Gradient descent-based power generation side spot quotation optimizing method and device for power market

Abstract

The invention relates to the related technical field of power dispatching, in particular to a gradient descent-based power generation side spot quotation optimizing method and device for a power market. The method comprises the following steps: inputting specific parameters of a power generation side main body power supply; constructing a quotation function based on the target frequency modulation time period combination; wherein the quotation function is composed of a P vector and a dQ vector, the P vector represents the quotation, and the dQ vector represents the length of the output section; performing a clearing settlement process, and calculating an optimization objective function according to the input data and the initialized quotation function; calculating a quotation gradient, and based on the quotation gradient, carrying out gradient updating to optimize a quotation function; wherein the gradient update of the P vector represents the influence of the quotation change on the total gross profit, and the gradient update of the dQ vector represents the influence of the output section change on the total gross profit; judging whether a convergence condition is met or not, if yes, stopping iteration, and if not, continuing gradient updating; and determining spot quotation corresponding to the target frequency modulation time period based on a gradient updating process.

Description

Technical Field

[0001] The present application relates to the technical field related to power dispatching, and specifically to a method and device for optimizing spot quotations on the power generation side based on gradient descent in the power market. Background Art

[0002] In the existing electricity spot market, the spot quotation strategies of power generation entities mainly rely on manual experience and simple statistical methods. These methods usually only consider the price scenario at a single time point, and lack a comprehensive analysis of the impact of spot prices at multiple time points throughout the day. In addition, the existing quotation strategies cannot flexibly adapt to changes in market rules, and when formulating spot quotation strategies, it is difficult to recommend output and quotations at the same time, resulting in low efficiency and accuracy of quotation strategies. Summary of the invention

[0003] In view of this, the embodiments of the present application are committed to providing a method and device for optimizing the spot quotation on the power generation side based on gradient descent in the power market, so as to optimize the spot quotation on the power generation side to assist in power dispatching.

[0004] The first aspect of the present application provides a method for optimizing spot prices on the power generation side in the power market based on gradient descent, comprising: Enter the specific parameters of the main power supply on the power generation side; Based on the target frequency modulation period combination, a quotation function is constructed; wherein the quotation function is composed of a P vector and a dQ vector, wherein the P vector represents the quotation, and the dQ vector represents the output section length; Carry out the clearing and settlement process, and calculate the optimization objective function based on the input data and the initialized quotation function; Calculate the quotation gradient, and based on the quotation gradient, perform gradient update to optimize the quotation function; wherein the gradient update of the P vector represents the impact of the quotation change on the total gross profit, and the gradient update of the dQ vector represents the impact of the output stage change on the total gross profit; Determine whether the convergence condition is met. If so, stop the iteration, otherwise continue the gradient update; Determining a spot price corresponding to the target frequency modulation period based on a gradient update process; If frequency modulation quotations are involved, first determine various frequency modulation time period combinations that meet the preset requirements; for each frequency modulation time period combination, take the frequency modulation time period combination as the target frequency modulation time period combination, and execute the above steps until the spot quotations corresponding to the various frequency modulation time period combinations are obtained respectively, and then determine the optimal frequency modulation time period combination and the corresponding spot quotation.

[0005] In some embodiments, the specific parameters include: plant power consumption rate, startup and shutdown time, medium and long-term holding quantity and price, variable cost function, installed capacity and upper and lower limits of output in different time periods.

[0006] In some embodiments, the optimization objective function is constructed based on at least one of total gross profit, revenue per kilowatt-hour, and gross profit per kilowatt-hour.

[0007] In some embodiments, the determining various frequency modulation time period combinations that meet preset requirements; Various frequency modulation time period combinations that meet the preset requirements are traversed to obtain all frequency modulation time period combinations that meet the preset requirements.

[0008] In some embodiments, it also includes: Display various frequency modulation time period combinations and corresponding spot quotes to assist relevant personnel in making decisions.

[0009] In some embodiments, the gradient-based updating process of determining a spot price corresponding to the target frequency modulation period includes: Post-process the quotation function to simplify the number of quotation function segments and determine the spot quotation; Among them, the principle of post-processing is: concentrate the quotation segment to the global highest point and local highest point of the quotation function; the length of the quotation output segment is divided into equal parts to the size of the local maximum value of the quotation function.

[0010] The present application also provides a device for optimizing spot price quotation on the power generation side based on gradient descent, comprising: An input module is used to input specific parameters of the main power supply on the power generation side; A construction module is used to construct a quotation function based on a target frequency modulation period combination; wherein the quotation function is composed of a P vector and a dQ vector, wherein the P vector represents the quotation and the dQ vector represents the output section length; A calculation module is used to perform a clearing and settlement process, calculate an optimization objective function according to input data and an initialized quotation function; calculate a quotation gradient, and based on the quotation gradient, perform a gradient update to optimize the quotation function; wherein the gradient update of the P vector represents the impact of the quotation change on the total gross profit, and the gradient update of the dQ vector represents the impact of the output segment change on the total gross profit; The judgment module is used to judge whether the convergence condition is met. If so, the iteration is stopped, otherwise the gradient update is continued; A determination module, configured to determine a spot price corresponding to the target frequency modulation period based on a gradient update process; The frequency modulation quotation module is used to first determine various frequency modulation time period combinations that meet the preset requirements if frequency modulation quotation is involved; for each frequency modulation time period combination, the frequency modulation time period combination is used as the target frequency modulation time period combination, and the above steps are executed until the spot quotations corresponding to the various frequency modulation time period combinations are obtained respectively, and then the optimal frequency modulation time period combination and the corresponding spot quotation are determined.

[0011] The present application also provides an electronic device, including: A processor, and a memory for storing a program executable by the processor; The processor is used to implement the above-mentioned method for optimizing the spot price quotation of the power generation side based on gradient descent in the power market by running the program in the memory.

[0012] The present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the processor executes the above-mentioned method for optimizing the spot price of the power generation side based on gradient descent in the power market.

[0013] The present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the above-mentioned method for optimizing spot prices on the power generation side based on gradient descent in the electricity market.

[0014] The present application provides a method for optimizing the spot quotation of the power generation side based on gradient descent in the power market, which includes inputting specific parameters of the main power source of the power generation side; constructing a quotation function based on a target frequency modulation period combination; wherein the quotation function is composed of a P vector and a dQ vector, wherein the P vector represents the quotation and the dQ vector represents the length of the output section; performing a clearing and settlement process, calculating the optimization target function according to the input data and the initialized quotation function; calculating the quotation gradient, and optimizing the quotation function by performing a gradient update based on the quotation gradient; wherein the gradient update of the P vector represents the impact of the quotation change on the total gross profit, and the gradient update of the dQ vector represents the impact of the output section change on the total gross profit; judging whether the convergence condition is met, if so, stopping the iteration, otherwise continuing the gradient update; determining the spot quotation corresponding to the target frequency modulation period based on the gradient update process; if a frequency modulation quotation is involved, first determining various frequency modulation period combinations that meet the preset requirements; for each frequency modulation period combination, taking the frequency modulation period combination as the target frequency modulation period combination, executing the above steps until the spot quotations corresponding to the various frequency modulation period combinations are obtained respectively, and then determining the optimal frequency modulation period combination and the corresponding spot quotation. In this way, by inputting the specific parameters of the power supply of the main power source on the power generation side and constructing the quotation function based on the target frequency regulation period combination, the scientific calculation method is used to replace the shortcomings of manual experience in the existing technology. This method can more accurately quantify the degree of influence between different power generation side quotation strategies, and provide a more accurate basis for the quotation decision of the main power generation side in the spot market. The process of calculating the quotation gradient and updating the gradient can adjust the quotation function in real time according to the market situation. The gradient update of the P vector and the dQ vector respectively reflects the impact of the quotation and output segment changes on the total gross profit, so that the quotation can be optimized with the dynamic changes of the market, and the flexibility and adaptability of the quotation strategy are improved. For situations involving frequency regulation quotations, by determining various frequency regulation period combinations that meet the preset requirements and performing the optimization step for each combination, the optimal frequency regulation period combination and the corresponding spot quotation can be found. This method can adapt to different market conditions and frequency regulation needs, so that the main power generation side can also formulate the optimal quotation strategy in the frequency regulation market. This method avoids the problem of over-reliance on manual experience in the existing technology and reduces the cost of traders learning spot rules and spot transactions. Traders do not need to consider too many complex variables and manual experience factors as before, which simplifies the decision-making process. By judging the convergence conditions to decide whether to stop the iteration, the calculation efficiency can be improved while ensuring the optimization effect of the quotation. It avoids unnecessary iterative calculations, saves computing resources and time, and enables the power generation side to obtain the optimal quotation strategy more quickly and respond to market changes in a timely manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 It is a flow chart of a method for optimizing spot prices on the power generation side in the electricity market based on gradient descent provided by an embodiment of the present application.

[0017] Figure 2 It is a schematic diagram of some cases of the method provided by an embodiment of the present application.

[0018] Figure 3 It is a schematic diagram of some cases of the method provided by an embodiment of the present application.

[0019] Figure 4 It is a schematic diagram of some cases of the method provided by an embodiment of the present application.

[0020] Figure 5 It is a schematic diagram of some cases of the method provided by an embodiment of the present application.

[0021] Figure 6 It is a schematic diagram of a partial process of a method provided in an embodiment of the present application.

[0022] Figure 7 It is a schematic diagram of some cases of the method provided by an embodiment of the present application.

[0023] Figure 8 It is a schematic diagram of some cases of the method provided by an embodiment of the present application.

[0024] Fig. 9 It is a schematic diagram of some cases of the method provided by an embodiment of the present application.

[0025] Fig.10 It is a schematic diagram of some cases of the method provided by an embodiment of the present application.

[0026] Fig.11 It is a schematic diagram of a partial process of a method provided in an embodiment of the present application.

[0027] Fig.12 It is a structural schematic diagram of a device for optimizing spot prices on the power generation side based on gradient descent provided in one embodiment of the present application.

[0028] Fig.13 It is a schematic diagram of the structure of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0029] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] In the spot market, market players mainly on the power generation side and the power consumption side will declare the spot information of the trading day in advance, including the power generation and consumption curve, power generation quotation, etc. The market clearing algorithm calculates the settlement electricity price of the power generation side and the power consumption side at each time point during the trading day, as well as the power generation and consumption curves of each power generation and power consumption entity. The principle is mainly to minimize the social electricity cost. At the same time, more spot market models now include auxiliary service markets such as frequency regulation, peak regulation, and standby for clearing.

[0031] Therefore, the market clearing results determine the settlement of electricity revenue of the power generation side. Therefore, the spot quotation strategy is the key to the profitability of the power generation side. When declaring spot transactions, traders need to master complex market clearing and settlement rules, such as the electric energy market, and auxiliary service market rules mainly based on frequency regulation, peak regulation, and standby, as well as the rules for connecting markets. At the same time, traders need to process a large amount of market information in order to formulate appropriate spot strategies. By recommending quotation strategies through algorithms, traders can effectively reduce the workload of market information analysis and improve overall profitability.

[0032] In order to solve the above problems, the present application provides a solution to build a set of clearing and settlement functions under given market clearing and settlement rules. In the process of calculating the clearing and settlement, the power generation side quotation also calculates the gradient at the same time. In the process of gradient back propagation, the change of each section of the power generation side quotation for the target value (such as total gross profit) can be calculated. Then the quotation is concentrated on the maximum value of the quotation function after the gradient is updated. This is because the curve contains the gradient information, which means that if the output section length is increased at the local maximum value, the target value will be increased the most. Therefore, the quotation function curve will increase the output section length of these local maximum values, thereby increasing the target value.

[0033] After introducing the basic principles of the present application, various non-limiting embodiments of the present application will be described in detail with reference to the accompanying drawings.

[0034] Exemplary Methods Figure 1 1 is a flow chart of a method for optimizing spot price quotation on the power generation side based on gradient descent in the power market provided by an embodiment of the present application. Figure 1 As shown, the method includes the following contents.

[0035] Step S110, inputting specific parameters of the main power supply on the power generation side; In this step, the system will collect and input all necessary parameters related to the power generation side. These parameters may include but are not limited to the power consumption rate of the plant, the start-up and shutdown time, the medium and long-term position quantity and price, the variable cost function, the installed capacity, and the upper and lower limits of the output in different time periods. These data are the basis for the subsequent optimization of the quotation strategy, ensuring that the algorithm can calculate based on accurate actual conditions.

[0036] Step S120, constructing a quotation function based on the target frequency modulation period combination; wherein the quotation function is composed of a P vector and a dQ vector, wherein the P vector represents the quotation, and the dQ vector represents the output section length; This step involves constructing a bid function based on the target frequency regulation period combination. The bid function consists of a P vector and a dQ vector, where the P vector represents different bid levels and the dQ vector represents the corresponding output section length. The construction of this function is key to the optimization process because it is directly related to the revenue of the power generation side entity.

[0037] Step S130, performing a clearing and settlement process, calculating an optimization objective function based on input data and an initialized quotation function; Specifically, the system will perform the clearing and settlement process based on the input data and the initialized quotation function. This includes calculating the optimization objective function, such as total gross profit, per-kilowatt-hour revenue, or per-kilowatt-hour gross profit. This process simulates the actual clearing and settlement mechanism of the electricity market, ensuring the practicality and effectiveness of the quotation strategy. Step S140, calculating the bid gradient, and based on the bid gradient, performing gradient update to optimize the bid function; wherein the gradient update of the P vector represents the impact of the bid change on the total gross profit, and the gradient update of the dQ vector represents the impact of the output stage change on the total gross profit; The system will calculate the gradient of the bid function, which reflects the impact of the bid change on the total gross profit. The gradient update of the P vector shows the impact of the bid change, while the gradient update of the dQ vector shows the impact of the output segment change. This step is the core of optimizing the bid function through the gradient descent method, aiming to find the bid strategy that maximizes the profit.

[0038] Step S150, determine whether the convergence condition is met, if so, stop iteration, otherwise continue gradient update; The system will determine whether the quotation function meets the preset convergence conditions. If it does, such as the change in the quotation and the output section length is less than a certain threshold, the iteration will stop. If it does not, the system will continue to perform gradient updates until the optimal quotation strategy is found.

[0039] Step S160, determining a spot price corresponding to the target frequency modulation period based on a gradient update process; Specifically, once the quotation function converges, the system will determine the spot quotation corresponding to the target frequency regulation period combination. This quotation is the quotation that is expected to maximize the revenue of the power generation side entity under the current market conditions.

[0040] Step S170, if frequency modulation quotation is involved, first determine various frequency modulation time period combinations that meet the preset requirements; for each frequency modulation time period combination, take the frequency modulation time period combination as the target frequency modulation time period combination, and execute the above steps until the spot quotations corresponding to the various frequency modulation time period combinations are obtained respectively, and then determine the optimal frequency modulation time period combination and the corresponding spot quotation.

[0041] That is, if a frequency modulation quotation is required, the system will first determine all frequency modulation time combinations that meet the preset requirements. Then, for each frequency modulation time combination, repeat the above steps to find the optimal spot quotation corresponding to each combination. Finally, the system will compare the quotations of different frequency modulation time combinations to determine the optimal frequency modulation time combination and the corresponding spot quotation.

[0042] These steps together constitute a complete optimization process, which aims to provide a scientific, efficient and market-adaptive spot quotation strategy for power generation entities. Through this method, power generation entities can better participate in the power spot market and improve their economic benefits and market competitiveness.

[0043] In this way, by inputting the specific parameters of the power supply of the main power source on the power generation side and constructing the quotation function based on the target frequency regulation period combination, the scientific calculation method is used to replace the shortcomings of manual experience in the existing technology. This method can more accurately quantify the degree of influence between different power generation side quotation strategies, and provide a more accurate basis for the quotation decision of the main power generation side in the spot market. The process of calculating the quotation gradient and updating the gradient can adjust the quotation function in real time according to market conditions. The gradient update of the P vector and the dQ vector respectively reflects the impact of the quotation and output segment changes on the total gross profit, so that the quotation can be optimized with the dynamic changes of the market, and the flexibility and adaptability of the quotation strategy are improved. For situations involving frequency regulation quotations, by determining various frequency regulation period combinations that meet the preset requirements and performing the optimization step for each combination, the optimal frequency regulation period combination and the corresponding spot quotation can be found. This method can adapt to different market conditions and frequency regulation needs, so that the main power generation side can also formulate the optimal quotation strategy in the frequency regulation market. This method avoids the problem of over-reliance on manual experience in the existing technology and reduces the cost of traders learning spot rules and spot transactions. Traders do not need to consider too many complex variables and manual experience factors as before, which simplifies the decision-making process. By judging the convergence conditions to decide whether to stop the iteration, the calculation efficiency can be improved while ensuring the optimization effect of the quotation. It avoids unnecessary iterative calculations, saves computing resources and time, and enables the power generation side to obtain the optimal quotation strategy more quickly and respond to market changes in a timely manner.

[0044] The specific parameters include: plant power consumption rate, startup and shutdown time, medium and long-term position quantity and price, variable cost function, installed capacity and upper and lower limits of output in different time periods. The specific parameters input include the following: Among them, plant power consumption rate: refers to the proportion of electric energy consumed by power generation equipment during operation to the total power generation. This parameter is used to calculate the actual power generation capacity and economic benefits of the power generation side entity. Start-up and shutdown time: refers to the time required for power generation equipment to go from the shutdown state to full operation, and the time required from the operating state to the shutdown state. This parameter affects the scheduling flexibility and response speed of the power generation side. Medium and long-term position quantity and price: includes the power generation and the corresponding market price in the medium and long term. These data help the power generation side entity evaluate its market positioning and profit expectations. Variable cost function: describes the cost function that changes with the output during the power generation process. This function is usually related to factors such as the operating efficiency and fuel cost of the generator set, and can help optimize the quotation strategy. Installed capacity: refers to the maximum power generation capacity of the power generation equipment, usually in megawatts (MW). This parameter is the basis for determining the amount of electricity that the power generation side entity can provide in the market. Output upper and lower limits for different time periods: refers to the minimum and maximum output limits of power generation equipment in different time periods. These limits ensure that the power generation side entities can follow the rules and technical constraints of the power market when formulating bidding strategies. By inputting these specific parameters, the system can more accurately reflect the actual operating conditions of the power generation side entities, thus providing a solid foundation for the subsequent construction and optimization of bidding functions.

[0045] In some embodiments, the optimization objective function is constructed based on at least one of total gross profit, revenue per kilowatt-hour, and gross profit per kilowatt-hour.

[0046] The optimization objective function is a key indicator used to evaluate and optimize the spot quotation strategy of the power generation side. The optimization objective function can be constructed based on the following key financial indicators: Total gross profit: The total gross profit refers to the difference between the total revenue and total cost obtained by the power generation side entity through spot market transactions in a certain period (usually one day). The total revenue includes medium- and long-term electricity charges, day-ahead deviation electricity charges, real-time deviation electricity charges, and frequency regulation electricity charges, while the total cost includes variable costs, fixed costs, etc. The total gross profit is an important indicator to measure the profitability of the power generation side entity in the spot market. Revenue per kilowatt-hour: Revenue per kilowatt-hour refers to the revenue generated by the power generation side entity for each kilowatt-hour of electricity generated. This indicator reflects the power generation efficiency and market pricing ability of the power generation side entity. By optimizing the revenue per kilowatt-hour, the power generation side entity can improve its competitiveness in the market. Gross profit per kilowatt-hour: Gross profit per kilowatt-hour refers to the gross profit generated for each kilowatt-hour of electricity generated, that is, the revenue per kilowatt-hour minus the cost of kilowatt-hour. This indicator reflects the profitability of the power generation side entity in more detail because it is directly related to the profitability of each unit of electricity. When constructing the optimization objective function, at least one of the above indicators can be selected as the target. In actual operation, one or more indicators may be selected as optimization targets based on the operating strategy and market environment of the power generation entity.

[0047] For example: If the power generation entity is more concerned about overall profitability, it may choose total gross profit as the optimization target. If the power generation entity hopes to improve power generation efficiency and market competitiveness, it may choose per-kilowatt-hour revenue or per-kilowatt-hour gross profit as the optimization target. In step S130, the system will perform the clearing and settlement process based on the input data and the initialized bidding function, combined with the selected optimization objective function. This process involves simulating the clearing and settlement mechanism of the electricity market to calculate the expected benefits of the power generation entity under different bidding strategies. In this way, the system can evaluate the effectiveness of different bidding strategies and select the optimal bidding strategy to maximize the selected optimization target.

[0048] Calculating the quotation gradient (step S140) is also to optimize this objective function. The system will calculate the impact of the quotation change (P vector) and the output segment change (dQ vector) on the optimization objective function, and then update the quotation function through the gradient descent method to iteratively approach the optimal solution. This process continues until the convergence condition is met (step S150), thereby determining the optimal spot quotation strategy.

[0049] The determining of various frequency modulation time period combinations meeting the preset requirements includes: traversing various frequency modulation time period combinations meeting the preset requirements to obtain all frequency modulation time period combinations meeting the preset requirements.

[0050] First, clearly define what is a frequency regulation time period combination that "meets the preset requirements". These requirements may include, but are not limited to: Price threshold for frequency regulation services. Minimum frequency regulation demand. Equipment availability and maintenance plan. Historical frequency regulation service performance data. Market rules and regulatory restrictions. Collect all relevant data, including market data, equipment status, historical performance, etc., to support the evaluation of preset requirements. Generate initial combinations: Generate all possible frequency regulation time period combinations based on the operating cycle of the power market. This may involve dividing a day into multiple periods, such as peak, flat and off periods. Filter combinations that meet the requirements: Evaluate each initial combination and filter out qualified combinations based on the preset requirements. For example, if the frequency regulation price of a period is lower than the preset threshold, the period may be excluded. Traverse and record: Traverse all frequency regulation time period combinations that meet the preset requirements and record detailed information for each combination, including period division, expected frequency regulation demand and price, etc. The detailed steps of the traversal process include: Initialize traversal parameters: Set the parameters of the traversal process, such as time range, price range, etc. Iterative check: Iteratively check each possible time period combination and conduct detailed market and equipment status analysis for each combination. Record detailed information: For each combination that meets the requirements, record its detailed information, including time period, expected frequency modulation demand, price, etc. Generate report: Organize the traversal results into a report or database for subsequent analysis and decision-making.

[0051] In some embodiments, it also includes: displaying each frequency modulation time period combination and the corresponding spot quotation to assist relevant personnel in making decisions.

[0052] Displaying each frequency modulation period combination and the corresponding spot price is a key link in decision support. The following is a detailed description of this step: Generate frequency modulation time combination report: summarize all frequency modulation time combinations obtained through traversal, including the time period division and corresponding market conditions of each combination, such as demand, price, etc. Calculate the spot price of each combination: for each frequency modulation time combination, use the optimization method determined in the previous step to calculate the optimal spot price. These quotes are obtained based on the gradient descent algorithm and the optimization objective function. Analyze the economic benefits of the quote: conduct an economic benefit analysis of the spot price of each frequency modulation time combination, including expected benefits, costs and profits. Develop decision support materials: organize the frequency modulation time combination and the corresponding spot price into easy-to-understand charts, tables or reports to help decision makers quickly grasp the pros and cons of various options. Provide decision suggestions: based on the economic benefit analysis, provide suggestions on which frequency modulation time combinations may bring the greatest benefits. At the same time, it can also provide risk assessment to help decision makers understand the market risks that different strategies may face. Interactive decision support tool: develop an interactive tool or software that allows decision makers to adjust parameters and view the changes in spot prices and expected benefits of different frequency modulation time combinations in real time according to different market conditions and corporate strategies. Real-time update and adjustment: According to market dynamics and changes in internal corporate conditions, frequency modulation time period combinations and spot quotations are updated in real time to ensure the timeliness and accuracy of decision-making information. Detailed data analysis and suggestions are provided to help decision makers make more scientific and reasonable decisions. It enables the entities on the power generation side to flexibly respond to market changes and quickly adjust quotation strategies. Optimized spot quotation strategies help improve the total revenue of entities on the power generation side. Through risk assessment, it helps decision makers understand the risks of different strategies and make more robust decisions. This step is the output stage of the entire optimization method. Its purpose is to transform complex data analysis results into practical and feasible decision recommendations to assist relevant personnel in making effective decisions.

[0053] In some embodiments, the gradient-based update process determines the spot price corresponding to the target frequency modulation period, including: post-processing the quotation function to simplify the number of quotation function segments to determine the spot price; wherein the principle of post-processing is: concentrating the quotation segments to the global highest point and the local highest point of the quotation function; the quotation output segment length is divided into equal parts to the size of the local maximum value of the quotation function.

[0054] The post-processing step is to simplify the complex quotation function to facilitate practical operation and decision-making. The following is a detailed description of this step: The spot price is determined based on the process of gradient update. Gradient update iteration: During the gradient descent process, the quotation function (composed of P vector and dQ vector) is updated iteratively, where P vector represents the quotation and dQ vector represents the corresponding output segment length. Each iteration aims to optimize the quotation function to maximize the total gross profit or other preset optimization goals. Convergence judgment: Determine whether the quotation function meets the convergence conditions, such as whether the quotation change or the output segment length change is lower than the preset threshold. If it is satisfied, the iteration is stopped; if it is not satisfied, the iteration is continued until the convergence conditions are met. Post-processing of the quotation function Simplify the number of quotation function segments: The purpose of post-processing is to simplify the quotation function and reduce the number of quotation segments for practical operation and decision-making. This is usually achieved by merging or deleting quotation segments with shorter output segment lengths. Concentrate quotation segments: Concentrate the quotation segments to the global and local highest points of the quotation function. These points represent the locations where the quotation contributes the most to the total gross profit at different output levels. Proportional division of output segment length: The division of the quotation output segment length should be proportional to the size of the quotation function at the local maximum. This means that in areas with higher gradients (i.e., areas where changes in bids have a greater impact on total gross profit), the output segment length should be longer to maximize total gross profit. Post-processing principles Global maximum point: The global maximum point refers to the bid point that contributes the most to total gross profit within the entire bid range. Concentrating the bid segments to this point ensures that the maximum benefit is obtained at the highest bid level. Local maximum point: The local maximum point refers to the point where the bid contributes the most to total gross profit at a specific output level. These points may be different at different output levels, so they need to be considered separately. Proportional division: The division of the output segment length should be proportional to the size of the bid function at the local maximum. This means that in areas with higher gradients, the output segment length should be longer to ensure that the bid in these areas can maximize the total gross profit. Benefits Simplify the decision-making process: By simplifying the bid function, decision makers can more intuitively understand the potential impact of different bid strategies, so as to make faster and more accurate decisions. Improve the adaptability of the bid strategy: The simplified bid function is easier to adjust and adapt to market changes, which improves the market adaptability of the power generation side. Optimize resource allocation: By concentrating the bid segments to key points, power generation resources can be allocated more effectively and resource utilization efficiency can be improved. Improve market competitiveness: Optimized spot quotation strategies help improve the market competitiveness of power generation entities and increase their profits in the electricity spot market.

[0055] This post-processing step is an important part of the entire optimization method. It transforms the complex gradient update process into a practical bidding strategy, providing strong decision support for the power generation side.

[0056] The solution provided by this application is described below in conjunction with specific embodiments: 1) Input data: Specific parameters of the main power supply on the power generation side: plant power consumption rate (comprehensive, actual, %), start-up and shutdown time, medium- and long-term position price (position price at various times throughout the day, position unit is MWh, price unit is yuan / MWh), variable cost function (c=f(q), cost is yuan / MWh, q is the power generation output, unit is MW), installed capacity (MW). Upper and lower limits of output in different time periods (MW). Empty quotation table.

[0057] For the case of upper and lower limits of output in different time periods, please refer to Figure 2 .

[0058] The quotation limit on the power generation side (presented in the form of a quotation table): (multiple sections, each section of data consists of starting output, ending output, and quotation). If there are relevant output division requirements, or rigid quotation constraints (such as filling in 0 yuan / MWh in the first section to ensure startup), the final output can take into account the constraint. The starting output of the first section must be the minimum technical output (the minimum value of the lower limit of each period), and the ending output of the last section must be the maximum value of the upper limit of each period. For specific cases, please refer to Figure 3 (matching this application Figure 2 ).

[0059] 2) Output data: Power generation side spot strategy: Under the given power generation side quotation constraints, recommend the optimal frequency regulation declaration period combination and the corresponding spot quotation.

[0060] Example: Shanxi is divided into five frequency modulation periods, namely 00:00-06:00 (early morning), 06:00-12:00 (morning peak), 12:00-16:00 (noon), 16:00-21:00 (evening peak), 21:00-24:00 (late night). If frequency modulation is recommended, the output of this period is True, and if it is not recommended, the output is False.

[0061] Prediction results of different bidding strategies on the power generation side (strategy comparison scheme): 1) No frequency modulation scheme The lower output limit of this unit is 150MW, and the upper output limit is 350MW. Under the restriction of no frequency regulation throughout the day, the main quoted output is concentrated on 332 yuan / MWh, but 358 and 401 are also the local best quotes, so some quotes will be scattered in these two quotes. The quotation constraints here are 0 for 150-151MW and 1500 for 330-350MW. Figure 4 .

[0062] 2) Plan for frequency regulation application on the power generation side It is recommended that frequency regulation can be applied for at all times of the day. The main output section is concentrated at 282 yuan / MWh, and some output sections apply for 398-399 yuan / MWh, which mainly corresponds to the minimum and maximum compensation prices in Shanxi frequency regulation quantity and price compensation rules, maximizing the frequency regulation quantity and price compensation income. Figure 5 .

[0063] based on Figure 4 and Figure 5 By comparing the contents, it can be seen that the FM scheme should be selected on that day.

[0064] Furthermore, the implementation process of the solution provided in this application is divided into a gradient descent quotation process and a frequency modulation period selection process.

[0065] The purpose of the gradient descent quotation process is to optimize the quotation given the frequency modulation period.

[0066] The frequency modulation period selection process is to try different frequency modulation period declarations to optimize quotations, and then select the best combination of frequency modulation period and quotation.

[0067] Reference Figure 6 , the gradient descent quotation process is as follows: The quotation function consists of two vectors, namely the P vector and the dQ vector. The specific quotation function can be expressed as: dQ = B(P), which specifically means the relationship between the lengths of the output segments allocated under each quotation. dQ=end output-start output. As shown in the figure, the upper left figure is the general format of the quotation table, and the lower left figure is the function graph of the quotation table. The right side is the quotation function format of the quotation table. The vector quotation function format actually lacks the information on the lower limit of the output in different time periods. During the optimization process, the upper and lower limits of the output will be calculated as quotation boundaries, but will not affect the gradient optimization. Therefore, this information will be completed in the input data, and the information on the lower limit of the output in different time periods will be considered during the clearing process. Refer to Figure 7 .

[0068] The quotation function can choose any value for initialization, as long as it satisfies the constraint ΣdQ = (upper output limit - lower output limit). In the spot quotation rules, the number of quotation segments generally cannot exceed 10 segments. However, there is no such restriction in the gradient descent quotation process. Considering the continuity of the quotation, the number of quotation segments can be set to more. Therefore, the initialization quotation vector can be divided into one quotation segment for every 1 yuan / MWh according to the upper and lower limit prices. For example, the spot quotation of Shanxi Electric Power is limited to 0-1500 yuan / MWh. The initialization quotation segment can be divided into 1501 segments, and the P value of each quotation segment is an integer yuan / MWh. The dQ vector initialization can be (upper output limit - lower output limit) / number of segments. For example, the upper output limit of a unit is 350MW and the lower limit is 150MW. Then the dQ output corresponding to each quotation segment is (350-150) / 1501. In this way, the quotation can be selected more continuously. The quotation segments with shorter output lengths will be filtered out during the post-processing process, and the remaining output segment lengths can be adjusted proportionally. Refer to Figure 8 .

[0069] Based on the input data and the initialized quotation function, the optimization target is calculated. Generally, the total gross profit, electricity revenue, and electricity gross profit are used as any of these revenue indicators as the optimization target. The following optimization target will take the total gross profit as an example.

[0070] The clearing and settlement process can be represented by the function: R = g(P, dQ, I). R is the optimization target (total gross profit), and I is the set of all other relevant data (including medium- and long-term position volume and price, day-ahead real-time price, variable cost, etc.). g(.) is the function of the clearing and settlement process. The clearing process is to calculate the day-ahead and real-time output of each time period according to the power market clearing rules after the quotation table and the day-ahead real-time price of each time period are given, including the clearing of electric energy and the clearing of frequency regulation, as well as the output limit of the upper and lower limits of the output at each time.

[0071] The settlement process is based on the day-ahead and real-time output and medium- and long-term positions calculated in the clearing process, and the income is calculated according to the settlement rules. The medium- and long-term electricity charges, day-ahead deviation electricity charges, real-time deviation electricity charges, frequency regulation electricity charges, total costs and online electricity consumption are calculated separately, and then the total gross profit, kilowatt-hour revenue, kilowatt-hour gross profit and other income targets for the whole day are calculated.

[0072] The clearing rules and settlement rules can be flexibly adjusted according to the settlement methods of different provinces.

[0073] The quote gradient and convergence process are as follows: Calculate the gradient by back propagation:

[0074] Perform gradient update on the P and dQ vectors of the quotation function:

[0075]

[0076] Subscript i is the quotation segment index, n is the number of iterations, and α is the coefficient of gradient update. Since the information contained in the gradient is the key and the initial quotation function is not very important, α can be selected to be a relatively large value, such as 0.3.

[0077] The gradient update of the P vector can be understood as how much the total gross profit will change if a certain quotation increases by 1 yuan or decreases by 1 yuan. After convergence, if the distribution of the P vector is concentrated on certain quotations, it means that increasing the output of this quotation can increase the total gross profit.

[0078] The gradient update of the dQ vector can be understood as how much the total gross profit will change if the output corresponding to a certain quotation segment increases by 1MW. In quotation segments with higher gradients, the increase in total gross profit will be greater if the output is increased by 1MW, so the output should be more concentrated on these quotation segments. After convergence, the output segment length of the quotation segment corresponding to the dQ vector will also be correspondingly higher. Therefore, the gradient provides optimization information.

[0079] Since there is a constraint ΣdQ = (upper limit of output - lower limit of output), the dQ vector will be adjusted twice after the gradient update. After the secondary adjustment, the absolute value information will be lost, but the relative size information of the dQ vectors in different quotation segments will be retained. In the post-processing process, since only the position of the local maximum of the dQ vector and the relative size of the local maximum are needed, the absolute value information is not important.

[0080] The convergence criteria are:

[0081]

[0082] n is the nth iteration, i is the number of segments, and ε is a small number but greater than 0. Generally, εp = 0.1, which means that the maximum change in the quotation does not exceed 0.1 yuan / MWh; εq = the minimum value of the initial dQ * 0.1, which means that the output segment adjustment length is less than 0.1 times the minimum initial dQ value, which means convergence (ensuring convergence accuracy). When P and dQ meet the convergence condition at the same time, the iteration stops, but if it does not converge after a certain number of iterations, it stops and outputs the quotation according to the quotation function after stopping.

[0083] For the continuity of quotation calculation, the number of quotation function segments will be more than the actual maximum number of quotation segments that can be submitted. Therefore, the quotation function will be simplified to a certain extent (reducing the number of function segments). In the process of calculating the quotation by the algorithm, the quotation segments will be concentrated on the quotation of the global highest point (post-review mode) and the quotation of the local highest point (pre-quotation mode). And the declared price where the quotation segments are relatively concentrated.

[0084] The mathematical meaning of quotation segment concentration is that, since the total output segment length is fixed, the total target value can be maximized only when the output segment length is concentrated on the maximum gradient value.

[0085] The length of the quotation output segment is equal to the size of the local maximum of the quotation function.

[0086] The meaning of the local maximum is that, within a certain adjacent interval, only by increasing the length of the output segment with the local maximum quotation can the total gross profit be maximized, because the gradient of the dQ vector represents the increase in the total gross profit by increasing the output by 1MW in a certain output segment, and the final value of the dQ vector depends on the gradient update. The local maximum contains information about price difference arbitrage and risk control. Since arbitrage can only be obtained by declaring within certain quotation segments, the local maximum is generally within the quotation segment where arbitrage can be obtained.

[0087] Reference Fig. 9 , P vector post-processing: where the x-axis is the quotation segment and the y-axis is the quotation. Under the initialized P vector, the quotation segment and the quotation are one-to-one corresponding, but in the frequency modulation process, due to the frequency modulation quantity and price compensation income, every change in the quotation of any segment of the P vector will increase the frequency modulation quantity and price compensation income, thereby increasing the overall gross profit. Most of the quotation segments will be concentrated on the two dotted line quotations, corresponding to the upper and lower limits of the frequency modulation quantity and price compensation quotation, thereby maximizing the frequency modulation quantity and price compensation income.

[0088] In the electricity energy quotation, since frequency regulation quantity and price compensation is not involved, the gradient of the P vector is 0 and the P vector is not updated.

[0089] The post-processing process will combine and accumulate the output section lengths (dQ) of the same quotation in different quotation sections.

[0090] Reference Fig.10 , dQ vector post-processing: the x-axis is the quotation segment (quotation), and the y-axis is the dQ vector length. After merging according to the P vector, the quotation where the dotted line is located is the quotation segment corresponding to the local maximum of the dQ vector (where the quotation segment corresponding to the left dotted line is the global maximum). Post-processing only retains the output length corresponding to the dotted line, and adjusts the size of the dotted line in proportion to the upper and lower limits of the operation, and the dQ value of the remaining output segment length is set to 0.

[0091] Reference Fig.11 , the FM time selection process is as follows: In the process of gradient descent quotation, the initial frequency modulation period selection combination needs to be given in advance before quotation optimization can be performed. The frequency modulation selection process is mainly a process of cyclically iterating through the frequency modulation period selection to select the optimal frequency modulation period. The specific process is: Initialize the frequency modulation time period combination. Here you can initialize multiple frequency modulation time period combinations, such as no frequency modulation all day, frequency modulation all day, flat frequency modulation, valley frequency modulation, peak frequency modulation, etc.

[0092] Given an initial frequency modulation combination, the quote is calculated by gradient descent method (Process 3.1).

[0093] After obtaining the quotation results of gradient descent, calculate the benefits of frequency modulation in each frequency modulation period through the clearing and settlement process. Select the period when the frequency modulation has an increase in power consumption to declare frequency modulation, and output the combination results of the frequency modulation period.

[0094] Traverse the initialized frequency modulation period combination, perform gradient descent and frequency modulation period combination calculation respectively, and output the results respectively.

[0095] Output results: The final output quote and frequency modulation period are the optimal quotes of the results calculated for multiple initial frequency modulation periods. Or the results calculated for the initial frequency modulation period are output at the same time as a basis for strategy comparison.

[0096] In the solution provided in this application, a set of clearing and settlement functions is built under given market clearing and settlement rules. In the process of calculating the clearing and settlement, the power generation side quotation also calculates the gradient at the same time. In the process of gradient back propagation, the change of each section of the power generation side quotation for the target value (such as total gross profit) can be calculated. Then the quotation is concentrated on the maximum value of the quotation function after the gradient is updated. This is because the curve contains the gradient information, which means that if the output section length is increased at the local maximum value, the target value will be increased the most. Therefore, the quotation function curve will increase the output section length of these local maximum values, thereby increasing the target value.

[0097] For the existing technology, if it is an ex post mode, the quotation corresponding to the global maximum value is consistent with the optimal electricity energy quotation of the existing technology. If it is an ex ante mode, the advantages of this method are: 1) This method can recommend frequency regulation quotations, but the existing technology cannot provide frequency regulation quotation recommendations. 2) This method can recommend quotations and output segment lengths at the same time. The existing technology can only recommend quotations after a given output segment division (if the existing technology does not have an output segment recommendation, it can only be divided by average), and this method recommends the output segment length in a more flexible and concise way. 3) It can consider multiple price scenarios more quickly, and output the gradient under comprehensive consideration based on multiple price scenarios without the need for cyclic calculations, and the calculation speed is faster.

[0098] Exemplary Devices The device embodiments of the present application can be used to execute the method embodiments of the present application. For details not disclosed in the device embodiments of the present application, please refer to the method embodiments of the present application.

[0099] Fig.12 FIG. 1 is a block diagram of a device for optimizing spot price quotation at the power generation side based on gradient descent provided by an embodiment of the present application. Fig.12 As shown, the device comprises: An input module 121 is used to input specific parameters of the main power supply on the power generation side; A construction module 122 is used to construct a quotation function based on the target frequency modulation period combination; wherein the quotation function is composed of a P vector and a dQ vector, wherein the P vector represents the quotation and the dQ vector represents the output section length; The calculation module 123 is used to perform the clearing and settlement process, calculate the optimization target function according to the input data and the initialized quotation function; calculate the quotation gradient, and based on the quotation gradient, perform gradient update to optimize the quotation function; wherein the gradient update of the P vector represents the impact of the quotation change on the total gross profit, and the gradient update of the dQ vector represents the impact of the output stage change on the total gross profit; A judgment module 124 is used to judge whether the convergence condition is met. If so, the iteration is stopped, otherwise, the gradient update is continued; A determination module 125, configured to determine a spot price corresponding to the target frequency modulation period based on a gradient update process; The frequency modulation quotation module 126 is used to first determine various frequency modulation time period combinations that meet preset requirements if frequency modulation quotation is involved; for each frequency modulation time period combination, the frequency modulation time period combination is used as the target frequency modulation time period combination, and the above steps are executed until the spot quotations corresponding to the various frequency modulation time period combinations are obtained respectively, and then the optimal frequency modulation time period combination and the corresponding spot quotation are determined.

[0100] Exemplary Electronic Devices Below, reference Fig.13 To describe an electronic device according to an embodiment of the present application. Fig.13 A block diagram of an electronic device according to an embodiment of the present application is illustrated.

[0101] like Fig.13 As shown, the electronic device 1300 includes one or more processors 1310 and a memory 1320 .

[0102] The processor 1310 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 1300 to perform desired functions.

[0103] The memory 1320 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, a random access memory (RAM) and / or a cache memory (cache), etc. The non-volatile memory may include, for example, a read-only memory (ROM), a hard disk, a flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 1310 may run the program instructions to implement the above-described method for optimizing the spot price quotation of the power market based on gradient descent on the generation side of the various embodiments of the present application and / or other desired functions. Various contents such as category correspondences may also be stored in the computer-readable storage medium.

[0104] In one example, the electronic device 1300 may further include: an input device 1330 and an output device 1340 , and these components are interconnected via a bus system and / or other forms of connection mechanisms (not shown).

[0105] In addition, the input device 1330 may also include, for example, a keyboard, a mouse, an interface, etc. The output device 1340 may output various information to the outside, including analysis results, etc. The output device 1340 may include, for example, a display, a speaker, a printer, a communication network and a remote output device connected thereto, etc.

[0106] Of course, to simplify, Fig.13 Only some of the components in the electronic device related to the present application are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, the electronic device may further include any other appropriate components according to specific application conditions.

[0107] Exemplary computer program products and computer-readable storage media In addition to the above-mentioned methods and devices, an embodiment of the present application may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps of the method for optimizing the spot price of the power market based on gradient descent on the power generation side according to various embodiments of the present application described in the above "Exemplary Method" section of this specification.

[0108] The computer program product may be written in any combination of one or more programming languages ​​to write program codes for performing the operations of the embodiments of the present application, including object-oriented programming languages, such as Java, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as an independent software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0109] In addition, an embodiment of the present application may also be a computer-readable storage medium having computer program instructions stored thereon, and when the computer program instructions are executed by a processor, the processor executes the steps of the method for optimizing the spot price of the power generation side based on gradient descent in the electricity market according to various embodiments of the present application described in the above "Exemplary Method" section of this specification.

[0110] The computer readable storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can include, for example, but is not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0111] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. A method for optimizing spot prices on the power generation side in the power market based on gradient descent, characterized in that: include: Enter the specific parameters of the main power supply on the power generation side; Construct a quotation function based on the target frequency modulation period combination; The quotation function is composed of a P vector and a dQ vector, wherein the P vector represents the quotation and the dQ vector represents the output section length; Carry out the clearing and settlement process, and calculate the optimization objective function based on the input data and the initialized quotation function; Calculate the quotation gradient, and based on the quotation gradient, perform gradient update to optimize the quotation function; wherein the gradient update of the P vector represents the impact of the quotation change on the total gross profit, and the gradient update of the dQ vector represents the impact of the output stage change on the total gross profit; Determine whether the convergence condition is met. If so, stop the iteration, otherwise continue the gradient update; Determining a spot price corresponding to the target frequency modulation period based on a gradient update process; If frequency modulation quotations are involved, first determine various frequency modulation time period combinations that meet the preset requirements; for each frequency modulation time period combination, take the frequency modulation time period combination as the target frequency modulation time period combination, and execute the above steps until the spot quotations corresponding to the various frequency modulation time period combinations are obtained respectively, and then determine the optimal frequency modulation time period combination and the corresponding spot quotation.

2. The method for optimizing spot prices of power generation side based on gradient descent in the power market according to claim 1 is characterized in that: The specific parameters include: plant power consumption rate, start-up and shutdown time, medium- and long-term holding quantity and price, variable cost function, installed capacity and upper and lower limits of output in different time periods.

3. The method for optimizing spot prices on the power generation side based on gradient descent in the power market according to claim 1 or 2, characterized in that: in, The optimization objective function is constructed based on at least one of the total gross profit, the revenue per kilowatt-hour and the gross profit per kilowatt-hour.

4. The method for optimizing spot prices of power generation side based on gradient descent in the power market according to claim 1 is characterized in that: The determination of various frequency modulation time period combinations that meet preset requirements includes: Various frequency modulation time period combinations that meet the preset requirements are traversed to obtain all frequency modulation time period combinations that meet the preset requirements.

5. The method for optimizing the spot price quotation of the power generation side based on gradient descent in the power market according to claim 1 is characterized in that: Also includes: Display various frequency modulation time period combinations and corresponding spot quotes to assist relevant personnel in making decisions.

6. The method for optimizing the spot price quotation of the power generation side based on gradient descent in the power market according to claim 1 is characterized in that: The process based on gradient updating, determining the spot price corresponding to the target frequency modulation period, includes: Post-process the quotation function to simplify the number of quotation function segments and determine the spot quotation; Among them, the principle of post-processing is: concentrate the quotation segment to the global highest point and local highest point of the quotation function; the length of the quotation output segment is divided into equal parts to the size of the local maximum value of the quotation function.

7. A device for optimizing spot price quotation on the power generation side based on gradient descent, characterized in that: include: An input module is used to input specific parameters of the main power supply on the power generation side; A construction module, used to construct a quotation function based on a target frequency modulation period combination; The quotation function is composed of a P vector and a dQ vector, wherein the P vector represents the quotation and the dQ vector represents the output section length; The calculation module is used to perform the clearing and settlement process and calculate the optimization objective function based on the input data and the initialized quotation function; Calculate the quotation gradient, and based on the quotation gradient, perform gradient update to optimize the quotation function; wherein the gradient update of the P vector represents the impact of the quotation change on the total gross profit, and the gradient update of the dQ vector represents the impact of the output stage change on the total gross profit; The judgment module is used to judge whether the convergence condition is met. If so, the iteration is stopped, otherwise the gradient update is continued; A determination module, configured to determine a spot price corresponding to the target frequency modulation period based on a gradient update process; The frequency modulation quotation module is used to first determine various frequency modulation time period combinations that meet the preset requirements if frequency modulation quotation is involved; for each frequency modulation time period combination, the frequency modulation time period combination is used as the target frequency modulation time period combination, and the above steps are executed until the spot quotations corresponding to the various frequency modulation time period combinations are obtained respectively, and then the optimal frequency modulation time period combination and the corresponding spot quotation are determined.

8. An electronic device, characterized in that: include: A processor, and a memory for storing a program executable by the processor; The processor is used to implement the method for optimizing the spot price of the power generation side in the power market based on gradient descent as described in any one of claims 1 to 6 by running the program in the memory.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, enables the processor to execute the method for optimizing the spot price of power generation side based on gradient descent in the power market according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method for optimizing the spot price quotation of the power generation side in the power market based on gradient descent according to any one of claims 1 to 6 is implemented.

Images