CN111984906B - Probability power flow rapid calculation method considering correlation of photovoltaic and load time sequence - Google Patents
Probability power flow rapid calculation method considering correlation of photovoltaic and load time sequence Download PDFInfo
- Publication number
- CN111984906B CN111984906B CN202010733457.9A CN202010733457A CN111984906B CN 111984906 B CN111984906 B CN 111984906B CN 202010733457 A CN202010733457 A CN 202010733457A CN 111984906 B CN111984906 B CN 111984906B
- Authority
- CN
- China
- Prior art keywords
- load
- probability
- time sequence
- photovoltaic
- photovoltaic output
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Physics (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Computational Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Probability & Statistics with Applications (AREA)
- Algebra (AREA)
- Software Systems (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Databases & Information Systems (AREA)
- Operations Research (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Photovoltaic Devices (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
A probability power flow rapid calculation method considering correlation of photovoltaic and load time sequence belongs to the technical field of power control. The invention aims to divide a day into a plurality of time periods for research, establish a photovoltaic output and load time sequence related model, sample, divide a typical scene by using a fuzzy C-means clustering algorithm for probability flow calculation, and effectively improve the calculation speed on the premise of ensuring the accuracy of the probability flow by taking the correlation of the photovoltaic and the load time sequence into consideration. Firstly, dividing a day into 24 time periods, respectively establishing probability density distribution models of photovoltaic output and load by adopting a self-adaptive diffusion nuclear density estimation method, improving the local adaptability of the probability models, and describing the correlation between the photovoltaic output and the load by using a Copula theory; and dividing the photovoltaic output and load scenes by using a fuzzy C-means clustering method, and carrying out probability load flow calculation by using a scene clustering center and scene occurrence probability instead of a Monte Carlo simulation process. The method and the device greatly reduce the calculation times, can improve the accuracy of the probability model, and have popularization value.
Description
Technical Field
The invention belongs to the technical field of power control.
Background
Photovoltaic power generation has the advantages of cleanliness, reproducibility and the like, and has been rapidly developed in recent years. By the end of 2019, the total capacity of the photovoltaic accumulation installation in China reaches 20430 kilowatts, and the same ratio is increased by 17.3%. In order to accurately evaluate the influence of the continuous increase of the permeability of the power grid of new energy sources such as photovoltaic power generation and the like on the running state of the power system, the probability power flow calculation method of the power system is widely researched. The photovoltaic power generation has larger dependence on the solar radiation intensity and obvious time sequence change characteristic. In addition, the traditional method generally utilizes a Monte Carlo simulation method to calculate the probability power flow, and the calculation efficiency is low. Therefore, the method has important significance in carrying out probability power flow calculation aiming at the time sequence change characteristics of photovoltaic output and load and improving the probability power flow calculation speed.
In recent years, expert scholars at home and abroad have conducted intensive research on a probability flow calculation method. The calculation method of the probability trend can be roughly classified into an analytical method and an analog method, wherein the analog method is widely paid attention to because of simplicity and easiness. The early-stage basis of probability power flow calculation by using an analog method is to establish a probability density model of novel energy sources such as photovoltaic output and wind power, and the probability density model is mainly divided into a parameter model and a non-parameter model at present. The parametric model refers to probability density distribution, such as beta distribution, t-location distribution, gaussian distribution and the like, of new energy sources by using an existing function. The non-parameter distribution is to select the optimal bandwidth from the data itself, and fit the probability distribution of the random variable. And secondly, carrying out probability power flow calculation by considering the correlation among random variables such as photovoltaic output, wind power, load and the like, and mainly applying Copula theory, an equal probability principle, a Cholesky decomposition method, a Bayesian network, NATAF conversion and other methods to the processing of the correlation among the random variables. The current research on the probability power flow calculation is focused on the influence of the correlation among wind power, photovoltaic and load on the probability power flow calculation, but the time sequence change characteristic of random variables is not considered, and the obtained probability power flow calculation result has limited meaning on the system reference. The existing method utilizes the Monte Carlo simulation process to generate a large number of samples to calculate the probability power flow, and the calculation time is long.
Disclosure of Invention
The invention aims to divide a day into a plurality of time periods for research, establish a photovoltaic output and load time sequence related model, sample, divide a typical scene by using a fuzzy C-means clustering algorithm for probability flow calculation, and effectively improve the calculation speed on the premise of ensuring the accuracy of the probability flow by taking the correlation of the photovoltaic and the load time sequence into consideration.
The method comprises the following steps:
step 1: establishing a random variable time sequence related probability model and sampling
Fitting the photovoltaic output and load data by adopting an adaptive diffusion kernel density estimation method to perform adaptive diffusion kernel density estimation:
wherein: x is X i ∈[0,1]For normalization of measured data, t=h 2 >0; f (x; t) represents diffusion kernel density estimation, K D (. Cndot.) represents the diffusion kernel, x, y are random variables within the kernel definition domain, a (x) and p (x) are fundamental control parameters of the linear diffusion process, where
For the pilot estimation of photovoltaic output and load probability density, a (x) =p (x) λ ，λ∈[0,1]The method comprises the steps of carrying out a first treatment on the surface of the Given initial conditions and boundary conditions:
wherein: delta (X-X) i ) Is a random variable X i Dirac measure of (a);
setting x and y constraint conditions:
wherein:
in the case of a linear diffusion partial differential operator,
an accompanying partial differential operator of L (;
giving an adaptive diffusion kernel density estimation an adaptive optimal bandwidth h:
let x be i 、y i For the ith hour photovoltaic output and load data, F 1i (x i )、F 2i (y i ) Edge probability distributions of photovoltaic output and load for the ith hour respectively;
step 1-3: establishing a time sequence joint probability density model of photovoltaic output and load by using Copula theory:
combined probability distribution G of photovoltaic output and load at ith hour i (x i ,y i ) Represented as
G i (x i ,y i )＝C i [F 1i (x i ),F 2i (y i )] (7)
Wherein: c (C) i Connecting Copula functions for both of the i-th hour;
step 1-4: using the i-th hour photovoltaic output and load joint probability distribution G i (x i ,y i ) Sampling to obtain a time sequence related sample of 24-hour photovoltaic output and load;
step 2: fuzzy C-means clustering of time sequence photovoltaic output and load
Step 2-1: photovoltaic output and load data sample matrix X obtained for sampling
Determination of fuzzy C-means cluster number C (C)>1) Power exponent m (m)>1) Determining a membership termination tolerance epsilon and an initial membership matrix U (0) ＝[u ik (0) ]；
Step 2-2, calculating a scene clustering center matrix V= [ V ] of photovoltaic and load data by using a formula (9) i (L) ]：
Wherein: v i The method is characterized in that the method is an i-th scene cluster center, and L is algorithm iteration times;
step 2-3: substituting the clustering center matrix and the history membership matrix into the formula (10) and applying the clustering center matrix and the history membership matrix to the objective function J (L) Optimizing and correcting membership matrix
Wherein: j (J) (L) D is the objective function of the clustering algorithm ik (L) ＝||x k -v i (L) ||；
Step 2-4: terminating tolerance ε in membershipThe iteration ending condition of the index judgment algorithm is adopted; if max { |u ik (L) -u ik (L -1) The I is smaller than the membership termination tolerance epsilon, and clustering is finished; otherwise, returning to the step 2-2, and continuing algorithm iteration;
step 2-5: ending algorithm iteration, and determining a final membership matrix U of photovoltaic and load, a scene clustering center matrix V and the total number n of random variable data in various scenes i And calculates the occurrence probability P of various scenes i ＝n i N, n is the sample size;
step 3: time sequence probability load flow calculation method
Step 3-1: substituting the photovoltaic output and load time sequence related data of each scene into the data (11) for time sequence probability load flow calculation according to the obtained scenes of the steps
Wherein: p (P) i 、Q i Respectively represent the active power and the reactive power of the node i injection system [21-22] ，U i 、U j Respectively represent the voltages of nodes i and j, θ ij The phase angle difference of the nodes i and j is represented by n, the node number is represented by Gij and Bij, and the system conductance and susceptance are represented by the system conductance and susceptance respectively;
step 3-2: and (3) calculating by using the formula (12) to obtain a final time sequence probability power flow calculation result.
Wherein: e (Y) i ) Representing the result of the probabilistic load flow calculation, F (v) i ) Expressed in terms of cluster center v i As a result of the tidal current calculation at the time of the input amount.
According to the invention, the whole-day photovoltaic output and load data are divided into 24 time periods, the probability density model of the whole-day photovoltaic output and the load data is established by using the self-adaptive diffusion nuclear density estimation method, and the correlation of the whole-day photovoltaic output and the load data along with the time sequence change is described by using the Copula theory, so that the accuracy of the probability model can be improved.
According to the invention, the fuzzy C-means clustering algorithm is adopted to perform cluster analysis on photovoltaic output and load, and the scene clustering center and the scene occurrence probability are utilized to replace the Monte Carlo simulation process to perform probability flow calculation, so that the probability flow calculation accuracy can be ensured, and the calculation speed of the time sequence probability flow can be effectively improved.
According to the method, the photovoltaic output and the load time sequence change characteristics are considered in the probability power flow calculation, the probability power flow calculation result can be given out at intervals of hours, more reliable references are provided for judging node voltage out-of-limit and branch power flow overload, and the method has popularization value.
Drawings
FIG. 1 is a graph of time-series photovoltaic output probability density;
FIG. 2 is a timing load probability density diagram;
FIG. 3 is a random variable fuzzy C-means clustering result;
fig. 4 is a flow chart of a time series probability flow quick calculation.
Detailed Description
With the rapid development of photovoltaic power generation, the operation uncertainty of a power system is continuously increased. In order to accurately evaluate the influence of the continuously increased grid permeability of new energy sources such as photovoltaic power generation on the running state of the power system, the running stability of the power system is improved. Aiming at the problems that the dependence of photovoltaic power generation on solar radiation intensity is large, the characteristic of obvious time sequence change is provided, the calculation efficiency of the probability tide calculation method of the traditional method is low, and the like. The invention provides a probability load flow rapid calculation method considering the correlation between photovoltaic and load time sequence. The method is characterized in that the calculation speed is effectively improved on the premise of ensuring the accuracy of the probability flow.
The following detailed description of specific embodiments of the invention refers to the accompanying drawings
Step 1: establishing a random variable time sequence related probability model and sampling
Step 1-1: establishing a random variable adaptive diffusion kernel density model
Fitting the photovoltaic output and load data by adopting an adaptive diffusion kernel density estimation method, wherein the adaptive diffusion kernel density estimation is as shown in formula (1):
wherein: x is X i ∈[0,1]For normalization of measured data, t=h 2 >0; f (x; t) represents diffusion kernel density estimation, K D (. Cndot.) represents the diffusion kernel, x, y are random variables within the kernel definition domain. a (x) and p (x) are fundamental control parameters of a linear diffusion process, where
For the pilot estimation of photovoltaic output and load probability density, a (x) =p (x) λ ，λ∈[0,1]。
Given initial conditions and boundary conditions as in formulas (2) - (3):
wherein: delta (X-X) i ) Is a random variable X i Dirac measure of (a).
To obtain a unique solution for the adaptive diffusion kernel density estimation, x and y constraint conditions are set:
wherein:
in the case of a linear diffusion partial differential operator,
is the companion partial differential operator of L ().
The adaptive optimal bandwidth h is given by the adaptive diffusion kernel density estimation as formula (5):
the probability density diagrams of the time sequence photovoltaic output and the load are respectively obtained according to the probability density diagrams shown in fig. 1 and 2.
Step 1-2: cumulative probability distribution of time-series photovoltaic output versus load:
accumulating the self-adaptive diffusion nuclear density distribution model in the step 1 to respectively obtain the accumulated probability distribution F of the photovoltaic output and the load 1i (x i )、F 2i (y i )
Let x be i 、y i For the ith hour photovoltaic output and load data, F 1i (x i )、F 2i (y i ) The photovoltaic output and the edge probability distribution of the load at the i hour are respectively.
Step 1-3: establishing a time sequence joint probability density model of photovoltaic output and load by using Copula theory:
combined probability distribution G of photovoltaic output and load at ith hour i (x i ,y i ) Can be expressed as
G i (x i ,y i )＝C i [F 1i (x i ),F 2i (y i )] (7)
Wherein: c (C) i The Copula function is connected for both of the i-th hour.
Step 1-4: using the i-th hour photovoltaic output and load joint probability distribution G i (x i ,y i ) Sampling to obtain the time sequence phase of 24-hour photovoltaic output and loadAnd closing the sample.
Step 2: fuzzy C-means clustering of time sequence photovoltaic output and load
Step 2-1: photovoltaic output and load data sample matrix X obtained for sampling
Determination of fuzzy C-means cluster number C (C)>1) Power exponent m (m)>1) Determining a membership termination tolerance epsilon and an initial membership matrix U (0) ＝[u ik (0) ]。
Step 2-2, calculating a scene clustering center matrix V= [ V ] of photovoltaic and load data by using a formula (9) i (L) ]：
Wherein: v i And L is the iterative times of the algorithm for the i-th scene cluster center.
Step 2-3: substituting the clustering center matrix and the history membership matrix into the formula (10) and applying the clustering center matrix and the history membership matrix to the objective function J (L) And (5) optimizing and correcting the membership matrix.
Wherein: j (J) (L) D is the objective function of the clustering algorithm ik (L) ＝||x k -v i (L) ||。
Step 2-4: and judging the iterative ending condition of the algorithm by taking the membership termination tolerance epsilon as an index. If max { |u ik (L) -u ik (L -1) And the I is smaller than the membership termination tolerance epsilon, and clustering is finished. Otherwise, returning to the step 2-2, and continuing algorithm iteration.
Step 2-5: ending algorithm iteration, and determining a final membership matrix U, a scene clustering center matrix V and various types of photovoltaic and loadTotal number of random variable data n in scene i And calculates the occurrence probability P of various scenes i ＝n i N, n is the sample size. The fuzzy C-means clustering result of the time sequence photovoltaic output and the load is shown in figure 3.
Step 3: time sequence probability load flow calculation method
Step 3-1: and substituting the photovoltaic output and load time sequence related data of each scene into the data (11) for the obtained scenes of the steps to perform time sequence probability load flow calculation.
Wherein: p (P) i 、Q i Respectively represent the active power and the reactive power of the node i injection system [21-22] ，U i 、U j Respectively represent the voltages of nodes i and j, θ ij The phase angle difference of the nodes i and j is that of the node number, and the Gij and the Bij respectively represent the system conductance and the susceptance.
Step 3-2: and (3) calculating by using the formula (12) to obtain a final time sequence probability power flow calculation result.
Wherein: e (Y) i ) Representing the result of the probabilistic load flow calculation, F (v) i ) Expressed in terms of cluster center v i As a result of the tidal current calculation at the time of the input amount. The flow chart of the proposed time sequence probability power flow rapid calculation method is shown in fig. 4.
Building a model and performing simulation verification:
building a time sequence probability power flow rapid calculation method model by using an RT-LAB simulation platform
Substituting data, performing time sequence probability power flow rapid calculation simulation, and verifying the accuracy and the rapidity of the method. Taking 10000 Monte Carlo time sequence probability power flow calculation results as accurate results, taking 13 hours as an example, and table 1 gives partial node voltage average value and branch power flow average value comparison conditions of Monte Carlo time sequence probability power flow calculation (method 1) and fuzzy C-average value clustering time sequence probability power flow calculation (method 2). Table 2 shows the partial node voltage standard deviation and the branch current standard deviation comparison conditions of the method 1 and the method 2.
TABLE 1 branch tidal current and node Voltage mean comparison
TABLE 2 branch tidal current vs. node Voltage standard deviation
As can be seen from comparison of the power flow calculation results of the method 2 and the method 1 in tables 1 and 2, as the clustering number increases from 5 to 50, the mean value and standard deviation calculation result of the node voltage and the branch power flow of the method 2 are more and more similar to the calculation result of the method 1, which indicates that the accuracy of the time sequence probability power flow calculation method proposed herein can be improved by properly increasing the clustering number.
Comparing the tidal current calculation result with the tidal current calculation result of the method 1 when the clustering number c=50 in the method 2 in tables 1 and 2 respectively, it can be seen that the probability tidal current calculation result and the accurate result keep consistent when the clustering number c=50. This illustrates that the method presented herein enables accurate time series probability flow calculations.
The rapidity of the proposed method is verified.
Table 3 gives the calculated time comparisons of method 1 and method 2. In table 3, the time required for running the method 2 clustering numbers c=5 and 50 is 306.46s and 353.13s respectively, which is far smaller than the time 5807.26s required for running the method 1, which indicates that the method provided herein can effectively improve the time sequence probability power flow calculation speed on the premise of ensuring the probability power flow calculation accuracy.
Table 3 run time comparison
The advantage of probability power flow taking into consideration chronology is verified.
In order to analyze the influence of the time sequence change characteristics of the photovoltaic output and the load on the probability power flow calculation, the time sequence probability power flow calculation of 6 hours in the early morning, 12 hours in the midday, 18 hours in the evening and 24 hours in the late night is selected in this section, and the time sequence probability power flow calculation result is compared with the untimely time sequence probability power flow calculation result. Table 4 shows the comparison between the partial branch time sequence probability power flow and the non-time sequence probability power flow. u (u) l-i 、σ l-i The mean value and standard deviation (i=1, 8 and 18) of the power flow of the branch i are respectively obtained.
Table 4 timing versus non-timing branch flow table
Taking the branch 1 as an example, it can be seen from table 4 that the average power flow of the branch 1 is 48.5069MW and the standard deviation is 13.8031MW when the time sequence is not counted, and the average power flow of the branch 1 is 2.4927MW, 41.1702MW, 22.9272MW and 4.1519MW when the time sequence is counted in the branches 6h, 12h, 18h and 24h, and the standard deviations are 5.6029MW, 7.9509MW, 7.4537MW and 6.3474MW respectively, so that the power flow of the branch 1 has different probability distribution characteristics from 6h in the early morning to 24h in the late night in the day, and the probability distribution characteristics are larger than the calculation results of the time sequence. This is due to the large data difference between photovoltaic output and load day and night, and between hours in the daytime, resulting in large branch tidal current changes in each period. The probability power flow calculation is carried out by the untimely sequence, so that the variation range of branch power flow in one day can be approximately given, and the referential significance for real-time scheduling operation and stability evaluation of the system is limited. And the time sequence probability power flow calculation method is considered, the probability distribution characteristics of the branch power flow can be given out at the intervals of hours, and more reliable reference information is provided for judging whether the power flow of the system line is overloaded.
Table 5 shows the results of partial node time sequence probability power flow and non-time sequence probability power flow calculation. u (u) v-j 、σ v-j The mean and standard deviation of the voltage at node j (j=4, 14, 24), respectively.
TABLE 5 timing and non-timing node voltage contrast table
Taking the node 4 as an example, it is known from Table 5 that the voltage average value of the untimely sequence node 4 is 0.9743, the standard deviation is 2.83E-03, the voltage average values of the nodes 4 at the 6 th, 12h, 18h and 24h in the time sequence are 0.9822, 0.9757, 0.9788 and 0.9819 respectively, the standard deviations are 6.37E-04, 1.23E-03, 9.94E-04 and 6.97E-04 respectively, and the voltage of the node 4 is uniformly distributed with different probability from 6h in the early morning to 24h in the late evening in one day, and the probability distribution is larger than the untimely sequence power flow calculation result. This is due to the large changes in photovoltaic output and load data from day to night, causing node voltage fluctuations for each period. In the probability power flow calculation, the time sequence change characteristics of the photovoltaic output and the load are considered, the change range of the node voltage in one day can be accurately given in 24 time periods, and whether the node voltage is out of limit or not is more beneficial to judging.
Claims (1)
1. A probability power flow rapid calculation method considering the correlation between photovoltaic and load time sequence is characterized by comprising the following steps: the method comprises the following steps:
step 1: establishing a random variable time sequence related probability model and sampling
Fitting the photovoltaic output and load data by adopting an adaptive diffusion kernel density estimation method to perform adaptive diffusion kernel density estimation:
wherein: x is X i ∈[0,1]For normalization of measured data, t=h 2 >0; f (x; t) represents diffusion kernel density estimation, K D (. Cndot.) represents the diffusion kernel, x, y are random variables within the kernel definition domain, a (x) and p (x) are fundamental control parameters of the linear diffusion process, whichIn (a)
For the pilot estimation of photovoltaic output and load probability density, a (x) =p (x) λ ，λ∈[0,1]；
Given initial conditions and boundary conditions:
wherein: delta (X-X) i ) Is a random variable X i Dirac measure of (a);
setting x and y constraint conditions:
wherein:
in the case of a linear diffusion partial differential operator,
an accompanying partial differential operator of L (;
giving an adaptive diffusion kernel density estimation an adaptive optimal bandwidth h:
let x be i 、y i For the ith hour photovoltaic output and load data, F 1i (x i )、F 2i (y i ) Edge probability distributions of photovoltaic output and load for the ith hour respectively;
step 1-3: establishing a time sequence joint probability density model of photovoltaic output and load by using Copula theory:
combined probability distribution G of photovoltaic output and load at ith hour i (x i ,y i ) Represented as
G i (x i ,y i )＝C i [F 1i (x i ),F 2i (y i )] (7)
Wherein: c (C) i Connecting Copula functions for both of the i-th hour;
step 1-4: using the i-th hour photovoltaic output and load joint probability distribution G i (x i ,y i ) Sampling to obtain a time sequence related sample of 24-hour photovoltaic output and load;
step 2: fuzzy C-means clustering of time sequence photovoltaic output and load
Step 2-1: photovoltaic output and load data sample matrix X obtained for sampling
Determination of fuzzy C-means cluster number C (C)>1) Power exponent m (m)>1) Determining a membership termination tolerance epsilon and an initial membership matrix U (0) ＝[u ik (0) ]；
Step 2-2, calculating a scene clustering center matrix V= [ V ] of photovoltaic and load data by using a formula (9) i (L) ]：
Wherein: v i The method is characterized in that the method is an i-th scene cluster center, and L is algorithm iteration times;
step 2-3: substituting the clustering center matrix and the history membership matrix into the formula (10) and applying the clustering center matrix and the history membership matrix to the objective function J (L) Proceeding withOptimizing, correcting membership matrix
Wherein: j (J) (L) D is the objective function of the clustering algorithm ik (L) ＝||x k -v i (L) ||；
Step 2-4: judging the iterative ending condition of the algorithm by taking the membership termination tolerance epsilon as an index; if max { |u ik (L) -u ik (L-1) The I is smaller than the membership termination tolerance epsilon, and clustering is finished; otherwise, returning to the step 2-2, and continuing algorithm iteration;
step 2-5: ending algorithm iteration, and determining a final membership matrix U of photovoltaic and load, a scene clustering center matrix V and the total number n of random variable data in various scenes i And calculates the occurrence probability P of various scenes i ＝n i N, n is the sample size;
step 3: time sequence probability load flow calculation method
Step 3-1: substituting the photovoltaic output and load time sequence related data of each scene into the data (11) for time sequence probability load flow calculation according to the obtained scenes of the steps
Wherein: p (P) i 、Q i Respectively representing the active power and the reactive power of the node i injection system, U i 、U j Respectively represent the voltages of nodes i and j, θ ij The phase angle difference of the nodes i and j is represented by n, the node number is represented by G ij 、B ij Respectively representing system conductance and susceptance;
step 3-2: a final time sequence probability power flow calculation result is obtained by using the formula (12),
wherein: e (Y) i ) Representing the result of the probabilistic load flow calculation, F (v) i ) Expressed in terms of cluster center v i As a result of the tidal current calculation at the time of the input amount.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010733457.9A CN111984906B (en) | 2020-07-27 | 2020-07-27 | Probability power flow rapid calculation method considering correlation of photovoltaic and load time sequence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010733457.9A CN111984906B (en) | 2020-07-27 | 2020-07-27 | Probability power flow rapid calculation method considering correlation of photovoltaic and load time sequence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111984906A CN111984906A (en) | 2020-11-24 |
| CN111984906B true CN111984906B (en) | 2023-06-23 |
Family
ID=73445778
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010733457.9A Active CN111984906B (en) | 2020-07-27 | 2020-07-27 | Probability power flow rapid calculation method considering correlation of photovoltaic and load time sequence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111984906B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112541546B (en) * | 2020-12-14 | 2022-08-02 | 国网江苏省电力有限公司经济技术研究院 | Photovoltaic power station typical scene generation method based on multi-scene model |
| CN114756808A (en) * | 2022-04-08 | 2022-07-15 | 广西电网有限责任公司 | Time sequence probability load flow calculation method based on nonparametric kernel density estimation |
| CN114723171A (en) * | 2022-05-04 | 2022-07-08 | 吕承昊 | Photovoltaic characteristic curve prediction method based on probability map model and scene classification |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107104442A (en) * | 2017-04-20 | 2017-08-29 | 河海大学 | The computational methods of Probabilistic Load containing wind power plant of meter and parameter fuzzy |
| CN107591830A (en) * | 2017-10-23 | 2018-01-16 | 国网能源研究院有限公司 | A kind of electricity power planing method for being suitable to Renewable Energy Development at high proportion |
| CN107681685A (en) * | 2017-09-13 | 2018-02-09 | 国网甘肃省电力公司电力科学研究院 | A kind of Probabilistic Load computational methods for considering photovoltaic non-linear dependencies |
| CN110112790A (en) * | 2019-06-06 | 2019-08-09 | 南方电网科学研究院有限责任公司 | Method, device, equipment and storage medium for evaluating operating efficiency of active power distribution network |
-
2020
- 2020-07-27 CN CN202010733457.9A patent/CN111984906B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107104442A (en) * | 2017-04-20 | 2017-08-29 | 河海大学 | The computational methods of Probabilistic Load containing wind power plant of meter and parameter fuzzy |
| CN107681685A (en) * | 2017-09-13 | 2018-02-09 | 国网甘肃省电力公司电力科学研究院 | A kind of Probabilistic Load computational methods for considering photovoltaic non-linear dependencies |
| CN107591830A (en) * | 2017-10-23 | 2018-01-16 | 国网能源研究院有限公司 | A kind of electricity power planing method for being suitable to Renewable Energy Development at high proportion |
| CN110112790A (en) * | 2019-06-06 | 2019-08-09 | 南方电网科学研究院有限责任公司 | Method, device, equipment and storage medium for evaluating operating efficiency of active power distribution network |
Non-Patent Citations (1)
| Title |
|---|
| 黄伟 ; 陈雪 ; 林怀德 ; 羿应棋 ; 容慧娴 ; 黄春艳 ; .考虑光伏不确定性的配电网负荷特性概率评估.广东电力.2018,(06),全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111984906A (en) | 2020-11-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111984906B (en) | Probability power flow rapid calculation method considering correlation of photovoltaic and load time sequence | |
| Liu et al. | Prediction of short-term PV power output and uncertainty analysis | |
| CN106529814B (en) | Distributed photovoltaic ultra-short term prediction method based on Adaboost clustering and Markov chain | |
| CN113572206B (en) | Wind power output interval prediction method | |
| CN112633578B (en) | Cascade reservoir group optimal scheduling method under influence of diversion and water diversion engineering | |
| CN110380444B (en) | Capacity planning method for distributed wind power orderly access to power grid under multiple scenes based on variable structure Copula | |
| Hu et al. | Adaptive confidence boundary modeling of wind turbine power curve using SCADA data and its application | |
| CN111082446B (en) | Energy storage optimal configuration method considering battery self-consumption | |
| CN107679687A (en) | A kind of photovoltaic output modeling method and Generation System Reliability appraisal procedure | |
| CN105426989A (en) | EEMD and combined kernel RVM-based photovoltaic power short-term prediction method | |
| US20230048056A1 (en) | Method for describing power output of a cluster of wind and solar power stations considering time-varying characteristics | |
| CN110350535A (en) | A kind of meter and large-scale wind power and the grid-connected distribution network voltage probability of stability appraisal procedure of electric car | |
| CN106355511A (en) | Active power distribution network reconstruction method taking new energy and electric vehicle access into consideration | |
| CN112332416A (en) | Power distribution network load prediction method and system based on photovoltaic electric heating load correlation | |
| Tan et al. | An improved modeling method for uncertainty of renewable energy power generation considering random variables correlation | |
| CN112232714B (en) | Deep learning-based risk assessment method for distribution network under incomplete structural parameters | |
| CN116316869A (en) | River basin water-light medium-term complementary scheduling method considering both daily electric power and daytime electric power | |
| CN105068591A (en) | Maximum power point tracking method for partially shielded photovoltaic array | |
| Zhang et al. | Research on the prediction of solar energy generation based on measured environmental data | |
| Jin et al. | A hybrid prediction framework based on deep learning for wind power | |
| Ma | Short-Term Prediction of Photovoltaic Power Generation Based on LMD Permutation Entropy and Singular Spectrum Analysis [J] | |
| Mahmoud et al. | Classification of solar variability using K-means method for the evaluation of solar photovoltaic systems performance | |
| CN117713108A (en) | Affine analysis recursion-based load supply risk assessment method | |
| Zhang et al. | Research on Joint Operation of Wind and Solar Storage System Based on Genetic Algorithm | |
| Xiang et al. | CEEMDAN-LSTM-RF-RBF short-term wind power forecast based on incomplete data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |