WO2023023901A1

WO2023023901A1 - Method for predicting medium- and long-term centralized bidding clearing price in electricity market

Info

Publication number: WO2023023901A1
Application number: PCT/CN2021/114123
Authority: WO
Inventors: 韦仲康; 朱天博; 邢劲; 牛家强; 张斓曦
Original assignee: 冀北电力交易中心有限公司
Priority date: 2021-08-23
Filing date: 2021-08-23
Publication date: 2023-03-02

Abstract

Provided in the present invention is a method for predicting a medium- and long-term centralized bidding clearing price in an electricity market. By means of predicting a clearing price of a medium- and long-term transaction in an electricity market, the actual value of electricity in production and circulation processes is accurately reflected, and a relationship between supply and demand of the electricity is adjusted, such that the economic benefits of electric power enterprises are improved. In addition, the price of electricity is determined by the market, which can facilitate mutual competition among the electric power enterprises, and enhance the vitality of the enterprises; and the price of electricity being determined by the market can also encourage electricity consumers to actively participate in automatic peak shaving and valley filling, so as to make an electricity consumption structure more rational, thereby reducing waste.

Description

A prediction method for mid- and long-term centralized bidding clearing prices in the electricity market

technical field

The invention belongs to the technical field of computer data analysis and adjustment applied to the electric power market, and relates to a method for predicting mid- and long-term centralized bidding clearing prices in the electric power market.

Background technique

In the electricity market environment, electricity price is not only a signal of supply and demand in the electricity market, but also an economic lever to control transactions in the electricity market. It is of great significance for all participants in the market to accurately predict the future market electricity price based on the relevant historical data of the electricity market.

The medium and long-term electricity price forecast can provide a good reference for the formulation of production plans of power generation enterprises and the long-term investment of power investors, and provide an objective basis for the regulatory authorities to formulate and implement effective regulatory measures. Power supply balance. The medium- and long-term centralized bidding price is a unified price that reflects the supply and demand relationship of electricity commodities in the electricity market, and provides a scientific basis for promoting healthy, stable, and orderly competition and development of the market.

In addition to the characteristics of multi-periodicity, high volatility, and multiple price peaks implied by short-term electricity prices, medium and long-term electricity prices are also affected by many uncertain factors, resulting in unclear changes in electricity prices. However, forecasting the medium and long-term electricity price is an indispensable and important aspect in the construction of electricity market.

Contents of the invention

Based on the above-mentioned defects and demands of the prior art, the present invention proposes a method for predicting the clearing price of mid- and long-term centralized bidding in the electric power market. By predicting the clearing price of mid- and long-term transactions in the electric power market, it can correctly reflect the power in the production and circulation The actual value in the process can adjust the supply and demand relationship of electricity and improve the economic benefits of electric power enterprises. In addition, the electricity price determined by the market can promote the competition among electric enterprises and enhance the vitality of the enterprise; the electricity price determined by the market can also promote electricity consumers Actively participate in automatic peak shaving and valley filling to make the power consumption structure more reasonable and reduce waste.

The present invention proposes a prediction method for mid- and long-term centralized bidding clearing prices in the electricity market, which includes the following steps:

Step 1: Collect the influencing factors associated with the clearing price, establish a database of influencing factors, and store the collected influencing factors in the database;

Step 2: Perform data preprocessing on the influencing factors in the database, and select the most relevant influencing factors;

Step 3: According to the selected influencing factors with the greatest correlation as the input data, use the ARIMA forecasting method in the time series forecasting method, use the historical time series electricity price related data and the historical cleared electricity price data for statistical analysis, and predict the future Clear the price of electricity.

In order to better realize the present invention, further, the operation of the step 3 is:

First, the stationarity test and processing of the input data are carried out;

Then, use the autocorrelation coefficient and partial autocorrelation coefficient of the sample to identify the ARIMA model; after the ARIMA model identification is completed, determine each parameter in the ARIMA model;

Finally, after the ARIMA model is determined, the ARIMA model is tested, so that the ARIMA model passes the parameter test and the residual white noise test, and after the model prediction accuracy is improved, the data is predicted.

In order to better realize the present invention, further, the specific operation of the step 3 is:

Step 3.1: Perform stationarity inspection and processing. First, observe the stationarity of the input time series data by drawing or by the autocorrelation diagram and partial autocorrelation diagram of the input data, and when analyzing and processing non-stationary data, use the difference method , that is, by calculating the numerical difference between time series t and time t-1, a new and more stable time series is obtained;

Step 3.2: Perform ARIMA model identification, combined with time series, using the autocorrelation coefficient ρ _k and partial autocorrelation coefficient of the input data

Identify the corresponding AR(p) model, MR(q) model and ARMR(p,q) model in the ARIMA model for identification; where p is the order of the autoregressive item, and q is the order of the moving average item;

The specific identification relationship is shown in the following table:

Step 3.3: Estimate the parameters of the ARIMA model, determine the value of the autoregressive item order p and the sequence difference order q by observing the characteristics of the autocorrelation diagram and partial autocorrelation diagram of the sequence, and select the optimal The best value; the value of the sequence difference order d is determined by the difference order of the time series;

The information amount criterion includes the Akaike information criterion AIC that can judge the complexity and degree of fitting of the ARIMA model, and the specific calculation formula is:

AIC=2k-2ln(L)

Among them, k is the number of model parameters, and L is the likelihood function;

Step 3.4: Verify the ARIMA model, which specifically includes two parts of verification: one part is to decompress the significance of the parameter estimation of the ARIMA model in step 3.3; the other part is to verify the randomness of the residual sequence , that is, to judge whether the built ARIMA model is advisable by performing residual white noise test and parametric test on the ARIMA model;

Step 3.5: Use the ARIMA model to fit the cleared electricity price with the input data after the data screening, and obtain the predicted data of the cleared electricity price.

In order to better realize the present invention, further, the specific operation of the step 2 is:

Select the influencing factors related to the electricity price in the database, and use the stepwise logistic regression algorithm to conduct correlation analysis on the selected influencing factors. By selecting the correlation between each influencing factor and the electricity price, select the influencing factor with the best correlation as the input variable in step 3.

In order to better realize the present invention, further, said step 2 includes the following steps:,

Step 2.1: Build a stepwise logistic regression algorithm model:

The stepwise logistic regression algorithm model does not directly model the value of the dependent variable Y, but calculates the probability of which type of result the dependent variable Y belongs to by establishing a linear regression function, and judges whether the dependent variable Y belongs to according to the probability. Which type of regression; the conditional probability distribution of the stepwise logistic regression model is:

Among them, x is an independent variable, that is, different influencing factors. The set of influencing factors x is expressed as x=(x ₁ ,x ₂ ,x ₃ ,...,x _n ,1) ^T , and ω is called different influencing factors The weight vector corresponding to x, the weight vector ω is expressed as ω=(ω ₁ ,ω ₂ ,ω ₃ ,...,ω _n ,b) ^T , b is called the bias, and Y is the value of 0 or 1 The dependent variable, P represents the probability of different results of Y; through transformation, the above function can be written as:

Convert the probability P in the above into a linear regression model about X, and get the following formula:

Step 2.2: Significant test of the regression coefficient The regression coefficient is the parameter that represents the influence of the independent variable on the dependent variable in the regression equation. Use the model regression coefficient β _i and the standard error Sβ _i of the regression coefficient to perform a significant test of the regression coefficient. The specific mathematical formula defined as:

When the value of the significant test is less than 0.05, it is considered that the corresponding independent variable and

There is a significant linear relationship; otherwise, it is considered that the corresponding independent variable and

There is no significant linear relationship, and independent variables without significant linear relationship cannot enter the stepwise logistic regression algorithm model;

Step 2.3: Screen the independent variable. After completing the significant test statistics of the regression coefficient, select the independent variable whose significance is less than 0.05 to enter the stepwise logistic regression algorithm model; The weight of some independent variables is investigated. If the weight of an independent variable is lower than before, it is considered that its contribution in the model is reduced, and it will be eliminated; the independent variable left by the final screening will be used as input in step 3. input data.

In order to better realize the present invention, further, the influencing factors include four aspects: generation side, power consumption side, grid side and other aspects.

In order to better realize the present invention, further, the influence factors included in the power generation side include power supply structure, fuel price, unit category, power generation enterprise HHI, grid electricity price, environmental protection and low-carbon degree, maximum available capacity of the system, and power generation enterprise year-on-year growth rate.

In order to better realize the present invention, further, the influence factors included in the aspect of the power consumption side include user power load demand, year-on-year growth rate of power users and power sales company agent users, power purchaser HHI index, and catalog electricity price.

In order to better realize the present invention, further, the influence factors included in the power grid side include power grid flow information, historical clearing prices, system maximum price limits, and system minimum price limits.

In order to better realize the present invention, further, the other aspects include weather factors, weekends and holidays, retention ratios, and deviation assessment factors.

Compared with the prior art, the present invention has the following advantages and beneficial effects:

The present invention correctly reflects the actual value of electric power in the process of production and circulation by predicting the clearing price of electricity in the medium and long-term transactions in the electric power market, adjusts the relationship between supply and demand of electric power, and improves the economic benefits of electric power enterprises. In addition, the electricity price is determined by the market, which can Promote competition among power companies and enhance business vitality; determining electricity prices by the market can also encourage power consumers to actively participate in automatic peak-shaving and valley-filling, making the power consumption structure more reasonable and reducing waste.

Description of drawings

Fig. 1 is a schematic structural diagram of the complete idea of the present invention.

Detailed ways

In order to illustrate the technical solutions of the embodiments of the present invention more clearly, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. It should be understood that the described embodiments are only Some, but not all, embodiments of the present invention should not be considered as limiting the scope of protection. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "arrangement", "connection" and "connection" should be understood in a broad sense, for example, it can be a fixed connection or a detachable connection. Connected, or integrally connected; it can be mechanically connected, or electrically connected; it can also be directly connected, or indirectly connected through an intermediary, and it can be the internal communication of two components. Those of ordinary skill in the art can understand the specific meanings of the above terms in the present invention in specific situations.

Example 1:

This embodiment proposes a method for predicting the price of mid- and long-term centralized competitive bidding in the electricity market, including the following steps:

Example 2:

This embodiment is based on the above-mentioned embodiment 1, and the purpose of the present invention is to build a mid- and long-term centralized bidding clearing electricity price prediction system in the electricity market. Through the prediction of the clearing price of electricity in the medium and long-term transactions in the electricity market, it can correctly reflect the actual value of electricity in the process of production and circulation, adjust the relationship between supply and demand of electricity, and improve the economic benefits of electricity companies. The competition between enterprises enhances the vitality of enterprises; the determination of electricity prices by the market can also encourage electricity consumers to actively participate in automatic peak-shaving and valley-filling, so that the structure of electricity consumption tends to be more reasonable and waste is reduced.

In order to achieve the above object, the present invention is achieved through the following technical solutions:

Step 1: Database establishment: According to the generation side, power consumption side, grid side, etc., the influencing factors related to the clearing price are screened out, and the input database is established to prepare the data for the later medium and long-term clearing price forecast.

Step: 2: Data processing: use the logistic regression algorithm to screen the input data.

Step: 3: Data forecasting: Using the ARIMA forecasting method in the time series forecasting method, using the past time series electricity price-related data and clearing electricity price data for statistical analysis, inferring the future clearing electricity price.

Step S100 is further explored, the factors affecting the price are screened out respectively, a database is established, and the factors affecting the system clearing electricity price are entered in the system by means of configuration.

(1) Power generation side

Power supply structure: Aiming at the substituting role of the unit, when there is a problem with the thermal power unit, if there is no substituting unit, the price of electricity will soar.

Fuel price: the energy required for power generation, such as coal, natural gas, etc. If the fuel price rises, it will lead to an increase in the power generation cost of the power generation company, an increase in the on-grid electricity price, and an increase in the final market clearing price; if the fuel price decreases, it will cause a decrease in the power generation cost of each power generation company, and the reduction in the on-grid electricity price will eventually reduce the market clearing price .

Unit category: A, B, and C three-category units. A-type units refer to generating units that have not yet obtained the qualification for direct transactions with the user side, and only have annual planned power generation; B-type units refer to generators that have obtained the qualification for direct transactions with the user side. Units can have both annual planned power generation and transactional power; C-type generators do not have annual planned power generation, but only transactional power. There is an upper limit for the annual direct trading power of Class B units, and there is no upper limit for Class C units.

Power generation company HHI: It is used to measure changes in the market share of power generation companies. The larger the HHI index, the higher the degree of monopoly, and the greater the possibility of abuse of monopoly power in the market, resulting in higher clearing electricity prices.

On-grid electricity price: The clearing price will increase with the increase of the on-grid electricity price and decrease with the decrease of the on-grid electricity price.

Environmental protection and low-carbon level: When adopting high-low matching clearing or marginal clearing, when the electricity sellers declare the same price, they will first be sorted according to energy-saving and low-carbon power scheduling, and if they are in the same order, they will be allocated according to the proportion of declared electricity.

The maximum available capacity of the system: generally equal to the installed capacity of the system minus the maintenance capacity, but it is not ruled out that some power generators declare the available units as unavailable in order to raise the electricity price. The system supply and demand index can be formed with the electricity load.

Year-on-year growth rate of power generation companies: The growth of power generation companies has improved the competitiveness of power generation companies in the market. With the increase of market competitiveness, the clearing electricity price will change accordingly.

(2) Power consumption side

User demand for electricity load: For a given total installed capacity of power generation, the more high-cost units are called, the higher the market clearing price will be when the load demand increases; when the demand for electricity load decreases, the competition on the power generation side The fuller it will be, the lower the market clearing price will be.

The year-on-year growth rate of power users and agency users of electricity sales companies: The increase in the number of power users and agency retail users of power sales companies will increase the demand for electricity loads, and subsequently cause changes in market clearing prices.

Power purchaser HHI index: It is used to measure changes in the market share of power purchasers. The larger the HHI index, the higher the degree of monopoly, and the greater the possibility of abuse of monopoly power in the market, resulting in lower electricity prices.

Catalog electricity price: It will affect the willingness of electricity purchasers to participate in the electricity market.

(3) Grid side

Grid flow information

Historically cleared electricity prices: there is a strong correlation between electricity prices before and after periods

System maximum and minimum price: It will affect the quotations of the power generation side and power users.

(4) Others

Weather factors: will affect load changes

Weekends and holidays: changes that will affect load

retention ratio

Deviation assessment factors

Policy reasons

Further, the step S200 performs data processing, artificially selects the influencing factors related to the electricity price in the database, uses the stepwise logistic regression algorithm to conduct correlation analysis on the artificially screened factors, and selects the relevant factors by selecting the correlation between each factor and the electricity price. The best factors are used as input variables.

(1) Principle of stepwise logistic regression algorithm

Logistic regression does not directly model the value of the dependent variable Y, but calculates the probability of which type of result the dependent variable Y belongs to by establishing a linear regression function, and judges which type of regression the dependent variable Y belongs to according to the probability . The conditional probability distribution for the logistic regression model is:

Where, x is an independent variable, x=(x ₁ ,x ₂ ,x ₃ ,...,x _n ,1) ^T , ω=(ω ₁ ,ω ₂ ,ω ₃ ,...,ω _n ,b ) ^T , ω is called the weight vector, b is called the bias, Y is the dependent variable with a value of 0 or 1, and P indicates the probability of different results of Y. The above function can be written by transformation as:

(2) Significant test of the regression coefficient: the regression coefficient is the parameter that represents the influence of the independent variable on the dependent variable in the regression equation. Using the model regression coefficient β _i and the standard error Sβ _i of the regression coefficient to perform a significant test of the regression coefficient, its mathematics defined as:

When the value is less than 0.05, the independent variable is considered to be consistent with

There is a significant linear relationship, otherwise, the independent variable is considered to be related to

There is no significant linear relationship, the independent variable cannot enter the model

(3) Independent variable screening: After completing the significant test statistics of the regression coefficient, select the independent variable whose significance is less than 0.05 to enter the model. When adding a variable, the weight of the existing independent variable in the model will be inspected. If the weight of an independent variable is lower than before, it is considered that its contribution in the model is reduced, and the variable will be eliminated.

After the above steps, the independent variables related to the dependent variable are selected for prediction.

Further, in the step S300 of electricity price forecasting, after the independent variables related to the dependent variable are selected through the above method, the electricity price forecasting work is performed on the related independent variables through the ARIMA model. First, test and process the stationarity of the data, and then use the autocorrelation coefficient and partial autocorrelation coefficient of the sample to identify the model. After the model identification is completed, determine the three parameters in the model. After the model is determined, the model is tested. Make the model pass the parameter test and residual white noise test to improve the prediction accuracy of the model and accurately predict the data.

(1) Stationarity inspection and processing

First, observe the stationarity of the input time series data by drawing or through the autocorrelation diagram and partial autocorrelation diagram of the data. Since most random processes are not stationary, when studying non-stationary data, the difference method is first used, that is, by calculating the numerical difference between time series t and t-1, a new, more stable time series.

(2) Model identification

The ARIMA model is a time series forecasting method, which includes three parts: AR, I, and MA. AR means autoregressive model; I means integer order; MA means moving average model. In the identification of the model, it is necessary to use the autocorrelation coefficient ρ _k and the partial autocorrelation coefficient of the sample

Identify AR(p), MR(q), and ARMR(p,q).

The formula for calculating the autocorrelation coefficient is:

Among them, Y _t represents the time series value at time t,

represents the mean of the time series.

The formula for calculating the partial autocorrelation coefficient is:

Among them, ρ _k represents the autocorrelation coefficient of the time series.

According to the above calculation formula, the autocorrelation coefficient and partial autocorrelation coefficient of the sample data can be calculated, and then pattern recognition can be performed according to the following table 1.

Table 1

According to the autocorrelation coefficient and the schematic diagram drawn by the autocorrelation coefficient, judge which model belongs to the above model.

(3) Parameter estimation

In the ARIMA model, it is necessary to determine the order p of the autoregressive term, the order d of the sequence difference, and the order q of the moving average term. The values of p and q are determined by observing the characteristics of the autocorrelation diagram and partial autocorrelation diagram of the sequence, and the optimal value is selected through the information amount criterion, while the d value is determined by the difference order of the time series.

The commonly used information criterion includes the Akaike information criterion AIC, which can judge the complexity and fitting degree of the model, and its calculation formula is expressed as:

AIC=2k-2ln(L)

Among them, k is the number of model parameters, and L is the likelihood function.

(4) Model checking

The model test mainly includes two parts, one is to decompress the significance of parameter estimates, and the other is to test the randomness of the residual sequence, that is, to judge the value of the model by performing the residual white noise test and parametric test on the model. Whether it is advisable to build a model.

(5) Result output

In the ARIMA model constructed through the above process, the independent variables after data screening are put into the model to fit the electricity price, and the electricity price forecast data is obtained.

Other parts of this embodiment are the same as those of Embodiment 1 above, so details are not repeated here.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention in any form. Any simple modifications and equivalent changes made to the above embodiments according to the technical essence of the present invention all fall within the scope of the present invention. within the scope of protection.

Claims

A method for predicting the price of mid- and long-term centralized competitive bidding in the electricity market, characterized in that it includes the following steps:

Step 1: Collect the influencing factors associated with the clearing price, establish a database of influencing factors, and store the collected influencing factors in the database;

Step 2: Perform data preprocessing on the influencing factors in the database, and select the most relevant influencing factors;

Step 3: According to the selected influencing factors with the greatest correlation as the input data, use the ARIMA forecasting method in the time series forecasting method, use the historical time series electricity price related data and the historical cleared electricity price data for statistical analysis, and predict the future Clear the price of electricity.
A method for predicting the price of mid- and long-term centralized competitive bidding in the electricity market as claimed in claim 1, wherein the operation of the step 3 is:

First, the stationarity test and processing of the input data are carried out;

Then, use the autocorrelation coefficient and partial autocorrelation coefficient of the sample to identify the ARIMA model; after the ARIMA model identification is completed, determine each parameter in the ARIMA model;

Finally, after the ARIMA model is determined, the ARIMA model is tested, so that the ARIMA model passes the parameter test and the residual white noise test, and after the model prediction accuracy is improved, the data is predicted.
A method for forecasting the mid- and long-term centralized bidding clearing price in the electricity market according to claim 1 or 2, characterized in that, the specific operation of the step 3 is:

Step 3.1: Perform stationarity inspection and processing. First, observe the stationarity of the input time series data by drawing or by the autocorrelation diagram and partial autocorrelation diagram of the input data, and when analyzing and processing non-stationary data, use the difference method , that is, by calculating the numerical difference between time series t and time t-1, a new and more stable time series is obtained;

Step 3.2: Perform ARIMA model identification, combined with time series, using the autocorrelation coefficient ρ k and partial autocorrelation coefficient of the input data
Identify the corresponding AR(p) model, MR(q) model and ARMR(p,q) model in the ARIMA model for identification; where p is the order of the autoregressive item, and q is the order of the moving average item;

The specific identification relationship is shown in the following table:

Model AR(p) MA(q) ARMA(p,q) autocorrelation coefficient smear Censored after q order smear partial autocorrelation coefficient Censored after p order smear smear

Step 3.3: Estimate the parameters of the ARIMA model, determine the value of the autoregressive item order p and the sequence difference order q by observing the characteristics of the autocorrelation diagram and partial autocorrelation diagram of the sequence, and select the optimal The best value; the value of the sequence difference order d is determined by the difference order of the time series;

The information amount criterion includes the Akaike information criterion AIC that can judge the complexity and degree of fitting of the ARIMA model, and the specific calculation formula is:

AIC=2k-2ln(L)

Among them, k is the number of model parameters, and L is the likelihood function;

Step 3.4: Verify the ARIMA model, which specifically includes two parts of verification: one part is to decompress the significance of the parameter estimation of the ARIMA model in step 3.3; the other part is to verify the randomness of the residual sequence , that is, to judge whether the built ARIMA model is advisable by performing residual white noise test and parametric test on the ARIMA model;

Step 3.5: Use the ARIMA model to fit the cleared electricity price with the input data after the data screening, and obtain the predicted data of the cleared electricity price.
The method for predicting the price of mid- and long-term centralized competitive bidding in the electricity market as claimed in claim 1, wherein the specific operation of the step 2 is as follows:

Select the influencing factors related to the electricity price in the database, and use the stepwise logistic regression algorithm to conduct correlation analysis on the selected influencing factors. By selecting the correlation between each influencing factor and the electricity price, select the influencing factor with the best correlation as the input variable in step 3.
A method for predicting the price of mid- and long-term centralized competitive bidding in the electricity market as claimed in claim 4, wherein said step 2 comprises the following steps:,

Step 2.1: Build a stepwise logistic regression algorithm model:

The stepwise logistic regression algorithm model does not directly model the value of the dependent variable Y, but calculates the probability of which type of result the dependent variable Y belongs to by establishing a linear regression function, and judges whether the dependent variable Y belongs to according to the probability. Which type of regression; the conditional probability distribution of the stepwise logistic regression model is:

Among them, x is an independent variable, that is, different influencing factors. The set of influencing factors x is expressed as x=(x 1 ,x 2 ,x 3 ,...,x n ,1) T , and ω is called different influencing factors The weight vector corresponding to x, the weight vector ω is expressed as ω=(ω 1 ,ω 2 ,ω 3 ,...,ω n ,b) T , b is called the bias, and Y is the value of 0 or 1 The dependent variable, P represents the probability of different results of Y; through transformation, the above function can be written as:

Convert the probability P in the above into a linear regression model about X, and get the following formula:

Step 2.2: Significant test of the regression coefficient The regression coefficient is the parameter that represents the influence of the independent variable on the dependent variable in the regression equation. Use the model regression coefficient β i and the standard error Sβ i of the regression coefficient to perform a significant test of the regression coefficient. The specific mathematical formula defined as:

When the value of the significant test is less than 0.05, it is considered that the corresponding independent variable and
There is a significant linear relationship; otherwise, it is considered that the corresponding independent variable and
There is no significant linear relationship, and independent variables without significant linear relationship cannot enter the stepwise logistic regression algorithm model;

Step 2.3: Screen the independent variable. After completing the significant test statistics of the regression coefficient, select the independent variable whose significance is less than 0.05 to enter the stepwise logistic regression algorithm model; The weight of some independent variables is investigated. If the weight of an independent variable is lower than before, it is considered that its contribution in the model is reduced, and it will be eliminated; the independent variable left by the final screening will be used as input in step 3. input data.
The method for predicting the price of mid- and long-term centralized competitive bidding in the electricity market according to claim 1, wherein the influencing factors include four aspects: power generation side, power consumption side, power grid side and other aspects .
A method for predicting mid- and long-term centralized competitive bidding prices in the electricity market as claimed in claim 6, wherein the influencing factors on the power generation side include power supply structure, fuel price, unit category, power generation company HHI, and grid connection. Electricity price, environmental protection and low-carbon level, maximum available capacity of the system, year-on-year growth rate of power generation companies.
The method for predicting the price of mid- and long-term centralized bidding in the electricity market as claimed in claim 6, wherein the influence factors included in the electricity consumption side include user electricity load demand, electricity users and agents of electricity sales companies Year-on-year growth rate of users, HHI index of electricity purchasers, and catalog electricity prices.
A method for predicting mid- and long-term centralized bidding clearing prices in the electricity market according to claim 6, wherein the influence factors included in the power grid side include power grid flow information, historical clearing prices, system ceiling prices, System minimum price.
The method for predicting mid- and long-term centralized competitive bidding prices in the electricity market according to claim 6, wherein said other factors include weather factors, weekends and holidays, retention ratios, and deviation assessment factors.