CN110994703A - Frequency modulation capacity demand allocation method considering multiple frequency modulation resources - Google Patents

Frequency modulation capacity demand allocation method considering multiple frequency modulation resources Download PDF

Info

Publication number
CN110994703A
CN110994703A CN201911209072.6A CN201911209072A CN110994703A CN 110994703 A CN110994703 A CN 110994703A CN 201911209072 A CN201911209072 A CN 201911209072A CN 110994703 A CN110994703 A CN 110994703A
Authority
CN
China
Prior art keywords
frequency modulation
agc
period
resource
energy storage
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.)
Granted
Application number
CN201911209072.6A
Other languages
Chinese (zh)
Other versions
CN110994703B (en
Inventor
周前
胡泽春
刘礼恺
程亮
朱寰
吴盛军
汪成根
赵静波
陈哲
岑炳成
黄成�
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.)
Tsinghua University
State Grid Corp of China SGCC
State Grid Jiangsu Electric Power Co Ltd
Electric Power Research Institute of State Grid Jiangsu Electric Power Co Ltd
Original Assignee
Tsinghua University
State Grid Corp of China SGCC
State Grid Jiangsu Electric Power Co Ltd
Electric Power Research Institute of State Grid Jiangsu 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 Tsinghua University, State Grid Corp of China SGCC, State Grid Jiangsu Electric Power Co Ltd, Electric Power Research Institute of State Grid Jiangsu Electric Power Co Ltd filed Critical Tsinghua University
Priority to CN201911209072.6A priority Critical patent/CN110994703B/en
Publication of CN110994703A publication Critical patent/CN110994703A/en
Application granted granted Critical
Publication of CN110994703B publication Critical patent/CN110994703B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H02J3/38—Arrangements for parallely feeding a single network by two or more generators, converters or transformers
    • H02J3/46—Controlling of the sharing of output between the generators, converters, or transformers
    • H02J3/48—Controlling the sharing of the in-phase component
    • 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
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06312—Adjustment or analysis of established resource schedule, e.g. resource or task levelling, or dynamic rescheduling
    • 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
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06315—Needs-based resource requirements planning or 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
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H02J3/28—Arrangements for balancing of the load in a network by storage of energy
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S10/00—Systems supporting electrical power generation, transmission or distribution
    • Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications

Landscapes

  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Engineering & Computer Science (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Tourism & Hospitality (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Marketing (AREA)
  • General Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • Quality & Reliability (AREA)
  • Development Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Operations Research (AREA)
  • Educational Administration (AREA)
  • Power Engineering (AREA)
  • Health & Medical Sciences (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 provides a frequency modulation capacity demand allocation method considering various frequency modulation resources, and belongs to the field of automatic power generation control of a power system. Firstly, calculating substitution ratios of energy storage frequency modulation resources to traditional frequency modulation resources under different energy storage frequency modulation resource occupation ratios in a model driving mode; then, for any AGC (automatic gain control) assessment period in the future, calculating the frequency modulation capacity requirement after the energy storage frequency modulation resource in the period is converted into the traditional frequency modulation resource according to historical data in a data driving mode, and acquiring all frequency modulation resource allocation combination schemes for enabling the total frequency modulation resource capacity after replacement to reach the frequency modulation capacity requirement according to the replacement ratio; and finally, selecting the scheme which enables the total frequency modulation resource cost to be the lowest from all the combination schemes as the frequency modulation capacity demand allocation scheme of the future time period. The invention can take the difference of the adjustment characteristics of various frequency modulation resources into consideration and fully exert the excellent adjustment performance of the energy storage frequency modulation resources.

Description

Frequency modulation capacity demand allocation method considering multiple frequency modulation resources
Technical Field
The invention belongs to the field of Automatic Generation Control (AGC) of an electric power system, and particularly relates to a frequency modulation capacity demand allocation method considering multiple frequency modulation resources.
Background
The output of the renewable energy has stronger fluctuation, and the amplitude and the frequency of the active power imbalance of the power system are in an increasing trend along with the access of a large amount of renewable energy. Compared with the traditional renewable energy, the stored energy has a faster regulation speed, so that the frequency modulation efficiency is higher, namely, the stored energy frequency modulation resource with unit capacity can replace frequency modulation resources (hereinafter referred to as the traditional frequency modulation resources) such as thermal power, hydropower and the like with multiple unit capacities on the premise of not damaging the frequency modulation performance.
At present, most of electric power markets still mainly use traditional frequency modulation resources, the existing method establishes the correlation among frequency modulation performance, traditional frequency modulation resource capacity and net load fluctuation according to historical data of an AGC control area, and then calculates the capacity demand of the traditional frequency modulation resources according to the prediction result of the future net load standard deviation. In recent years, more and more energy storage devices participate in frequency modulation service, the regulation capacity of energy storage frequency modulation resources is greatly different from that of traditional frequency modulation resources, and the energy storage frequency modulation resources with unit capacity can equivalently replace the traditional frequency modulation resources with multiple unit capacities, so that the energy storage frequency modulation resources and the traditional frequency modulation resources need to be distinguished when the frequency modulation capacity demand is calculated, and the energy storage frequency modulation resources and the traditional frequency modulation resources cannot be distinguished when the frequency modulation capacity demand is calculated by the conventional method.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides a frequency modulation capacity demand allocation method considering various frequency modulation resources. The invention can take the difference of the adjustment characteristics of various frequency modulation resources into consideration and fully exert the excellent adjustment performance of the energy storage frequency modulation resources.
The invention provides a frequency modulation capacity demand allocation method considering various frequency modulation resources, which is characterized by comprising the following steps of:
1) calculating the substitution ratio of the energy storage frequency modulation resources to the traditional frequency modulation resources under different energy storage frequency modulation resource occupation ratios; the method comprises the following specific steps:
1-1) taking a working day before 15 and a non-working day after 15 every month in the previous year as typical days, and obtaining 24 typical days to form a typical day set; dividing all typical days into corresponding AGC typical examination time periods according to the AGC examination unit time length gamma;
for each AGC typical evaluation period, selecting the load power, wind power generation power and photovoltaic power generation power data of each discrete time point in the period, and calculating the net load of each discrete time point in the period:
Figure BDA0002297642220000021
wherein q represents the typical evaluation period number of the AGC,
Figure BDA0002297642220000022
respectively the net load and the load power at the kth discrete time point in the qth AGC typical examination period,
Figure BDA0002297642220000023
wind power generation power and photovoltaic power generation power are respectively at the kth discrete time point in the time period; the sampling period of the discrete time point data is the same as the instruction period tau of the AGC, each AGC assessment period comprises K-gamma/tau AGC instruction periods, and K is the total number of samples of the discrete time point in each period;
calculating the net load fluctuation of each discrete time point
Figure BDA0002297642220000024
The expression is as follows:
Figure BDA0002297642220000025
wherein ,
Figure BDA0002297642220000026
for the net load mean value of the qth AGC typical examination period, the calculation expression is as follows:
Figure BDA0002297642220000027
within each AGC typical examination periodNet load fluctuations at all discrete time points
Figure BDA0002297642220000028
The constructed sequence constitutes the net load fluctuation of the time period
Figure BDA0002297642220000029
1-2) establishing an optimization model to solve the problem of AGC typical examination period net load fluctuation
Figure BDA00022976422200000210
When the ratio of the energy storage frequency modulation resource capacity to the total frequency modulation capacity of the system is η, the frequency modulation capacity requirement of the AGC control area
Figure BDA00022976422200000211
The optimization model is based on
Figure BDA00022976422200000212
And η as input variables to minimize the FM capacity requirement of the AGC control region
Figure BDA00022976422200000213
For optimization purposes, the expression is as follows:
Figure BDA00022976422200000214
s.t.
Figure BDA0002297642220000031
ACE[k-1]＝BΔf[k-1]
Figure BDA0002297642220000032
Figure BDA0002297642220000033
Figure BDA0002297642220000034
Figure BDA0002297642220000035
Figure BDA0002297642220000036
Figure BDA0002297642220000037
Figure BDA0002297642220000038
πg[k]＝min{max{π[k],ΔPg m[k]},ΔPg M[k]}
πs[k]＝π[k]-πg[k]
ΔPg[k]＝πg[k]
Ps[k]＝min{max{πs[k],Ps m[k]},Ps M[k]}
e[k]＝e[k-1]-Ps[k]δ
Figure BDA0002297642220000039
Figure BDA00022976422200000310
Figure BDA00022976422200000311
Figure BDA00022976422200000312
where, Δ f is the system frequency deviation,
Figure BDA00022976422200000313
is A in ACE examination index2The limit value specified by the index, B represents a system frequency response constant, IACE represents a PI filtering integral term of ACE,
Figure BDA00022976422200000314
the upper limit of the climbing rate of the traditional frequency modulation resource is shown, lambda is the maximum time allowed by the frequency modulation resource to reach the maximum frequency modulation capacity,
Figure BDA00022976422200000315
in order to store the capacity of the frequency modulation resource,
Figure BDA00022976422200000316
the maximum adjustable power of the conventional fm resource for the kth discrete time point,
Figure BDA0002297642220000041
for the kth discrete time point the maximum tunable power downwards of the conventional fm resource,
Figure BDA0002297642220000042
storing the upward maximum adjustable power of the frequency modulation resource for the kth discrete time point,
Figure BDA0002297642220000043
storing the maximum downward adjustable power of the frequency modulation resource for the kth discrete time point, eM and emRespectively the maximum value and the minimum value of the electric quantity of the energy storage resource, e [ k ]]Storing energy resource electricity quantity for kth discrete time point, pi [ k [ [ k ]]Adjusting the power requirement, pi, for the kth discrete-time systemg[k] and πs[k]The regulation power, delta P, respectively allocated to the conventional units and energy storage resources at the kth discrete time pointg[k] and Ps[k]The regulated power r actually born by the traditional frequency modulation resource and the energy storage resource respectivelyg[k] and rnl[k]Respectively the change rate of the traditional frequency modulation resource and the net load at the kth discrete time point, a [ k ]]Calculating the constant required for frequency deviation for the k discrete time point, D is the system load damping constantH is a system inertia constant;
1-3) selecting any AGC typical assessment time period q, taking the value of the energy storage frequency modulation resource ratio η from 0% to 100% by taking 1% as a step length, and fluctuating according to the net load corresponding to the time period
Figure BDA0002297642220000044
Calculating the frequency modulation capacity requirement of the system in the period when the energy storage frequency modulation resource occupation ratio is η by using the model established in the step 1-2)
Figure BDA0002297642220000045
1-4) repeating the steps 1-3), calculating the energy storage frequency modulation resource ratio of η for each AGC typical assessment time period in the step 1-1), and obtaining the corresponding frequency modulation capacity requirement of the system in each time period
Figure BDA0002297642220000046
The substitution ratio of the energy storage frequency modulation resource relative to the traditional frequency modulation resource under different η is calculated according to the following formula:
Figure BDA0002297642220000047
wherein Q is the total number of typical evaluation periods of AGC;
2) selecting any AGC (automatic gain control) assessment time period in the future as an F time period, and calculating the frequency modulation capacity requirement after the energy storage frequency modulation resource in the time period is converted into the traditional frequency modulation resource according to historical data; the method comprises the following specific steps:
2-1) collecting historical data of the past N years in an automatic generation control AGC control area; the historical data includes: load power per minute, wind power generation power per minute, photovoltaic power generation power per minute, and A of each AGC (automatic gain control) examination period2Indexes, the capacity of traditional frequency modulation resources and the capacity of energy storage frequency modulation resources in each AGC assessment period;
2-2) according to the traditional frequency modulation resource capacity of each AGC assessment period in the step 2-1)
Figure BDA0002297642220000048
And the capacity P of energy storage frequency modulation resources hObtaining the energy storage frequency modulation resource proportion η corresponding to each AGC examination period in the historical datahA value; h is a time interval number of the historical data;
Figure BDA0002297642220000051
according to the result of step 1), ηhThe substitution ratio of the corresponding energy storage frequency modulation resources under the value relative to the traditional frequency modulation resources is converted into the traditional frequency modulation resources from the energy storage frequency modulation resources of each AGC (automatic gain control) assessment period in the historical data, and the total frequency modulation capacity of each AGC assessment period in the converted historical data is obtained
Figure BDA0002297642220000052
The conversion for each AGC qualification period is performed according to the following equation:
Figure BDA0002297642220000053
2-3) calculating the load standard deviation delta in each AGC examination period in historical datah,l：
Figure BDA0002297642220000054
Wherein Z represents the number of discrete time points in each AGC examination period by taking 1 minute as a sampling period, Z is the Z-th discrete time sampling point,
Figure BDA0002297642220000055
load power for the z-th minute in time period h;
Figure BDA0002297642220000056
the load mean value in the time period h is calculated as follows:
Figure BDA0002297642220000057
calculating the standard deviation of the photovoltaic power generation power in each AGC examination period in historical data:
Figure BDA0002297642220000058
wherein ,
Figure BDA0002297642220000059
for the photovoltaic power generation power of the z minute in the period h,
Figure BDA00022976422200000510
the calculation method is the average value of the photovoltaic power generation power in the time period h:
Figure BDA00022976422200000511
calculating the average value of the wind power generation power in each AGC assessment period in historical data:
Figure BDA00022976422200000512
wherein ,
Figure BDA00022976422200000513
the power generated by the wind power generation in the z minute in the time period h;
2-4) comparing A of each AGC examination period in historical data2,
Figure BDA0002297642220000061
δh,l,δh,pvAnd
Figure BDA0002297642220000062
forming samples corresponding to the time period, and screening all samples in the historical data according to the following formula:
Figure BDA0002297642220000063
the specific screening method comprises the following steps: for each sample i in the historical data, counting the frequency modulation capacity after the conversion except i in the historical data
Figure BDA0002297642220000064
Is greater than
Figure BDA0002297642220000065
And the frequency modulation shows an absolute value
Figure BDA0002297642220000066
Is greater than
Figure BDA0002297642220000067
The proportion of the sample (c); if the proportion is higher than a set sample judgment threshold value l, judging that the sample i is a normal sample and reserving the normal sample, and if not, deleting the normal sample; after all samples are processed, a set consisting of all normal samples is obtained;
2-5) establishing an interval prediction model based on the load standard deviation of the extreme learning machine ELM: input of the model
Figure BDA0002297642220000068
The vector is formed by the load standard deviations of AGC assessment time periods which are the same as i every day in M days before the prediction time period i and 2 adjacent AGC assessment time periods before and after the time period; weight matrix k from input layer to hidden layerlAnd an offset vector blFor randomly generated numbers with the value between 0 and 1, each unit in the hidden layer contains an activation function sigma, and the weight from the hidden layer to the output layer
Figure BDA0002297642220000069
Optimized generation is carried out, the output of the model is the mapping corresponding to the input to the output of the model in the time period i
Figure BDA00022976422200000610
Comprises the following steps:
Figure BDA00022976422200000611
randomly extracting 75% of samples from the set consisting of the normal samples obtained in the step 2-4) to form a training set I of the prediction modellForming a weight optimization model from a hidden layer to an output layer of the extreme learning machine as follows:
Figure BDA00022976422200000612
Figure BDA00022976422200000613
Figure BDA00022976422200000614
Figure BDA00022976422200000615
Figure BDA00022976422200000616
wherein, α represents the quantile,
Figure BDA00022976422200000617
for the standard deviation of the load in the ith AGC assessment period in the training set,
Figure BDA00022976422200000618
and
Figure BDA00022976422200000619
is an auxiliary variable;
solving the weight optimization model pair weight
Figure BDA00022976422200000620
Optimizing to obtain an interval prediction model of the trained load standard deviation;
2-6) establishing an interval prediction model of the standard deviation of the photovoltaic power generation power based on the extreme learning machine: input of the model
Figure BDA00022976422200000621
The vector is formed by the standard deviations of the photovoltaic power generation power in the same time interval as i every day and in each 2 adjacent AGC assessment time intervals before and after the time interval in M days before the prediction time interval i; weight matrix k from input layer to hidden layerpvAnd an offset vector bpvFor randomly generated numbers with the value between 0 and 1, each unit in the hidden layer contains an activation function sigma, and the weight from the hidden layer to the output layer
Figure BDA0002297642220000071
Through optimized generation, the output of the model is the quantile of the standard deviation of the photovoltaic power generation power in the time period i
Figure BDA0002297642220000072
And
Figure BDA0002297642220000073
the predicted value of (2); (ii) a Mapping of the model input to output
Figure BDA0002297642220000074
Comprises the following steps:
Figure BDA0002297642220000075
randomly extracting 75% of samples from the set consisting of the normal samples obtained in the step 2-4) to form a training set I of the prediction modelpvForming a weight optimization model from a hidden layer to an output layer of the extreme learning machine as follows:
Figure BDA0002297642220000076
Figure BDA0002297642220000077
Figure BDA0002297642220000078
Figure BDA0002297642220000079
Figure BDA00022976422200000710
wherein, α represents the quantile,
Figure BDA00022976422200000711
for the standard deviation of the photovoltaic power generation power in the ith AGC examination period in the training set,
Figure BDA00022976422200000712
and
Figure BDA00022976422200000713
is an auxiliary variable;
solving the optimized model pair weights
Figure BDA00022976422200000714
Optimizing to obtain an interval prediction model of the standard deviation of the trained photovoltaic power generation power;
2-7) establishing an interval prediction model of wind power generation based on an extreme learning machine: input of the model
Figure BDA00022976422200000715
Inputting a weight matrix k from the layer to the hidden layer for predicting the wind power generation amount of the AGC examination time period which is the same as i every day in M days before the time period i and 2 adjacent AGC examination time periods before and after the same time periodwAnd an offset vector bwFor randomly generated numbers with the value between 0 and 1, each unit in the hidden layer contains an activation function sigma, and the weight from the hidden layer to the output layer
Figure BDA00022976422200000716
Through optimized generation, the output of the model is the quantile of the wind power generation power in the time period i
Figure BDA00022976422200000717
And
Figure BDA00022976422200000718
the predicted value of (2); mapping of the model input to output
Figure BDA00022976422200000719
Comprises the following steps:
Figure BDA00022976422200000720
randomly extracting 75% of samples from the set consisting of the normal samples obtained in the step 2-4) as a training set I of the prediction modelwThe weight optimization model from the hidden layer to the output layer of the extreme learning machine is formed as follows:
Figure BDA0002297642220000081
Figure BDA0002297642220000082
Figure BDA0002297642220000083
Figure BDA0002297642220000084
Figure BDA0002297642220000085
wherein, α represents the quantile,
Figure BDA0002297642220000086
for the mean value of the wind power generation power in the ith AGC examination period in the training set,
Figure BDA0002297642220000087
and
Figure BDA0002297642220000088
is an auxiliary variable;
solving the optimized model pair weights
Figure BDA0002297642220000089
Optimizing to obtain a trained interval prediction model of the wind power generation capacity;
2-8) selecting any AGC (automatic gain control) assessment period in the future as a period F, inputting the AGC assessment period which is the same as the F every day in M days before the day of the F and the load standard deviation of each 2 AGC assessment periods before and after the same period into the model trained in the step 2-5)
Figure BDA00022976422200000810
Obtaining the predicted value of the interval of the F time interval load standard deviation
Figure BDA00022976422200000811
Inputting the photovoltaic power generation power standard deviation of the AGC assessment time period which is the same as that of the F every day in M days before the day of the F and 2 AGC assessment time periods before and after the same time period into the model trained in the step 2-6)
Figure BDA00022976422200000812
Obtaining a predicted value of a region of the standard deviation of the photovoltaic power generation power in the F time period
Figure BDA00022976422200000813
Inputting the wind power generation power of the AGC assessment time period which is the same as that of the F every day for M days before the day of the F and 2 AGC assessment time periods before and after the same time period into the extreme learning machine model trained in the step 2-7)
Figure BDA00022976422200000814
Obtaining the predicted value of the interval of the standard deviation of the wind power generation power in the F time interval
Figure BDA00022976422200000815
2-9) setting F time interval to convert into initial frequency modulation capacity of traditional frequency modulation resource
Figure BDA00022976422200000816
2-10) obtaining all normal samples in the step 2-4), and calculating the frequency modulation capacity after conversion according to the following formula
Figure BDA00022976422200000817
Belong to the interval
Figure BDA00022976422200000818
Probability ratio of frequency modulation up to standard to down to standard in sample
Figure BDA00022976422200000819
wherein ΔPr' is the range quantity of the frequency modulation capacity interval;
Figure BDA00022976422200000820
2-11) judgment
Figure BDA0002297642220000091
Whether or not it is greater than the confidence ratio gammaLimitOr
Figure BDA0002297642220000092
Whether the maximum value of the converted frequency modulation capacity corresponding to all samples in the set consisting of the normal samples obtained in the step 2-4) is equal to or not
Figure BDA0002297642220000093
If at least one of the two conditions is satisfied, then
Figure BDA0002297642220000094
Converting the energy storage frequency modulation resource in the F time period into the frequency modulation capacity requirement of the traditional frequency modulation resource, and then entering the step 3); otherwise, it orders
Figure BDA0002297642220000095
Increasing by 20MW, and then returning to the step 2-10);
γLimitand calculating according to the confidence α that the frequency modulation performance reaches the standard, wherein the expression is as follows:
Figure BDA0002297642220000096
3) according to the substitution ratio of the energy storage frequency modulation resource obtained in the step 1) and the traditional frequency modulation resource, after the energy storage frequency modulation resource is equivalently substituted into the traditional frequency modulation resource in the F time period, the total capacity of the substituted frequency modulation resource reaches the capacity obtained by calculation in the step 2-9)
Figure BDA0002297642220000097
The frequency modulation resource allocation combination scheme of (1) is to find all solutions satisfying the following formula
Figure BDA0002297642220000098
wherein
Figure BDA0002297642220000099
The capacity of the frequency modulation resource is stored for the F time period,
Figure BDA00022976422200000910
for the conventional fm resource capacity in the F period:
Figure BDA00022976422200000911
4) cost C according to energy storage frequency modulation resourcesAnd cost C of conventional frequency modulation resourcesgSelecting the combination schemes obtained from step 3) to obtain the total cost
Figure BDA00022976422200000912
The lowest optimal combination scheme, which is the frequency modulation capacity demand allocation scheme of the F time period selected in the step 2).
The invention has the characteristics and beneficial effects that:
according to the method, the frequency modulation capacity is distributed among novel frequency modulation resources such as traditional frequency modulation resources and energy storage resources in a data driving and model driving fusion mode, the excellent adjusting performance of the energy storage frequency modulation resources is fully exerted, the rationality of a frequency modulation resource distribution scheme is improved, and the adjusting effect of automatic power generation control of a power system is improved.
Detailed Description
The present invention provides a method for allocating frequency modulation capacity requirements considering various frequency modulation resources, and the following describes the present invention in further detail with reference to specific embodiments.
The invention provides a frequency modulation capacity demand allocation method considering various frequency modulation resources, which comprises the following steps:
1) and calculating the substitution ratio of the energy storage frequency modulation resources and the traditional frequency modulation resources under different energy storage frequency modulation resource occupation ratios in a model-driven manner. The method comprises the following specific steps:
1-1) taking a working day before 15 th of each month in the previous year and a non-working day after 15 th of each month as typical days, forming a typical day set (the typical day set of the embodiment totally comprises 2 × 12 — 24 typical days), and dividing all the typical days into 24 × 4 — 2304 typical assessment periods according to the AGC assessment unit time length Γ (which can be checked according to the AGC assessment criterion, and the embodiment is 15 min).
For each typical AGC assessment period (the number is denoted by q, q is 1,2 … 2304), the load power, the wind power generation power and the photovoltaic power generation power data of each discrete time point in the period recorded by the AGC control area are selected, and the net load of each discrete time point in the period is calculated:
Figure BDA0002297642220000101
wherein ,
Figure BDA0002297642220000102
respectively the net load and the load power at the kth discrete time point in the qth AGC typical examination period,
Figure BDA0002297642220000103
wind power generation power and photovoltaic power generation power are respectively at the kth discrete time point in the time period. The sampling period of the discrete time point data is the same as the instruction period tau of the AGC, each AGC assessment period comprises K-gamma/tau AGC instruction periods, tau can be determined according to an AGC control area, and K is the total number of samples of the discrete time point in each period. Calculating the net load fluctuation of each discrete time point
Figure BDA0002297642220000104
The expression is as follows:
Figure BDA0002297642220000105
wherein ,
Figure BDA0002297642220000106
for the net load mean value of the qth AGC typical examination period, the calculation expression is as follows:
Figure BDA0002297642220000107
net load fluctuation of all discrete time points in each AGC typical examination period
Figure BDA0002297642220000108
The constructed sequence constitutes the net load fluctuation of the time period
Figure BDA0002297642220000109
1-2) establishing an optimization model to solve the problem of AGC typical examination period net load fluctuation
Figure BDA00022976422200001010
(by
Figure BDA00022976422200001011
Formed sequence), when the ratio of the capacity of the energy storage frequency modulation resource to the total frequency modulation capacity of the system is η (hereinafter referred to as the energy storage frequency modulation resource ratio), the frequency modulation of the AGC control areaCapacity requirement
Figure BDA00022976422200001012
The optimization model is based on
Figure BDA00022976422200001013
And η as input variables to minimize the FM capacity requirement of the AGC control region
Figure BDA0002297642220000111
For optimization purposes, the other symbols represent system constants or intermediate state quantities, which can be calculated from the input quantities and the system constants. The expression is as follows:
Figure BDA0002297642220000112
s.t.
Figure BDA0002297642220000113
ACE[k-1]＝BΔf[k-1]
Figure BDA0002297642220000114
Figure BDA0002297642220000115
Figure BDA0002297642220000116
Figure BDA0002297642220000117
Figure BDA0002297642220000118
Figure BDA0002297642220000119
Figure BDA00022976422200001110
πg[k]＝min{max{π[k],ΔPg m[k]},ΔPg M[k]}
πs[k]＝π[k]-πg[k]
ΔPg[k]＝πg[k]
Ps[k]＝min{max{πs[k],Ps m[k]},Ps M[k]}
e[k]＝e[k-1]-Ps[k]δ
Figure BDA00022976422200001111
Figure BDA00022976422200001112
Figure BDA00022976422200001113
Figure BDA00022976422200001114
where, Δ f is the system frequency deviation,
Figure BDA0002297642220000121
is A in ACE examination index2The limit value specified by the index, B represents a system frequency response constant, IACE represents a PI filtering integral term of ACE,
Figure BDA0002297642220000122
the upper limit of the climbing rate of the traditional frequency modulation resource is shown, lambda is the maximum time allowed by the frequency modulation resource to reach the maximum frequency modulation capacity,
Figure BDA0002297642220000123
in order to store the capacity of the frequency modulation resource,
Figure BDA0002297642220000124
the maximum adjustable power (positive value) of the conventional fm resource at the kth discrete time point is up,
Figure BDA0002297642220000125
the maximum adjustable power (negative value) of the conventional frequency modulation resource at the kth discrete time point is downward,
Figure BDA0002297642220000126
storing the upward maximum adjustable power (positive value) of the frequency modulation resource for the kth discrete time point,
Figure BDA0002297642220000127
for the kth discrete time point energy storage frequency modulation resource down maximum adjustable power (negative value), eM and emRespectively the maximum value and the minimum value of the electric quantity of the energy storage resource, e [ k ]]Storing energy resource electric quantity, pi, for the kth discrete time pointg[k] and πs[k]The regulation power, delta P, respectively allocated to the conventional units and energy storage resources at the kth discrete time pointg[k] and Ps[k]The regulated power r actually born by the traditional frequency modulation resource and the energy storage resource respectivelyg[k] and rnl[k]Respectively the change rate of the traditional frequency modulation resource and the net load at the kth discrete time point, a [ k ]]And calculating a constant required for the frequency deviation for the kth discrete time point, wherein D is a system load damping constant, and H is a system inertia constant.
1-3) selecting any AGC typical assessment time period q, taking the value of the energy storage frequency modulation resource ratio η from 0% to 100% by taking 1% as a step length, and fluctuating according to the net load corresponding to the time period
Figure BDA0002297642220000128
Calculating the frequency modulation capacity requirement of the system in the period when the energy storage frequency modulation resource occupation ratio is η by using the model established in the step 1-2)
Figure BDA0002297642220000129
1-4) repeating the steps 1-3), calculating the energy storage frequency modulation resource ratio of η for each AGC typical assessment time period in the step 1-1), and obtaining the corresponding frequency modulation capacity requirement of the system in each time period
Figure BDA00022976422200001210
The substitution ratio of the energy storage frequency modulation resource relative to the traditional frequency modulation resource under different η is calculated according to the following formula:
Figure BDA00022976422200001211
wherein Q is the total number of typical checking periods of AGC.
2) Selecting any AGC (automatic gain control) assessment time period in the future as an F time period, and calculating the frequency modulation capacity requirement after the energy storage frequency modulation resource in the time period is converted into the traditional frequency modulation resource according to historical data in a data driving mode; the method comprises the following specific steps:
2-1) collecting historical data of the last N years (the value range of N is 2-3) in an automatic generation control AGC control area, wherein the historical data comprises: load power per minute, wind power generation power per minute, photovoltaic power generation power per minute, and A of each AGC (automatic gain control) examination period2Indexes, the capacity of traditional frequency modulation resources and the capacity of energy storage frequency modulation resources in each AGC assessment period;
2-2) numbering according to each AGC assessment time interval (numbered by h, wherein h is the time interval number of historical data) in the step 2-1); ) Capacity of conventional fm resources
Figure BDA0002297642220000131
And energy storage frequency modulation resource capacity
Figure BDA0002297642220000132
Obtaining η energy storage frequency modulation resource proportion corresponding to each AGC examination period in historical datahThe value:
Figure BDA0002297642220000133
according to the result of step 1), ηhThe substitution ratio of the corresponding energy storage frequency modulation resources under the value relative to the traditional frequency modulation resources is converted into the traditional frequency modulation resources from the energy storage frequency modulation resources of each AGC (automatic gain control) assessment period in the historical data, and the total frequency modulation capacity of each AGC assessment period in the converted historical data is obtained
Figure BDA0002297642220000134
The conversion for each AGC qualification period is performed according to the following equation:
Figure BDA0002297642220000135
2-3) calculating the load standard deviation delta in each AGC examination period in historical datah,l：
Figure BDA0002297642220000136
Wherein Z represents the number of discrete time points in each AGC examination period by taking 1 minute as a sampling period, Z is the Z-th discrete time sampling point,
Figure BDA0002297642220000137
load power for the z-th minute in time period h;
Figure BDA0002297642220000138
the load mean value in the time period h is calculated as follows:
Figure BDA0002297642220000139
calculating the standard deviation of the photovoltaic power generation power in each AGC examination period in historical data:
Figure BDA00022976422200001310
wherein ,
Figure BDA00022976422200001311
for the photovoltaic power generation power of the z minute in the period h,
Figure BDA00022976422200001312
the method is characterized in that the average value of the photovoltaic power generation power in the AGC (automatic gain control) evaluation period in historical data is calculated in the following way:
Figure BDA0002297642220000141
calculating the average value of the wind power generation power in each AGC assessment period in historical data:
Figure BDA0002297642220000142
wherein ,
Figure BDA0002297642220000143
the power generated by the wind power generation in the z minute in the time period h;
2-4) comparing A of each AGC examination period in historical data2,
Figure BDA0002297642220000144
δh,l,δh,pvAnd
Figure BDA0002297642220000145
forming samples corresponding to the time period, and screening all samples in the historical data according to the following formula:
Figure BDA0002297642220000146
the specific implementation method comprises the following steps: for each sample i in the historical data, counting the frequency modulation capacity after the conversion except i in the historical data
Figure BDA0002297642220000147
Is greater than
Figure BDA0002297642220000148
And frequency modulatedExpress absolute value
Figure BDA0002297642220000149
Is greater than
Figure BDA00022976422200001410
The proportion of the sample (c); if the proportion is higher than a set sample judgment threshold value l (l is 0.01-0.05%), judging that the sample i is a normal sample and reserving the normal sample, otherwise deleting the normal sample; and after all the samples are processed, obtaining a set consisting of all the normal samples.
2-5) establishing an interval prediction model based on the load standard deviation of an Extreme Learning Machine (ELM): input of the model
Figure BDA00022976422200001411
Inputting a weight matrix k from a layer to a hidden layer for predicting the AGC examination time period which is the same as i every day in M days before the time period i (M is 5-7) and the load standard deviation in 2 adjacent AGC examination time periods before and after the time periodlAnd an offset vector blFor randomly generated numbers with the value between 0 and 1, each unit in the hidden layer contains an activation function sigma (adopting a signoid function), and the weight from the hidden layer to the output layer
Figure BDA00022976422200001412
Through optimized generation, the output of the model is quantile of time interval i load standard deviation
Figure BDA00022976422200001413
And
Figure BDA00022976422200001414
the predicted values of (corresponding to upper and lower quantiles, respectively)
Figure BDA00022976422200001415
Andα) (ii) a Mapping of the model input to output
Figure BDA00022976422200001416
Comprises the following steps:
Figure BDA00022976422200001417
randomly extracting 75% of samples from the set consisting of the normal samples obtained in the step 2-4) to form a training set I of the prediction modellForming a weight optimization model from a hidden layer to an output layer of the extreme learning machine as follows:
Figure BDA00022976422200001418
Figure BDA00022976422200001419
Figure BDA00022976422200001420
Figure BDA00022976422200001421
Figure BDA00022976422200001422
wherein, α represents the quantile,
Figure BDA0002297642220000151
for the standard deviation of the load in the ith AGC assessment period in the training set,
Figure BDA0002297642220000152
and
Figure BDA0002297642220000153
is an auxiliary variable;
solving the optimized model pair weights
Figure BDA0002297642220000154
And optimizing to obtain a trained interval prediction model of the standard deviation of the load.
2-6) establishing an interval prediction model of the standard deviation of the photovoltaic power generation power based on the extreme learning machine: input of the model
Figure BDA0002297642220000155
The vector is formed by the standard deviations of the photovoltaic power generation power in the same time interval as i in M days before the prediction time interval i and in each of 2 adjacent AGC assessment time intervals before and after the time interval; weight matrix k from input layer to hidden layerpvAnd an offset vector bpvFor randomly generated numbers with the value between 0 and 1, each unit in the hidden layer contains an activation function sigma (adopting a signoid function), and the weight from the hidden layer to the output layer
Figure BDA0002297642220000156
Through optimized generation, the output of the model is the quantile of the standard deviation of the photovoltaic power generation power in the time period i
Figure BDA0002297642220000157
And
Figure BDA0002297642220000158
the predicted values of (corresponding to upper and lower quantiles, respectively)
Figure BDA0002297642220000159
Andα). Mapping of the model input to output
Figure BDA00022976422200001510
Comprises the following steps:
Figure BDA00022976422200001511
randomly extracting 75% of samples from the set consisting of the normal samples obtained in the step 2-4) to form a training set I of the prediction modelpvForming a weight optimization model from a hidden layer to an output layer of the extreme learning machine as follows:
Figure BDA00022976422200001512
Figure BDA00022976422200001513
Figure BDA00022976422200001514
Figure BDA00022976422200001515
Figure BDA00022976422200001516
wherein, α represents the quantile,
Figure BDA00022976422200001517
for the standard deviation of the photovoltaic power generation power in the ith AGC examination period in the training set,
Figure BDA00022976422200001518
and
Figure BDA00022976422200001519
is an auxiliary variable;
solving the optimized model pair weights
Figure BDA00022976422200001520
And optimizing to obtain an interval prediction model of the standard deviation of the trained photovoltaic power generation power.
2-7) establishing an interval prediction model of wind power generation based on an extreme learning machine: input of the model
Figure BDA00022976422200001521
Inputting a weight matrix k from the layer to the hidden layer for predicting the wind power generation amount of the AGC examination time period which is the same as i every day in M days before the time period i and 2 adjacent AGC examination time periods before and after the same time periodwAnd an offset vector bwFor randomly generated numbers with the value between 0 and 1, each unit in the hidden layer contains an excitationLive function σ (using signoid function), weights of hidden layer to output layer
Figure BDA0002297642220000161
Through optimized generation, the output of the model is the quantile of the wind power generation power in the time period i
Figure BDA0002297642220000162
And
Figure BDA0002297642220000163
the predicted values of (corresponding to upper and lower quantiles, respectively)
Figure BDA0002297642220000164
Andα) (ii) a Mapping of the model input to output
Figure BDA0002297642220000165
Comprises the following steps:
Figure BDA0002297642220000166
randomly extracting 75% of samples from the set consisting of the normal samples obtained in the step 2-4) as a training set I of the prediction modelwForming a weight optimization model from a hidden layer to an output layer of the extreme learning machine as follows:
Figure BDA0002297642220000167
Figure BDA0002297642220000168
Figure BDA0002297642220000169
Figure BDA00022976422200001610
Figure BDA00022976422200001611
wherein, α represents the quantile,
Figure BDA00022976422200001612
for the mean value of the wind power generation power in the ith AGC examination period in the training set,
Figure BDA00022976422200001613
and
Figure BDA00022976422200001614
is an auxiliary variable;
solving the optimized model pair weights
Figure BDA00022976422200001615
Optimizing to obtain a trained interval prediction model of the wind power generation capacity;
2-8) selecting any AGC (automatic gain control) assessment period in the future as a period F, inputting the AGC assessment period which is the same as the F every day in M days before the day of the F and the load standard deviation of each 2 AGC assessment periods before and after the same period into the model trained in the step 2-5)
Figure BDA00022976422200001616
Obtaining the predicted value of the interval of the F time interval load standard deviation
Figure BDA00022976422200001617
Inputting the photovoltaic power generation power standard deviation of the AGC assessment time period which is the same as that of the F every day in M days before the day of the F and 2 AGC assessment time periods before and after the same time period into the model trained in the step 2-6)
Figure BDA00022976422200001618
Obtaining a predicted value of a region of the standard deviation of the photovoltaic power generation power in the F time period
Figure BDA00022976422200001619
Inputting the wind power generation power of the AGC assessment time period which is the same as that of the F every day for M days before the day of the F and 2 AGC assessment time periods before and after the same time period into the extreme learning machine model trained in the step 2-7)
Figure BDA00022976422200001620
Obtaining the predicted value of the interval of the standard deviation of the wind power generation power in the F time interval
Figure BDA00022976422200001621
2-9) setting F time interval to convert into initial frequency modulation capacity of traditional frequency modulation resource
Figure BDA00022976422200001622
(unit: MW);
2-10) obtaining all normal samples in the step 2-4), and calculating the frequency modulation capacity after conversion according to the following formula
Figure BDA00022976422200001623
Belong to the interval
Figure BDA0002297642220000171
Probability ratio of frequency modulation up to standard to down to standard in sample
Figure BDA0002297642220000172
wherein
Figure BDA0002297642220000173
The range amount of the frequency modulation capacity interval is set to 20MW in the present embodiment;
Figure BDA0002297642220000174
2-11) judgment
Figure BDA0002297642220000175
Whether or not it is greater than the confidence ratio gammaLimitOr
Figure BDA0002297642220000176
Whether the maximum value is equal to the maximum value in the converted frequency modulation capacity corresponding to all samples in the set consisting of the normal samples obtained in the step 2-4)
Figure BDA0002297642220000177
If at least one of the two conditions is satisfied, then
Figure BDA0002297642220000178
Converting the energy storage frequency modulation resource in the F time period into the frequency modulation capacity requirement of the traditional frequency modulation resource, and then entering the step 3); otherwise, it orders
Figure BDA0002297642220000179
Increasing by 20MW, and then returning to the step 2-10);
γLimitand calculating according to the confidence α that the frequency modulation performance reaches the standard, wherein the expression is as follows:
Figure BDA00022976422200001710
3) according to the substitution ratio of the energy storage frequency modulation resource obtained in the step 1) and the traditional frequency modulation resource, after the energy storage frequency modulation resource is equivalently substituted into the traditional frequency modulation resource in the F time period, the total capacity of the substituted frequency modulation resource reaches the capacity obtained by calculation in the step 2-9)
Figure BDA00022976422200001711
The frequency modulation resource allocation combination scheme of (1) is to find all solutions satisfying the following formula
Figure BDA00022976422200001712
wherein
Figure BDA00022976422200001713
The capacity of the frequency modulation resource is stored for the F time period,
Figure BDA00022976422200001714
for the conventional fm resource capacity in the F period:
Figure BDA00022976422200001715
4) cost C according to energy storage frequency modulation resourcesAnd cost C of conventional frequency modulation resourcesgSelecting the combination schemes obtained from step 3) to obtain the total cost
Figure BDA00022976422200001716
The lowest optimal combination scheme, which is the frequency modulation capacity demand allocation scheme of the F time period selected in the step 2).

Claims (1)

1. A method for demand allocation of frequency modulation capacity considering a plurality of frequency modulation resources is characterized by comprising the following steps:
1) calculating the substitution ratio of the energy storage frequency modulation resources to the traditional frequency modulation resources under different energy storage frequency modulation resource occupation ratios; the method comprises the following specific steps:
1-1) taking a working day before 15 and a non-working day after 15 every month in the previous year as typical days, and obtaining 24 typical days to form a typical day set; dividing all typical days into corresponding AGC typical examination time periods according to the AGC examination unit time length gamma;
for each AGC typical evaluation period, selecting the load power, wind power generation power and photovoltaic power generation power data of each discrete time point in the period, and calculating the net load of each discrete time point in the period:
Figure FDA0002297642210000011
wherein q represents the typical evaluation period number of the AGC,
Figure FDA0002297642210000012
respectively the net load and the load power at the kth discrete time point in the qth AGC typical examination period,
Figure FDA0002297642210000013
respectively in the time intervalk discrete time points are wind power generation power and photovoltaic power generation power; the sampling period of the discrete time point data is the same as the instruction period tau of the AGC, each AGC assessment period comprises K-gamma/tau AGC instruction periods, and K is the total number of samples of the discrete time point in each period;
calculating the net load fluctuation of each discrete time point
Figure FDA00022976422100000113
The expression is as follows:
Figure FDA0002297642210000014
wherein ,
Figure FDA0002297642210000016
for the net load mean value of the qth AGC typical examination period, the calculation expression is as follows:
Figure FDA0002297642210000015
net load fluctuation of all discrete time points in each AGC typical examination period
Figure FDA0002297642210000017
The constructed sequence constitutes the net load fluctuation of the time period
Figure FDA0002297642210000018
1-2) establishing an optimization model to solve the problem of AGC typical examination period net load fluctuation
Figure FDA0002297642210000019
When the ratio of the energy storage frequency modulation resource capacity to the total frequency modulation capacity of the system is η, the frequency modulation capacity requirement of the AGC control area
Figure FDA00022976422100000111
The optimization model is based on
Figure FDA00022976422100000110
And η as input variables to minimize the FM capacity requirement of the AGC control region
Figure FDA00022976422100000112
For optimization purposes, the expression is as follows:
Figure FDA0002297642210000021
s.t.
Figure FDA0002297642210000022
ACE[k-1]＝BΔf[k-1]
Figure FDA0002297642210000023
Figure FDA0002297642210000024
Figure FDA0002297642210000025
Figure FDA0002297642210000026
Figure FDA0002297642210000027
Figure FDA0002297642210000028
Figure FDA0002297642210000029
Figure FDA00022976422100000214
πs[k]＝π[k]-πg[k]
ΔPg[k]＝πg[k]
Ps[k]＝min{max{πs[k],Ps m[k]},Ps M[k]}
e[k]＝e[k-1]-Ps[k]δ
Figure FDA00022976422100000210
Figure FDA00022976422100000211
Figure FDA00022976422100000212
Figure FDA00022976422100000213
where, Δ f is the system frequency deviation,
Figure FDA00022976422100000215
is A in ACE examination index2The limit value specified by the index, B represents a system frequency response constant, IACE represents a PI filtering integral term of ACE,
Figure FDA00022976422100000216
the upper limit of the climbing rate of the traditional frequency modulation resource is shown, lambda is the maximum time allowed by the frequency modulation resource to reach the maximum frequency modulation capacity,
Figure FDA0002297642210000034
in order to store the capacity of the frequency modulation resource,
Figure FDA0002297642210000032
the maximum adjustable power of the conventional fm resource for the kth discrete time point,
Figure FDA0002297642210000033
for the kth discrete time point, the maximum adjustable power, P, downwards of the conventional FM resources M[k]Storing the upward maximum adjustable power, P, of the FM resource for the kth discrete time points m[k]Storing the maximum downward adjustable power of the frequency modulation resource for the kth discrete time point, eM and emRespectively the maximum value and the minimum value of the electric quantity of the energy storage resource, e [ k ]]Storing energy resource electricity quantity for kth discrete time point, pi [ k [ [ k ]]Adjusting the power requirement, pi, for the kth discrete-time systemg[k] and πs[k]The regulation power, delta P, respectively allocated to the conventional units and energy storage resources at the kth discrete time pointg[k] and Ps[k]The regulated power r actually born by the traditional frequency modulation resource and the energy storage resource respectivelyg[k] and rnl[k]Respectively the change rate of the traditional frequency modulation resource and the net load at the kth discrete time point, a [ k ]]Calculating a constant required by frequency deviation for the kth discrete time point, wherein D is a system load damping constant, and H is a system inertia constant;
1-3) selecting any AGC typical assessment time period q, taking the value of the energy storage frequency modulation resource ratio η from 0% to 100% by taking 1% as a step length, and fluctuating according to the net load corresponding to the time period
Figure FDA0002297642210000037
Calculating the frequency modulation capacity requirement of the system in the period when the energy storage frequency modulation resource occupation ratio is η by using the model established in the step 1-2)
Figure FDA0002297642210000038
1-4) repeating the steps 1-3), and calculating the energy storage frequency modulation resource ratio of η for each AGC typical assessment period in the step 1-1) by the systemEach time interval frequency modulation capacity requirement is obtained correspondingly
Figure FDA0002297642210000039
The substitution ratio of the energy storage frequency modulation resource relative to the traditional frequency modulation resource under different η is calculated according to the following formula:
Figure FDA0002297642210000031
wherein Q is the total number of typical evaluation periods of AGC;
2) selecting any AGC (automatic gain control) assessment time period in the future as an F time period, and calculating the frequency modulation capacity requirement after the energy storage frequency modulation resource in the time period is converted into the traditional frequency modulation resource according to historical data; the method comprises the following specific steps:
2-1) collecting historical data of the past N years in an automatic generation control AGC control area; the historical data includes: load power per minute, wind power generation power per minute, photovoltaic power generation power per minute, and A of each AGC (automatic gain control) examination period2Indexes, the capacity of traditional frequency modulation resources and the capacity of energy storage frequency modulation resources in each AGC assessment period;
2-2) according to the traditional frequency modulation resource capacity of each AGC assessment period in the step 2-1)
Figure FDA00022976422100000310
And the capacity P of energy storage frequency modulation resources hObtaining the energy storage frequency modulation resource proportion η corresponding to each AGC examination period in the historical datahA value; h is a time interval number of the historical data;
Figure FDA0002297642210000041
according to the result of step 1), ηhThe substitution ratio of the corresponding energy storage frequency modulation resources under the value relative to the traditional frequency modulation resources is converted into the traditional frequency modulation resources from the energy storage frequency modulation resources of each AGC (automatic gain control) assessment period in the historical data, and the converted energy storage frequency modulation resources of each AGC assessment period in the historical data are obtainedTotal modulation capacity P ofｒ n′(ii) a The conversion for each AGC qualification period is performed according to the following equation:
Figure FDA0002297642210000042
2-3) calculating the load standard deviation delta in each AGC examination period in historical datah,l：
Figure FDA0002297642210000043
In the formula, Z represents the number of discrete time points in each AGC examination period by taking 1 minute as a sampling period, Z is the Z th discrete time sampling point, and P isl h[z]Load power for the z-th minute in time period h; pl h，avgThe load mean value in the time period h is calculated as follows:
Figure FDA0002297642210000044
calculating the standard deviation of the photovoltaic power generation power in each AGC examination period in historical data:
Figure FDA0002297642210000045
wherein ,
Figure FDA00022976422100000411
for the photovoltaic power generation power of the z minute in the period h,
Figure FDA00022976422100000412
the calculation method is the average value of the photovoltaic power generation power in the time period h:
Figure FDA0002297642210000046
calculating the average value of the wind power generation power in each AGC assessment period in historical data:
Figure FDA0002297642210000047
wherein ,
Figure FDA00022976422100000413
the power generated by the wind power generation in the z minute in the time period h;
2-4) comparing A of each AGC examination period in historical data2,Pｒ n′；δh,l,δh,pvAnd
Figure FDA0002297642210000059
forming samples corresponding to the time period, and screening all samples in the historical data according to the following formula:
Figure FDA0002297642210000051
the specific screening method comprises the following steps: for each sample i in the historical data, counting the frequency modulation capacity after the conversion except i in the historical data
Figure FDA00022976422100000510
Is greater than
Figure FDA00022976422100000511
And the frequency modulation shows an absolute value
Figure FDA00022976422100000512
Is greater than
Figure FDA00022976422100000513
The proportion of the sample (c); if the ratio is higher than the set sample determination threshold
Figure FDA00022976422100000515
Then the sample i is judged to be a normal sample andreserving, otherwise, deleting; after all samples are processed, a set consisting of all normal samples is obtained;
2-5) establishing an interval prediction model based on the load standard deviation of the extreme learning machine ELM: input of the model
Figure FDA00022976422100000514
The vector is formed by the load standard deviations of AGC assessment time periods which are the same as i every day in M days before the prediction time period i and 2 adjacent AGC assessment time periods before and after the time period; weight matrix k from input layer to hidden layerlAnd an offset vector blFor randomly generated numbers with the value between 0 and 1, each unit in the hidden layer contains an activation function sigma, and the weight from the hidden layer to the output layer
Figure FDA00022976422100000516
Optimized generation is carried out, the output of the model is the mapping corresponding to the input to the output of the model in the time period i
Figure FDA00022976422100000517
Comprises the following steps:
Figure FDA0002297642210000052
randomly extracting 75% of samples from the set consisting of the normal samples obtained in the step 2-4) to form a training set I of the prediction modellForming a weight optimization model from a hidden layer to an output layer of the extreme learning machine as follows:
Figure FDA0002297642210000053
Figure FDA0002297642210000054
Figure FDA0002297642210000055
Figure FDA0002297642210000056
Figure FDA0002297642210000057
wherein, α represents the quantile,
Figure FDA00022976422100000518
for the standard deviation of the load in the ith AGC assessment period in the training set,
Figure FDA00022976422100000519
and
Figure FDA00022976422100000520
is an auxiliary variable;
solving the weight optimization model pair weight
Figure FDA00022976422100000521
Optimizing to obtain an interval prediction model of the trained load standard deviation;
2-6) establishing an interval prediction model of the standard deviation of the photovoltaic power generation power based on the extreme learning machine: input of the model
Figure FDA00022976422100000522
The vector is formed by the standard deviations of the photovoltaic power generation power in the same time interval as i every day and in each 2 adjacent AGC assessment time intervals before and after the time interval in M days before the prediction time interval i; weight matrix k from input layer to hidden layerpvAnd an offset vector bpvFor randomly generated numbers with the value between 0 and 1, each unit in the hidden layer contains an activation function sigma, and the weight from the hidden layer to the output layer
Figure FDA0002297642210000068
Through optimized generation, moduleThe output of the model is quantile of standard deviation of photovoltaic power generation power in the time period i
Figure FDA00022976422100000610
And
Figure FDA0002297642210000069
the predicted value of (2); (ii) a Mapping of the model input to output
Figure FDA00022976422100000611
Comprises the following steps:
Figure FDA0002297642210000061
randomly extracting 75% of samples from the set consisting of the normal samples obtained in the step 2-4) to form a training set I of the prediction modelpvForming a weight optimization model from a hidden layer to an output layer of the extreme learning machine as follows:
Figure FDA0002297642210000062
Figure FDA0002297642210000063
Figure FDA0002297642210000064
Figure FDA0002297642210000065
Figure FDA0002297642210000066
wherein α represents quantile, δi pvξ standard deviation of photovoltaic power generation power in the ith AGC examination period in the training seti pv，α，θi pv，αAnd
Figure FDA00022976422100000614
is an auxiliary variable;
solving the optimized model pair weights
Figure FDA00022976422100000615
Optimizing to obtain an interval prediction model of the standard deviation of the trained photovoltaic power generation power;
2-7) establishing an interval prediction model of wind power generation based on an extreme learning machine: input of the model
Figure FDA00022976422100000616
Inputting a weight matrix k from the layer to the hidden layer for predicting the wind power generation amount of the AGC examination time period which is the same as i every day in M days before the time period i and 2 adjacent AGC examination time periods before and after the same time periodwAnd an offset vector bwFor randomly generated numbers with the value between 0 and 1, each unit in the hidden layer contains an activation function sigma, and the weight from the hidden layer to the output layer
Figure FDA00022976422100000617
Through optimized generation, the output of the model is the quantile of the wind power generation power in the time period i
Figure FDA00022976422100000619
And
Figure FDA00022976422100000618
the predicted value of (2); mapping of the model input to output
Figure FDA00022976422100000620
Comprises the following steps:
Figure FDA0002297642210000067
randomly extracting 75% of samples from the set consisting of the normal samples obtained in the step 2-4) as a training set I of the prediction modelwThe weight optimization model from the hidden layer to the output layer of the extreme learning machine is formed as follows:
Figure FDA0002297642210000071
Figure FDA0002297642210000072
Figure FDA0002297642210000073
Figure FDA0002297642210000074
Figure FDA0002297642210000075
wherein α represents quantile, Pi w，avgFor the mean value of the wind power generation power in the ith AGC examination period in the training set,
Figure FDA00022976422100000710
and
Figure FDA00022976422100000711
is an auxiliary variable;
solving the optimized model pair weights
Figure FDA00022976422100000712
Optimizing to obtain a trained interval prediction model of the wind power generation capacity;
2-8) selecting any AGC (automatic gain control) assessment period in the future as a period F, and performing AGC assessment at the same AGC assessment period as F every day on M days before the day of F and 2 AGC assessment periods before and after the same periodInputting the standard deviation of the load of the section into the model trained in the step 2-5)
Figure FDA00022976422100000713
Obtaining the predicted value of the interval of the F time interval load standard deviation
Figure FDA0002297642210000076
Inputting the photovoltaic power generation power standard deviation of the AGC assessment time period which is the same as that of the F every day in M days before the day of the F and 2 AGC assessment time periods before and after the same time period into the model trained in the step 2-6)
Figure FDA00022976422100000714
Obtaining a predicted value of a region of the standard deviation of the photovoltaic power generation power in the F time period
Figure FDA0002297642210000077
Inputting the wind power generation power of the AGC assessment time period which is the same as that of the F every day for M days before the day of the F and 2 AGC assessment time periods before and after the same time period into the extreme learning machine model trained in the step 2-7)
Figure FDA00022976422100000715
Obtaining the predicted value of the interval of the standard deviation of the wind power generation power in the F time interval
Figure FDA0002297642210000078
2-9) setting F time interval to convert into traditional frequency modulation resource and then obtaining initial frequency modulation capacity Pr F′＝20MW；
2-10) obtaining all normal samples in the step 2-4), and calculating the converted frequency modulation capacity P according to the following formular h′Belong to the interval
Figure FDA0002297642210000079
The probability ratio gamma (P) of the frequency modulation reaching standard and failing to reach standard in the sampler F′), wherein ΔPr' is the range quantity of the frequency modulation capacity interval;
Figure FDA0002297642210000081
2-11) determination of gamma (P)r F′) Whether or not it is greater than the confidence ratio gammaLimitOr Pr F′Whether the maximum value of the converted frequency modulation capacity corresponding to all samples in the set consisting of the normal samples obtained in the step 2-4) is equal to or not
Figure FDA0002297642210000087
If at least one of the two conditions is satisfied, Pr F′Converting the energy storage frequency modulation resource in the F time period into the frequency modulation capacity requirement of the traditional frequency modulation resource, and then entering the step 3); otherwise, let Pr F′Increasing by 20MW, and then returning to the step 2-10);
γLimitand calculating according to the confidence α that the frequency modulation performance reaches the standard, wherein the expression is as follows:
Figure FDA0002297642210000082
3) according to the substitution ratio of the energy storage frequency modulation resource obtained in the step 1) and the traditional frequency modulation resource, after the energy storage frequency modulation resource is equivalently substituted into the traditional frequency modulation resource in the F time period, the total capacity of the substituted frequency modulation resource reaches the P obtained by calculation in the step 2-9)r F′The frequency modulation resource allocation combination scheme of (1) is to find all solutions satisfying the following formula
Figure FDA0002297642210000084
wherein Ps FThe capacity of the frequency modulation resource is stored for the F time period,
Figure FDA0002297642210000085
for the conventional fm resource capacity in the F period:
Figure FDA0002297642210000083
4) cost C according to energy storage frequency modulation resourcesAnd cost C of conventional frequency modulation resourcesgSelecting the combination schemes obtained from step 3) to obtain the total cost
Figure FDA0002297642210000086
The lowest optimal combination scheme, which is the frequency modulation capacity demand allocation scheme of the F time period selected in the step 2).
CN201911209072.6A 2019-11-30 2019-11-30 Frequency modulation capacity demand allocation method considering multiple frequency modulation resources Active CN110994703B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911209072.6A CN110994703B (en) 2019-11-30 2019-11-30 Frequency modulation capacity demand allocation method considering multiple frequency modulation resources

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911209072.6A CN110994703B (en) 2019-11-30 2019-11-30 Frequency modulation capacity demand allocation method considering multiple frequency modulation resources

Publications (2)

Publication Number Publication Date
CN110994703A true CN110994703A (en) 2020-04-10
CN110994703B CN110994703B (en) 2023-05-23

Family

ID=70089113

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911209072.6A Active CN110994703B (en) 2019-11-30 2019-11-30 Frequency modulation capacity demand allocation method considering multiple frequency modulation resources

Country Status (1)

Country Link
CN (1) CN110994703B (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112350344A (en) * 2020-05-25 2021-02-09 清华大学 Energy storage system-thermal power generating unit combined frequency modulation control method considering frequency modulation performance examination
CN112821424A (en) * 2021-01-29 2021-05-18 国网辽宁省电力有限公司大连供电公司 Power system frequency response analysis method based on data-model fusion drive
CN112838621A (en) * 2021-01-22 2021-05-25 上海交通大学 Electric power system frequency modulation capacity realization method considering new energy growth
CN113011030A (en) * 2021-03-23 2021-06-22 南方电网科学研究院有限责任公司 CPS 1-based frequency modulation capacity allocation method and device and storage medium
CN115189371A (en) * 2022-08-03 2022-10-14 东南大学溧阳研究院 Power curve dynamic matching-based auxiliary frequency modulation capacity calculation method for power system

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110247406A (en) * 2019-05-05 2019-09-17 清华大学 A kind of frequency regulation capacity based on conditional probability optimization method at times

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110247406A (en) * 2019-05-05 2019-09-17 清华大学 A kind of frequency regulation capacity based on conditional probability optimization method at times

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
HONGKAI LIAO等: "Design of AGC and primary frequency regulation monitoring and management system" *

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112350344A (en) * 2020-05-25 2021-02-09 清华大学 Energy storage system-thermal power generating unit combined frequency modulation control method considering frequency modulation performance examination
CN112350344B (en) * 2020-05-25 2022-03-25 清华大学 Energy storage system-thermal power generating unit combined frequency modulation control method
CN112838621A (en) * 2021-01-22 2021-05-25 上海交通大学 Electric power system frequency modulation capacity realization method considering new energy growth
CN112821424A (en) * 2021-01-29 2021-05-18 国网辽宁省电力有限公司大连供电公司 Power system frequency response analysis method based on data-model fusion drive
CN112821424B (en) * 2021-01-29 2023-07-14 国网辽宁省电力有限公司大连供电公司 Power system frequency response analysis method based on data-model fusion driving
CN113011030A (en) * 2021-03-23 2021-06-22 南方电网科学研究院有限责任公司 CPS 1-based frequency modulation capacity allocation method and device and storage medium
CN115189371A (en) * 2022-08-03 2022-10-14 东南大学溧阳研究院 Power curve dynamic matching-based auxiliary frequency modulation capacity calculation method for power system
CN115189371B (en) * 2022-08-03 2023-04-07 东南大学溧阳研究院 Power curve dynamic matching-based auxiliary frequency modulation capacity calculation method for power system

Also Published As

Publication number Publication date
CN110994703B (en) 2023-05-23

Similar Documents

Publication Publication Date Title
CN110994703A (en) Frequency modulation capacity demand allocation method considering multiple frequency modulation resources
CN104600747B (en) The Operation of Electric Systems optimization method that coordinated operation risk and wind energy are dissolved
CN107039977A (en) With the uncertain collection construction method of the power system Robust Scheduling of the minimum target of integrated cost
CN107482692B (en) Active control method, device and system for wind power plant
CN112072703B (en) Flexibility evaluation method and system suitable for new energy access planning
CN113452033B (en) Method for controlling voltage of photovoltaic power distribution network with high proportion and partitioned and autonomous and storage medium
CN117239740B (en) Optimal configuration and flexibility improvement method and system for virtual power plant system
CN112561252B (en) Reactive power combination evaluation method for power grid in new energy-containing region
CN116388245A (en) Method for configuring energy storage capacity of optical storage and charging integrated power station and related equipment
CN107994592A (en) Power budget method when one kind improves electric-net valley based on heat-storing device
CN114330938B (en) Distributed energy storage planning method and system for power distribution network
CN114243794A (en) Wind, light and water multi-energy complementary system flexibility requirement quantification and coordination optimization method
CN116826710A (en) Peak clipping strategy recommendation method and device based on load prediction and storage medium
CN110247406B (en) Frequency modulation capacity time-interval optimization method based on conditional probability
CN113991640B (en) Thermal power-based multi-energy complementary energy base energy configuration planning method
CN117439090B (en) Flexible resource allocation or scheduling method taking flexible adjustment coefficient as index
CN113947499A (en) Water, wind and light planning scheme set optimization method based on hierarchical optimization model
CN118017573A (en) Source network charge storage resource planning method considering flexible supply and demand balance
CN112117772B (en) Energy storage fast optimization configuration method for considering output resistor plug under new energy access
CN117895468A (en) Cooperative control method and system for source network load storage
CN117196884A (en) Virtual power plant optimization control method and system
CN116993022A (en) Method and device for overhauling unit and allocating hydroelectric power
CN114819362A (en) Power grid power load balancing method for wind-solar power system
CN105048491A (en) Multi-stage wind power accepted range calculating method based on unit combination and economic dispatching
CN112348706A (en) Power supply planning method and device considering wind-solar cross-space-time complementary potential

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