provide certain data reference for decision-making personnel. The control variables corresponding to this solution are:

X＝[68 17 39.929 20 29.904]

4.3 installed capacity analysis under different wind farm access points

By comparing the installed wind power capacities of the wind power plants accessed from the node 7 and the node 28, when the wind power plants are under the same load margin requirement, the access points of the wind power plants are different, the standard deviation of the installed wind power capacity and the load margin is different, and the installed wind power capacities are different. The method is caused by a grid structure of the system, and has important reference significance for site selection and grid-connected planning of the wind power plant. Therefore, when the wind power plant is planned, the optimal wind power access point can be determined by calculating and comparing the wind power installed capacity of each access point, so that the system can accept more wind power on the premise of meeting the static voltage stability standard.

4.4 influence of wind power transmission line impedance on installed wind power capacity

Taking the example that the wind power plant is accessed into the system from the node 7, the transmission impedance of the wind power transmission line is increased to 1.5 times of the original impedance, other parameters are unchanged, the Pareto solution of the wind power generator capacity optimization model under the condition is solved, and the Pareto solution is compared with the result under the original impedance condition, so that the result shown in fig. 13 is obtained:

project figure 13 to f₁-f₂Planar, resulting in fig. 14. It can be known that when the installed capacities of the wind power are the same, the static voltage stability margin of the system with the higher impedance of the wind power transmission line is smaller, and the phenomenon is more obvious along with the increase of the installed capacity of the wind power. This is because the larger the line impedance is, the larger the reactive loss of the line after the large-scale wind power grid connection is, and the larger the influence on the static voltage stability of the system after the grid connection is.

Similarly, when the lower limit value of the set load margin is 0.2, the installed capacity of the wind power under the condition of 1.5 times of the original line impedance is 58MW, and the installed capacity of the wind power is reduced compared with 63MW under the condition of the original impedance parameter, which is caused by the reduction of the static voltage stability margin of the system due to the increase of the impedance of the wind power transmission line. Therefore, when the installed wind power capacity under large-scale wind power grid connection is calculated, it is meaningful to consider the static voltage stability of the system under the wind power grid connection line.

Project figure 13 to f₁-f₃Planar, resulting in FIG. 15. As can be seen from fig. 15 in combination with fig. 14, under the same installed wind power capacity, the static voltage stability margin of the system with the larger impedance of the wind power transmission line is smaller, but the standard deviation of the load margin after considering the randomness of the load increasing direction is also smaller, that is, the fluctuation risk of the load margin at this time is also smaller, so that a certain load margin fluctuation risk reference is provided for a decision maker.

In summary, in the invention, in the context of an alternating current/direct current system, randomness of a load increase direction during continuous power flow load margin calculation is considered, a multi-objective wind power installed capacity optimization model considering system static voltage stability is established, based on an IEEE30 node alternating current/direct current system example, the model is solved through a hybrid intelligent algorithm, a Pareto solution of the multi-objective optimization model is obtained, and therefore, a mutual relation among objective functions is obtained, wind power installed capacities under different wind power plant access points and different output line impedances are compared, and therefore, a certain reference is provided for wind power plant planning. In the process of actually determining the wind power installed capacity, a decision maker can balance the relationship between the wind power installed capacity and the load margin expected value and the standard deviation representing the fluctuation risk of the load margin expected value according to different specific operation requirements, so that a compromise solution under different requirements is obtained.

Claims

1. A wind power installed capacity calculation method considering randomness of a load growth direction is characterized by firstly establishing an alternating current-direct current continuous power flow model containing wind power, calculating a load growth coefficient of an unbalanced area considering the randomness, establishing a wind power installed capacity optimization model, then solving the model by using a hybrid intelligent algorithm formed by combining an extreme learning machine and a less-parameter multi-target backbone particle swarm algorithm, and calculating and comparing the wind power installed capacities under different wind power access points and different output line impedances.

2. The wind power installed capacity calculation method considering the randomness of the load growth direction as claimed in claim 1, wherein an alternating current-direct current continuous power flow model containing a wind power system is established, and the continuous power flow equation of the system is as follows:

(1+k_Gi)P_Gi0+P_Wi-(1+k_PLi)P_Li0＝P_i(x)

Q_Gi0+Q_Wi-(1+k_QLi)Q_Li0＝Q_i(x)

In the calculation of the continuous power flow, each load increase coefficient k_PLiAnd k_QLiConstituting the direction of load growth.

Assuming that m regions are provided, assuming that the load increase direction in each region is the same, the load increase coefficient of the jth region is:

wherein K_A，jFor the expected value of the load increase coefficient of the jth area, assuming that the load increase coefficients in the same area are the same; p_A，jPredicted for region j to obtainTotal load P of all nodes after planning year_A，j，0The current total load for zone j.

For the load nodes after partitioning, considering the fluctuation of load increment, the active growth coefficient of the nodes in each area is listed as a random variable, and the load growth fluctuation meets the normal distribution, namely K_A，j～N(μ_j，σ_j)。

3. The wind power installed capacity calculation method considering the randomness of the load growth direction as claimed in claim 2, wherein a wind power installed capacity optimization model is established, and the maximum wind power installed capacity, the maximum load margin expected value and the maximum load margin standard deviation inverse value are taken as objective functions.

a. Maximum installed wind power capacity

b. Maximum expected value of system load margin

max f₂＝E(λ)

In the formula, E (×) represents the expectation function.

c. Maximum system load margin standard deviation inverse value

In order to reduce the risk caused by the fluctuation of the load margin caused by the fluctuation of the growing direction, the maximum inverse value of the standard deviation of the load margin is also taken as an objective function, which is equivalent to the maximum inverse value thereof, and a certain fluctuation range reference of the expected value of the load margin can be provided for a decision maker:

max f₃＝1/Std(λ)

4. The wind power installed capacity calculation method considering the randomness of the load growth direction as claimed in claim 3, wherein the objective function values are estimated based on an extreme learning machine, and a hybrid intelligent algorithm for solving the optimization model is formed together with a multi-objective backbone particle swarm algorithm. The concrete solving steps are as follows:

a. randomly generated 1 particle, the control variable of the system:

(N_W1，…，N_Wm，C_W1，…C_Wm，P_G1，…P_Gi)