CN111177881A - Random production simulation method for power system containing photo-thermal-photovoltaic power generation - Google Patents

Random production simulation method for power system containing photo-thermal-photovoltaic power generation Download PDF

Info

Publication number
CN111177881A
CN111177881A CN201911178697.0A CN201911178697A CN111177881A CN 111177881 A CN111177881 A CN 111177881A CN 201911178697 A CN201911178697 A CN 201911178697A CN 111177881 A CN111177881 A CN 111177881A
Authority
CN
China
Prior art keywords
variation
thermal
photo
power generation
random
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.)
Pending
Application number
CN201911178697.0A
Other languages
Chinese (zh)
Inventor
张珍珍
陈钊
王明松
高鹏飞
赵龙
周强
王定美
汪宁渤
李津
马志程
吕清泉
张金平
张彦琪
韩旭杉
马彦宏
张睿骁
张健美
丁坤
韩自奋
黄蓉
马明
周识远
张艳丽
董海鹰
安玫
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
State Grid Corp of China SGCC
State Grid Gansu Electric Power Co Ltd
Electric Power Research Institute of State Grid Gansu Electric Power Co Ltd
Original Assignee
State Grid Corp of China SGCC
State Grid Gansu Electric Power Co Ltd
Electric Power Research Institute of State Grid Gansu Electric Power Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by State Grid Corp of China SGCC, State Grid Gansu Electric Power Co Ltd, Electric Power Research Institute of State Grid Gansu Electric Power Co Ltd filed Critical State Grid Corp of China SGCC
Priority to CN201911178697.0A priority Critical patent/CN111177881A/en
Publication of CN111177881A publication Critical patent/CN111177881A/en
Pending legal-status Critical Current

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
    • 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
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • Human Resources & Organizations (AREA)
  • General Physics & Mathematics (AREA)
  • Economics (AREA)
  • Theoretical Computer Science (AREA)
  • Strategic Management (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Analysis (AREA)
  • Development Economics (AREA)
  • Operations Research (AREA)
  • Health & Medical Sciences (AREA)
  • Pure & Applied Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • Mathematical Physics (AREA)
  • Tourism & Hospitality (AREA)
  • Mathematical Optimization (AREA)
  • Marketing (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Computational Mathematics (AREA)
  • Educational Administration (AREA)
  • Software Systems (AREA)
  • Quality & Reliability (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Game Theory and Decision Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Databases & Information Systems (AREA)
  • Algebra (AREA)
  • Probability & Statistics with Applications (AREA)
  • Public Health (AREA)
  • Water Supply & Treatment (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Supply And Distribution Of Alternating Current (AREA)

Abstract

The invention belongs to the field of reliability analysis of an electric power system containing new energy, and particularly relates to random production simulation of an electric power system containing photo-thermal-photovoltaic power generation.

Description

Random production simulation method for power system containing photo-thermal-photovoltaic power generation
Technical Field
The invention belongs to the field of reliability analysis of an electric power system containing new energy, and relates to a photo-thermal-photovoltaic power generation-containing electric power system random production simulation improvement algorithm, in particular to a photo-thermal-photovoltaic power generation-containing electric power system random production simulation method based on a pseudo sequential Monte Care improvement algorithm.
Background
Solar power generation is increasingly prominent in power supply systems with safety, cleanness, high efficiency, high sustainability and other advantages. The photo-thermal power station has good adjustability and controllability, and can replace peak regulation of a thermal power generating unit when the photo-thermal ratio reaches a certain degree. However, as the photovoltaic power generation permeability is continuously increased, the characteristics of intermittency, randomness, uncertainty and the like of photovoltaic output bring great challenges to the safe and stable operation of a power system, so that the problems of photovoltaic power generation access and consumption become main obstacles for limiting the development of photovoltaic power generation. Therefore, the photo-thermal power generation and the large-scale photovoltaic power generation are coordinated, the intermittent and random renewable energy can be consumed, and a source-end comprehensive energy power system taking the renewable energy as a main energy source is constructed.
In order to realize the photo-thermal-photovoltaic power generation coordinated planning operation, a power system production simulation method is needed to analyze the technical and economic characteristics of a power supply planning scheme, however, the power supply planning method aiming at the traditional power supply (thermal power) is difficult to be used for new energy power supply planning. On one hand, the influence of the load and new energy output time sequence characteristic change on the technical and economic indexes of system operation is difficult to accurately and sensitively reflect by the continuous load curve-based power system production simulation calculation, so that the estimation of the system operation cost generates deviation. On the other hand, the planning method based on the electric quantity balance cannot consider the resource space-time characteristics and the power system acceptance capacity of the new energy power supply, so that the estimation of the new energy grid-connected operation condition is optimistic. Therefore, a power supply planning method and a production simulation calculation method based on time sequence power balance need to be established.
Disclosure of Invention
The invention provides a random production simulation method for a power system containing photo-thermal-photovoltaic power generation, which comprises the following steps:
step A: reading photovoltaic output at each time interval, according with expected values and original data of attribute parameters of each unit, and sequencing thermal power units according to economy;
and B: obtaining a state continuous curve of the unit according to the pseudo sequential Monte Carlo, and determining the uncertainty of the new energy load error in each time period to obtain an actual curve;
and C: the VR-SVI algorithm is used for realizing the following processes:
c1, inputting the number D of original state sets, the number N of samples x and the learning rate Ptlocal variation parameters phi and gamma, initialization hyper-parameters alpha and η and global variation parameter epsilon(0)If the iteration count variable t is 0, sliding a window queue SW and obtaining a window size R;
c2, randomly sampling N system states, x, from the state set DtJudging the size values of t and R, and classifying and analyzing the type of the sliding window SW according to the size values, {1,2, …, N };
c3, initialization gammad,kCalculating local variation parameters for 1, k ∈ {1,2, …, R }, and calculating local variation parameters
Figure BDA0002290695800000021
And gammad,kUntil the local variation parameters phi and gamma converge; wherein gamma isd,kAcquiring a local variation parameter corresponding to the sample d when the window size is k;
c4 calculating random noise
Figure BDA0002290695800000022
And storing, judging the values of t and R, performing classification analysis according to the values, and calculating the new noise item value
Figure BDA0002290695800000023
C5, calculating and updating the expected outage probability index expectation E [ H ]LOLP]Expected power shortage index expected E [ H ]EENS]Random gradient of
Figure BDA0002290695800000024
And a total variation parameter epsilon;
and D, judging whether the variation coefficient β or all the variation parameters Epsilon are converged by adopting a pseudo-sequential improved simulation algorithm until the convergence of the variation coefficient β or all the variation parameters Epsilon is finished.
In C2 of the step C, if t<R, then slide window SW ← SW + xtOtherwise SW ← SW-SWt-R、SW←SW+xt。
In step C3, calculating local variation parameters
Figure BDA0002290695800000025
And gammad,kIf the local variation parameters Φ and γ converge, C4 is performed; otherwise return to re-execution C3.
In step C4, if t is judged to be the same as R, then<R, then calculate
Figure BDA0002290695800000031
And is provided with
Figure BDA0002290695800000032
Otherwise, calculating
Figure BDA0002290695800000033
And is provided with
Figure BDA0002290695800000034
and D, judging whether the variation coefficient β or all the variation parameters epsilon are converged by adopting a pseudo sequential improved simulation algorithm, ending the operation flow if the variation coefficient β or all the variation parameters epsilon are converged, and executing C5 of the step C if the variation coefficient β or all the variation parameters epsilon are not converged.
the evaluation flow of the variation coefficient β is that a power system containing n elements is subjected to random production simulation by adopting a pseudo sequential simulation algorithm:
(1) generating n random numbers according to the unavailability lambda of each elementiJudging whether an element fails or not;
(2) if the element m fails, the system fault state x is obtainedmRecording as a fault event m, otherwise, turning to the step (1);
(3) generating a random number ζ, calculating the duration D of the system in the fault state by using the formula (1)mAnalyzing the power failure time of the affected load point according to the type of the fault element and the network topology structure;
Figure BDA0002290695800000035
(4) carrying out forward and backward simulation on the fault state by adopting state transition sampling to obtain the adjacent state of the fault state, namely a complete system power failure sequence I, wherein the I consists of all fault states between two normal states;
(5) analyzing each fault state in the system power failure sequence I to obtain the power failure sequence I of each load pointj；
(6) Calculating a test function related to the fault subsequence and a required system reliability index;
(7) and (4) determining the coefficient of variation β, stopping if a threshold value is met, and returning to the step (1) if the coefficient of variation β is not met.
A device containing a method for simulating random production of a photo-thermal-photovoltaic power generation power system is characterized by comprising the following steps: the method comprises the following steps:
a reading module: the system is used for reading the photovoltaic output, the expected value and the original data of the attribute parameters of each unit in each time period and sequencing thermal power units according to economy;
an uncertainty analysis module: the system is used for obtaining a state continuous curve of the unit according to the pseudo sequential Monte Carlo, determining the uncertainty of the new energy load error in each time period, and obtaining an actual curve;
a sampling module: for reducing the number of samples using a reduced variance VR-SVI algorithm;
and the judging module is used for judging whether the variation coefficient β or all the variation parameters epsilon are converged by adopting a pseudo-sequential improved simulation algorithm until the convergence of the variation coefficient β or all the variation parameters epsilon is finished.
The invention has the beneficial effects that: the pseudo sequential Monte Carlo improved algorithm is adopted to carry out random production simulation on the power system containing photo-thermal-photovoltaic power generation, the calculation efficiency of the power system is improved by combining the pseudo sequential Monte Carlo improved algorithm with an analytical method, and the convergence speed is improved by adopting the technology with small variance, so that the calculation time consumed by the pseudo sequential Monte Carlo method is obviously reduced, and the reliability is improved.
Drawings
FIG. 1 is a flow chart of a random production simulation;
FIG. 2 is a schematic diagram of a sample power outage sequence;
FIG. 3 is a schematic view of a photo-thermal-photovoltaic power generation system;
Detailed Description
The technical scheme of the invention is further explained by specific embodiments in the following with the accompanying drawings:
example 1
The invention provides a random production simulation method of a photo-thermal-photovoltaic power generation power system based on an improved algorithm, which comprises the following steps as shown in figure 1:
step A: reading photovoltaic output at each time interval, according with expected values and original data of attribute parameters of each unit, and sequencing thermal power units according to economy;
and B: obtaining a state continuous curve of the unit according to the pseudo sequential Monte Carlo, and determining the uncertainty of the new energy load error in each time period to obtain an actual curve;
and C: the VR-SVI algorithm is used for realizing the following processes:
c1: inputting the number D of original state sets, the number N of samples x, and the learning rate Ptlocal variation parameters phi and gamma, initialization hyper-parameters alpha and η and global variation parameter epsilon(0)If the iteration count variable t is 0, sliding a window queue SW and obtaining a window size R;
c2, randomly sampling N system states, x, from the state set DtJudging the size values of t and R, and classifying and analyzing the type of the sliding window SW according to the size values, {1,2, …, N }; if t is<R, then slide window SW ← SW + xtOtherwise SW ← SW-SWt-R、SW←SW+xt。
C3, initialization gammad,kCalculating local variation parameters for 1, k ∈ {1,2, …, R }, and calculating local variation parameters
Figure BDA0002290695800000051
And gammad,kUntil the local variation parameters phi and gamma converge; wherein gamma isd,kAcquiring a local variation parameter corresponding to the sample d when the window size is k; calculating local variation parameters
Figure BDA0002290695800000052
And gammad,kIn the process, if the local variation parameters Φ and γ converge, C4 is executed; otherwise return to re-execution C3.
C4 calculating random noise
Figure BDA0002290695800000053
And storing, judging the values of t and R, and classifying, analyzing and calculating new noise item value according to the values
Figure BDA0002290695800000054
When the values of t and R are judged, if t is<R, then calculate
Figure BDA0002290695800000055
And is provided with
Figure BDA0002290695800000056
Otherwise, calculating
Figure BDA0002290695800000057
And is provided with
Figure BDA0002290695800000058
C5, calculating and updating the expected outage probability index expectation E [ H ]LOLP]Expected power shortage index expected E [ H ]EENS]Random gradient of
Figure BDA0002290695800000059
And a total variation parameter epsilon;
and D, judging whether the variation coefficient β or all the variation parameters Epsilon are converged by adopting a pseudo sequential improved simulation algorithm, ending the operation flow if the variation coefficient β or all the variation parameters Epsilon are converged, and executing C5 of the step C until the variation coefficient β or all the variation parameters Epsilon are converged.
Example 2
1 random production simulation
1.1 pseudo sequential Monte Carlo method
In the method, the non-sequential Monte Carlo sampling is carried out on the system state, and only the fault adjacent state subsequence describing the whole power failure process is subjected to sequential Monte Carlo simulation, so that the time sequence information of the off-load state subsequence contributing to the system reliability index is obtained. The basic idea of random production simulation by adopting a pseudo-sequential simulation algorithm on a power system containing n elements is as follows:
(1) generating n random numbers according to the unavailability lambda of each elementiJudging whether an element fails or not;
(2) if the element m fails, the system fault state x is obtainedmAnd is recorded as a fault event m, otherwise,turning to the step (1);
(3) generating a random number ζ, calculating the duration D of the system in the fault state by using the formula (1)mAnalyzing the power failure time of the affected load point according to the type of the fault element and the network topology structure;
Figure BDA0002290695800000061
(4) forward and backward simulations are performed on the fault state using state transition sampling, respectively, to obtain the neighboring states of the fault state, i.e. a complete system outage sequence I, which consists of all fault states between two normal states, as shown in fig. 2.
(5) Analyzing each fault state in the system power failure sequence I to obtain the power failure sequence I of each load pointj；
(6) And calculating a test function related to the fault subsequence and a required system reliability index.
(7) evaluating the corresponding coefficient of variation beta, stopping if a threshold is met, and returning to the step (1) if the threshold is not met.
Similar to the nonsequential monte carlo method, the expected value of the test function F is calculated using the average of F (x (t)) N samples obtained in the above procedure.
Figure BDA0002290695800000071
Since in the Monte Carlo simulation, state x is a random variable with a probability distribution P (x), F (x) and
Figure BDA00022906958000000711
also random variable, therefore, according to the central limit theorem, variance can be used to measure the error of the estimate:
Figure BDA0002290695800000072
where V (F) is the variance of the test function.
In pseudo-MonteIn the Carlo method, the expected outage probability index is expected to be E [ H ]LOLP]And the expected electric energy supply shortage index expected E [ H ]EENS]Can be respectively expressed as:
Figure BDA0002290695800000073
Figure BDA0002290695800000074
wherein HLOLP(xi) And HEENS(xi) Is a sampling state xiThe test results corresponding to the reliability index are shown below:
Figure BDA0002290695800000075
Figure BDA0002290695800000076
wherein m isiAn off-load status subsequence generated for an off-load status; ps() load shedding power in a certain state; d (-) is the duration of a state; x is the number offIs a set of unloaded states.
the coefficient of variation β is often used to measure the convergence of the simulation:
Figure BDA0002290695800000077
wherein, Delta is a convergence threshold value,
Figure BDA0002290695800000078
for the desired estimated value
Figure BDA0002290695800000079
The variance of (8) can be obtained by:
Figure BDA00022906958000000710
equation (9) shows that, given a convergence threshold, the only way to reduce the number of samples is to reduce the variance of the sampled random number.
Therefore, aiming at the problem that noise influences convergence speed when a large-scale data set is processed by a Stochastic Variable Inference (SVI) algorithm (unit states are represented by noise in the text), a SVI (variable Reduced SVI) algorithm with Reduced variance is provided based on an SVI algorithm framework, and the SVI algorithm is used for reliability index Inference of a power system Stochastic production simulation model for large-scale data sampling, so that sampling variance is Reduced to the maximum extent, and sampling efficiency is improved.
1.2 VR-SVI Algorithm
The (VR-SVI) algorithm introduces the concept of a sliding window, reduces the noise of the random gradient by adopting a sliding window method in the iterative process, reduces the variance of the random gradient, and enables the direction and the value of the random gradient to be as close to the real gradient as possible, thereby improving the performance of the random variational inference algorithm.
Definition 1: unit state data xt. Grouping original data sets, each group of data x of random samplingt＝{d1,d2…, dM as a unit data.
Definition 2: the window SW is slid. For positive integer R (R ∈ N ×), raw state set D ═ x1,x2,…xRThe window size is R, and the window moves forward by 1 unit data size for each iteration, and the window SW is a sliding window.
Because the SVI algorithm adopts a mode of randomly acquiring partial samples to update local variation parameters, the random gradient is ensured
Figure BDA0002290695800000081
The noise is large, different from the traditional batch variational reasoning, all samples are repeatedly used for calculating the random gradient, a part of samples are collected to calculate the random gradient in each iteration of the SVI algorithm, and after the global variational parameters are updated, the samples which are used only once are discarded, so that the random gradient deviates from the true gradient.The sliding window SW is used herein to retain the samples collected for iterations, x for t iterations when t ≦ RtEntry window SW, D ═ x1,x2,…xtUntil a sliding window SW is formed; when t is>During R, moving out t-R unit data at the leftmost side of the sliding window SW, and collecting newly xtAnd entering a sliding window to form a new sliding window data set. Sample information contained in the noise term is expanded while the high efficiency of random variational reasoning is kept.
After the sliding window SW retains the past collected sample information, the weighted sum of the noise terms of each unit data in the sliding window can be used to replace the noise term of the unit data of the current iteration. Because each data unit is randomly acquired, and in the original data set, the respective states of the data units are the same, and the influence of each data unit in the sliding window can be equally considered, the weight average of each noise term is 1/R, and the value of a new noise term is obtained by calculation, wherein the calculation formula is as follows:
Figure BDA0002290695800000091
as can be seen from the equation (10), the sliding window phase change enlarges the sample information, so as to improve the calculation method of the noise term, reduce the noise of random gradient, and enable the noise term calculated by adopting the sliding window
Figure BDA0002290695800000095
The method is closer to the value in batch variation estimation, noise is reduced, so that the random gradient is more biased to the true gradient compared with the SVI algorithm, the variance of the random gradient is reduced compared with the SVI algorithm, and finally the convergence speed of the algorithm is increased; and when the sliding window moves backwards, the window sample information changes correspondingly, and the calculation of the next noise term contains the sample information acquired last time, so that the noise is smoothed.
To avoid the increase in time complexity caused by repeatedly calculating the noise term of each unit data of the sliding window, the VR-SVI algorithm can store the current x at each iterationtNoise term of
Figure BDA0002290695800000092
Gas is used for next iterative computation
Figure BDA0002290695800000093
The algorithm takes the spatial cost of the sliding window size to store the noise term of a single bit of data in the window. The VR-SVI algorithm is implemented as follows:
(1) inputting the number D of original state sets, the number N of samples x, and the learning rate rhotinitializing local variation parameters phi and gamma, initializing hyper-parameters α and η and a global variation parameter epsilon (0), setting an iteration count variable t as 0, sliding a window queue SW and obtaining a window size R;
(2) randomly sampling N system states, x, from a set of states D t1,2, …, N, if t<R, then slide window SW ← SW + xtOtherwise SW ← SW-SWt-R、SW←SW+xt；
(3) Initializing gamma d,k1, k ∈ {1,2, …, R }, since d ∈ xtAnd N is equal to {1,2, …, N }, calculating local variation parameters
Figure BDA0002290695800000094
And gammad,kUntil the local variation parameters phi and gamma converge;
(4) since k ∈ {1,2, …, R }, random noise is calculated
Figure BDA0002290695800000101
And store if t<R, then calculate
Figure BDA0002290695800000102
And is provided with
Figure BDA0002290695800000103
Otherwise, calculating
Figure BDA0002290695800000104
And is provided with
Figure BDA0002290695800000105
(5) Computing
Figure BDA0002290695800000106
And e until all the variation parameters e converge.
1.3 analysis of variance reduction
To facilitate proof of random gradients in the algorithm
Figure BDA0002290695800000107
The variance of (2) is reduced, defining a true gradient gtIt conforms to E [ gt]＝EBi[gt]And a full gradient is introduced:
Figure BDA0002290695800000108
wherein the content of the first and second substances,
Figure BDA0002290695800000109
indicates when the sample xiThe size N of the initial state set is equal to the size D of the whole initial state set; . Order to
Figure BDA00022906958000001010
It is possible to obtain:
Figure BDA00022906958000001011
the following steps are provided:
Figure BDA00022906958000001012
substituting the formulae (10), (11) and (13) into the formula (11):
Figure BDA00022906958000001013
wherein the content of the first and second substances,
Figure BDA00022906958000001014
the term is related to the value of ε;
Figure BDA00022906958000001015
is a random variable whose average approaches zero. The random variational reasoning process can know that the two items have no correlation, so that the following results are obtained:
Figure BDA00022906958000001016
expanding equation (12) to:
Figure BDA00022906958000001017
since the variance of the random gradient in the SVI is very small in the R iterations of the sliding window, it can be considered as invariant,
Figure BDA0002290695800000111
the following can be obtained:
Figure BDA0002290695800000112
the following binding formulae (12), (15) and (17):
Figure BDA0002290695800000113
from the reasoning process, the variance between the random gradient and the full gradient in the VR-SVI algorithm provided by the method is 1/R of the SVI algorithm, and the variance between the random gradient and the full gradient is smaller than that in the SVI algorithm, which shows that the method provided by the method can achieve the purposes of reducing the variance and rapidly converging.
2 simulation of uncertainty factors of system
Fig. 3 is a schematic diagram of a photo-thermal-photovoltaic power generation system, and it can be found that the photo-thermal-photovoltaic complementary power generation system can be mainly divided into a power grid load, a photo-thermal power generation system, a photovoltaic power generation system, a heat storage system, a conventional unit and the like, wherein the heat storage system mainly adopts a lava energy storage unit.
In a photo-thermal-photovoltaic power generation system, the photo-thermal power generation mainly utilizes a heat storage system to stabilize photovoltaic output fluctuation, stores energy when photovoltaic power generation is carried out greatly, and provides energy when photovoltaic power generation is insufficient, so that impact on a power grid is mainly reduced, and the electric energy quality is improved.
2.1 photo-thermal-photovoltaic power generation output model
(1) Photovoltaic power generation output model
And expressing the uncertainty of the photovoltaic output by the uncertainty of the prediction error, and defining the photovoltaic actual output as the superposition of the photovoltaic output prediction value and the prediction error. The formula is as follows:
Figure BDA0002290695800000114
wherein the content of the first and second substances,
Figure BDA0002290695800000115
actual output at photovoltaic time t;
Figure BDA0002290695800000116
for the random deviation of the photovoltaic at the time t, theoretical research and engineering practice show that: photovoltaic output prediction error
Figure BDA0002290695800000121
The obeyed probability distribution may be approximated as a normal distribution. And a large number of researches show that the expected value of the photovoltaic prediction error distribution is 0, and the standard deviation is related to the predicted time scale and the installed capacity and is recorded as:
Figure BDA0002290695800000122
wherein, PHIAnd the actual output at the photovoltaic time t.
(2) Photo-thermal power generation output model
According to the charging and discharging characteristics of the photo-thermal power generation heat storage device, the operating time and the operating times of the photo-thermal power generation multi-state model are extracted by using Monte Carlo, probability statistics is carried out on the stored energy output, and the approximate probability distribution of the photo-thermal power station power is obtained and recorded as Fpt(X). Wherein X represents the output power of the energy storage device for photo-thermal power generation, and X existence discharge (X) can be obtained according to the output characteristic of the energy storage device for photo-thermal power generation>0) And charging (X)<0) Two states. Therefore, the generator and the load can be used for equivalently replacing the photo-thermal power generation output in the two states, and further the corresponding output/load power probability distribution is obtained:
Figure BDA0002290695800000123
Figure BDA0002290695800000124
in the probability density distribution functions represented by the formulae (21) and (22), Fpt_gene、Fpt_loadAnd respectively representing the power probability distribution of an equivalent generator and an equivalent load corresponding to the optical thermal power station configured with stored energy in the discharging state and the charging state.
(3) Photo-thermal-photovoltaic power generation output model
In the photo-thermal-photovoltaic power generation system, a photo-thermal power station with certain heat storage capacity carries out smooth control on photovoltaic power generation output, fully utilizes the photo-thermal-photovoltaic power generation time-sharing energy complementation characteristic, and effectively reduces the combined output fluctuation of the system. Therefore, the photothermal-photovoltaic output model is mainly based on the energy storage control method of the low-pass filtering principle, and the method comprises the following steps:
Figure BDA0002290695800000131
Ppt_ref(t+1)＝Ppv_ref(t)-Ppv(t) (24)
wherein, Ppv_refIs a photovoltaic output reference value; ppt_refThe reference value of photo-thermal output is obtained; t isdFor calculating the period, T is a filtering time constant; (t-1) and (t +1) indicate the previous and subsequent periods of the sampling time t, respectively.
After the uncertainty of photovoltaic output and the complementary characteristics of the photo-thermal power station configured with heat storage are considered, the influence of weather factors on the output of the photo-thermal-photovoltaic power generation system is mainly analyzed. In the photo-thermal-photovoltaic power generation system, it is assumed that weather patterns caused by weather factors are mainly divided into the following two cases: 1) when the illumination is sufficient: the photovoltaic output is gradually increased after 7:30 in the morning, the photo-thermal output is gradually reduced until the photovoltaic output is maximum at 13 noon, the photo-thermal output is corresponding to the minimum photo-thermal output, the photovoltaic output is gradually reduced after 16:30, the photo-thermal output is gradually increased until the photo-thermal output is maximum at 21 pm, and the operation is continued until 7:30 in the morning next day, so that the photo-thermal-photovoltaic power generation system keeps stable operation in a circulating manner. 2) When the illumination is insufficient: when the time of insufficient illumination or continuous cloudy days is within the adjustable range of the energy storage capacity of the photo-thermal power station, the photo-thermal power station configured with energy storage can stabilize the photovoltaic output fluctuation when the load changes, so that the system can stably run; on the contrary, if the illumination is insufficient or the continuous cloudy time exceeds the adjustable range of the energy storage capacity of the photo-thermal power station, a conventional unit needs to be loaded to keep the system stable in operation.
Therefore, through the analysis of each meteorological model and based on the above (1) (2), the output probability model of the available photothermal-photovoltaic can be represented by the following piecewise function through the conditional probability:
Figure BDA0002290695800000132
Figure BDA0002290695800000133
in the formula, Ppt_pv_gene(X)、Ppt_pv_load(X) respectively representing power probability distribution of an equivalent generator and an equivalent load corresponding to photo-thermal-photovoltaic combined output in two states of charging and discharging; fpt is approximate probability distribution of photo-thermal output; f. ofpvApproximate probability distribution for photovoltaic output;
Figure BDA0002290695800000134
the photo-thermal-photovoltaic combined output probability distribution under the discharge state and the charge state is respectively expressed and converted according to the definition of a conditional probability density function.
In the formulas (25) and (26), when the photothermal-photovoltaic power generation output model is described, the load change of the photothermal regulation photovoltaic output response in the system is mainly described according to the charging and discharging characteristics of the energy storage device for photothermal power generation, namely when the X is greater than 0, and the load peak period and the insufficient illumination condition in a fine day are mainly described; when X is less than 0, the light heat in the system keeps the lowest output operation, and the photovoltaic responds to the load change, so that the system stably operates.
2.2 load uncertainty handling
With regard to a method for handling uncertainty of a load in an electric power system, it is often used that an actual load is represented by a superposition of a predicted load and a random deviation of the load. I.e. the actual load at any time t can be calculated as follows:
Figure BDA0002290695800000141
wherein the content of the first and second substances,
Figure BDA0002290695800000149
actual output at the moment of load t;
Figure BDA0002290695800000142
the original load at the moment t of the load is obtained;
Figure BDA0002290695800000143
for the load random deviation, the deviation is obedient mean value of 0 and variance of
Figure BDA0002290695800000144
Normally distributing random variables. As can be seen from a summary of the study,
Figure BDA0002290695800000145
the standard deviation of (d) can be calculated by:
Figure BDA0002290695800000146
2.3 component uncertainty handling
Conventional genset components are described using a two-state model. Using duration of stateThe sampling method simulates the state transition and circulation process of each element and considers the normal operation duration tau1And fault repair time τ2Following the exponential distribution, the calculation formula is as follows:
Figure BDA0002290695800000147
wherein the content of the first and second substances,
Figure BDA0002290695800000148
is a random number which is uniformly distributed; t isTMTTFIs the average working time; t isTMTTRMean repair time.
A device containing a method for randomly simulating production of a photo-thermal-photovoltaic power generation power system is characterized by comprising the following steps:
a reading module: the system is used for reading the photovoltaic output, the expected value and the original data of the attribute parameters of each unit in each time period and sequencing thermal power units according to economy;
an uncertainty analysis module: the system is used for obtaining a state continuous curve of the unit according to the pseudo sequential Monte Carlo, determining the uncertainty of the new energy load error in each time period, and obtaining an actual curve;
a sampling module: for reducing the number of samples using a reduced variance VR-SVI algorithm;
and the judging module is used for judging whether the variation coefficient β or all the variation parameters epsilon are converged by adopting a pseudo-sequential improved simulation algorithm until the convergence of the variation coefficient β or all the variation parameters epsilon is finished.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solutions of the present invention and not for limiting the same, and although the present invention is described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications and equivalents may be made to the embodiments of the invention without departing from the spirit and scope of the invention, which is to be covered by the claims.

Claims (8)

1. The method for simulating random production of the power system containing photo-thermal-photovoltaic power generation is characterized by comprising the following steps of:
step A: reading original data of photovoltaic output, expected values and unit attribute parameters at each time interval, and sequencing thermal power units according to economy;
and B: obtaining a state continuous curve of the unit according to the pseudo sequential Monte Carlo by using the sequencing data, and determining the uncertainty of the new energy load error at each time interval to obtain an actual curve;
and C: according to the actual curve, reducing the sampling times by utilizing a VR-SVI algorithm with reduced variance;
and D, judging whether the variation coefficient β or all the variation parameters Epsilon are converged by adopting a pseudo-sequential improved simulation algorithm until the convergence of the variation coefficient β or all the variation parameters Epsilon is finished.
2. The method for stochastic production simulation of an electric power system comprising photo-thermal-photovoltaic power generation as claimed in claim 1, wherein: in the step C, the VR-SVI algorithm is realized as follows:
c1: inputting the number D of original state sets, the number N of samples x, and the learning rate Ptlocal variation parameters phi and gamma, initialization hyper-parameters alpha and η and global variation parameter epsilon(0)If the iteration count variable t is 0, sliding a window queue SW and obtaining a window size R;
c2, randomly sampling N system states from the number D of the original state set, xtJudging the size values of t and R, and classifying and analyzing the type of the sliding window SW according to the size values, {1,2, …, N };
c3, initialization gammad,kCalculating local variation parameters for 1, k ∈ {1,2, …, R }, and calculating local variation parameters
Figure FDA0002290695790000011
And gammad,kUntil the local variation parameters phi and gamma converge; wherein gamma isd,kAcquiring a local variation parameter corresponding to the sample d when the window size is k;
c4 calculating random noise
Figure FDA0002290695790000012
And storing, judging the values of t and R, performing classification analysis according to the values, and calculating the new noise item value
Figure FDA0002290695790000013
C5, calculating and updating the expected outage probability index expectation E [ H ]LOLP]Expected power shortage index expected E [ H ]EENS]Random gradient of
Figure FDA0002290695790000014
And an overall variation parameter epsilon.
3. The method for stochastic production simulation of an electric power system comprising photo-thermal-photovoltaic power generation as claimed in claim 2, wherein: in C2 of the step C, if t<R, then slide window SW ← SW + xtOtherwise SW ← SW-SWt-R、SW←SW+xt。
4. The method for stochastic production simulation of an electric power system comprising photo-thermal-photovoltaic power generation as claimed in claim 2, wherein: in step C3, calculating local variation parameters
Figure FDA0002290695790000021
And gammad,kIf the local variation parameters Φ and γ converge, C4 is performed; otherwise return to re-execution C3.
5. The method for stochastic production simulation of an electric power system comprising photo-thermal-photovoltaic power generation as claimed in claim 2, wherein: in step C4, if t is judged to be the same as R, then<R, then calculate
Figure FDA0002290695790000022
And is provided with
Figure FDA0002290695790000023
Otherwise, calculating
Figure FDA0002290695790000024
And is provided with
Figure FDA0002290695790000025
6. the method for simulating random production of the power system with photo-thermal-photovoltaic power generation of claim 1, wherein the step D adopts a pseudo-sequential improved simulation algorithm to judge whether the variation coefficient β or all the variation parameters epsilon are converged, if the variation coefficient β or all the variation parameters epsilon are converged, the operation process is ended, and if the variation coefficient β or all the variation parameters epsilon are not converged, the step of calculating and updating the expected outage probability index expectation EHH is executedLOLP]Expected power shortage index expected E [ H ]EENS]Random gradient of
Figure FDA0002290695790000026
And C5 for the total variation parameter epsilonctep C.
7. the method for simulating random production of a power system containing photo-thermal-photovoltaic power generation as claimed in claim 1, wherein the evaluation process of the coefficient of variation β is to perform random production simulation on a power system containing n elements by using a pseudo sequential simulation algorithm:
(1) generating n random numbers according to the unavailability lambda of each elementiJudging whether an element fails or not;
(2) if the element m fails, the system fault state x is obtainedmRecording as a fault event m, otherwise, turning to the step (1);
(3) generating a random number ζ, calculating the duration D of the system in the fault state by using the formula (1)mAnalyzing the power failure time of the affected load point according to the type of the fault element and the network topology structure;
Figure FDA0002290695790000031
(4) carrying out forward and backward simulation on the fault state by adopting state transition sampling to obtain the adjacent state of the fault state;
(5) analyzing each fault state in the system power failure sequence I to obtain the power failure sequence I of each load pointj；
(6) Calculating a test function related to the fault subsequence and a required system reliability index;
(7) and (4) determining the coefficient of variation β, stopping if a threshold value is met, and returning to the step (1) if the coefficient of variation β is not met.
8. A device containing a method for randomly simulating production of a photo-thermal-photovoltaic power generation power system is characterized by comprising the following steps:
a reading module: the system is used for reading the photovoltaic output, the expected value and the original data of the attribute parameters of each unit in each time period and sequencing thermal power units according to economy;
an uncertainty analysis module: the system is used for obtaining a state continuous curve of the unit according to the pseudo sequential Monte Carlo, determining the uncertainty of the new energy load error in each time period, and obtaining an actual curve;
a sampling module: for reducing the number of samples using a reduced variance VR-SVI algorithm;
and the judging module is used for judging whether the variation coefficient β or all the variation parameters epsilon are converged by adopting a pseudo-sequential improved simulation algorithm until the convergence of the variation coefficient β or all the variation parameters epsilon is finished.
CN201911178697.0A 2019-11-27 2019-11-27 Random production simulation method for power system containing photo-thermal-photovoltaic power generation Pending CN111177881A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911178697.0A CN111177881A (en) 2019-11-27 2019-11-27 Random production simulation method for power system containing photo-thermal-photovoltaic power generation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911178697.0A CN111177881A (en) 2019-11-27 2019-11-27 Random production simulation method for power system containing photo-thermal-photovoltaic power generation

Publications (1)

Publication Number Publication Date
CN111177881A true CN111177881A (en) 2020-05-19

Family

ID=70650088

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911178697.0A Pending CN111177881A (en) 2019-11-27 2019-11-27 Random production simulation method for power system containing photo-thermal-photovoltaic power generation

Country Status (1)

Country Link
CN (1) CN111177881A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115189402A (en) * 2022-07-27 2022-10-14 国网甘肃省电力公司经济技术研究院 Photo-thermal-photovoltaic-wind power combined output probability modeling method
CN116720324A (en) * 2023-05-15 2023-09-08 中铁第四勘察设计院集团有限公司 Traction substation key equipment fault early warning method and system based on prediction model

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115189402A (en) * 2022-07-27 2022-10-14 国网甘肃省电力公司经济技术研究院 Photo-thermal-photovoltaic-wind power combined output probability modeling method
CN115189402B (en) * 2022-07-27 2023-08-18 国网甘肃省电力公司经济技术研究院 Photo-thermal-photovoltaic-wind power combined output probability modeling method
CN116720324A (en) * 2023-05-15 2023-09-08 中铁第四勘察设计院集团有限公司 Traction substation key equipment fault early warning method and system based on prediction model

Similar Documents

Publication Publication Date Title
Wu et al. A novel hybrid system based on multi-objective optimization for wind speed forecasting
CN109902801B (en) Flood collective forecasting method based on variational reasoning Bayesian neural network
CN110059844B (en) Energy storage device control method based on ensemble empirical mode decomposition and LSTM
Wang et al. A novel combined model based on hybrid optimization algorithm for electrical load forecasting
CN110705743B (en) New energy consumption electric quantity prediction method based on long-term and short-term memory neural network
CN111401755B (en) Multi-new-energy output scene generation method, device and system based on Markov chain
CN111898828A (en) Hydroelectric power generation prediction method based on extreme learning machine
CN113554466B (en) Short-term electricity consumption prediction model construction method, prediction method and device
CN114462718A (en) CNN-GRU wind power prediction method based on time sliding window
CN111008790A (en) Hydropower station group power generation electric scheduling rule extraction method
CN108960485A (en) One provenance-lotus interacts the online dictionary learning probability optimal load flow method under electricity market
CN111177881A (en) Random production simulation method for power system containing photo-thermal-photovoltaic power generation
CN115986728A (en) Power grid situation prediction method considering uncertainty factors and terminal
CN112381282A (en) Photovoltaic power generation power prediction method based on width learning system
CN115759415A (en) Power consumption demand prediction method based on LSTM-SVR
CN111815039A (en) Weekly scale wind power probability prediction method and system based on weather classification
CN116345555A (en) CNN-ISCA-LSTM model-based short-term photovoltaic power generation power prediction method
Liu et al. Research on wind speed prediction of wind power system based on GRU deep learning
CN114971090A (en) Electric heating load prediction method, system, equipment and medium
CN114219126A (en) Method for predicting power supply load of small hydropower station permeable regional network based on residual correction
CN109447843B (en) Photovoltaic output prediction method and device
CN115759343A (en) E-LSTM-based user electric quantity prediction method and device
CN114021847A (en) Short-term power load prediction method and system based on ISGU (inverse transformation and aggregation) hybrid model
Shen et al. An interval analysis scheme based on empirical error and mcmc to quantify uncertainty of wind speed
Alomoush et al. Residential Power Load Prediction in Smart Cities using Machine Learning Approaches

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