Power market transaction optimization decision-making system for micro-grid assets

Abstract

The invention relates to an electricity market transaction optimization decision-making system for micro-grid assets, and belongs to the technical field of electricity market transaction and financial risk management. The system obtains risk event signals such as market electricity price, plan power, financial data and extreme weather in real time; the risk quantification module constructs a risk factor reflecting the comprehensive financial risk level based on the quantified potential market opportunity income and the plan deviation penalty cost; the strategy decision and weight generation module dynamically generates weight coefficients of income and deviation penalty through a nonlinear mapping function according to the financial risk, and constructs a real-time optimization objective function with income maximization as an objective after risk adjustment. Finally, the system solves the function to determine an optimal transaction decision. According to the invention, the complex market financial risk is quantified as a dynamic decision basis, so that the transaction strategy can be adaptively adjusted between financial loss avoidance and arbitrage opportunity chasing.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power market transaction and financial risk management, in particular to a power market transaction optimization decision system for micro-grid assets. BACKGROUND

[0002] Under the background of power market transaction, the micro-grid operator as a market subject, its dispatching decision directly affects its economic benefits and risk exposure in various transactions such as spot market, ancillary service market, etc. In the prior art, the dispatching strategy is usually based on a fixed financial objective function, such as single pursuit of cost minimization or revenue maximization; such static financial model cannot effectively cope with the complex financial risks composed of real-time fluctuations in market prices and sudden events, and it is difficult to dynamically and quantitatively balance between avoiding high deviation penalties and capturing instantaneous market arbitrage opportunities, resulting in market transaction strategies that are either too conservative and miss profits or face significant economic losses due to insufficient risk assessment; Therefore, how to construct a dispatching method that can quantitatively and comprehensively assess risks in real time and dynamically adjust optimization targets, thereby improving the economic efficiency and robustness of micro-grid operation in a complex and changing market environment, is a key technical problem that needs to be solved at present. SUMMARY

[0003] To solve the above technical problems, the present application provides a power market transaction optimization decision system for micro-grid assets, specifically, the technical solution of the present application comprises: A data acquisition module for acquiring real-time financial data and risk event signals related to power market transactions, the financial data at least including real-time market price signals and day-ahead planned power curves, and the risk event signals at least including regional extreme weather event warning signals; A risk quantification module for quantifying potential market opportunity revenue and potential planned deviation penalty cost based on real-time market price signals; the risk quantification module is also used to combine regional extreme weather event warning signals and real-time market price signals to construct a risk factor reflecting the level of comprehensive financial risk; A strategy decision module for calculating a transaction strategy tendency index based on potential market opportunity revenue, potential planned deviation penalty cost, and comprehensive risk factor; A weight generation module for dynamically generating revenue weight coefficients and deviation penalty weight coefficients through a pre-set nonlinear mapping function based on the transaction strategy tendency index; A target construction module for constructing a real-time optimization objective function with risk-adjusted revenue maximization as the target based on the revenue weight coefficients and the deviation penalty weight coefficients; An optimization solving module is configured to solve a real-time optimization objective function to determine a set of optimal power market trading strategies, which at least include a bidding curve or a regulation power for market bidding.

[0004] Preferably, the risk quantification module constructs a risk factor reflecting the comprehensive financial risk level, including: Based on the regional extreme weather event early warning signal, an external shock amplification term is determined; Based on the mean and standard deviation of the real-time market price signal within a preset sliding time window, a standard variation coefficient is determined. The standard variation coefficient is used to describe the endogenous relative volatility of the market price. According to the product of the external shock amplification term and the standard variation coefficient, a comprehensive risk factor is determined.

[0005] Preferably, the strategy decision module calculates a trading strategy tendency index, including: Based on the difference between the potential market opportunity income and the potential plan deviation penalty cost, the expected net income is determined. Based on the sum of the potential market opportunity income and the potential plan deviation penalty cost, the total economic scale is determined. The total economic scale is multiplied by the comprehensive risk factor to obtain the risk-adjusted total economic scale. By calculating the ratio of the expected net income to the risk-adjusted total economic scale, the trading strategy tendency index is obtained.

[0006] Preferably, the weight generation module uses a logistic function as a nonlinear mapping function to map the trading strategy tendency index to a preset interval to generate a revenue weight coefficient and a deviation penalty weight coefficient; wherein the sum of the revenue weight coefficient and the deviation penalty weight coefficient is always 1.

[0007] Preferably, the target construction module constructs a real-time optimization objective function with the goal of maximizing risk-adjusted revenue, including: The revenue weight coefficient is multiplied by the expected actual market income generated after executing the dispatching action to obtain a weighted market income. The deviation penalty weight coefficient is multiplied by the expected actual deviation penalty generated after executing the dispatching action to obtain a weighted deviation penalty. The difference between the weighted market income and the weighted deviation penalty is maximized to construct a real-time optimization objective function with the goal of maximizing risk-adjusted revenue.

[0008] Preferably, the risk prediction module is configured to call a pre-trained probability prediction model for any candidate dispatching action to output a probability density function of the actual plan deviation power caused by the action.

[0009] Preferably, the risk prediction module is further configured to: determine, based on the probability density function and a preset nonlinear penalty function, an expected deviation penalty cost corresponding to the candidate dispatch action by integral operation, and take the expected deviation penalty cost as an expected actual deviation penalty; wherein the nonlinear penalty function is used to describe the relationship between the penalty cost and the deviation electric quantity.

[0010] Preferably, the dispatch mode decision module is configured to compare the trading strategy tendency index with preset conservative mode threshold and aggressive mode threshold, and select the current dispatch mode from the conservative mode, the balanced mode and the aggressive mode.

[0011] Preferably, the specific operation of the dispatch mode decision module comprises: in response to the trading strategy tendency index being less than the conservative mode threshold, triggering the conservative mode, and setting the return weight coefficient to zero and the deviation penalty weight coefficient to one; in response to the trading strategy tendency index being greater than the aggressive mode threshold, triggering the aggressive mode, and setting the return weight coefficient to one and the deviation penalty weight coefficient to zero; in response to the trading strategy tendency index being between the conservative mode threshold and the aggressive mode threshold, triggering the balanced mode, and adopting the return weight coefficient and the deviation penalty weight coefficient generated by the weight generation module.

[0012] Compared with the prior art, the present application has the following beneficial effects: 1. The present system unifies the price fluctuation of the electricity market and external impact events into a comprehensive financial risk assessment framework, realizing dynamic and comprehensive quantification of market transaction risks; compared with the traditional dispatch method relying on only a single technology or economic indicator, the present system can more accurately assess financial risks in complex market environments, providing a decision basis for formulating scientific transaction strategies; 2. The present system can dynamically adjust the emphasis on market returns and deviation penalties in the transaction strategy according to the quantified financial risks; by generating a transaction strategy tendency index to real-time weigh potential returns and risk costs, the transaction strategy can adaptively and smoothly adjust between aggressive profit-seeking and conservative risk-avoiding, overcoming the problem of rigid transaction decision models in the prior art that cannot adapt to market dynamics; 3. The present system introduces three transaction modes of conservative, balanced and aggressive, and automatically switches based on the transaction strategy tendency index, making the macro strategy of the microgrid participating in market transactions more explicit and intelligent; when market opportunities are huge and financial risks are low, the system can switch to the aggressive mode to maximize transaction returns; when financial risks are high, the system automatically switches to the conservative mode to focus on avoiding penalties and risk hedging; 4. The system builds a real-time optimization objective function with the core of maximizing risk-adjusted financial returns, and can more accurately quantify potential deviation penalties by introducing a probability prediction model; this refined financial modeling and solving mechanism ensures that the final output decision instructions can effectively improve the overall economic efficiency and financial robustness of market participants in a volatile market environment while meeting the compliance of the power grid. BRIEF DESCRIPTION OF DRAWINGS

[0013] The application will be further explained below in conjunction with the accompanying drawings and embodiments: Figure 1 is a structural diagram of the system of the application. DETAILED DESCRIPTION

[0014] In order to make the purpose, technical scheme and advantages of the application more clear and explicit, the application will be further described in detail below in conjunction with specific embodiments.

[0015] Embodiment 1: Please refer to Figure 1 A power market transaction optimization decision system for micro-grid assets, comprising: A data acquisition module for acquiring real-time financial data and risk event signals related to power market transactions, wherein the financial data at least includes real-time market price signals and day-ahead planned power curves, and the risk event signals at least include regional extreme weather event warning signals; A risk quantification module for quantifying potential market opportunity returns and potential plan deviation penalty costs based on real-time market price signals; the risk quantification module is also used to combine regional extreme weather event warning signals and real-time market price signals to construct a risk factor reflecting the comprehensive financial risk level; A strategy decision module for calculating a transaction strategy tendency index according to potential market opportunity returns, potential plan deviation penalty costs and comprehensive risk factors; A weight generation module for dynamically generating return weight coefficients and deviation penalty weight coefficients through a pre-set nonlinear mapping function based on the transaction strategy tendency index; A target construction module for constructing a real-time optimization objective function with the goal of maximizing risk-adjusted returns based on the return weight coefficients and the deviation penalty weight coefficients; An optimization solving module for solving the real-time optimization objective function to determine a set of optimal power market transaction strategies, wherein the power market transaction strategies at least include a bid curve or an adjusted power for market bidding.

[0016] The embodiment of the application provides a power market transaction optimization decision system for micro-grid assets; the system contains a closed-loop, adaptive dispatching decision process, which is specifically realized by the following modules working cooperatively: A data acquisition module, which aims to provide real-time, multi-dimensional data input for subsequent risk quantification and strategy decision-making, is the perception basis of the entire optimization scheduling system; in this system, a data preprocessing and verification module is provided after the data acquisition module to identify and process abnormal market price signals, ensuring the rationality of the data input into the risk quantification module; in addition, in the weight generation module, the calculated comprehensive risk factor and the trading strategy tendency index are set with reasonable upper and lower limits to prevent the weight coefficient from appearing saturation or mutation due to extreme input, ensuring smooth transition of the decision; in this embodiment, the module realizes concurrent and high-frequency acquisition of multiple data sources through standard industrial communication protocol interfaces such as Modbus / TCP, OPC-UA or DNP3, etc.; the module acquires real-time price signals of multiple markets in which the microgrid participates, mainly including real-time electricity price of the electricity market , service price of the ancillary service market , and carbon price of the carbon trading market ; the module receives day-ahead planning power curves with scheduling time interval resolution from the superior grid dispatching center; the module subscribes and receives regional extreme weather event warning signals through an external information interface, which can be realized based on RESTful API, and the received data format is usually JSON; the signal is defined as an extreme weather warning level factor ; the variable is used to quantify the severity level of external major disturbance events, and its basic form can be simplified as a binary state, for example represents warning activation, represents the normal state, while in a refined model, different values can be taken for different warning levels, and the source is the official warning information issued by the authoritative meteorological department or grid operator; A risk quantification module, which aims to convert abstract market risks and environmental risks into calculable and comparable mathematical indicators; in this embodiment, based on the real-time market price signals obtained by the data acquisition module, the module prospectively quantifies the potential revenue and potential cost that the microgrid faces in the next scheduling period if it conducts speculative scheduling; specifically, it calculates potential market opportunity revenue and potential planning deviation penalty cost ; potential market opportunity revenue refers to the maximum theoretical revenue that the microgrid can obtain if it uses its entire adjustable power to participate in the ancillary service market without considering the penalty risk; potential planning deviation penalty cost refers to the theoretical penalty cost that the microgrid may incur if it performs the above maximum power adjustment action and deviates from the day-ahead plan; Among them, the current total adjustable power It is the maximum responsive power calculated in real time based on the current operating status of each controllable unit within the microgrid; In this embodiment, and The calculation formula can be specifically defined as follows:

[0017]

[0018] in, The current service prices in the assisted services market, For scheduling time intervals, The penalty function related to the deviation in electricity consumption is defined by the power grid; the risk quantification module is also used to construct a comprehensive risk factor by combining regional extreme weather event warning signals with real-time market price signals. The purpose of this factor is to quantify the impact of external emergencies and the market's endogenous price volatility in a unified manner. The core purpose of the strategy decision-making module is to generate a macro-level decision-making instruction that can guide subsequent weight allocation based on the return, cost, and risk indicators output by the risk quantification module. In this embodiment, the module calculates a trading strategy preference index. To achieve this function; the trading strategy favors indexes. It refers to a dimensionless indicator used to measure the expected net return per unit of risk-adjusted economic size; its function is to determine the appropriateness of speculative allocation at present, thereby deciding whether the allocation strategy should be more aggressive or conservative. The weight generation module aims to generate the trading strategy preference index output by the strategy decision module. This is transformed into precise mathematical weights for constructing a specific optimization objective function; in this embodiment, the module dynamically generates the revenue weight coefficients through a preset nonlinear mapping function. And deviation penalty weight coefficient ; The objective construction module aims to construct a real-time optimization objective function that reflects the current risk preference, based on the weight coefficients dynamically generated by the weight generation module. This module is based on and The value of is adjusted and optimized in real time to adjust the emphasis on market returns and deviation penalties in the objective, thus constructing a dynamic objective with the maximization of risk-adjusted comprehensive returns as the core. The optimization solution module aims to solve the real-time optimization objective function constructed by the objective function construction module to determine the optimal electricity market trading strategy within the current scheduling cycle. In this embodiment, the module employs a numerical optimization algorithm, given the objective function... The specific form may be nonlinear. In this embodiment, efficient nonlinear programming solvers such as sequential quadratic programming or interior-point methods can be used to find the optimal solution; the adjustment power of each controllable unit in the microgrid is used as the decision variable. Under the premise of satisfying power grid security constraints, solve for... The maximum value is obtained to arrive at the optimal electricity market trading strategy. The optimal electricity market trading strategy This is the final output of the system, specifically a set of optimal business decision data to support bidding activities in the electricity market; for further explanation, the trading strategy... This should include at least a bid curve or regulated electricity volume used for market bidding. For example, this electricity market trading strategy. The determined optimal regulation power value and the corresponding price (i.e., the point on the bid curve) can be packaged into a standardized data format for automatic generation and submission to the higher-level electricity market trading platform for corresponding ancillary service bids or clearing power declarations in the real-time energy market. The function of this system terminates at generating this business strategy data, rather than executing physical power control.

[0019] Compared to existing technologies, this invention can perceive multi-dimensional risks such as market price fluctuations and extreme weather impacts in real time, quantify them into dynamic decision-making criteria, and then reconstruct and optimize the objective function in real time. This enables the microgrid's scheduling strategy to shift from a fixed cost minimization or revenue maximization model to a dynamic and smooth optimization model that balances risk avoidance and revenue pursuit. Thus, while ensuring grid compliance, this system primarily focuses on optimizing economic risks in the market environment. In specific engineering applications, the system's output commands also need to be verified by a verification module based on equipment operating status and safety constraints to address other operational risks such as equipment failures and significant load forecast deviations. This improves the overall economic efficiency and operational robustness of the microgrid in complex and volatile market environments.

[0020] Example 2: The risk quantification module constructs risk factors that reflect the overall level of financial risk, including: Based on regional extreme weather event warning signals, determine the external shock amplification term; The standard coefficient of variation is determined based on the mean and standard deviation of real-time market price signals within a preset sliding time window. The standard coefficient of variation is used to describe the endogenous relative volatility of market prices. The comprehensive risk factor is determined according to the product of the external impact amplification term and the standard coefficient of variation.

[0021] This embodiment is a further description of the specific implementation of the risk quantification module of embodiment 1, which reflects the risk factor of the comprehensive financial risk level ; in order to uniformly quantify the impact of external extreme weather events and endogenous market volatility, this embodiment introduces a comprehensive risk factor , which is calculated as follows:

[0022] , wherein, is the extreme weather warning level factor, which corresponds to the official published weather warning level, for example, no warning , blue warning , yellow warning , orange warning , etc.; in this way, the severity of external impact can be quantified more finely; α is the extreme weather influence weight coefficient, which is used to quantify the amplification effect of external impact on market volatility, and is a dimensionless positive real number; its source is determined by statistical analysis of the historical calibration data set; specifically, the market price volatility amplification factor in the historical calibration data set is defined as , which is calculated as the ratio of the market price standard deviation during the occurrence of an extreme weather event to the standard deviation in the same length of time window before the event. By linear regression analysis or direct averaging of the values of all similar extreme weather events in the data set, the parameter is calibrated; is the standard deviation of the market comprehensive price in the past preset sliding time window, with the unit of yuan / MWh, which is calculated by the module based on the price data collected by the data collection module in real time; the length of the sliding time window is preset to 1 hour; the length is selected to balance the sensitivity to short-term high-frequency fluctuations of market price and the stability to long-term trend, and through historical data analysis, a 1-hour window can better reflect the market risk characteristics in the upcoming dispatching period; is the mean value of the market comprehensive price in the past preset sliding time window, with the unit of yuan / MWh, which is calculated by the module based on the price data collected by the data collection module in real time; wherein the market comprehensive price is an equivalent price calculated by weighting according to the current main market activities of the microgrid, for example , wherein the weight coefficients we, wa, and wc are determined according to the operation strategy of the microgrid and the proportion of the transaction volume of each market; The calculation logic of the formula consists of two parts; one is to determine the external impact amplification term based on regional extreme weather event warning signals ; the second is to determine the standard variation coefficient based on the mean and standard deviation of real-time market price signals within the preset sliding time window ; the standard variation coefficient is a standardized measure of dispersion in statistics, which describes the endogenous relative volatility of market prices; finally, according to the product of the external impact amplification term and the standard variation coefficient, the comprehensive risk factor is determined ; through this design, not only can reflect the market's own volatility, but also can reasonably amplify the volatility when external impact comes, so that the subsequent decision can predictively increase the weight of risk aversion.

[0023] Embodiment 3: The strategy decision module calculates the transaction strategy tendency index, including: determine the expected net income based on the difference between the potential market opportunity income and the potential plan deviation penalty cost; determine the total economic scale based on the sum of the potential market opportunity income and the potential plan deviation penalty cost; multiply the total economic scale by the comprehensive risk factor to get the risk-adjusted total economic scale; obtain the transaction strategy tendency index by calculating the ratio of the expected net income to the risk-adjusted total economic scale.

[0024] This embodiment is a further description of the specific implementation of the strategy decision module of embodiment 1 to calculate the transaction strategy tendency index ; in order to establish a decision basis for dynamic balance between pursuing income and avoiding risk, this embodiment introduces the transaction strategy tendency index , which is calculated as follows:

[0025] Among them, is the potential market opportunity income, with unit of yuan, calculated by the risk quantification module based on real-time market prices; is the potential plan deviation penalty cost, with unit of yuan, estimated by the risk quantification module based on grid penalty rules; is the comprehensive risk factor, dimensionless, calculated by the risk quantification module; to prevent the denominator from being zero, a small normal number is added in the calculation, and the formula is modified as ; when tends to 0, it means that the economic scale of the dispatching action is very small, at this time can be defined as 0, indicating neutral tendency; The calculation logic of this index is based on: potential market opportunity returns. Penalty costs for potential plan deviations The difference determines the expected net income. Based on potential market opportunity benefits Penalty costs for potential plan deviations The sum of these factors determines the total economic scale. Total economic scale refers to the total amount of funds involved in executing this dispatch action; combining total economic scale with comprehensive risk factors Multiply by each other to obtain the risk-adjusted total economic size. The trading strategy preference index is obtained by calculating the ratio of expected net return to the risk-adjusted total economic size. The physical meaning of this index is the net return that a unit of risk-adjusted economic size can generate.

[0026] Example 4: The weight generation module uses the logistic function as a non-linear mapping function to map the trading strategy tendency index to a preset range to generate the return weight coefficient and the deviation penalty weight coefficient; wherein, the sum of the return weight coefficient and the deviation penalty weight coefficient is always 1.

[0027] This embodiment is a further explanation of the specific implementation of the weight generation module in Embodiment 1; in order to favor the trading strategy towards the index. The weights are smoothly and reasonably mapped to weight coefficients. In this embodiment, the logistic function is used as the nonlinear mapping function. The weight generation formula is as follows:

[0028]

[0029] in, The trading strategy bias index is dimensionless and is calculated by the strategy decision module. The return weighting coefficient represents the degree of importance attached to pursuing market returns in the optimization objective. It is a dimensionless variable with a value between 0 and 1. The bias penalty weight coefficient represents the degree of importance attached to avoiding bias penalties in the optimization objective. It is a dimensionless variable with a value between 0 and 1. k and η0 are adjustable parameters of the logistic function, both of which are dimensionless; Control the steepness of the curve. It represents the equilibrium point for decision-making; its origin is determined through backtesting and optimization of historical scheduling data; specifically, it involves defining a performance index to evaluate the effectiveness of historical scheduling. by solving an optimization problem to determine the optimal value of ; for example, numerical optimization methods such as grid search algorithm or gradient ascent algorithm can be used to solve the optimization problem; the performance indicator may be defined as the cumulative net income during the historical data backtest period, or the risk-adjusted return, i.e. the ratio of total net income to return volatility; The logic of this module is to take the trading strategy inclination index as input, map it to the interval of 0 to 1 through a logistic function, to generate the return weight coefficient ; at the same time, generate the deviation penalty weight coefficient , and ensure that the sum of the return weight coefficient and the deviation penalty weight coefficient is always 1.

[0030] Embodiment 5: The target construction module constructs a real-time optimization target function with the goal of maximizing risk-adjusted return, including: multiply the return weight coefficient by the expected actual market return generated after executing the scheduling action to obtain the weighted market return; multiply the deviation penalty weight coefficient by the expected actual deviation penalty generated after executing the scheduling action to obtain the weighted deviation penalty; maximize the difference between the weighted market return and the weighted deviation penalty, and construct a real-time optimization target function with the goal of maximizing risk-adjusted return; in this embodiment, the above function can be specifically defined as follows: The expected actual market return function is defined as the income obtained by participating in the market for auxiliary services, etc. after the microgrid executes the power regulation action , for example , where is the service price of the auxiliary service market, is the regulation power provided to the auxiliary service market, is the dispatch time interval; The expected actual deviation penalty function is defined as the penalty caused by deviating from the day-ahead planned power due to the execution of the power regulation action ; it can be initially constructed as , where is the actual total power after the action is executed, is the penalty function specified by the grid; where the actual total power can be obtained by superimposing the power of each unit in the microgrid, for example , where is the load power at the current time, u is the output of controllable units such as renewable energy, and the output of energy storage (discharge is positive); To further improve the physical fidelity of the model, a cost term related to device wear and tear can be introduced into the objective function; for example, for energy storage units, the charging and discharging actions caused by the cyclic aging cost are added to the penalty term, and the corrected objective function is ; At the same time, when solving the optimization, the upper and lower limits of the power of each unit, the climbing rate, the state of charge range, etc. must be taken as rigid constraints; This embodiment is a further description of the specific implementation of the target construction module of embodiment 1 to construct a real-time optimization objective function with risk-adjusted maximum return as the target. To construct an optimization target that can adapt to the current market risk, this embodiment designs the following real-time optimization objective function:

[0031] wherein, is a decision variable representing a specific market trading action, such as the regulated power to be bid (in units of MW) or the key parameters of the bidding curve; is the return weight coefficient and the deviation penalty weight coefficient calculated by the weight generation module, dimensionless; is the expected actual market return function, indicating the expected actual market return after executing the dispatch action , in yuan; is the expected actual deviation penalty function, indicating the expected actual deviation penalty after executing the dispatch action , in yuan; The construction logic of this objective function is as follows: multiply the return weight coefficient by the expected actual market return after executing the dispatch action to get the weighted market return; multiply the deviation penalty weight coefficient by the expected actual deviation penalty after executing the dispatch action to get the weighted deviation penalty; maximize the difference between the weighted market return and the weighted deviation penalty to construct a real-time optimization objective function with risk-adjusted maximum return as the target.

[0032] Embodiment 6: The risk prediction module is configured to call a pre-trained probability prediction model for any candidate dispatch action, and output the probability density function of the actual plan deviation power caused by the action.

[0033] The risk prediction module is further configured to: The expected deviation penalty cost corresponding to the candidate scheduling action is determined through integral operation based on the probability density function and the preset nonlinear penalty function, and the expected deviation penalty cost is taken as the expected actual deviation penalty. The nonlinear penalty function is used to describe the relationship between the penalty cost and the deviation electric quantity.

[0034] This embodiment is a preferred implementation of the system of embodiment 5, and the expected actual deviation penalty is quantified more finely by adding a risk prediction module. The system further includes a risk prediction module, which aims to introduce a probability prediction model to evaluate the risk of each potential scheduling action in a more scientific way. For any candidate scheduling action , the module calls a pre-trained probability prediction model to output the probability density function of the actual planning deviation electric quantity caused by the action ; in this embodiment, the probability prediction model uses a long short-term memory network based on deep learning, which can effectively learn the dynamic characteristics in time series data, so as to predict the probability distribution of future deviation electric quantity according to the current system state and control instruction. To convert the uncertainty to be predicted into a determined cost item for optimization, this embodiment uses the expected value calculation method:

[0035] Among them, is the expected deviation penalty cost, that is, the penalty cost under the mathematical expectation corresponding to the execution of the action , with unit of yuan; is the conditional probability density function, which is derived from the probability prediction model in this module, which is trained based on historical operation data and related features. Specifically, the input features of the model can include: candidate scheduling action , current load level, renewable energy output prediction value, market price signal , and comprehensive risk factor , etc. is a nonlinear penalty function used to describe the relationship between the penalty cost and the deviation electric quantity, which is derived from the official penalty rules published by the grid operator; The logic of the module is to determine the expected deviation penalty cost corresponding to the candidate scheduling action through integral operation based on the probability density function and the preset nonlinear penalty function ; the expected deviation penalty cost is taken as the expected actual deviation penalty ​the expected actual deviation penalty as defined in embodiment 5 By this step, the system completes the decision-making closed loop from macroscopic tendency judgment to microscopic action optimization.

[0036] Embodiment 7: The scheduling mode decision module is configured to compare the trading strategy tendency index with preset conservative mode threshold and aggressive mode threshold, and select the current scheduling mode from the conservative mode, balanced mode and aggressive mode.

[0037] The specific operation of the scheduling mode decision module includes: In response to the trading strategy tendency index being less than the conservative mode threshold, triggering the conservative mode, and setting the return weight coefficient to zero and the deviation penalty weight coefficient to one; In response to the trading strategy tendency index being greater than the aggressive mode threshold, triggering the aggressive mode, and setting the return weight coefficient to one and the deviation penalty weight coefficient to zero; In response to the trading strategy tendency index being between the conservative mode threshold and the aggressive mode threshold, triggering the balanced mode, and using the return weight coefficient and the deviation penalty weight coefficient generated by the weight generation module.

[0038] This embodiment is a preferred control strategy implementation of the system of embodiment 1, by adding a scheduling mode decision module, the system can switch to the optimal macroscopic operation mode in different risk level market environment; The system further comprises a scheduling mode decision module, which compares the trading strategy tendency index with preset conservative mode threshold and aggressive mode threshold , and selects the current scheduling mode from the conservative mode, balanced mode and aggressive mode; the conservative mode threshold and the aggressive mode threshold refer to the critical values for dividing different scheduling modes; the setting logic is based on the risk preference of the operator, and is calibrated through historical data backtesting; for example, can be set as the value corresponding to when the lower limit of the preset confidence interval of the expected speculative net return is less than zero, to ensure that when entering the conservative mode, speculative behavior has a high probability of causing loss; The specific operation of the scheduling mode decision module includes: In response to the trading strategy tendency index being less than the conservative mode threshold , the system triggers the conservative mode; in this mode, the module forcibly sets the return weight coefficient to zero and the deviation penalty weight coefficient to one; at this time, the real-time optimization objective function degenerates to ; in response to the trading strategy inclination index greater than the aggressive mode threshold , the system triggers the aggressive mode; in this mode, the module forces the return weight coefficient to one, and forces the deviation penalty weight coefficient to zero; at this time, the real-time optimization objective function degenerates to ; in response to the trading strategy inclination index between the conservative mode threshold and the aggressive mode threshold , the system triggers the balanced mode; in this mode, the system uses the weight generation module to solve using the complete real-time optimization objective function with the dynamically generated return weight coefficient and the deviation penalty weight coefficient .

[0039] It should be noted that the above examples are only used to illustrate the technical solutions of the present application and not to limit it. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced by equivalents without departing from the spirit and scope of the present application.​

Claims

1. A power market trading optimization decision-making system for microgrid assets, characterized in that, Applications in the electricity market for financial transaction decisions and risk management of microgrid assets include: The data acquisition module is used to acquire financial data and risk event signals related to electricity market transactions in real time. The financial data includes at least real-time market price signals and day-ahead planned power curves. The risk event signals include at least regional extreme weather event warning signals. The risk quantification module is used to quantify potential market opportunity returns and potential planning deviation penalty costs based on real-time market price signals. The risk quantification module is also used to combine regional extreme weather event warning signals with real-time market price signals to construct risk factors that reflect the overall level of financial risk. The strategy decision-making module is used to calculate the trading strategy preference index based on potential market opportunity returns, potential plan deviation penalty costs, and comprehensive risk factors. The weight generation module is used to dynamically generate return weight coefficients and deviation penalty weight coefficients based on the trading strategy preference index through a preset non-linear mapping function. The objective construction module is used to construct a real-time optimization objective function based on the return weighting coefficient and the deviation penalty weighting coefficient, with the goal of maximizing risk-adjusted return. An optimization solution module is used to solve the real-time optimization objective function to determine a set of optimal electricity market trading strategies, which include at least a bid curve or regulation of electricity volume for market bidding.

2. The power market transaction optimization decision-making system for microgrid assets according to claim 1, characterized in that, The risk quantification module constructs risk factors that reflect the overall level of financial risk, including: Based on regional extreme weather event warning signals, determine the external shock amplification term; The standard coefficient of variation is determined based on the mean and standard deviation of real-time market price signals within a preset sliding time window. The standard coefficient of variation is used to describe the endogenous relative volatility of market prices. The comprehensive risk factor is determined by multiplying the external shock amplification term by the standard coefficient of variation.

3. The power market transaction optimization decision-making system for microgrid assets according to claim 1, characterized in that, The strategy decision module calculates the trading strategy preference index, including: The expected net income is determined based on the difference between the potential market opportunity gains and the potential planning deviation penalty costs. The total economic size is determined by the sum of potential market opportunity gains and potential planning deviation penalty costs. Multiplying the total economic size by the comprehensive risk factor yields the risk-adjusted total economic size. The trading strategy preference index is obtained by calculating the ratio of expected net return to the risk-adjusted total economic size.

4. The power market transaction optimization decision-making system for microgrid assets according to claim 1, characterized in that, The weight generation module uses the logistic function as a non-linear mapping function to map the trading strategy tendency index to a preset range to generate a return weight coefficient and a deviation penalty weight coefficient; wherein the sum of the return weight coefficient and the deviation penalty weight coefficient is always 1.

5. The power market transaction optimization decision-making system for microgrid assets according to claim 1, characterized in that, The objective construction module constructs a real-time optimized objective function aimed at maximizing risk-adjusted return, including: The weighted market return is obtained by multiplying the return weighting coefficient by the expected actual market return generated after the scheduling action is performed. The weighted deviation penalty is obtained by multiplying the deviation penalty weighting coefficient by the expected actual deviation penalty generated after the scheduling action is executed. To maximize the difference between the weighted market return and the weighted bias penalty, a real-time optimization objective function is constructed with the goal of maximizing the risk-adjusted return.

6. The power market transaction optimization decision-making system for microgrid assets according to claim 5, characterized in that, Also includes: The risk prediction module is used to call a pre-trained probabilistic prediction model for any candidate scheduling action and output the probability density function of the actual planned power deviation caused by the action.

7. The power market transaction optimization decision-making system for microgrid assets according to claim 6, characterized in that, The risk prediction module is also used for: Based on the probability density function and the preset nonlinear penalty function, the expected deviation penalty cost corresponding to the candidate scheduling action is determined by integral operation, and the expected deviation penalty cost is used as the expected actual deviation penalty. The nonlinear penalty function is used to describe the relationship between penalty cost and deviation power.

8. The power market transaction optimization decision-making system for microgrid assets according to claim 1, characterized in that, Also includes: The scheduling mode decision module is used to compare the trading strategy preference index with the preset conservative mode threshold and aggressive mode threshold, and select the current scheduling mode from conservative mode, balanced mode and aggressive mode.

9. The power market transaction optimization decision-making system for microgrid assets according to claim 8, characterized in that, The specific operations of the scheduling mode decision module include: When the trading strategy tendency index is less than the conservative mode threshold, the conservative mode is triggered, and the return weight coefficient is set to zero, while the deviation penalty weight coefficient is set to one. When the trading strategy tendency index exceeds the aggressive mode threshold, aggressive mode is triggered, and the profit weight coefficient is set to one, while the deviation penalty weight coefficient is set to zero. In response to the trading strategy tendency index falling between the conservative mode threshold and the aggressive mode threshold, an equilibrium mode is triggered, and the return weight coefficient and deviation penalty weight coefficient generated by the weight generation module are used.

Images