1. A method for arranging the electric quantity purchased by medium and long-term electric power markets in different time scales is characterized by comprising the following steps:

s1, establishing a risk assessment calculation model aiming at the prediction error, measuring the risk through the conditional risk value CVaR, and determining the basic form of the CVaR;

s2, establishing an annual transaction optimization model and a monthly transaction optimization model based on conditional risk values, introducing risk cost into an optimal economic optimization function, and performing linear transformation on nonlinear constraint conditions to realize optimal solution;

And S3, performing rolling optimization in different time scale markets, and determining the power purchasing plan of the medium and long term markets of the power.

2. The method according to claim 1, wherein a risk assessment calculation model for prediction errors is established in step S1, risk is measured by conditional risk value CVaR, and a basic form of CVaR is determined;

at a given confidence level β (β ∈ (0, 1)), the value of the risk value VaR is:

f_VaR(x)＝min{α∈R；∫_{f(x，y)≤α}ρ(y)dy≥β} (1)；

wherein: f (x, y) is a loss function of investment; x is formed by RⁿAn n-dimensional portfolio scenario vector; y is formed by R^mThe random vector is an m-dimensional random vector and represents a random factor in the medium and long-term market of the power; α is a threshold value of allowable loss, and ρ (y) is a probability density function of y;

the expression for conditional risk value CVaR is:

will f is_CVaR(x) Simplified and converted to the following forms, using F_β(x, α) to denote CVaR:

converting the above equation (3) into a discretization form, and substituting the discretization expression into sample data calculation:

[f(x，y_k)-α]⁺＝max[f(x，y_k)-α，0] (5)；

wherein the content of the first and second substances,as an estimate of CVaR, y_kThe sample data of the kth group has N groups.

3. The method according to claim 1, wherein the step S2 is implemented by establishing an annual transaction optimization model based on conditional risk values, introducing risk costs into an economically optimal annual transaction optimization function through risk analysis of load prediction and uncertainty of clean energy output, and performing linear transformation on nonlinear constraint conditions, and specifically includes:

(1) Setting an objective function

Setting an annual transaction objective function including electricity purchasing cost and risk cost, wherein the expression is as follows:

wherein: lambda [ alpha ]_iThe price of the on-line electricity of the ith clean energy in the preferential power generation is set; lambda [ alpha ]_jThe price of electricity of an electric power plant outside the jth zone; lambda [ alpha ]_xBilateral agreement electricity prices for the xth power generator in the annual market; lambda [ alpha ]_m ^tPredicted average electricity prices for the monthly market for t months; lambda [ alpha ]_zThe online electricity price of the non-marketized power generation enterprise of the z type power generation type; respectively representing the electric quantity of the ith clean energy in the t month, the electric quantity of a j district external power plant and the x generationBilateral agreement electric quantity of an electric company in a yearly market, electric quantity of an x-th electric generator in a monthly market, and electric quantity of a z-th non-marketized electric generation enterprise of an electric generation type; rho is a risk aversion coefficient, the more conservative decision makers select the rho to be larger, and the more aggressive decision makers select the rho to be smaller;the item is the total amount of the foreign area electricity contract;

(2) constructing inequality constraint conditions and linearizing nonlinear constraint

In the proposed annual transaction optimization model, the CVaR has the following inequality constraint conditions, and the estimated value of the CVaR can be obtained through Monte Carlo simulation:

where R is the maximum value of CVaR and the subscript "k" indicates that the identifier is the value of the kth sample;

Introduction of an intermediate quantity z_kLinearize equation (7) as:

z_k＝[C_{annual_k}-α]⁺≥0 (9)；

z_kC_{annual_k}-α (11)；

wherein λ is_m，kIs the kth sample of the monthly market forecast average electricity prices for t months in the monte carlo simulation; the predicted price of the monthly market has prediction error which can be represented as normal distribution with the mean value of 0, and the predicted price has an expression lambda_m，k～(λ_m，σ_m ²)，λ_mThe value of (c) needs to be obtained by prediction;

the following constraints exist for the amount of power:

(3) constructing equality constraint condition and linearizing non-linear constraint

The optimization function for annual transactions has the following equality constraints:

wherein the content of the first and second substances,is the predicted load value for t months;is the kth set of samples in a load prediction monte carlo sample for t months; predicting the existence of errorsDifference, error can be represented by a normal distribution with a mean value of 0, thenExistence relationship The predicted value of the electric quantity of the ith clean energy in t months is shown,the ith group of samples in the monte carlo sampling for the electricity quantity prediction of the ith clean energy in t months have errors in prediction, and the errors can be represented by normal distribution with the mean value of 0, so that the method is used for predicting the electricity quantity of the ith group of samples in the monte carlo sampling

4. The method according to claim 1, wherein the step S2 is implemented by establishing a monthly transaction optimization model based on conditional risk values, introducing risk costs into an economically optimal monthly transaction optimization function by using risk analysis of uncertainty of load prediction and clean energy output according to a result of prediction, and performing linear transformation on nonlinear constraint conditions, and specifically comprises:

(1) Setting an objective function

Setting a monthly transaction objective function including electricity purchasing cost and risk cost, wherein the expression is as follows:

wherein λ is_iIs the ith clean energy in the preferential power generationThe power price of the source on the internet; lambda [ alpha ]_jThe price of electricity of an electric power plant outside the jth zone; lambda [ alpha ]_xBilateral agreement electricity prices for the xth power generator in the annual market; lambda [ alpha ]_bmA centralized bid price for the pre-month market for t months; lambda [ alpha ]_im，xListing prices for the x-th generator market within the month; lambda [ alpha ]_zThe online electricity price of the non-marketized power generation enterprise of the z type power generation type;the electric quantity of the ith clean energy, the electric quantity of a j-th out-of-district power plant, the bilateral agreement electric quantity of the x-th power generator in the annual market, the electric quantity of the x-th power generator in the monthly market and the electric quantity of the z-th power generation type non-marketized power generation enterprise are respectively represented in the period tau; the switching value is used for indicating whether the power plant outputs 0 in the period tau, and the value is 0 or 1; rho₂Rho chosen by the more conservative decision maker for the risk aversion factor₂Larger, more aggressive decision maker-chosen ρ₂The smaller;

(2) constructing inequality constraint conditions and linearizing nonlinear constraint

In the proposed pre-monthly transaction optimization model, the CVaR has the following inequality constraint conditions, and the estimated value of the CVaR can be obtained through Monte Carlo simulation:

Wherein R is₂Is CVaR₂The subscript "k" indicates that the identifier is the value of the kth sample;

introduction of an intermediate quantity z_kEquation (19) is linearized as:

z_k＝[C_{month_k}-α]⁺≥0 (21)；

z_k≥C_{month_k}-α (23)；

the following constraints exist for the amount of electricity in each time period:

there is also a hill climbing constraint:

(3) constructing equality constraint condition and linearizing non-linear constraint

The optimization function for a pre-month transaction has the following equality constraints:

in addition to the constraints, the transaction results must be checked for security to obtain a final pre-monthly transaction scheme.

5. The method according to claim 1, wherein the step S3 of performing rolling prediction for multiple times in a month, determining the monthly transaction power by introducing risk cost into an economically optimal monthly optimization function through risk analysis of uncertainty of load prediction and clean energy output, and performing linear transformation on nonlinear constraint conditions to realize optimization solution, specifically comprises:

(1) establishing an objective function

The objective function expression for transactions within the month is as follows:

different from the objective function of the pre-month transaction, the decomposition of the electric quantity curve of the out-of-area incoming call, the annual transaction and the pre-month transaction is completed, The values of the three terms have been determined;

(2) constructing inequality constraint conditions and linearizing nonlinear constraint

In the proposed intra-month trading model, the CVaR has the following inequality constraint conditions, and the estimated value of the CVaR can be obtained through Monte Carlo simulation:

wherein R is₃Is CVaR₃The subscript "k" indicates that the identifier is the value of the kth sample;

introduction of an intermediate quantity z_kEquation (37) is linearized as:

z_k＝[C_{month_k′}-α]⁺≥0 (39)；

the following constraints exist for the amount of electricity in each time period:

there is also a hill climbing constraint:

(3) constructing equality constraint condition and linearizing non-linear constraint

The optimization function for monthly transactions has the following equality constraints:

in addition to the constraints, the transaction results must be checked for security to obtain a final intra-month transaction scheme.

Background

The electric power trading center is used as an organizer of the current electric power medium-term and long-term market, and electric power purchasing quantity of different trading varieties needs to be reasonably arranged in order to ensure balance of electric power supply and demand and ensure connection of electric power trading of different time scales. The power grid company is an organizer of non-marketized electric quantity at the present stage, and making a reasonable combination of marketized electric quantity and non-marketized electric quantity is an important link for ensuring electric power and electric quantity balance.

At present, the electric power is still in the transition period of market reformation, so not all electric quantity is market electric quantity, the market electric quantity mainly takes coal power as main power, and other power generation types are almost non-market electric quantity except a small amount of long-term water power. In addition, as the electric power spot market is only in a part of provincial market test points, a plurality of provinces only exist in a medium-term and long-term market at present.

Chronologically, the power medium and long term transactions include annual transactions and monthly transactions. Before annual power generation, the priority power generation is determined, mainly the electric quantity of clean energy and the power coming from an outside area. The target object of annual transaction is the next year electricity quantity, mainly bilateral negotiation transaction. The monthly transaction is divided into pre-monthly and intra-monthly (multi-day) transactions, the subject matter of the pre-monthly transaction is the next-month electric quantity, and the pre-monthly transaction is mainly completed in a centralized bidding mode; the target object of the monthly transaction is the residual quantity of electricity in the month, and the monthly transaction is mainly completed in a listing transaction mode and is used as a supplement for the monthly transaction. After the introduction of the electric power spot market, the curve decomposition of the electric quantity is one of the keys of the medium-long term market and the spot market.

Disclosure of Invention

The invention aims to provide a method for arranging the electricity purchasing quantity of the medium-long term power market at different time scales aiming at the defects of the background technology, and the risk brought by the errors of load, clean energy prediction and electricity price prediction is considered through condition risk value analysis. And introducing the risk cost into an optimization function taking economic optimization as a target, and performing linear transformation on the nonlinear constraint condition to improve the solving efficiency so as to realize the overall arrangement of annual contract electric quantity, monthly centralized transaction electric quantity and monthly branding transaction electric quantity.

The invention provides a method for arranging the electric quantity purchased in different time scales in a medium-long term electric power market, which comprises the following steps:

s1, establishing a risk assessment calculation model aiming at the prediction error, measuring the risk through the conditional risk value CVaR, and determining the basic form of the CVaR;

s2, establishing an annual transaction optimization model and a monthly transaction optimization model based on conditional risk values, introducing risk cost into an optimal economic optimization function, and performing linear transformation on nonlinear constraint conditions to realize optimal solution;

and S3, performing rolling optimization in different time scale markets, and determining the power purchasing plan of the medium and long term markets of the power.

Further preferably, a risk assessment calculation model for the prediction error is established in step S1, the risk is measured by the conditional risk value CVaR, and the basic form of the CVaR is determined;

at a given confidence level β (β ∈ (0,1)), the value of the risk value VaR is:

f_VaR(x)＝min{α∈R；∫_f(x,y)≤αρ(y)dy≥β} (1)；

wherein: f (x, y) is a loss function of investment; x is formed by RⁿAn n-dimensional portfolio scenario vector; y is formed by R^mThe random vector is an m-dimensional random vector and represents a random factor in the medium and long-term market of the power; α is a threshold value of allowable loss, and ρ (y) is a probability density function of y;

The expression for conditional risk value CVaR is:

will f is_CVaR(x) Simplified and converted to the following forms, using F_β(x, α) to denote CVaR:

converting the above equation (3) into a discretization form, and substituting the discretization expression into sample data calculation:

[f(x,y_k)-α]⁺＝max[f(x,y_k)-α,0] (5)；

wherein the content of the first and second substances,as an estimate of CVaR, y_kThe sample data of the kth group has N groups.

Further preferably, in step S2, an annual transaction optimization model based on conditional risk values is established, risk analysis on load prediction and uncertainty of clean energy output is performed, risk cost is introduced into an economic optimal annual transaction optimization function, and nonlinear constraint conditions are subjected to linear transformation to realize optimal solution, which specifically includes:

(1) setting an objective function

Setting an annual transaction objective function including electricity purchasing cost and risk cost, wherein the expression is as follows:

wherein: lambda [ alpha ]_iThe price of the on-line electricity of the ith clean energy in the preferential power generation is set; lambda [ alpha ]_jThe price of electricity of an electric power plant outside the jth zone; lambda [ alpha ]_xBilateral agreement electricity prices for the xth power generator in the annual market; lambda [ alpha ]_m ^tPredicted average electricity prices for the monthly market for t months; lambda [ alpha ]_zThe online electricity price of the non-marketized power generation enterprise of the z type power generation type; respectively representing the electric quantity of the ith clean energy in t month, the electric quantity of a power plant outside the jth district, the bilateral agreement electric quantity of the x generator in the annual market, the electric quantity of the x generator in the monthly market and the electric quantity of the z-th non-marketized power generation enterprise of the power generation type; rho is a risk aversion coefficient, the more conservative decision makers select the rho to be larger, and the more aggressive decision makers select the rho to be smaller; The term is the total amount of the out-of-district electricity contract.

(2) Constructing inequality constraint conditions and linearizing nonlinear constraint

In the proposed annual transaction optimization model, the CVaR has the following inequality constraint conditions, and the estimated value of the CVaR can be obtained through Monte Carlo simulation:

where R is the maximum value of CVaR and the subscript "k" indicates that the identifier is the value of the kth sample;

introduction of an intermediate quantity z_kLinearizing the formula (7) into

z_k＝[C_{annual_k}-α]⁺≥0 (9)；

z_k≥C_{annual_k}-α (11)；

Wherein λ is_m,kIs the kth sample of the monthly market forecast average electricity prices for t months in the monte carlo simulation; the predicted price of the monthly market has prediction error which can be represented as normal distribution with the mean value of 0, and the predicted price has an expression lambda_m,k～(λ_m,σ_m ²)，λ_mThe value of (c) needs to be obtained by prediction;

the following constraints exist for the amount of power:

(3) constructing equality constraint condition and linearizing non-linear constraint

The optimization function has the following equality constraints

Wherein the content of the first and second substances,is the predicted load value for t months;is the kth set of samples in a load prediction monte carlo sample for t months; the prediction has an error, which can be represented by a normal distribution with a mean value of 0, thenExistence relationship The predicted value of the electric quantity of the ith clean energy in t months is shown,the ith group of samples in the monte carlo sampling for the electricity quantity prediction of the ith clean energy in t months have errors in prediction, and the errors can be represented by normal distribution with the mean value of 0, so that the method is used for predicting the electricity quantity of the ith group of samples in the monte carlo sampling

Further preferably, the step S2 is to establish a monthly transaction optimization model based on conditional risk values, introduce risk cost into an economically optimal monthly transaction optimization function by using risk analysis of uncertainty of load prediction and clean energy output according to a result of prediction again, and perform linear transformation on nonlinear constraint conditions to implement optimization solution, specifically including:

(1) setting an objective function

Setting a monthly transaction objective function including electricity purchasing cost and risk cost, wherein the expression is as follows:

wherein λ is_iThe price of the on-line electricity of the ith clean energy in the preferential power generation is set; lambda [ alpha ]_jThe price of electricity of an electric power plant outside the jth zone; lambda [ alpha ]_xBilateral agreement electricity prices for the xth power generator in the annual market; lambda [ alpha ]_bmA centralized bid price for the pre-month market for t months; lambda [ alpha ]_im,xListing prices for the x-th generator market within the month; lambda [ alpha ]_zThe online electricity price of the non-marketized power generation enterprise of the z type power generation type;the electric quantity of the ith clean energy, the electric quantity of a j-th out-of-district power plant, the bilateral agreement electric quantity of the x-th power generator in the annual market, the electric quantity of the x-th power generator in the monthly market and the electric quantity of the z-th power generation type non-marketized power generation enterprise are respectively represented in the period tau; The switching value is used for indicating whether the power plant outputs 0 in the period tau, and the value is 0 or 1; rho₂Rho chosen by the more conservative decision maker for the risk aversion factor₂Larger, more aggressive decision maker-chosen ρ₂The smaller.

(2) Constructing inequality constraint conditions and linearizing nonlinear constraint

In the proposed pre-monthly transaction model, the CVaR has the following inequality constraint conditions, and the estimated value of the CVaR can be obtained through Monte Carlo simulation:

wherein R is₂Is CVaR₂The subscript "k" indicates that the identifier is the value of the kth sample;

introduction of an intermediate quantity z_kEquation (19) is linearized as:

z_k＝[C_{month_k}-α]⁺≥0 (21)；

z_k≥C_{month_k}-α (23)；

the following constraints exist for the amount of electricity in each time period:

there is also a hill climbing constraint:

(3) constructing equality constraint condition and linearizing non-linear constraint

The optimization function has the following equality constraints:

in addition to the constraints, the transaction results must be checked for security to obtain a final pre-monthly transaction scheme.

Further preferably, in step S3, performing rolling prediction for multiple times in a month, introducing risk cost into an economically optimal monthly transaction optimization function by using risk analysis of load prediction and uncertainty of clean energy output, and performing linear transformation on a nonlinear constraint condition to implement optimization solution to determine monthly branding transaction electric quantity, specifically including:

(1) Establishing an objective function

The objective function is expressed as follows:

different from the objective function of the pre-month transaction, the decomposition of the electric quantity curve of the out-of-area incoming call, the annual transaction and the pre-month transaction is completed,the values of the three terms have been determined.

(2) Constructing inequality constraint conditions and linearizing nonlinear constraint

In the proposed intra-month trading model, the CVaR has the following inequality constraint conditions, and the estimated value of the CVaR can be obtained through Monte Carlo simulation:

wherein R is₃Is CVaR₃The subscript "k" indicates that the identifier is the value of the kth sample;

introduction of an intermediate quantity z_kLinearizing equation (37) to

z_k＝[C_{month_k}′-α]+≥0 (39)；

The following constraints exist for the amount of electricity in each time period:

there is also a hill climbing constraint:

(3) constructing equality constraint condition and linearizing non-linear constraint

The intra-month trade optimization function has the following equality constraints:

in addition to the constraints, the transaction results must be checked for security to obtain a final intra-month transaction scheme.

Has the advantages that:

the invention provides an electric power purchase arrangement method for medium and long-term electric power markets at different time scales, which is used for establishing a risk assessment calculation model aiming at load prediction, clean energy output prediction and electricity price prediction errors and measuring risks through conditional risk value (CVaR). And establishing an optimization model based on the condition risk value, introducing the risk cost into an optimal economic optimization function, and performing linear transformation on the nonlinear constraint condition to improve the solving efficiency and realize optimization solution. By arranging annual electric quantity and monthly electric quantity in a comprehensive mode and carrying out curve decomposition on the monthly electric quantity, an economic optimal scheme is finally obtained, and reference is provided for an organizer of an electric power market.

Drawings

Fig. 1 is a flow chart of a method for scheduling the electric quantity purchased in different time scales of a medium-long term electric power market.

Detailed Description

The technical scheme of the invention is explained in detail in the following with reference to the attached drawings.

The method for scheduling the electric quantity purchased in different time scales in the medium-and long-term electric power market disclosed by the application is shown in fig. 1 and comprises the following two steps.

The method comprises the following steps: and establishing a risk assessment calculation model aiming at the prediction error, measuring the risk through a conditional risk value (CVaR), and determining the basic form of the CVaR.

Let f (x, y) be a loss function of investment. Wherein x ∈ RⁿAn n-dimensional portfolio scenario vector; y is formed by R^mIs an m-dimensional random vector and represents a random factor in the power medium and long-term market. At a given confidence level β (β ∈ (0,1)), the value of the risk value VaR is

f_VaR(x)＝min{α∈R；∫_f(x,y)≤αρ(y)dy≥β} (1)；

Where α is the threshold for allowable loss and ρ (y) is the probability density function of y.

The conditional risk value CVaR is expressed as

In general, f can be_CVaR(x) Simplified and converted to the following forms, using F_β(x, α) to denote CVaR:

in the situation described in the present invention, it is usually difficult to obtain the probability density function of y accurately, and solving the continuous non-linear problem is complicated and takes a long time. Therefore, the above equation needs to be converted into a discretization expression and then substituted into the sample data calculation

[f(x,y_k)-α]⁺＝max[f(x,y_k)-α,0] (5)；

In the above formula, the first and second carbon atoms are,as an estimate of CVaR, y_kThe sample data of the kth group has N groups.

Step two: and establishing an annual transaction optimization model based on the condition risk value, introducing the risk cost into an economic optimal optimization function by utilizing risk analysis of load prediction and uncertainty of clean energy output, and realizing optimization solution by carrying out linear transformation on nonlinear constraint conditions.

1) Setting an objective function

Before annual transaction is executed, firstly, determining the priority power generation amount, wherein the clean energy power amount is required to be completely consumed and is non-marketized power amount due to the existence of a renewable energy consumption guarantee mechanism; out-of-range calls are also one of the boundary conditions for market-oriented transactions in provinces. In the annual market transaction execution process, the annual market electricity price is the electricity price of bilateral transaction and is a constant. The marginal electricity price of the centralized bidding trading of the market before the month is changed, and the electricity quantity and the electricity price of the market need to be predicted when the market trading decision of the year is made. The annual contract is an electric quantity contract, and the contract is divided into months. The rest of the load demand capacity is performed by the non-marketable capacity.

Setting an annual transaction objective function including electricity purchasing cost and risk cost, wherein the expression is as follows

In the above formula, λ_iIs the grid price of the ith (total I) clean energy (such as photovoltaic, wind power and the like) in the prior power generation, lambda_jFor the electricity prices, λ, of electric power plants outside the jth (total J) zone_xBilateral agreement electricity price, lambda, for the X (X total) th generator in the annual market_m ^tPredicted average electricity price, λ, for a monthly market of t months_zThe online price of the non-marketable power generation enterprise of the z type of power generation.The electric quantity of the ith clean energy in t month, the electric quantity of a power plant outside the jth district, the bilateral agreement electric quantity of the x-th power generator in the annual market, the electric quantity of the x-th power generator in the monthly market and the electric quantity of the z-th power generation type non-marketized power generation enterprise are respectively represented. ρ is a risk aversion coefficient, and the more conservative decision makers select the ρ to be larger, and the more aggressive decision makers select the ρ to be smaller. Wherein the content of the first and second substances,the item is the total amount of the foreign area electricity contract which is used as the boundary of the electricity purchasing arrangement and is a fixed value.

2) Constructing inequality constraint conditions and linearizing nonlinear constraint

In the proposed annual transaction optimization model, CVaR has the following inequality constraints. An estimate of CVaR can be obtained by monte carlo simulations. The subscript "k" indicates that the identifier is the value of the kth sample.

In the above formula, R is the maximum value of CVaR. Wherein

To linearize equation (7), an intermediate quantity z is introduced_k. Equation (7) can be linearized

z_k＝[C_{annual_k}-α]⁺≥0 (9)；

z_k≥C_{annual_k}-α (11)；

In the above formula, λ_m,kIs the kth sample of the monthly market forecast average electricity prices for t months in the monte carlo simulation. The predicted price of the monthly market has prediction error which can be represented as normal distribution with the mean value of 0, and the predicted price has an expression lambda_m,k～(λ_m,σ_m ²)。λ_mThe value of (c) needs to be obtained by prediction.

The following constraints exist for the amount of power:

3) constructing equality constraint condition and linearizing non-linear constraint

The annual transaction optimization function of the present invention has the following equality constraints

Wherein the content of the first and second substances,is a predicted value of the load for the t month,the k group of samples in the Monte Carlo sampling of load prediction of t months exist, the prediction has errors, and the errors can be represented by normal distribution with the mean value of 0, so thatExistence relationship The predicted value of the electric quantity of the ith clean energy in t months is shown,the ith group of samples in the monte carlo sampling for the electricity quantity prediction of the ith clean energy in t months have errors in prediction, and the errors can be represented by normal distribution with the mean value of 0, so that the method is used for predicting the electricity quantity of the ith group of samples in the monte carlo sampling

Step three: and establishing a monthly transaction optimization model based on the condition risk value, introducing the risk cost into an optimal economic optimization function by utilizing the risk analysis of load prediction and uncertainty of clean energy output according to the result of secondary prediction, and realizing optimization solution by carrying out linear transformation on the nonlinear constraint condition.

1) Setting an objective function

Before the transaction before the month is executed, the load and the clean energy output of the month are predicted again. The transaction electricity price of the concentrated bidding transaction of the market before the month is the marginal electricity price. When monthly electric quantity is arranged, the electric quantity needs to be subjected to curve decomposition.

Setting a pre-month transaction objective function including electricity purchasing cost and risk cost, wherein the expression is as follows

In the above formula, λ_iIs the grid price of the ith (total I) clean energy (such as photovoltaic, wind power and the like) in the prior power generation, lambda_jFor the electricity prices, λ, of electric power plants outside the jth (total J) zone_xBilateral agreement electricity price, lambda, for the X (X total) th generator in the annual market_bmFor a concentrated bid electricity price, λ, for a pre-month market of t months_im,xThe listing price for the x-th generator market within the month. Lambda [ alpha ]_zThe online price of the non-marketable power generation enterprise of the z type of power generation.The electric quantity of the ith clean energy, the electric quantity of the jth out-of-district power plant, the bilateral agreement electric quantity of the xth power generator in the annual market, the electric quantity of the xth power generator in the monthly market and the electric quantity of the z power generation type non-marketized power generation enterprises which respectively represent the tau time period (total T time periods). The switching value is 0 or 1, and represents whether the power plant is outputting 0 in the period tau. Rho₂Rho chosen by the more conservative decision maker for the risk aversion factor₂Larger, more aggressive decision maker-chosen ρ₂The smaller.

2) Constructing inequality constraint conditions and linearizing nonlinear constraint

In the proposed pre-month trading model, CVaR has the following inequality constraints. An estimate of CVaR can be obtained by monte carlo simulations. The subscript "k" indicates that the identifier is the value of the kth sample.

In the above formula, R₂Is CVaR₂Is measured. Wherein

To linearize equation (7), an intermediate quantity z is introduced_k. Equation (19) can be linearized

z_k＝[C_{month_k}-α]⁺≥0 (21)；

z_k≥C_{month_k}-α (23)；

The following constraints exist for the amount of electricity in each time period:

there is also a hill climbing constraint:

3) constructing equality constraint condition and linearizing non-linear constraint

The pre-monthly transaction optimization function of the present invention has the following equality constraints:

in addition to the constraints, the transaction results must be checked for security to obtain a final pre-monthly transaction scheme.

Step four: and rolling and predicting for multiple times in the month, introducing risk cost into an economic optimal optimization function by utilizing risk analysis of uncertainty of load prediction and clean energy output, and realizing optimization solution by carrying out linear transformation on nonlinear constraint conditions to determine the monthly branding transaction electric quantity.

1) Setting an objective function

Before the transaction in each month (multiple days) is executed, the load and the clean energy output of the month are predicted again. The transaction electricity price of the concentrated bidding transaction of the market before the month is the marginal electricity price. When monthly electric quantity is arranged, the electric quantity needs to be subjected to curve decomposition.

Different from the objective function of the pre-month transaction, the decomposition of the electric quantity curve of the out-of-area incoming call, the annual transaction and the pre-month transaction is completed,the values of the three terms have been determined.

2) Constructing inequality constraint conditions and linearizing nonlinear constraint

In the proposed intra-month trading optimization model, CVaR has the following inequality constraints. An estimate of CVaR can be obtained by monte carlo simulations. The subscript "k" indicates that the identifier is the value of the kth sample.

In the above formula, R₃Is CVaR₃Is measured. Wherein

To linearize equation (7), an intermediate quantity z is introduced_k. Equation (19) can be linearized

z_k＝[C_{month_k}′-α]⁺≥0 (39)；

The following constraints exist for the amount of electricity in each time period:

there is also a hill climbing constraint:

3) constructing equality constraint condition and linearizing non-linear constraint

The intra-month transaction optimization function of the present invention has the following equality constraints:

in addition to the constraints, the transaction results must be checked for security to obtain a final intra-month transaction scheme.

In summary, the invention discloses a method for arranging the electricity purchasing quantity of the medium-long term electric power market at different time scales, belonging to the technical field of calculation, calculation or counting. The optimization solution is realized by risk analysis of load prediction uncertainty, introduction of risk cost into an economic optimal optimization function and linear transformation of nonlinear constraint conditions, and the algorithm can be used for overall arrangement of the electricity purchasing quantity of different time scales and different transaction varieties and improves the solution efficiency. The invention realizes the trading strategy of overall arrangement of annual contract electric quantity, monthly centralized trading electric quantity and monthly branding trading electric quantity, and provides reference for an organizer of an electric power market.