CN114614463A

CN114614463A - Virtual power plant transaction strategy considering multiple uncertainties

Info

Publication number: CN114614463A
Application number: CN202111544745.0A
Authority: CN
Inventors: 刘凌燕; 庄重; 郑文明; 刘晓婕; 孙宇; 陈子豪; 吴国奇; 柴婷逸; 岳立; 钱宇轩
Original assignee: Changzhou Power Supply Co of State Grid Jiangsu Electric Power Co Ltd; Beijing Zhongdian Feihua Communication Co Ltd
Current assignee: Changzhou Power Supply Co of State Grid Jiangsu Electric Power Co Ltd; Beijing Zhongdian Feihua Communication Co Ltd
Priority date: 2021-12-16
Filing date: 2021-12-16
Publication date: 2022-06-10

Abstract

The invention provides a reasonable trading strategy considering that a virtual power plant participates in an electric power market, which comprises the following steps: firstly, analyzing and modeling uncertainty factors of virtual power plants participating in market trading; and analyzing the income of the virtual power plant in each sub-market transaction, and modeling risk factors by adopting a CVaR method to finally obtain an investment portfolio model of the virtual power plant participating in the transaction. The trading strategy provided by the invention can provide effective reference for the virtual power plant to participate in market trading, so that the trading risk caused by uncertain factors is effectively relieved when the virtual power plant trades, the problem that effective coordination control is lacked among independent distributed energy power generation is solved, the stability of the system is ensured, the complementary utilization among energy sources is promoted, and engineering practical personnel can develop related research work according to the trading strategy.

Description

Virtual power plant transaction strategy considering multiple uncertainties

Technical Field

The invention belongs to the field of power markets of power systems, and mainly relates to a virtual power plant trading strategy considering various uncertainties, which is suitable for realizing optimized trading simulation of a virtual power plant.

Background

In recent years, the penetration rate of new energy power generation in the power grid is gradually increased due to the increasingly serious environmental problems caused by the conventional power generation and the progress of new energy power generation technology. However, uncertainty and volatility caused by new energy power generation interfere with the stability of the system. The virtual power plant can organically combine facilities such as a distributed generator set, controllable load and distributed energy storage, and the integrated regulation and control of various distributed energy sources can be realized through a matched regulation and control technology and a communication technology.

With the advance of the reform of the electric power market in China, the virtual power plant as a comprehensive electricity vendor gradually participates in the electric power market, can sell surplus electric quantity, and can purchase the electric quantity from the market when the electric power in the plant is insufficient or needs to be adjusted, so that the optimal regulation and control of resources and income are realized. In an increasingly competitive electricity market, power plants are faced with different supply market options, with different markets having unique market price and profitability fluctuation characteristics. Due to the complex and diversified structures of the virtual power plant, multi-dimensional uncertainty exists in market competition. How to adopt through reasonable transaction strategy, make virtual power plant participate in the transaction in electric power market better, on the basis of keeping electric power system stability, realize the maximize of trade interests, and then provide effective reference for improving renewable energy utilization, reduce the environmental problem that traditional power generation mode brought, become the problem that at present awaits a solution.

Disclosure of Invention

The invention aims to provide a virtual power plant trading strategy considering multiple uncertainties.

In order to achieve the above purpose, the technical scheme of the invention mainly comprises the following steps:

step A: analyzing and modeling uncertainty factors of main power parts (traditional generator sets, new energy generator sets and electric vehicles) of the virtual power plant participating in medium and long-term trading markets and spot markets (main day-ahead markets, intra-day markets and real-time balance markets);

the method comprises the following steps of firstly, taking part in market trading of a virtual power plant as main uncertain factors: wind power generation, photovoltaic power generation and electric vehicle charging and discharging are analyzed, and the method mainly comprises the following steps:

a1 simulating wind speed by using a Weibull distribution probability density function, and establishing a piecewise functional relation between the output power of the wind turbine generator and the wind speed;

step A2, simulating the probability of illumination intensity in a short time period by using beta distribution, and obtaining the output of the photovoltaic cell;

step A3, establishing a probability density function of the electric automobile at the charging and discharging moment in a V2G (Vehicle to Grid) mode;

step A4 is to generate corresponding time sequence by using autoregressive moving average model according to historical data for uncertainty factors, and repeating the operation to obtain sequence scenes of N future T periods.

And B, step B: and starting to analyze the income and risk of the virtual power plant participating in market trading, and establishing a trading combination optimization model.

The method mainly comprises the following steps:

step B1: analyzing the income of the virtual power plant participating in market trading; the virtual power plant participates in the income analysis of market trading, and is characterized in that: the income R of the virtual power plant participating in market trading comprises medium-long term contract income R_LR and RI of the spot market profit_R。

Step B2: analyzing the cost (mainly reflected as the power generation cost and the operation cost of the thermal power generating unit) of the virtual power plant participating in market trading;

step B3: analyzing the risk of the virtual power plant participating in market trading; risk analysis of virtual power plant participation in market trading will be characterized in that: a corresponding loss value is defined for the confidence level corresponding to the loss caused by each uncertainty. Because CVaR is convex and less additive, risk analysis was performed using multiple loss CVaR.

Step B4: and establishing a virtual power plant transaction combination optimization model based on the analysis of the steps B1-B3. The virtual power plant transaction combination optimization model is characterized in that: a profit objective function considering risk and cost is defined as a profit expectation function, and the constraint conditions comprise market constraint, unit operation constraint, system constraint and the like.

The invention has the beneficial effects that:

based on the background that the electric power marketization innovation is continuously promoted, the virtual power plant trading strategy considering various uncertainties is designed, uncertainty factors of the virtual power plant in each sub-market trading are fully analyzed, an investment combination trading model is established, and the virtual power plant trading strategy has reference significance for the virtual power plant to participate in market trading, reasonably control trading risks and improve self income.

Drawings

FIG. 1 is a graph of the output fluctuation trend of uncertainty factors over a day;

FIG. 2 is a diagram illustrating the optimization results of the transaction power under different risk aversion values;

FIG. 3 is a graph of profit versus risk for a transaction as a function of risk aversion;

FIG. 4 shows the results of the trade optimization at different new energy permeabilities;

FIG. 5 is a graph of trading profit versus risk as a function of new energy penetration;

FIG. 6 is a one-day internal charging and discharging optimized operation curve of the electric automobile.

Detailed Description

The invention will be further explained with reference to the drawings.

The present invention will be described in further detail with reference to specific embodiments.

A virtual power plant trading strategy considering multiple uncertainties specifically comprises the following steps:

step A: analyzing and modeling uncertainty factors of main power parts (traditional generator sets, new energy generator sets and electric automobiles) of the virtual power plant participating in market trading:

step A1 wind power generation analysis

Simulating the wind speed by using a Weibull distribution probability density function:

wherein,

k is a shape parameter, which is a dimensionless quantity; c is an R degree parameter, and the dimension and the speed of the R degree parameter are the same;

the functional relationship between the output power and the wind speed of the wind turbine generator can be specifically expressed by the following piecewise function:

wherein

v_ciCut-in wind speed of wind turbine

v_coCut-out wind speed of wind turbine generator

v_rRating of wind turbineWind speed

P_r ^WRated power of wind turbine

Step A2 photovoltaic power generation analysis

The probability of the illumination intensity in a short period of time is modeled by a beta distribution, and the probability density function is as follows:

wherein,

r-actual light intensity in time period/W.m^-2；

r_max-maximum light intensity/Wm in selected time period^-2；

a, b-the position parameter and the shape parameter of the beta distribution;

the parameters in the above formula can be obtained by fitting historical meteorological data, so that the illumination intensity distribution position parameter a and the shape parameter b can be determined.

The photovoltaic cell output can be converted by the formula (3):

wherein,

the output/MW of the photovoltaic cell module at the moment P (t) -t;

a-area of unit cell module/m²；

η -cell conversion efficiency, temperature dependent;

i (t) is the solar irradiation intensity/MJ.m at the time t^-2。

Step A3 electric vehicle charging and discharging analysis

In the V2G (Vehicle to Grid) mode, the probability density function of the electric Vehicle at the charging and discharging time is as follows:

respectively, the probability density of charging and discharging of the electric vehicle, [ t ]_c1,t_c2]∪[t_c3,t_c4]For the load trough period, [ t_d1,t_d2]∪[t_d3,t_d4]During peak load periods.

Step A4 uncertainty factor modeling

And generating a corresponding time sequence by adopting an autoregressive moving average model according to historical data, and repeatedly operating to obtain sequence scenes of N future T time periods. The stationary time series obeying the ARMA model can be expressed as

Wherein,

p-coefficient of regression part

q-coefficient of moving average part

ε_tNoise terms, generally expressed as independent normal terms

θ_t-model parameters

A scenario is defined herein as an uncertainty factor value. To improve the accuracy of the solution, a sufficient number of scenes need to be generated. However, the problem scale becomes large and the solution is difficult due to too many scenes. For this reason, a fast predecessor scene cut subtraction [11] is employed herein to reduce the number of scenes to a specified number with guaranteed accuracy. And finally, converting the related sequence scene into an uncertainty factor scene sequence of a research foundation according to the relation curves of wind speed and output, illumination and output, charging and discharging of the electric vehicle and the spot price.

And B: and starting to analyze the income and risk of the virtual power plant participating in market trading, and establishing a trading combination optimization model.

The method mainly comprises the following steps:

step B1: analyzing the income of the virtual power plant participating in market trading;

the income R of the virtual power plant participating in market trading comprises medium-long term contract income R_LR and RI of the spot market profit_R. According to the research objects, the income R of the virtual power plant participating in the market trading comprises the medium-long term contract income R_LThe return of the spot market R_IAnd R_R。

(1) Medium and long term market revenue

The forward contracts may typically contract for one year to several years of power trading on a time scale. The prediction of new energy output on a long time scale lacks accuracy, and a year-round power contract trade is selected as a research object for medium-long term power trade. The medium and long term market gains are as follows:

T_Lmedium and long term contract duration

E^L-medium and long term contract electric quantity

Medium and long term contract price

(2) Spot market revenue

Fluctuating spot electricity prices and new energy output are the main uncertain factors affecting transactions among market factors. The virtual power plant sells or buys electric energy in the spot market according to the electric quantity condition of the power plant, the prediction condition of the short-term output of the new energy and the charging and discharging condition of the electric automobile.

(the day ahead market)

In the day-ahead market, market members can obtain more accurate and reliable data information, and the start-stop plan of medium and large-sized units needs to be made in the day-ahead time period, so the day-ahead market is very important for system operation and power trading.

② the domestic market

The virtual plant's return to the spot market can be expressed as follows:

T_Inumber of periods of market trade deadline within a day

-amount of electricity for market trading in the day

Electricity prices for market time t in days

(iii) real-time market

Before closing the gate, the VPP needs to trade in the equilibrium market according to the deviation of the electric quantity condition and the contract signing condition. The virtual power plant gains in the equilibrium market are as follows:

λ^R+,λ^R-penalty coefficient for unbalance of electric quantity

Day-ahead market price of electricity

The time period t requires an increased/decreased amount of power

Step B2: analyzing the cost (mainly embodied as the power generation cost and the operation cost of the thermal power generating unit) of the virtual power plant participating in market trading:

the transaction cost mainly takes the power generation cost into consideration. The power generation cost is the operation cost of the thermal power generating unit under the condition of neglecting the power generation cost of the hydroelectric generating unit and the new energy source unit. The power generation cost in the medium and long term contract transaction is determined according to the price of the power coal and the contract electric quantity; the power generation cost and the start-stop cost of the thermal power generating unit are considered in the power generation cost of the spot market. Cost function C of thermal power generation_hThe following were used:

P_h,toutput of thermal power generating unit h in time period t

a_h b_h c_h-operating consumption parameter of the power plant h, a having the unit $/(MW/h)²The unit of b is $/MWh, and the unit of c is $/h;

the starting and stopping cost of the thermal power generating unit is as follows:

0-1 Start-stop state variable of a unit h during a time period t

Cost of starting and stopping a unit h once

Step B3: analyzing the risk of the virtual power plant participating in market trading;

in the electric power market transaction, the risks mainly come from transaction risks brought by uncertain factors, namely new energy output fluctuation, electric vehicle charging and discharging behaviors and spot market price fluctuation in a spot market. Because there are many uncertainty factors causing loss in this study, different loss functions often correspond to different confidence levels, so we cannot change all loss functions into one loss to process, but define a corresponding loss value for the confidence level corresponding to each loss. Since CVaR is convex and less additive we used multiple loss CVaR for risk analysis here.

Let x be (x)₁,x₂,…x_n)^TThe proportion vector distributed for the electric quantity in each sub-market, n is the number of the sub-markets, in this case n is 4,

let y_iFor a random risk factor in the ith sub-market, with a probability density function of ρ (y), then the probability that f (x, y) exceeds the threshold α is:

the CVaR expression for a given confidence interval β ∈ (0,1) is as follows:

a confidence level is given to each risk loss function. For a given I confidence level, β_iE (0,1), I ═ 1,2, … I, a β -VaR loss value for decision vector x at this confidence level

The following were used:

on this basis, we define a β -CVaR loss value for the decision vector x

The following were used:

in addition, since the market risk as the power seller exists only in the case where the actual output value of the new energy is lower than the predicted average, the electric vehicle charge amount is higher than the predicted average, the power is purchased from the spot market at a higher than expected price, or the power is sold in the spot market at a lower than expected price. Therefore, defining the standard risk with variance alone does not fully meet practical requirements. Hence here we present the risk loss function as follows:

wherein,

real-time output predicted value of new energy output in time period t

Step B4: and establishing a virtual power plant transaction combination optimization model based on the analysis of the steps B1-B3. Defining a profit objective function considering risk and cost as a profit expectation function, and specifically expressing the following expressions:

π_ωprobability of occurrence of scene w

The constraints are as follows:

(ii) market constraints

In each sub-market trading period, the trading electric quantity cannot exceed a limit value:

while also satisfying trade contract constraints

E^L+E^D+E^I+E^R＝E (31)

② unit operation restriction

Unit output restraint:

and (4) the upper and lower output limits of the thermal power generating unit.

Upper limit of technical output value of wind turbine generator and photovoltaic generator

And (3) slope climbing rate constraint:

upward and downward climbing rate of thermal power generating unit

And power balance constraint:

P^H+P^W+P^PH-P^EVchar+P^EVdchar＝P_load (37)

third electric vehicle restraint

Charging and discharging efficiency of electric automobile

Example analysis

TABLE 1 thermal power generating unit parameters

The wind power photovoltaic long-term output prediction curve refers to historical wind speed data from 2011 to 2015 in Washington, and the power price is based on historical electricity price data from 2015 in Nord Pool power market in Nord in Nor, North Europe. The output predicted power of wind power generation and photovoltaic power generation in one day and the charge-discharge power condition of electric automobile and other value systems in one day are as shown in the attached figure 1:

except that uncertain factors in the model bring influence to trading strategies, the selection of a risk aversion value lambda, the penetration ratio of new energy and the like all influence the trading decision of VPP. Here we first verify the impact of risk aversion on the transaction decision outcome of the VPP. Calling a CPLEX solver in an MATLAB environment, obtaining a confidence level beta of 0.9 according to the model established above and the example data, and obtaining a real-time market unbalance penalty coefficient lambda by changing a risk aversion value lambda^R+＝0.85,λ^R-1.15. Setting the upper limit and the lower limit of the contract electric quantity proportion to be 90 percent and 10 percent respectively, and setting the new energy permeability r_neThe optimization solution in the case of 20% results in different electric quantity distribution and income cases as follows:

TABLE 2 trade optimization results (r) under different risk aversion values_ne＝20％)

The data results in Table 2 show the optimal values of the VPP trades for the assigned results in each of the sub-markets under the set-up scenarios. With the increase of the risk aversion index lambda, the VPP trading strategy as a power seller tends to be conservative, and the profit risk brought by the uncertainty of spot trading is preferably reduced by signing a medium-long term contract. From the profit expectation curve of fig. 4, it can be seen that in the present example, the profit risk is gradually increased with λ at the new energy permeability of 20%, and the profit is expected to have a trend of increasing first and then decreasing. The expected value of the benefit is highest at approximately λ 4.62, which is beneficial to the trader in terms of trading risk and benefit.

Below we verify the impact on the trading capacity and revenue as the new energy penetration ratio changes when λ is 4.62, the solution is as follows:

TABLE 3 New energy Permeability r_neEffect on VPP transactions (λ ═ 4.62)

From the results in table 3, we can see that under a certain risk aversion degree, with the increase of the penetration ratio of uncertainty factors, the profit of VPP in each sub-market is obviously increased, which is mainly related to the obvious reduction of power generation cost; however, when the penetration ratio of the new energy is increased to a certain degree, the instability of the system operation caused by the intermittence and the randomness of the new energy is increased sharply, the n-CVAR is also increased obviously, the operation cost of the thermal power generating unit is increased obviously, and the expected profit is reduced. In this example, under the current power generation technology, the profit value is the highest when the new energy penetration ratio is about 26.4%.

After the new energy is accessed into the system, the fluctuation of the generated output of the system is enhanced, and the market price of electricity will fluctuate within the day. Fig. 6 is a study on the influence of a new energy source (permeation rate is 26.4%) on the charging and discharging behaviors of an electric vehicle in one day.

From the results, after the new energy is accessed into the system, the charging power of the electric automobile is slightly reduced relative to the original power from 0 hour to the next morning, but as the wind power and photoelectric output are increased, the price of electricity is reduced, and the electric automobile gradually enters a charging mode from the original discharging to accumulate electric energy; at the peak time of late electricity utilization, the output of new energy is also reduced, and at the moment, the electric automobile can release the charged energy. Through the combined operation of new energy and the electric automobile, on one hand, the energy storage function of a V2G mode of the large-scale electric automobile is fully exerted, and the power generation space is increased for wind power through peak clipping and valley filling; on the other hand, the electric power market means is fully utilized, direct transaction is carried out on the large-scale electric automobile and the wind power, the charging price of the electric automobile can be reduced, and the positivity of the electric automobile as the power load to participate in wind power consumption is mobilized.

Claims

1. A virtual plant trading strategy that accounts for multiple uncertainties, characterized by:

the strategy mainly comprises the following steps:

2. A virtual plant trading strategy taking into account multiple uncertainties according to claim 1, characterized in that the analysis and modeling of virtual plant uncertainty factors:

the method mainly comprises the following steps:

firstly, main uncertain factors of virtual power plants participating in market trading are as follows: analyzing wind power generation, photovoltaic power generation and electric vehicle charging and discharging:

step A3, establishing a probability density function of the electric Vehicle at the charging and discharging moment in a V2G (Vehicle to Grid) mode;

3. A virtual power plant trading strategy taking into account multiple uncertainties according to claim 1, characterized in that the trading portfolio of virtual power plants optimizes the model:

the method mainly comprises the following steps:

step B2: analyzing the cost (mainly embodied as the power generation cost and the operation cost of the thermal power generating unit) of the virtual power plant participating in market trading;

step B4: and establishing a virtual power plant transaction combination optimization model based on the analysis of the steps B1-B3.

4. A virtual power plant trading strategy according to claim 3, wherein the revenue analysis of the participation of the virtual power plant in the market trading comprises the medium and long term contract revenue R_LR and RI of the spot market profit_R。

5. A virtual power plant trading strategy taking into account multiple uncertainties according to claim 3, characterized in that the risk analysis of virtual power plant participation in market trading: defining a corresponding loss value for the confidence level corresponding to the loss caused by each uncertainty, and performing risk analysis by using the multi-loss CVaR due to convexity and sub-additivity of the CVaR.

6. A virtual plant trading strategy taking into account multiple uncertainties according to claim 3, characterized in that the virtual plant trading portfolio optimization model: a profit objective function considering risk and cost is defined as a profit expectation function, and the constraint conditions comprise market constraint, unit operation constraint, system constraint and the like.