Day-ahead market clearing price interval prediction method considering source load uncertainty

Abstract

The invention belongs to the related technical field of power system calculation, and discloses a day-ahead market clearing price interval prediction method considering source load uncertainty, which comprises the following steps of: taking the source load uncertainty as a main variable; multi-dimensional parameter random distribution characteristics of power generation end unit output prediction and uncertainty of load end demand response effect are quantified, a source load output uncertainty interval is discretized, a discrete point is used as a constraint variable, and the minimum system power generation total cost is used as a target. Establishing a dynamic influence change model considering source load uncertainty and a day-ahead market clearing price interval, traversing each source load discrete output scene and substituting into the model, cyclically calculating node marginal electricity prices under different scenes, and counting and predicting a clearing electricity price interval as a prediction result. According to the method, the clearing price range under the influence of the source load uncertainty can be accurately calculated, the unit quotation income and the power system safety are improved, the quotation risk is reduced, and a reference is provided for a bidding strategy of a power generation manufacturer.

Description

Technical Field

[0001] The present invention belongs to the technical field related to power system calculation, and more specifically, relates to a method for predicting a day-ahead market clearing price range taking into account source and load uncertainty. Background Art

[0002] As the scale of market-based electricity transactions continues to grow annually, electricity trading models and pricing mechanisms are continuously optimized. By 2024, power trading centers nationwide will have organized and completed market transactions totaling 6,179.57 billion kWh, accounting for 62.7% of total electricity consumption and 76% of grid electricity sales. Within the State Grid region, market trading volume across power trading centers increased by 6.3% year-on-year. The scale of medium- and long-term contract transactions continues to expand, accounting for over 90% of the total. The spot electricity market is also steadily advancing, with some provinces already entering full operation. The steady expansion of the electricity market has facilitated the initial formation of a diversified competitive landscape in my country.

[0003] my country's current electricity spot market primarily utilizes a centralized market model, whereby transactions are conducted through full electricity declaration and centralized optimization and clearing, with the node marginal price (LMP) serving as the market clearing price. During market operation, power dispatch agencies optimize the unit operation mix, time-of-use generation output curves, and time-of-use node electricity prices based on the declared information of market participants and grid operating boundary conditions, incorporating the Safety Constrained Unit Commitment (SCUC) and Security Constrained Economic Dispatch (SCED) models. This pricing mechanism effectively reflects market supply and demand, electricity price costs, and grid congestion, providing important price signals for power companies in formulating power generation plans.

[0004] With the rapid development of the electricity market, predicting spot market clearing prices has become a research hotspot. Currently, methods for predicting electricity spot clearing prices primarily include data-driven machine learning models and physical-mechanism-based simulation clearing models. Data-driven machine learning models, such as blending ensemble learning, random forest regression, stacked LSTMs, and support vector machines (SVMs) optimized with evolutionary algorithms, can accurately predict electricity price trends. However, they are limited by their strong reliance on training data and their difficulty adapting to market fluctuations. Physical-mechanism-based simulation clearing models, which simulate the market clearing process using optimization algorithms such as SCED, can accurately reflect the supply and demand relationship and its evolution in the electricity market. However, in practice, due to their high model complexity and numerous parameters, they struggle to fully address the uncertainties inherent in both the power supply and load sides.

[0005] At the same time, due to the high uncertainty of the supply and demand relationship in the electricity market, how to improve the accuracy, stability, and applicability of clearing price forecasts has become one of the core issues of concern to power generation companies and power market operators. Current forecasting methods often fail to take into account the uncertainty of supply and demand, resulting in large deviations between the quotations of power generation entities and the expected results, affecting the optimal allocation of power resources and the benefits of market entities. Therefore, there is an urgent need for a power spot market clearing price forecasting technology that can comprehensively consider source and load uncertainty, based on economic supply and demand theory and the SCED model, and combined with methods such as Monte Carlo simulation, to improve the accuracy and reliability of forecasts, provide scientific decision-making support for power market participants, and promote the healthy development of the power market. Summary of the Invention

[0006] In response to the above defects or improvement needs of the prior art, the present invention provides a method for predicting the day-ahead market clearing price range taking into account the uncertainty of source and load, the method comprising: S1: by considering the uncertainty under the influence of multiple parameters of the unit's predicted output, the output uncertainty probability is calculated according to the random distribution characteristics of each parameter and the influence weight of each parameter; S2: according to the influence of the power market demand response mechanism, the demand elasticity model is adopted, and the uncertainty range of the predicted load demand is calculated according to the actual uncertainty factors of the power grid operation day; S3: Latin hypercube sampling is discretized within the unit output forecast range to determine the superposition step size of the output upper limit of the thermal, wind and solar units, and the upper limit of the load forecast range is calculated within the load forecast range. Within the range, the total power generation capacity of the bidding units is uniformly discretized, and the different output scenarios of the units are combined with the corresponding grid load demand forecast data for traversal; S4: through the unit start-up and shutdown plan, the source-load forecast results of each group are substituted, and the start-up units in each period are called to meet the load balance and the unit output combination under various constraints, and the total cost objective function value under this unit combination is calculated; S5: The solver is called to solve the unit output combination and the marginal cost of each node that meet the accuracy of the objective function; S6: The source-load scenario combinations under different situations are traversed, and the clearing price forecast range of the unit winning output range and marginal cost within the source-load uncertainty interval is calculated.

[0007] Furthermore, the various parameters for setting the predicted output of the unit in step S1 include:

[0008] The output power of the wind turbine is calculated by adding five parameters, namely wind direction and wind turbine status, to the original parameters of wind speed, air density, and swept area. The output power of the thermal power unit and photovoltaic unit is determined by comprehensively considering the influence of five parameters: solar radiation intensity, temperature, solar angle, photovoltaic unit efficiency, and effective area. The random distribution characteristics of these parameters are determined by Monte Carlo simulation, and the probability distribution function f(λ) of each parameter is obtained.

[0009] The output uncertainty probability is calculated by taking the probability distribution function of each output calculation parameter and the probability of its cumulative distribution function F(x) in the interval [-u,u] as the confidence interval of the parameter, calculating the accuracy probability of the parameter, and performing a weighted average of the accuracy probability according to the influence of each parameter to obtain the uncertainty range of the unit output.

[0010] Furthermore, the load demand uncertainty range described in step S2 is predicted by using a Gaussian mixture model (GMM) to calculate the load uncertainty range [-δ, δ] based on the influence of the power market demand response mechanism and the actual uncertainty factors of the power grid operation day using a demand elasticity model.

[0011] Furthermore, in step S3, the Latin hypercube sampling discretization within the unit output prediction range is performed by discretizing the output of the three types of thermal, wind and solar units participating in the bidding in an uncertain interval with an equal step size of l, and discretizing the load demand uncertainty interval with an equal step size according to the output units of the thermal, wind and solar units in different scenarios, so that the load demand increases with the increase of the total power generation capacity of all units, reflecting the change of grid load under the load demand response.

[0012] Furthermore, the unit start-up and shutdown plan described in step S4 is based on the perspective of hydropower. Since the electricity market information disclosure rules do not publish the specific start-up and shutdown plans of units of other operating manufacturers, it is impossible to obtain the start-up and shutdown plans of thermal, wind and solar power from the perspective of hydropower. In addition, in the actual simulation process, it is impossible to start from all units at the same time. For the prediction of thermal, wind and solar power, only the total output of thermal, wind and solar power in the node area is used as the unit output. Therefore, the unit start-up and shutdown plan only includes the start-up and shutdown plan of the hydropower unit, and the thermal, wind and solar power units are all in the start-up state by default. The various constraints such as the load balance constraint are based on the constraints of the actual electricity spot market clearing rules, including: system load balance constraints, unit output upper and lower limit constraints, system rotating reserve constraints, unit climbing constraints, line flow constraints, hydropower vibration zone constraints and hydropower coupling constraints.

[0013] Furthermore, the total cost objective function in step S4 is to minimize the total power generation cost of all generators in the simulated power system. When the IEEE30 DC power flow model is used, only the flow of active power of the line is considered, and the positive and negative power flow slack variables of the line and section are not considered. The total operating cost is minimized. In the calculation period T of N units, the total power generation cost of each unit is C i,t (i=1,2,…,N;t=1,2,…,T), the force P is obtained by taking the middle mark of time period t i,t The total cost of all units is expressed as follows:

[0014]

[0015] Where C T represents the total power generation cost of all units in the total calculation period, RMB; N1 represents the number of hydropower units; N e (e=2,3,4) represents the number of thermal, wind and solar units respectively; represents the power generation cost of the i-th hydropower unit in time period t, yuan; represents the expected power generation cost of the i-th thermal, wind, or solar unit in the n calculation intervals of time period t, in yuan; represents the expected average power of unit i in n calculation intervals of time period t, MW; is the predicted average bid output of unit i in the kth interval of time period t, MW; p i,t,k is the interval probability of unit i in the kth interval of time period t.

[0016] The marginal cost clearing on the power generation side mainly consists of three parts: power generation cost, start-up and shutdown cost, and upper and lower reserve cost. Since the power market information disclosure rules do not publish the specific start-up and shutdown plans of units of other operating manufacturers, it is impossible to obtain the start-up and shutdown plans of thermal, wind and solar power from the perspective of hydropower. In addition, in the actual simulation process, it is impossible to start from all units at the same time. For the prediction of thermal, wind and solar power, only the total output of thermal, wind and solar power in the node area is used as the unit output. Therefore, only the start-up and shutdown costs and upper and lower reserve costs of hydropower units are considered, and only the rotating reserve costs of other units are considered. The operating costs of hydropower, thermal, wind and solar power are calculated as follows:

[0017] Cost of hydropower unit:

[0018]

[0019] Where C i,t,Q represents the water consumption cost of the i-th hydropower unit in time period t, yuan / MW; C i,t,S represents the start-up and shutdown cost of the i-th hydropower unit in time period t, yuan; C i,t,RH It represents the cost of the upper and lower spinning reserve of the i-th hydropower unit in time period t, RMB.

[0020] Cost of thermal, wind and solar units:

[0021]

[0022] Where C i,t,F represents the operation and maintenance cost of the i-th thermal, wind and solar power generator set in period t, RMB; C i,t,RF It represents the cost of the upper and lower spinning reserves of the i-th thermal, wind and solar generator set in period t, RMB.

[0023] Furthermore, the calling of the solver for solving in step S5 may be performed by using a Cplex solver in Matlab.

[0024] Furthermore, in step S6, the winning output range of the unit and the clearing price prediction range of the marginal cost within the source-load uncertainty interval are obtained by calculating all source-load scenario combinations under Latin hypercube sampling, and by statistically calculating the clearing price and winning output under each scenario, to obtain the prediction range of the winning output and clearing price of the unit.

[0025] In general, compared with the prior art, the method for predicting the day-ahead market clearing price range considering source-load uncertainty provided by the present invention has the following beneficial effects:

[0026] Compared with the traditional single-point price forecasting model, the interval forecasting method can provide multi-interval price probability forecasts, and combined with the winning bid output forecast, it enables power generation companies to more comprehensively evaluate the changing trends of market prices. While considering the impact of source and load uncertainty on clearing prices, the forecast results are more in line with the actual market operation conditions, enhancing the adaptability of power generation companies to electricity price fluctuations.

[0027] The price range forecasting method combines the characteristics of the tiered quotation mechanism, enabling power generation companies to obtain more accurate reference information when making day-ahead market quotations and output range declarations. By comprehensively considering the fluctuations in the generation capacity of different types of power sources and grid load demand, this method can more accurately simulate the market clearing process, improve the competitiveness of power generation companies in the market, and provide a reference for power companies' bidding strategies in the spot market. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions adopted in the embodiments of the present invention, a brief description of the drawings cited in the embodiments of the present invention is now given. It should be noted that the drawings listed below are only used to illustrate some typical implementations of the present invention and do not constitute a limitation on the scope of protection of the present invention.

[0029] Figure 1 This is a step diagram of a method for predicting a day-ahead market clearing price range considering source-load uncertainty in the present application;

[0030] Figure 2 It is the uncertainty interval discrete diagram of the predicted output of thermal, wind and solar power units;

[0031] Figure 3 It is a diagram simulating the process of grid load demand forecasting;

[0032] Figure 4 This is a chart showing the forecast results of the winning bid output and clearing price range of hydropower units.

[0033] Figure 5 This is the expected curve of the winning output and clearing price forecast range of unit 2 DETAILED DESCRIPTION

[0034] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0035] The present invention provides a method for predicting the day-ahead market clearing price range considering the uncertainty of source and load. The specific steps are as shown in the attached figure. Figure 1 shown.

[0036] This embodiment uses the IEEE30-node system to simulate the node participation process of hydropower, thermal, wind and solar power units. Six hydropower units, two thermal power units, one wind farm and a photovoltaic power plant are set respectively. Considering the cascade hydropower coupling constraints in the process of clearing in Sichuan Province, the installed capacity ratio of 64.1% for hydropower, thermal, wind and solar power in 2025 is set according to the Sichuan Power Grid Development Plan (2022-2025): 16.6%: 6.0%: 13.3%.

[32] In the IEEE30 DC power flow model, the solution primarily relies on node power balance, considering the flow of active power. The precise calculation of voltage amplitude and phase angle is generally ignored. Therefore, slack variables for positive and negative power flows are not directly involved. The objective function aims to minimize the total power generation cost of hydropower, thermal power, wind power, and solar power. Within the above constraints, in accordance with the characteristics of the IEEE30 node system, power flow constraints at specific sections are not considered, only line power flow constraints. To avoid damage to hydropower units caused by frequent passage through vibration zones, the output of the hydropower units is guaranteed to be above 60% of the rated output.

[0037] With reference to the spot clearing rules, the hydropower units are set according to the installed capacity ratio of Sichuan Province. The six hydropower units are set as three upstream and three downstream, and the matching upstream and downstream reservoir data are set. The main calculation constraints are:

[0038] System load balance constraints, unit output upper and lower limit constraints, system spinning reserve constraints, unit ramp constraints, line flow constraints, hydropower vibration zone constraints, and hydropower coupling constraints

[0039] 1. System load balance constraint: Considering the load uncertainty under the power market demand response, a discrete interval of ±3MWh is given here.

[0040] 2. Unit output upper and lower limit constraints: Since actual thermal, wind and solar power output data is difficult to obtain, the output process of thermal, wind and solar power units is set according to their output characteristics and set installed capacity. The output uncertainty is set to [-0.5%, 0.5%] for thermal power units, [-7.5%, 7.5%] for wind farms, and [-5.0%, 5.0%] for photovoltaic power plants.

[0041] 3. Unit vibration zone constraints: The output of each hydropower station unit should avoid the vibration zone and ensure that the output of the hydropower unit is above 60% of the rated output of the unit.

[0042] 4. System spinning reserve constraint: The sum of the unit's output increase and decrease capabilities in each period must meet the upward and downward spinning reserve requirements required for operation.

[0043] 5. Unit ramp constraint: The unit ramp rate per unit time during the ramp-up or ramp-down process is less than the limit constraint.

[0044] 6. Line power flow constraints: The load transferred to the line by the output power of the relevant node generators in all simulated lines cannot exceed the upper and lower constraints of the line.

[0045] 7. Hydropower coupling constraint: The total inflow and outflow of the simulated upstream and downstream hydropower stations are balanced, that is, the sum of the outflow of the upstream hydropower station, the interval inflow, and the change in storage capacity is equal to the outflow of the downstream hydropower station.

[0046] (1) Determination of source and load parameters

[0047] The parameters of a hydropower unit mainly include the unit's node number, upper and lower output limits, and upper and lower ramp rates. Table 1 below is a partial parameter table of a hydropower unit:

[0048] Table 1 Some relevant parameters of hydropower units

[0049] Since the simulation data is hypothetical data, the output uncertainty and power generation cost adopt reference values. Figure 1 This is a discrete distribution diagram of the output of thermal, wind and solar units. The parameters of each thermal, wind and solar unit are shown in Table 2 below:

[0050] Table 2 Parameters of thermal, wind and solar power units

[0051] The simulation of the power grid load process uses the data given by the IEEE30 node system, but the unit output range needs to match the load demand. Therefore, the proportional coefficient is multiplied on the basis of the IEEE30 node load to obtain the power grid load demand process as shown in the attached figure. Figure 2The line power constraint is multiplied by the same proportionality factor, and a discrete interval of ±3 MWh is imposed on the grid load based on the load uncertainty under demand response. This yields the uncertainty range of the simulated demand load in the model, and the day-ahead market clearing forecast is calculated. Since the runoff data used is all 1-hour interval, the calculation period interval is set to 1 hour. Clearing forecast results are calculated for the next 24 time periods. For the initial states of the six upstream and downstream hydropower units, downstream unit 2 is set to shutdown, while the remaining units are all in operation.

[0052] (2) Calculation results

[0053] According to the input of the above source-load parameter data, under the multi-dimensional influence of thermal, wind and solar on electricity price clearing, the price range prediction model conforms to the applicable scenario of the Latin hypercube sampling method. Therefore, the output process of thermal, wind and solar units is evenly divided into 5 equal parts and sampled according to its uncertainty range. At the same time, the influence of grid load demand elasticity under the power market demand response is considered. Within the output uncertainty range imposed by the grid load, the grid load demand is made to increase evenly with the increase of the total power generation capacity of thermal, wind and solar within the interval. Finally, 125 groups of possible clearing processes are obtained, and the price range of the clearing results and the corresponding hydropower unit output allocation interval are analyzed. Statistical analysis of the price range is performed, and the statistical results of the clearing price and output forecast data are divided into 5 intervals, as shown in the attached figure. Figure 3 Figure (a) shows the calculation results of the output range of the hydropower unit with the lowest cost. Figure 3 Figure (b) shows the calculation results of the clearing price range of hydropower units under the condition of considering the uncertainty of source and load. The red range is the calculated clearing price range. Figure 4 The different data points in the middle interval are divided into 5 intervals, and the expected calculation of the interval points is performed to obtain the expected curve and probability of the electricity price and unit output interval, such as Figure 4 It is the expected curve of the output range and price forecast range of unit 2.

[0054] Based on the predicted clearing price ranges and winning output ranges of different units, price segmentation and output segmentation are carried out within the range of more reliable probability during the day-ahead trading process in the provincial or regional spot market. Currently, most provinces use a maximum of 10 output ranges and corresponding price declarations. The price range can be divided within the probability range based on factors such as the actual status of the hydropower station units, providing a reference for hydropower plant units to participate in day-ahead market price declarations and unit output declarations.

[0055] It will be easily understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for predicting the day-ahead market clearing price range considering source and load uncertainty, characterized by The method of predicting a clearing price range using source-load uncertainty includes: S1: By considering the uncertainty of the unit's predicted output under the influence of multiple parameters, the output uncertainty probability is calculated according to the random distribution characteristics of each parameter and the influence weight of each parameter; S2: Based on the influence of the power market demand response mechanism, the demand elasticity model is used to calculate the uncertainty range of the predicted load demand according to the actual uncertainty factors of the power grid operation day; S3: Perform Latin hypercube sampling discretization within the unit output forecast range to determine the superposition step size of the output upper limit of the thermal, wind, and solar units. Within the load forecast range, perform uniform discretization based on the total power generation capacity of the bidding units, and combine different unit output scenarios with the corresponding grid load demand forecast data. S4: Based on the unit start-up and shutdown plan, the source-load forecast results of each group are substituted, and the unit output combination that satisfies the load balance and various constraints of the start-up units in each period is called, and the total cost objective function value under this unit combination is calculated; S5: Call the solver to solve the unit output combination and the marginal cost of each node that meets the accuracy of the objective function; S6: Traverse the source-load scenario combinations under different circumstances, and calculate the clearing price prediction range of the unit's bid output range and marginal cost within the source-load uncertainty interval.

2. The method according to claim 1, characterized in that In step S1, the parameters for predicting the unit output are set as follows: the output power of the wind turbine is calculated by adding wind direction and wind turbine status to the original wind speed, air density, and swept area parameters; the output power of the thermal power unit and photovoltaic unit is calculated by comprehensively considering the influence of five parameters: solar radiation intensity, temperature, solar angle, photovoltaic unit efficiency, and effective area. The random distribution characteristics of the above parameters are determined by Monte Carlo simulation to obtain the probability distribution function f(λ) of each parameter: During the calculation process, the upper limit of the unit output is determined based on the predicted unit power generation capacity. Since the output of multi-source units is affected by multiple factors, resulting in uncertainty in the clearing forecast, this method adds multi-parameter influence coefficients to the original thermal, wind, and solar unit output calculation. Monte Carlo simulation is used to simulate the random distribution characteristics of the multi-dimensional parameters of thermal, wind, and solar units. The single-value predicted output in the traditional model is expressed as an uncertainty interval, and the corresponding clearing price forecast interval is calculated. The output calculation model of each type of unit is described as follows: 1) Wind turbine output prediction model: The output power of a wind turbine is closely related to wind speed. By adding wind direction and wind turbine status parameters to the original calculation, the output of the wind turbine can be expressed as: P w =5×10 -4 ρAv 3 C p C w (Δθ)C s Where, P w represents the output power of the wind turbine, MW; v is the real-time wind speed, m / s; ρ is the air density, kg / m 3 ; A is the fan swept area, m 2 ; C p is the fan output coefficient; C w is the wind direction correction coefficient, C w =cos(Δθ); C s The fan status is 0: maintenance or shutdown status, and 1: normal operation status. 2) Photovoltaic unit output prediction model: The output power of the photovoltaic unit is mainly related to the intensity of solar radiation. The output power of the photovoltaic unit P is affected by temperature, solar angle and photovoltaic unit efficiency. pv It can be expressed as: P pv =1×10 -3 or pv,r (1+β(TT r ))GA pv cos(θ s ) Where η pv,r is the efficiency of the photovoltaic unit at the reference temperature; β is the temperature coefficient, which indicates the change in the efficiency of the photovoltaic unit when the temperature rises by 1°C; T is the temperature in the current state, ℃; T r Indicates the reference standard ambient temperature, ℃; G is the light intensity, kW / m 2 ; A pv Effective area of photovoltaic unit, m 2 θ s The angle between solar radiation and the normal to the photovoltaic panel. 3) Thermal power unit output prediction model: The output of thermal power units has minimum and maximum output limits under normal operation. The up and down floating is mainly constrained by the output limit during the up and down ramp process, which can be expressed as: Where, P f is the current output of the thermal power unit, MW; P min 、P max are the minimum and maximum output constraints of thermal power units, MW; P u 、P d are the actual output of thermal power units climbing up and down, MW; P u,max 、P d，max They are the up and down ramp constraints of thermal power units, MW.

3. The method according to claim 1, characterized in that In step S1, the output uncertainty probability is calculated by taking the probability distribution function of each output calculation parameter and the probability of its cumulative distribution function F(x) in the interval [-u, u] as the confidence interval of the parameter, calculating the accuracy probability of the parameter, and performing a weighted average of the accuracy probability according to the influence of each parameter to obtain the uncertainty range of the unit output: The output prediction process of thermal wind and solar power units involves the influence of multi-dimensional parameters. The random distribution characteristics of the multi-dimensional influencing parameters of the unit output are analyzed through Monte Carlo simulation, and the weight coefficient α of the influencing factor is considered. i Determine the impact of different parameter inputs on the inaccuracy of the prediction results, and all parameters that affect the unit output λ i The accuracy probability of the weighted average can be used to obtain the corresponding unit output uncertainty range, which can avoid the large-scale output range calculation and the probability statistics to obtain the calculation amount of output prediction uncertainty, thereby obtaining the output uncertainty quantification based on the random parameter distribution, and then expressing the fixed output upper limit as the uncertainty range. Here, it is assumed that the parameter λ obeys the probability distribution of λ~(μ,σ under Monte Carlo simulation 2 ) is a normal distribution, and its probability density function is expressed as: Where, σ 2 is the parameter error; μ is the mean of its parameter normal distribution. Its cumulative distribution function F(x) is: Normalize λ to Z~N(0,1): Taking the probability of the interval in [-u,u] as the confidence interval of parameter λ, the accuracy probability of parameter λ is: The accuracy probability of the unit's predicted output can be expressed as: By using the above calculation method of output uncertainty, the uncertainty of the output of the i-th thermal wind and solar power plant can be expressed as 1-p pi , the output prediction range of the thermal, wind and solar units in period t can be obtained as 4. The method according to claim 1, wherein In step S2, the load demand uncertainty range is predicted by using a Gaussian mixture model (GMM) based on the influence of the power market demand response mechanism and the actual uncertainty factors of the power grid operation day, and the load uncertainty range [-δ, δ] is: Taking load uncertainty into account in the system load balance constraint during the clearing process: Where, P i,t represents the output of unit i in time period t, MW; T j,t represents the planned power of inter-provincial transmission tie line j in time period t (input is positive, output is negative), MW; NT is the total number of tie lines; D t is the total system load during period t, which can be expressed as: Where, d n,t represents the load of the nth computing node in time period t, in MW. Due to the influence of the power market's demand response mechanism, and according to supply-demand theory in economics, grid load demand varies in a correlated manner with generation capacity. When generation capacity exceeds demand, price reductions will cause a certain increase in demand; when generation capacity falls short of demand, the opposite occurs. Therefore, the model employs a demand elasticity model. Based on the actual uncertainty factors of the grid's daily operation, a load forecast uncertainty range of [-δ, δ] is given (the short-term forecast uncertainty range is generally within 1% of the total load). This allows grid load changes to be uniformly discretized within this range as the total hydropower, thermal, wind, and solar power generation capacity increases. This ensures that grid load demand exhibits demand elasticity with the total generation capacity of the units in the cyclic calculation.

5. The method according to claim 1, wherein In step S3, the Latin hypercube sampling discretization within the unit output prediction range is performed by discretizing the output of the three types of thermal, wind and solar units participating in the bidding in an uncertain interval with an equal step size l, and discretizing the load demand uncertainty interval with an equal step size according to the output units of the thermal, wind and solar units in different scenarios, so that the load demand increases with the increase of the total power generation capacity of all units, reflecting the change of grid load under load demand response.

6. The method according to claim 1, characterized in that In step S4, the unit start-up and shutdown plan is based on the perspective of hydropower. Since the electricity market information disclosure rules do not publish the specific start-up and shutdown plans of units of other operating manufacturers, it is impossible to obtain the start-up and shutdown plans of thermal, wind and solar units from the perspective of hydropower. In the actual simulation process, it is impossible to start from all units at the same time. For the prediction of thermal, wind and solar units, only the total output of thermal, wind and solar units in the node area is used as the unit output. Therefore, the unit start-up and shutdown plan only includes the start-up and shutdown plan of the hydropower unit, and the thermal, wind and solar units are defaulted to all started up.

7. The method according to claim 1, characterized in that In step S4, the various constraints such as the load balance constraint are based on the constraints of the actual electricity spot market clearing rules, including: system load balance constraint, unit output upper and lower limit constraints, system spinning reserve constraint, unit climbing constraint, line flow constraint, hydropower vibration zone constraint and hydropower coupling constraint.

8. The method according to claim 1, characterized in that In step S4, the total cost objective function is to minimize the total power generation cost of all generators in the simulated power system. When using the IEEE30 DC power flow model, only the flow of active power of the line is considered, and the positive and negative power flow slack variables of the line and section are not considered. The total operating cost is minimized. In the calculation period T of N units, the total power generation cost of each unit is C i,t (i=1,2,…,N;t=1,2,…,T), the force P is obtained by taking the middle mark of time period t i,t The total cost of all units is expressed as follows: Where C T represents the total power generation cost of all units in the total calculation period, RMB; N1 represents the number of hydropower units; N e (e=2,3,4) represents the number of thermal, wind and solar units respectively; represents the power generation cost of the i-th hydropower unit in time period t, yuan; represents the expected power generation cost of the i-th thermal, wind, or solar unit in the n calculation intervals of time period t, in yuan; represents the expected average power of unit i in n calculation intervals of time period t, MW; is the predicted average bid output of unit i in the kth interval of time period t, MW; p i,t,k is the interval probability of unit i in the kth interval of time period t. The marginal cost clearing on the power generation side mainly consists of three parts: power generation cost, start-up and shutdown cost, and upper and lower reserve cost. Since the power market information disclosure rules do not publish the specific start-up and shutdown plans of units of other operating manufacturers, it is impossible to obtain the start-up and shutdown plans of thermal, wind and solar power from the perspective of hydropower. In addition, in the actual simulation process, it is impossible to start from all units at the same time. For the prediction of thermal, wind and solar power, only the total output of thermal, wind and solar power in the node area is used as the unit output. Therefore, only the start-up and shutdown costs and upper and lower reserve costs of hydropower units are considered, and only the rotating reserve costs of other units are considered. The operating costs of hydropower, thermal, wind and solar power are calculated as follows: Cost of hydropower unit: Where C i,t,Q represents the water consumption cost of the i-th hydropower unit in time period t, yuan / MW; C i,t,S represents the start-up and shutdown cost of the i-th hydropower unit in time period t, yuan; C i,t,RH It represents the cost of the upper and lower spinning reserve of the i-th hydropower unit in time period t, RMB. Cost of thermal, wind and solar units: Where C i,t,F represents the operation and maintenance cost of the i-th thermal, wind and solar power generator set in period t, RMB; C i,t,RF It represents the cost of the upper and lower spinning reserves of the i-th thermal, wind and solar generator set in period t, RMB.

9. The method according to claim 1, characterized in that In step S6, the winning output range of the unit and the clearing price prediction range of the marginal cost within the source-load uncertainty interval are obtained by calculating all source-load scenario combinations under Latin hypercube sampling, and by statistically calculating the clearing price and winning output under each scenario, thereby obtaining the prediction range of the winning output and clearing price of the unit.

Images