A virtual power plant bidding optimization method and system considering bounded rationality

Abstract

The present application belongs to the technical field of market bidding decision, and particularly provides a virtual power plant bidding optimization method and system considering bounded rationality. The method comprises the following steps: S1: constructing a multi-dimensional anchor price; S2: dynamically adjusting the contribution weight of each component in the multi-dimensional anchor price; S3: realizing adaptive allocation of three-step bidding; S4: dynamically modifying the anchor price according to market clearing feedback information; S5: establishing a double-layer bidding optimization model; and S6: outputting the optimal bidding strategy of the virtual power plant. The present application constructs a multi-dimensional anchor price by comprehensively considering historical average price, marginal cost and competitor bidding, and introduces an adaptive weight mechanism, realizes adaptive allocation of bidding through a prospect theory gain-loss separation mechanism, and realizes dynamic updating of the anchor price through a Markov chain, and combines CVaR risk measurement to control tail loss risk, so that the bidding income and winning rate of the virtual power plant are effectively improved, and the uncertainty risk of the bidding strategy is reduced.

Description

Technical Field

[0001] This invention relates to the field of market bidding decision-making technology, specifically to a virtual power plant bidding optimization method and system that considers bounded rationality. Background Technology

[0002] With the continued rise in the installed capacity of renewable energy sources such as wind power and photovoltaics, virtual power plants, by coordinating and optimizing heterogeneous resources such as distributed power sources, energy storage systems, and controllable loads, have become an important way to promote the participation of new energy in the electricity market. In the current market, virtual power plants need to formulate scientific and reasonable bidding strategies to achieve a balance between profit optimization and risk avoidance.

[0003] In research on bidding strategies for virtual power plants in the electricity market, existing methods mostly model the behavior of market participants based on the assumption of perfect rationality, employing single-segment bidding or multi-segment bidding with a fixed proportion. However, actual decision-makers often exhibit bounded rationality characteristics, such as overestimating low-probability events, dependence on reference points, and loss aversion. The assumption of perfect rationality cannot accurately characterize the actual behavioral biases of decision-makers, leading to discrepancies between bidding strategies and actual market conditions, thus affecting the bidding returns and success rate of virtual power plants.

[0004] Regarding the pricing benchmark, existing methods typically use the predicted electricity price directly as the basis for pricing, without comprehensively considering multi-dimensional information such as historical market information, operating costs, and competitive landscape. Furthermore, the pricing benchmark is statically set and cannot adapt to the time-varying characteristics of the market environment, resulting in higher pricing risks.

[0005] In terms of volume allocation, existing methods mostly adopt single-segment volume allocation or fixed-proportion multi-segment volume allocation, which fails to dynamically adjust the volume allocation of each price segment according to market profit and loss expectations, thus limiting the bidding flexibility and profitability of virtual power plants under different market conditions. Summary of the Invention

[0006] The purpose of this section is to outline some aspects of the embodiments of the present invention and to briefly describe some preferred embodiments. Simplifications or omissions may be made in this section, as well as in the abstract and title of this application, to avoid obscuring the purpose of these documents; however, such simplifications or omissions should not be construed as limiting the scope of the invention.

[0007] To address the aforementioned technical problems, according to one aspect of the present invention, the present invention provides the following technical solution: A virtual power plant bidding optimization method considering bounded rationality includes the following steps: S1: Construct a multi-dimensional anchor price based on the anchoring effect, which combines the historical average market transaction price, the comprehensive marginal cost of virtual power plants, and the weighted bids of competitors, and use it as a reference benchmark for bid pricing; S2: Based on market volatility, an adaptive dynamic weighting mechanism is designed to dynamically adjust the contribution weight of each component in the multidimensional anchor price. S3: Based on prospect theory, a three-tiered reporting volume allocation mechanism with profit and loss separation is constructed. The profit and loss status is determined according to the deviation between the expected clearing price and the anchor price, so as to realize the adaptive allocation of the three-tiered reporting volume. S4: Based on the Markov chain, a dynamic update mechanism for the anchor price is constructed, and the anchor price is dynamically corrected according to the market clearing feedback information; S5: Establish a two-level bidding optimization model with CVaR risk-adjusted return maximization as the upper-level objective and market clearing and internal scheduling as the lower-level objectives; S6: The frost-ice optimization algorithm and the solver are used to solve the two-layer bidding optimization model and output the optimal bidding strategy for the virtual power plant.

[0008] As a preferred embodiment of the virtual power plant bidding optimization method considering bounded rationality described in this invention, the multi-dimensional anchored price in S1 includes three components: 1) The historical average market transaction price is expressed as: ; In the formula: For time period The historical average transaction price; For the first Daytime Market-clearing electricity prices; The length of the historical data window; 2) The comprehensive marginal cost of the virtual power plant is expressed as follows: ; In the formula: For time period The overall marginal cost of a virtual power plant; For the operation and maintenance costs of new energy infrastructure; This represents the cost ratio of gas turbines. The unit power generation cost of a gas turbine; For time period The scenery always contributes to its beauty; For time period The internal load of the virtual power plant; This is the output margin threshold; This is the load coverage threshold; 3) Competitor-weighted bid, its expression is: ; In the formula: For time period The competitor's weighted bid; It is the collection of other market participants in the market; and Participants During the period The estimated declared price and the estimated declared electricity volume.

[0009] As a preferred embodiment of the virtual power plant bidding optimization method considering bounded rationality described in this invention, the adaptive dynamic weighting mechanism in S2 is as follows: The anchor price is expressed as: ; In the formula: For time period The anchor price; , , These are the dynamic weighting coefficients for historical average price, marginal cost, and competitor quotations, respectively. Market volatility is defined as the coefficient of variation of historical prices: ; In the formula: For time period Market volatility.

[0010] As a preferred embodiment of the virtual power plant bidding optimization method considering bounded rationality described in this invention, the three-tiered bidding allocation mechanism in S3 is as follows: ; In the formula: For time period Price deviation; For time period The expected market clearing price; For time period The anchor price; Substituting the price deviation into the prospect theory value function: ; In the formula: For time period The value function value; This is the sensitivity decreasing coefficient, with a value range of (0,1); This is the loss aversion coefficient; Normalize the value function: ; In the formula: The value is the normalized value function value; This serves as the normalized benchmark for price deviations. Base score for each bid segment: ; In the formula: For time period No. The base score for the price range; For time period The Segment price; This is the width of the reference range for pricing. For the first The basic score bias of the segment satisfies ; Asymmetric adjustment of the base score based on profit and loss status: ; In the formula: The adjusted score; To adjust the strength coefficient; In a profitable state, the first The directional coefficient of the price range satisfies ; In a loss-making state, the first The directional coefficient of the price range satisfies and ; Normalized report volume ratios and report volumes for each segment: ; In the formula: For time period No. The proportion of reported volume for each price segment; For time period No. The declared electricity volume for each price range; For time period The total declared electricity volume.

[0011] As a preferred embodiment of the virtual power plant bidding optimization method considering bounded rationality described in this invention, the anchor price dynamic update mechanism in S4 is as follows: Define time period In the Markov state in the next iteration: ; In the formula: For time period In the Markov state in the next iteration; For time period In the The market clearing price of the next iteration; For time period In the The anchor price for the next iteration is set in state 1 to indicate a loss and state 2 to indicate a profit. Introducing continuous state counting to reflect the memory effect of Markov chains: ; In the formula: For time period In the Counting the consecutive states in each iteration; For maximum consecutive counts; Calculate the adjustment amount based on the current profit and loss status: ; In the formula: For time period In the The adjustment amount for the next iteration; For time period In the Price deviation in the next iteration; The base adjustment rate is for loss-making situations; Adjustment rate based on profitability; The loss aversion coefficient of the Markov chain; The relaxation factor is the aversion coefficient during periods of continuous losses. For continuous adjustments in the same state; Limit the adjustment amount: ; In the formula: This represents the maximum adjustment range in a single instance. Anchor price update formula: ; In the formula: The updated anchor price; To update the weights; This is the initial anchor price.

[0012] As a preferred embodiment of the virtual power plant bidding optimization method considering bounded rationality described in this invention, the two-layer bidding optimization model in S5 is as follows: 1) Upper-level virtual power plant bidding model: The upper-level model takes the virtual power plant operator as the decision-making entity, and the objective function is: ; In the formula: This is the risk-adjusted objective function value; For expected returns; A collection of benefits under different scenarios; This refers to the risk weighting coefficient. Confidence level Conditional Value at Risk (VaR); The expected return is the probability-weighted sum of the returns for each scenario: ; In the formula: Total number of scenes; For the scene The probability of occurrence; This represents the number of time periods in the scheduling cycle. For the scene Next period Market-clearing electricity prices; For the scene Next period The winning bid volume; For the scene Next period The internal operating costs of a virtual power plant; Within the discrete scenario framework, auxiliary variables are introduced to transform conditional value at risk into a linear constraint: ; In the formula: This is an estimate of the Value at Risk (VaR). For the scene Lower returns The deviation; For the scene The actual benefits; Confidence level; Upper-level decision variables include the total reported electricity volume for each time period. With three-stage pricing , , The following constraints must be met: ; In the formula: Minimum declared electricity volume; For time period Maximum schedulable capacity; , These are the lower and upper limits of the price quote, respectively. 2) Lower-level market clearing model: (Setting a time period) The system load requirement is After the bidding units are sorted by price, the first... The winning bid amount for each unit is: ; In the formula: For time period No. The winning bid volume of each bidding unit; For the first The declared electricity volume of each bidding unit; For time period The system load requirements; 3) Economic dispatch model within the lower-level virtual power plant: ; In the formula: This represents the total operating cost within the virtual power plant. , These are the output adjustment cost coefficients for wind power and solar power, respectively. , Time periods Adjustments in output for wind and solar power; For time period The price of natural gas; For time period Gas turbine output; For gas turbine power generation efficiency; This refers to the cost coefficient for energy storage charging and discharging losses. , Time periods Energy storage charging power and discharging power; , These are the scheduling cost coefficients for interruptible loads and time-shiftable loads, respectively. For time period Interruptible load reduction amount; For time period Adjustable load shifting amount.

[0013] As a preferred embodiment of the virtual power plant bidding optimization method considering bounded rationality described in this invention, the constraints of the internal economic dispatch model in S5 are as follows: 1) Power balance constraints: ; In the formula: , , Time periods Actual output of wind power, photovoltaic power, and gas turbines; For time period Virtual power plant internal base load; For time period Total winning bid volume for virtual power plants; 2) Wind and solar power output constraints: ; In the formula: For time period Upper limit of wind power output; For time period Photovoltaic power output limit; 3) Gas turbine output and ramping constraints: ; In the formula: For time period The gas turbine start-stop state variable takes a value of 0 or 1. , These represent the minimum and maximum output of the gas turbine, respectively. , These represent the maximum downward and upward ramp power of the gas turbine per unit time period; 4) Energy storage operation constraints: ; In the formula: For energy storage operating mode variables, Indicates the discharge mode. Indicates the charging mode; Indicates the state of charge; For time period The amount of energy stored at the end of the time; , These are the lower and upper limits of the state of charge, respectively; , These are charging efficiency and discharging efficiency, respectively. 5) Interruptible load and time-shiftable load constraints: ; In the formula: This represents the maximum proportion of interruptible load to base load. This represents the maximum proportion of time-shiftable loads to base loads.

[0014] As a preferred embodiment of the virtual power plant bidding optimization method considering bounded rationality described in this invention, the specific method for solving the two-layer bidding optimization model in step S6 using the frost-ice optimization algorithm in conjunction with the solver is as follows: Within the feasible space defined by the reported quantity and bid price, the positions of individual populations are randomly generated, and each individual corresponds to a complete set of bidding strategy vectors. The population is updated according to the update rules of the Frost Ice Optimization Algorithm. After the position update, constraint repair is performed. Based on the prospect theory mechanism, the total declared electricity is dynamically allocated into three segments of reported quantity, which are then assembled with the corresponding bid price to form a stepped bidding curve. The upper-level bidding strategy is passed to the lower level. First, market clearing is performed to output the clearing price and the winning bid volume of the virtual power plant for each time period. Then, the winning bid volume is used as a constraint to call the CPLEX solver to solve the internal optimization scheduling problem of the virtual power plant, and output the output allocation and operating cost of each device. All wind and solar scenarios are traversed to calculate the revenue of each scenario. The fitness value is calculated in combination with the conditional risk value. The global optimal solution is refreshed according to the greedy selection mechanism. At the end of each iteration, the lower-level clearing is performed once with the current global optimal strategy as input. The obtained clearing price is corrected according to the Markov chain update rule to adjust the anchor price. The iteration returns to the position update stage to continue until the maximum number of iterations is reached or convergence is achieved, and the final optimal bidding strategy is output.

[0015] A virtual power plant bidding optimization system considering bounded rationality includes: The anchor price construction module is used to construct a multi-dimensional anchor price based on the anchoring effect, the average historical market transaction price, the comprehensive marginal cost of virtual power plants, and the weighted bids of competitors. The adaptive weighting module is used to dynamically adjust the contribution weight of each component in the anchor price based on market volatility. The three-tiered reporting volume allocation module is used for the profit and loss separation mechanism based on prospect theory. It uses the anchor price as a reference point to determine the profit and loss status and realizes the adaptive allocation of the three-tiered reporting volume. The anchor price update module is used to dynamically adjust the anchor price based on market clearing feedback information using the Markov chain; The two-level optimization model module is used to establish a two-level bidding optimization model with maximizing CVaR risk-adjusted return as the upper-level objective and market clearing and internal scheduling as the lower-level objectives. The solver module is used to solve the two-layer bidding optimization model in collaboration with the frost-ice optimization algorithm and the CPLEX solver, and output the optimal bidding strategy for the virtual power plant.

[0016] Compared with existing technologies, the beneficial effects of this invention are as follows: By integrating the anchoring effect and prospect theory, the invention constructs a three-tiered bidding strategy that considers bounded rationality. It integrates historical average prices, marginal costs, and competitor quotations to construct a multi-dimensional anchoring price and introduces an adaptive weighting mechanism. Through the profit and loss separation mechanism of prospect theory, it achieves adaptive allocation of bidding volume and realizes dynamic updating of the anchoring price through Markov chains. Combined with CVaR risk measurement, it controls tail loss risk, effectively improving the bidding revenue and success rate of virtual power plants and reducing the uncertainty risk of bidding strategies. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and detailed embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 This is a flowchart illustrating the solution process in an embodiment of the present invention; Figure 2 This is a diagram of the day-ahead electricity market trading architecture in Embodiment 1 of the present invention; Figure 3 In Embodiment 1 of the present invention, the market operator sorts the quotations of each power generation entity from low to high to form a supply curve. Figure 4 This is a distribution diagram of the winning bid volume of various market entities under different scenarios in Embodiment 2 of the present invention; Figure 5 This is a breakdown diagram of the winning bids in the tiered bidding process of Embodiment 2 of the present invention; Figure 6 This is the internal power balance diagram of the virtual power plant in Embodiment 2 of the present invention; Figure 7 This is a diagram illustrating the dynamic anchor price update process in Embodiment 2 of the present invention. Detailed Implementation

[0018] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0019] Secondly, the present invention is described in detail with reference to the schematic diagrams. When detailing the embodiments of the present invention, for ease of explanation, the cross-sectional views illustrating the device structure may be partially enlarged, not according to the usual scale. Furthermore, the schematic diagrams are merely examples and should not limit the scope of protection of the present invention. In addition, actual fabrication should include three-dimensional spatial dimensions of length, width, and depth.

[0020] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0021] A virtual power plant bidding optimization method considering bounded rationality includes the following steps: A multi-dimensional anchor price is constructed based on the anchoring effect, which integrates historical market transaction average prices, the comprehensive marginal cost of virtual power plants, and the weighted bids of competitors, and serves as a reference benchmark for bid pricing. An adaptive dynamic weighting mechanism is designed based on market volatility to dynamically adjust the contribution weight of each component in the multidimensional anchor price. The multidimensional anchor price comprises three components: the average clearing price for each time period calculated based on historical transaction data; the comprehensive marginal cost of the virtual power plant determined according to segmented rules based on the supply and demand relationship between renewable energy output and load; the comprehensive bid from competitors calculated using a weighted average of bid prices; and a three-tiered volume allocation mechanism based on prospect theory that separates profit and loss, determining the profit and loss status based on the deviation between the expected clearing price and the anchor price, thereby achieving adaptive allocation of the three-tiered volume allocation. The adaptive dynamic weighting mechanism is described as follows: Market volatility is defined by the coefficient of variation of historical prices, and the market state is divided into three categories: high volatility, medium volatility, and low volatility. In the high volatility state, the weight of the historical average price is increased to play its role as a stabilizing anchor. In the low volatility state, the weight of marginal cost and competitor quotations is increased to pay more attention to its own costs and competitive situation. The final anchor price is the weighted sum of the three components: the historical market transaction average price, the comprehensive marginal cost of the virtual power plant, and the weighted quotation of competitors, according to the dynamic weights. The three-tiered bidding allocation mechanism is described as follows: Using the anchor price as a reference point, the price deviation is substituted into the prospective theoretical value function to determine the profit and loss status. This value function adopts a power function form in the profit domain and introduces a loss aversion coefficient in the loss domain to amplify negative effects. The basic winning bid score for each bidding segment relative to the anchor price is calculated, and the mid-price segment is given a higher basic weight bias. In a profitable state, the mid-price segment increases significantly, the low-price segment increases moderately, and the high-price segment decreases to lock in profits. In a loss-making state, the low-price segment increases significantly, and the high-price segment decreases sharply to avoid the risk of failed bids. The adjusted scores are normalized to the bidding volume ratio of each segment. A dynamic update mechanism for the anchor price is constructed based on Markov chains, and the anchor price is dynamically adjusted according to market clearing feedback information; The dynamic update mechanism for the anchor price defines the profit and loss status based on the relationship between the clearing price and the anchor price. It introduces continuous state counting to reflect the memory effect, incrementing the count while the state is maintained and resetting it when the state changes. The adjustment amount is calculated and limited based on the state and continuous count. The update formula integrates the current correction value with the initial anchor price to prevent excessive deviation. Establish a two-level bidding optimization model with maximizing CVaR risk-adjusted return as the upper-level objective and market clearing and internal scheduling as the lower-level objectives. The two-tier bidding optimization model is described as follows: The objective function of the upper-level model is the expected return minus the risk weight coefficient multiplied by the conditional value at risk (CVaR). The expected return is the probability-weighted sum of the returns for each scenario. The CVaR is transformed into a linear constraint by introducing auxiliary variables in the discrete scenario framework. The upper-level decision variables include the total declared electricity volume for each time period and the three-stage bidding price, satisfying the upper and lower limits of the declared volume and the increment and range constraints of the bidding price. The lower-level model includes a market clearing model and a virtual power plant internal economic dispatch model. The market clearing model uses a queuing method to call the bidding units one by one according to the bidding price from low to high to determine the winning bid volume and the marginal clearing price. The three-stage bidding curve of the virtual power plant is split into three independent bidding units to participate in the ranking and bidding. The internal economic dispatch model aims to minimize operating costs and coordinates the output of various resources under the premise of satisfying power balance constraints, wind power and photovoltaic output constraints, gas turbine output and ramping constraints, energy storage operation constraints, interruptible load constraints, and time-shiftable load constraints. The dual-layer bidding optimization model is solved in collaboration with the frost-ice optimization algorithm and the solver, and the optimal bidding strategy for the virtual power plant is output. The two-layer model solution method is described as follows: The Frost Ice Optimization Algorithm is used to search for the upper-layer bidding strategy. Each particle is encoded as a strategy vector containing the total declared electricity volume for each time period and the three-segment price. The lower layer calls the CPLEX solver to solve the market clearing and internal scheduling subproblems. At the end of each iteration, the current global optimal strategy is used as input to perform a lower-layer clearing once, and the resulting clearing price is used to correct the anchor price according to the Markov rule. The two layers alternate iterating until convergence, and the final optimal bidding strategy is output.

[0022] A virtual power plant bidding optimization system considering bounded rationality includes: The anchor price construction module is used to construct a multi-dimensional anchor price based on the anchoring effect, the average historical market transaction price, the comprehensive marginal cost of virtual power plants, and the weighted bids of competitors. The adaptive weighting module is used to dynamically adjust the contribution weight of each component in the anchor price based on market volatility. The three-tiered reporting volume allocation module is used for the profit and loss separation mechanism based on prospect theory. It uses the anchor price as a reference point to determine the profit and loss status and realizes the adaptive allocation of the three-tiered reporting volume. The anchor price update module is used to dynamically adjust the anchor price based on market clearing feedback information using the Markov chain; The two-level optimization model module is used to establish a two-level bidding optimization model with maximizing CVaR risk-adjusted return as the upper-level objective and market clearing and internal scheduling as the lower-level objectives. The solver module is used to solve the two-layer bidding optimization model in collaboration with the frost-ice optimization algorithm and the CPLEX solver, and output the optimal bidding strategy for the virtual power plant.

[0023] Example 1 Reference Figures 1 to 3 ,in Figure 2 The document demonstrates the transaction process where various market participants on the power generation side submit bids and declared electricity volumes to the market operator, and the winning bid volume and clearing price are determined after market clearing.

[0024] A virtual power plant bidding optimization method considering bounded rationality includes, Step 1: Construct a multidimensional anchored price based on the anchoring effect.

[0025] The anchor price is constructed by comprehensively considering three dimensions: historical market information, its own operating costs, and the competitive landscape. Specifically, it includes the following three components: The average clearing price for each period, calculated based on historical transaction data, is used as the first-dimensional anchoring component. Its expression is as follows: ; In the formula: For time period The historical average transaction price; For the first Daytime Market-clearing electricity prices; This is the length of the historical data window.

[0026] Considering that the marginal cost of a virtual power plant changes dynamically with the relationship between renewable energy output and load supply and demand, the operating cost of a virtual power plant is determined according to the following segmented rules: ; In the formula: For time period The overall marginal cost of a virtual power plant; For the operation and maintenance costs of new energy infrastructure; This represents the cost ratio of gas turbines. The unit power generation cost of a gas turbine; For time period The scenery always contributes to its beauty; For time period The internal load of the virtual power plant; This is the output margin threshold; This is the load coverage threshold.

[0027] Considering the differences in market influence among competitors, a weighted average of declared electricity volume is used to calculate the comprehensive price quote for competitors, and its expression is as follows: ; In the formula: For time period The competitor's weighted bid; It is the collection of other market participants in the market; and Participants During the period The estimated declared price and the estimated declared electricity volume.

[0028] Step 2: Design an adaptive dynamic weighting mechanism.

[0029] In step 1, the weights of the three price components in relation to the anchor price are adaptively adjusted based on market volatility. The final anchor price is expressed as: ; In the formula: For time period The anchor price; , , These are the dynamic weighting coefficients for historical average price, marginal cost, and competitor quotations, respectively. These dynamic weighting coefficients satisfy... .

[0030] Market volatility is defined as the coefficient of variation of historical prices: ; In the formula: For time period Market volatility. Market states are categorized into three types and corresponding weights are determined: when... When the time is in a high-fluctuation state, take , , ;when When the time is in a state of fluctuation, take , , ;when When the time is in a low-fluctuation state, take , , .

[0031] Step 3: Construct a three-tiered profit and loss separation mechanism for volume distribution based on prospect theory.

[0032] Using the anchored prices established in steps 1 and 2 as reference points in prospect theory, the time period is calculated. Price deviation: ; In the formula: For time period Price deviation; For time period The expected market clearing price; For time period The anchor price.

[0033] Substituting price deviations into the value function of prospect theory: ; In the formula: For time period The value function value; This is the sensitivity decreasing coefficient, with a value range of (0,1); is the loss aversion coefficient. When... At that time, the virtual power plant was in a state of expected profitability; when At that time, virtual power plants face the risk of losses.

[0034] Normalize the value function: ; In the formula: The value is the normalized value function value; This serves as the normalized benchmark for price deviations.

[0035] The virtual power plant adopts a three-tiered bidding strategy, allocating the total declared electricity volume according to low, medium, and high price segments, with each segment's bid meeting certain requirements. Calculate the base winning bid score for each bid segment relative to the anchor price: ; In the formula: For time period No. The base score for the price range; For time period The Segment price; This is the width of the reference range for pricing. For the first The base score bias for the segment. Since the median price segment is in a relatively balanced position between profit and winning bid, the bias parameter is set as follows: This assigns a higher base weight to the mid-price segment.

[0036] The base score is asymmetrically adjusted based on profit and loss status. Profit and loss statuses are as follows: ; In the formula: The adjusted score; To adjust the strength coefficient; In a profitable state, the first The directional coefficient of the price range satisfies ; In a loss-making state, the first The directional coefficient of the price range satisfies and .

[0037] After status adjustment, the scores are normalized into the reporting ratio and the final reporting volume for each segment: ; In the formula: For time period No. The reported volume ratio for each price segment meets the requirements. ; For time period No. The declared electricity volume for each price range; For time period The total declared electricity volume.

[0038] Step 4: Construct a dynamic update mechanism for the anchor price based on the Markov chain.

[0039] In the iterative solution of the two-level optimization model, the clearing price returned by the lower-level market clearing provides feedback signals for the dynamic adjustment of the anchor price. (Definition of time period) In the Markov state in the next iteration: ; In the formula: For time period In the Markov state in the next iteration; For time period In the The market clearing price of the next iteration; For time period In the The anchor price for the next iteration. State 1 represents a loss, and state 2 represents a profit.

[0040] Introducing continuous state counting to reflect the memory effect of Markov chains: ; In the formula: For time period In the Counting the consecutive states in each iteration; This is the maximum continuous count. The count increments while the state is maintained and resets when the state changes.

[0041] Calculate the adjustment amount based on the current profit and loss status: ; In the formula: For time period In the The adjustment amount for the next iteration; For time period In the Price deviation in the next iteration; The base adjustment rate is for loss-making situations; Adjustment rate based on profitability; The loss aversion coefficient of the Markov chain; The relaxation factor is the aversion coefficient during periods of continuous losses. This represents the adjustment rate increment for consecutive states.

[0042] To prevent excessive adjustments in a single instance, the adjustment amount is limited: ; In the formula: This represents the maximum adjustment range in a single instance.

[0043] Based on the update factor, the formula for updating the anchor price is: ; In the formula: The updated anchor price; To update the weights; This serves as the initial anchor price. Introducing an initial anchor fusion term prevents the anchor price from deviating excessively from a reasonable range during iteration.

[0044] Step 5: Establish a two-tier bidding optimization model.

[0045] (1) Bidding model for upper-level virtual power plants.

[0046] The upper-level model takes virtual power plant operators as the decision-making entities, and the objective function is to maximize risk-adjusted returns. ; In the formula: This is the risk-adjusted objective function value; For expected returns; A collection of benefits under different scenarios; This refers to the risk weighting coefficient. Confidence level Conditional Value at Risk (VaR).

[0047] The expected return is the probability-weighted sum of the returns for each scenario: ; In the formula: Total number of scenes; For the scene The probability of occurrence; This represents the number of time periods in the scheduling cycle. For the scene Next period Market-clearing electricity prices; For the scene Next period The winning bid volume; For the scene Next period The internal operating costs of the virtual power plant.

[0048] Within the discrete scenario framework, auxiliary variables are introduced to transform conditional value at risk into a linear constraint: ; In the formula: This is an estimate of the Value at Risk (VaR). For the scene Lower returns The deviation; For the scene The actual benefits; The confidence level.

[0049] Upper-level decision variables include the total reported electricity volume for each time period. With three-stage pricing , , The following constraints must be met: ; In the formula: Minimum declared electricity volume; For time period Maximum schedulable capacity; , These represent the lower and upper limits of the price quote, respectively.

[0050] (2) Lower-level market clearing model.

[0051] like Figure 3 As shown, the market operator ranks the bids of various power generators from low to high to form a supply curve, and determines the marginal clearing price at the intersection with the system load demand. (Time period is defined.) The system load requirement is After the bidding units are sorted by price, the first... The winning bid amount for each unit is: ; In the formula: For time period No. The winning bid volume of each bidding unit; For the first The declared electricity volume of each bidding unit; For time period The system load demand is considered. The clearing process terminates when the cumulative winning bids first reach or exceed the system demand, and the marginal clearing price is the bid price of the last winning bidder. The three-tiered bidding curve of the virtual power plant is split into three independent bidding units for ranked bidding.

[0052] (3) Economic dispatch model of the lower-level virtual power plant.

[0053] After market clearing, the internal economic dispatch of the virtual power plant aims to minimize operating costs, and its objective function is: ; In the formula: This represents the total operating cost within the virtual power plant. , These are the output adjustment cost coefficients for wind power and solar power, respectively. , Time periods Adjustments in output for wind and solar power; For time period The price of natural gas; For time period Gas turbine output; For gas turbine power generation efficiency; This refers to the cost coefficient for energy storage charging and discharging losses. , Time periods Energy storage charging power and discharging power; , These are the scheduling cost coefficients for interruptible loads and time-shiftable loads, respectively. For time period Interruptible load reduction amount; For time period Adjustable load shifting amount.

[0054] Internal scheduling must meet the following constraints: Power balance constraints: ; In the formula: , , Time periods Actual output of wind power, photovoltaic power, and gas turbines; For time period Virtual power plant internal base load; For time period Total winning bid volume for virtual power plants.

[0055] Wind and solar power output constraints: ; In the formula: For time period Upper limit of wind power output; For time period Photovoltaic power output limit.

[0056] Gas turbine output and ramping constraints: ; In the formula: For time period The gas turbine start-stop state variable takes a value of 0 or 1. , These represent the minimum and maximum output of the gas turbine, respectively. , These represent the maximum downward and upward ramp power of the gas turbine per unit time period, respectively.

[0057] Energy storage operation constraints: ; In the formula: For energy storage operating mode variables, Indicates the discharge mode. Indicates the charging mode; Indicates the state of charge; For time period The amount of energy stored at the end of the time; , These are the lower and upper limits of the state of charge, respectively; , These are charging efficiency and discharging efficiency, respectively.

[0058] Interruptible load and time-shiftable load constraints: ; In the formula: This represents the maximum proportion of interruptible load to base load. This represents the maximum proportion of time-shiftable loads to base loads.

[0059] Step 6: Solve the two-layer bidding optimization model.

[0060] The upper layer uses the Frost-Ice Optimization Algorithm to search for bidding strategies. Each particle is encoded as a strategy vector containing the total declared electricity volume for each time period and three price segments. The fitness function is CVaR risk-adjusted return. The lower layer completes market clearing and calls the CPLEX solver to solve the internal scheduling subproblem.

[0061] The specific solution process includes the following steps: 1) Randomly generate the positions of individuals in the population within the feasible space specified by the quantity and price, and each individual corresponds to a complete set of bidding strategy vectors.

[0062] 2) The population updates its position according to the frost-ice optimization algorithm update rules. After the position update, constraint repair is performed. According to step 3, the total declared electricity is dynamically allocated into three segments and assembled with the corresponding bid price into a stepped bidding curve.

[0063] 3) Pass the upper-level bidding strategy to the lower level. First, execute the market clearing to output the clearing price and the winning bid volume of the virtual power plant for each time period. Then, use the winning bid volume as a constraint to solve the internal optimization scheduling problem of the virtual power plant and output the output allocation and operating cost of each device.

[0064] 4) Traverse all scenic scenes to calculate the revenue of each scene, calculate the fitness value in combination with conditional risk value, and determine whether to refresh the global optimal solution based on the greedy selection mechanism.

[0065] 5) At the end of each iteration, a lower-level clearing is performed with the current global optimal strategy as input, and the resulting cleared electricity price is used to correct the anchor price according to the Markov chain update rule in step 4.

[0066] 6) Return to step 2) Continue iterating until the maximum number of iterations is reached or the improvement of the optimal solution for several consecutive generations is lower than the convergence threshold, and output the final optimal bidding strategy.

[0067] The above is an illustrative scheme of a three-tiered bidding optimization method for virtual power plants participating in the day-ahead market considering bounded rationality, according to this embodiment. It should be noted that the technical solution of this three-tiered bidding optimization system for virtual power plants participating in the day-ahead market considering bounded rationality belongs to the same concept as the technical solution of the aforementioned three-tiered bidding optimization method for virtual power plants participating in the day-ahead market considering bounded rationality. Details not described in detail in the technical solution of the three-tiered bidding optimization system for virtual power plants participating in the day-ahead market considering bounded rationality in this embodiment can be found in the description of the technical solution of the aforementioned three-tiered bidding optimization method for virtual power plants participating in the day-ahead market considering bounded rationality.

[0068] This embodiment considers a three-tiered bidding optimization system for virtual power plants participating in the day-ahead market based on bounded rationality, including: The anchor price construction module is used to construct a multi-dimensional anchor price based on the anchoring effect, the average historical market transaction price, the comprehensive marginal cost of virtual power plants, and the weighted bids of competitors. The adaptive weighting module is used to dynamically adjust the contribution weight of each component in the anchor price based on market volatility. The three-tiered reporting volume allocation module is used for the profit and loss separation mechanism based on prospect theory. It uses the anchor price as a reference point to determine the profit and loss status and realizes the adaptive allocation of the three-tiered reporting volume. The anchor price update module is used to dynamically adjust the anchor price based on market clearing feedback information using the Markov chain; The two-level optimization model module is used to establish a two-level bidding optimization model with maximizing CVaR risk-adjusted return as the upper-level objective and market clearing and internal scheduling as the lower-level objectives. The solver module is used to solve the two-layer bidding optimization model in collaboration with the frost-ice optimization algorithm and the CPLEX solver, and output the optimal bidding strategy for the virtual power plant.

[0069] Example 2 Reference Figures 4 to 7 Based on the previous embodiment, this embodiment provides an application example and a comparative example of a three-tiered bidding optimization method for virtual power plants participating in the day-ahead market that considers bounded rationality.

[0070] The virtual power plant in this embodiment is composed of wind power, photovoltaic power, gas turbines, energy storage, and flexible loads. The virtual power plant participates in day-ahead electricity market bidding as a price setter, and its bidding behavior will affect the market clearing price. The operating parameters of the internal resources of the virtual power plant are shown in Table 1.

[0071] Table 1. Internal Resource Configuration Parameters of the Virtual Power Plant

[0072] Other participants in the market include a 350MW thermal power plant, a 120MW wind-storage combined power plant, a 200MW microgrid, and a 70MW small virtual power plant. The electricity market clears in 1-hour increments, with a total of 24 time slots throughout the day.

[0073] To address the uncertainty in wind and solar power output, a generative adversarial network (GAN) approach was used to generate 1000 initial wind and solar power output scenarios, which were then reduced to 10 typical scenarios using K-means clustering. The relevant parameter settings in the behavioral decision-making model are shown in Table 2.

[0074] Table 2. Relevant parameters in the behavioral decision-making model

[0075] Based on the above parameter settings, the parameter values ​​in Tables 1 and 2 are input into the two-layer bidding optimization model established in Example 1. The Frost Ice Optimization Algorithm is used to search for the upper-layer bidding strategy, with the algorithm parameters set to population size 30, maximum number of iterations 300, and position update scaling factor 5; the lower layer calls the CPLEX solver to solve the market clearing and internal scheduling subproblems.

[0076] To verify the effectiveness of each module of the bidding optimization method proposed in Example 1, the following five comparative scenarios were analyzed based on the idea of ​​ablation experiments. Scenario 1: The virtual power plant directly uses the predicted electricity price as the bid and participates in market bidding with a single-segment quantity bid, serving as the benchmark scenario; Scenario 2: Based on Scenario 1, the bidding method is replaced with a static single-segment bid constructed based on the anchoring effect, while the quantity bid remains single-segment; Scenario 3: Based on Scenario 2, a Markov chain is introduced to dynamically update the anchor price, while the bid quantity bid remains single-segment; Scenario 4: Based on Scenario 3, the single-segment bid quantity bid is replaced with a three-tier structure, with fixed quantities bid in the low, medium, and high segments; Scenario 5: Based on Scenario 4, a profit and loss separation mechanism based on prospect theory is introduced, replacing the fixed proportion of the three-tier quantity bid with an adaptive allocation based on profit and loss expectations. Under the same parameter settings, the optimization scheduling comparison results for the five scenarios are shown in Table 3.

[0077] Table 3 Comparison of Market Clearing Results in Different Scenarios

[0078] As shown in Table 3, from Scenario 1 to Scenario 5, the expected revenue of the virtual power plant gradually increased from RMB 321,900 to RMB 514,900, and the success rate increased from 49.89% to 98.73%. Scenario 2, after introducing the anchoring effect, saw an increase of 19.3% in expected revenue, but also the highest standard deviation. Scenario 3, after introducing dynamic updates, reduced the standard deviation by 83.5%. Scenario 4, after introducing a three-tiered structure, saw the success rate jump to 84.33%. Scenario 5, after introducing profit and loss separation, saw an increase of 59.9% in expected revenue compared to the benchmark scenario.

[0079] The distribution of winning bids for each market participant under different bidding strategies at different time periods was obtained, such as... Figure 4 The diagram (where (a), (b), and (c) represent the distribution of winning bids for each market participant in different time periods under scenarios 1, 4, and 5, respectively) illustrates the market share of virtual power plants and other participants such as thermal power plants, wind-storage combined power plants, and microgrids across the 24 time periods throughout the day. Figure 4 It can be seen that when using the baseline strategy, virtual power plants only obtain a limited share of the bids during certain periods, while thermal power plants occupy the vast majority of the market space during peak periods. After adopting the three-tiered bidding strategy, the share of the bids won by virtual power plants in each period has increased significantly; after further introducing profit and loss separation, virtual power plants occupy a considerable share in all 24 periods, demonstrating the structural advantages of tiered bidding.

[0080] The decomposition of the winning bid volume in the tiered bidding process under the optimal strategy, as described in Example 1, is as follows: Figure 5 The image shows the allocation of winning bids for electricity across three price tiers—low-price, medium-price, and high-price—during different time periods throughout the day. Figure 5It can be seen that the low-price segment maintains a stable number of successful bids throughout the day, playing a role in ensuring a minimum cost; the mid-price segment, as the main source of profit, accounts for a relatively high proportion during peak electricity consumption periods; the high-price segment has a larger number of successful bids during periods of high electricity prices and a smaller number of successful bids during periods of low electricity prices in the early morning, reflecting the strategy's adaptive response to market prices.

[0081] The power balance within the virtual power plant under the optimal strategy, as described in Example 1, is obtained. Figure 6 The diagram shows the power output distribution of various resources, including wind power, solar power, gas turbines, energy storage, and flexible loads, throughout the day. Figure 6 It can be seen that wind power provides stable output throughout the day, forming the main source of electricity; photovoltaic power contributes incremental output during the daytime; gas turbines operate almost all day to fill the gap between wind and solar power output and the commitments made in the bidding; and energy storage charging and discharging occur in small amounts, undertaking minor power regulation during different time periods. The power remains balanced throughout the day, indicating that the commitments made in the bidding are indeed feasible in terms of resource allocation.

[0082] The update process of the anchor price after Markov chain correction in the method proposed in Example 1 is obtained, as follows: Figure 7 The diagram shows the offset and evolution trajectory between the initial and final anchor prices after Markov chain correction. (This is from...) Figure 7 It can be seen that the initial and final anchoring curves are highly consistent, with the offset being relatively limited in most periods. The adjustment is sometimes upward and sometimes downward, demonstrating a two-way correction characteristic. The average adjustment is only 0.004 yuan / kWh, ensuring the stability of the pricing benchmark and avoiding strategy oscillations caused by excessive correction.

[0083] Based on the above analysis results, the three-tiered bidding optimization method for virtual power plants considering bounded rationality proposed in Example 1 has significant advantages in terms of bidding revenue, success rate, and strategy robustness through the synergistic effects of multi-dimensional anchor price construction, adaptive weight adjustment, prospect theory profit and loss separation bidding quantity allocation, and Markov chain dynamic updates.

[0084] Although the present invention has been described above with reference to embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of the invention. In particular, as long as there is no structural conflict, the features in the disclosed embodiments can be combined with each other in any manner. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, the present invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A virtual power plant bidding optimization method considering bounded rationality, characterized in that, Includes the following steps: S1: Construct a multi-dimensional anchor price based on the anchoring effect, which combines the historical average market transaction price, the comprehensive marginal cost of virtual power plants, and the weighted bids of competitors, and use it as a reference benchmark for bid pricing; S2: Based on market volatility, an adaptive dynamic weighting mechanism is designed to dynamically adjust the contribution weight of each component in the multidimensional anchor price. S3: Based on prospect theory, a three-tiered reporting volume allocation mechanism with profit and loss separation is constructed. The profit and loss status is determined according to the deviation between the expected clearing price and the anchor price, so as to realize the adaptive allocation of the three-tiered reporting volume. S4: Based on the Markov chain, a dynamic update mechanism for the anchor price is constructed, and the anchor price is dynamically corrected according to the market clearing feedback information; S5: Establish a two-level bidding optimization model with CVaR risk-adjusted return maximization as the upper-level objective and market clearing and internal scheduling as the lower-level objectives; S6: The frost-ice optimization algorithm and the solver are used to solve the two-layer bidding optimization model and output the optimal bidding strategy for the virtual power plant.

2. The virtual power plant bidding optimization method considering bounded rationality according to claim 1, characterized in that, The multidimensional anchored price in S1 includes three components: 1) The historical average market transaction price is expressed as: ； In the formula: For time period The historical average transaction price; For the first Daytime Market-clearing electricity prices; The length of the historical data window; 2) The comprehensive marginal cost of the virtual power plant is expressed as follows: ； In the formula: For time period The overall marginal cost of a virtual power plant; For the operation and maintenance costs of new energy infrastructure; This represents the cost ratio of gas turbines. The unit power generation cost of a gas turbine; For time period The scenery always contributes to its beauty; For time period The internal load of the virtual power plant; This is the output margin threshold; This is the load coverage threshold; 3) Competitor-weighted bid, its expression is: ； In the formula: For time period The competitor's weighted bid; It is the collection of other market participants in the market; and Participants During the period The estimated declared price and the estimated declared electricity volume.

3. The virtual power plant bidding optimization method considering bounded rationality according to claim 1, characterized in that, The adaptive dynamic weighting mechanism in S2 is as follows: The anchor price is expressed as: ； In the formula: For time period The anchor price; , , These are the dynamic weighting coefficients for historical average price, marginal cost, and competitor quotations, respectively. Market volatility is defined as the coefficient of variation of historical prices: ； In the formula: For time period Market volatility.

4. The virtual power plant bidding optimization method considering bounded rationality according to claim 1, characterized in that, The three-tiered reporting allocation mechanism in S3 is as follows: ； In the formula: For time period Price deviation; For time period The expected market clearing price; For time period The anchor price; Substituting the price deviation into the prospect theory value function: ； In the formula: For time period The value function value; This is the sensitivity decreasing coefficient, with a value range of (0,1); This is the loss aversion coefficient; Normalize the value function: ； In the formula: The value is the normalized value function value; This serves as the normalized benchmark for price deviations. Base score for each bid segment: ； In the formula: For time period No. The base score for the price range; For time period The Segment price; This is the width of the reference range for pricing. For the first The basic score bias of the segment satisfies ; Asymmetric adjustment of the base score based on profit and loss status: ； In the formula: The adjusted score; To adjust the strength coefficient; In a profitable state, the first The directional coefficient of the price range satisfies ; In a loss-making state, the first The directional coefficient of the price range satisfies and ; Normalized report volume ratios and report volumes for each segment: ； In the formula: For time period No. The proportion of reported volume for each price segment; For time period No. The declared electricity volume for each price range; For time period The total declared electricity volume.

5. The virtual power plant bidding optimization method considering bounded rationality according to claim 4, characterized in that, The dynamic update mechanism for the anchor price in S4 is as follows: Define time period In the Markov state in the next iteration: ； In the formula: For time period In the Markov state in the next iteration; For time period In the The market clearing price of the next iteration; For time period In the The anchor price for the next iteration is set in state 1 to indicate a loss and state 2 to indicate a profit. Introducing continuous state counting to reflect the memory effect of Markov chains: ； In the formula: For time period In the Counting the consecutive states in each iteration; For maximum consecutive counts; Calculate the adjustment amount based on the current profit and loss status: ； In the formula: For time period In the The adjustment amount for the next iteration; For time period In the Price deviation in the next iteration; The base adjustment rate is for loss-making situations; Adjustment rate based on profitability; The loss aversion coefficient of the Markov chain; The relaxation factor is the aversion coefficient during periods of continuous losses. For continuous adjustments in the same state; Limit the adjustment amount: ； In the formula: This represents the maximum adjustment range in a single instance. Anchor price update formula: ； In the formula: The updated anchor price; To update the weights; This is the initial anchor price.

6. The virtual power plant bidding optimization method considering bounded rationality according to claim 1, characterized in that, The two-level bidding optimization model in S5 is as follows: 1) Upper-level virtual power plant bidding model: The upper-level model takes the virtual power plant operator as the decision-making entity, and the objective function is: ； In the formula: This is the risk-adjusted objective function value; For expected returns; A collection of benefits under different scenarios; This refers to the risk weighting coefficient. Confidence level Conditional Value at Risk (VaR); The expected return is the probability-weighted sum of the returns for each scenario: ； In the formula: Total number of scenes; For the scene The probability of occurrence; This represents the number of time periods in the scheduling cycle. For the scene Next period Market-clearing electricity prices; For the scene Next period The winning bid volume; For the scene Next period The internal operating costs of a virtual power plant; Within the discrete scenario framework, auxiliary variables are introduced to transform conditional value at risk into a linear constraint: ； In the formula: This is an estimate of the Value at Risk (VaR). For the scene Lower returns The deviation; For the scene The actual benefits; Confidence level; Upper-level decision variables include the total reported electricity volume for each time period. With three-stage pricing , , The following constraints must be met: ； In the formula: Minimum declared electricity volume; For time period Maximum schedulable capacity; , These are the lower and upper limits of the price quote, respectively. 2) Lower-level market clearing model: (Setting a time period) The system load requirement is After the bidding units are sorted by price, the first... The winning bid amount for each unit is: ； In the formula: For time period No. The winning bid volume of each bidding unit; For the first The declared electricity volume of each bidding unit; For time period System load requirements 3) Economic dispatch model within the lower-level virtual power plant: ； In the formula: This represents the total operating cost within the virtual power plant. , These are the output adjustment cost coefficients for wind power and solar power, respectively. , Time periods Adjustments in output for wind and solar power; For time period The price of natural gas; For time period Gas turbine output; For gas turbine power generation efficiency; This refers to the cost coefficient for energy storage charging and discharging losses. , Time periods Energy storage charging power and discharging power; , These are the scheduling cost coefficients for interruptible loads and time-shiftable loads, respectively. For time period Interruptible load reduction amount; For time period Adjustable load shifting amount.

7. The virtual power plant bidding optimization method considering bounded rationality according to claim 1, characterized in that, The constraints of the internal economic scheduling model in S5 are as follows: 1) Power balance constraints: ； In the formula: , , Time periods Actual output of wind power, photovoltaic power, and gas turbines; For time period Virtual power plant internal base load; For time period Total winning bid volume for virtual power plants; 2) Wind and solar power output constraints: ； In the formula: For time period Upper limit of wind power output; For time period Photovoltaic power output limit; 3) Gas turbine output and ramping constraints: ； In the formula: For time period The gas turbine start-stop state variable takes a value of 0 or 1. , These represent the minimum and maximum output of the gas turbine, respectively. , These represent the maximum downward and upward ramp power of the gas turbine per unit time period; 4) Energy storage operation constraints: ； In the formula: For energy storage operating mode variables, Indicates the discharge mode. Indicates the charging mode; Indicates the state of charge; For time period The amount of energy stored at the end of the time; , These are the lower and upper limits of the state of charge, respectively; , These are charging efficiency and discharging efficiency, respectively. 5) Interruptible load and time-shiftable load constraints: ； In the formula: This represents the maximum proportion of interruptible load to base load. This represents the maximum proportion of time-shiftable loads to base loads.

8. The virtual power plant bidding optimization method considering bounded rationality according to claim 5, characterized in that, The specific method for using the frost-ice optimization algorithm and solver in S6 to solve the two-layer bidding optimization model is as follows: Within the feasible space defined by the reported quantity and bid price, the positions of individual populations are randomly generated, and each individual corresponds to a complete set of bidding strategy vectors. The population is updated according to the update rules of the Frost Ice Optimization Algorithm. After the position update, constraint repair is performed. Based on the prospect theory mechanism, the total declared electricity is dynamically allocated into three segments of reported quantity, which are then assembled with the corresponding bid price to form a stepped bidding curve. The upper-level bidding strategy is passed to the lower level. First, the market clearing is performed to output the clearing price and the winning bid quantity of the virtual power plant for each time period. Then, the winning bid quantity is used as a constraint to call the CPLEX solver to solve the internal optimization scheduling problem of the virtual power plant, and output the output allocation and operating cost of each device. The algorithm iterates through all scenic scenes to calculate the revenue of each scene, calculates the fitness value based on the conditional risk value, and determines whether to refresh the global optimal solution based on the greedy selection mechanism. At the end of each iteration, the algorithm performs a lower-level clearing once with the current global optimal strategy as input, and corrects the anchor price according to the Markov chain update rule based on the cleared electricity price. The return position update phase continues iterating until the maximum number of iterations is reached or convergence occurs, at which point the final optimal bidding strategy is output.

9. A virtual power plant bidding optimization system considering bounded rationality, used to implement the virtual power plant bidding optimization method considering bounded rationality as described in any one of claims 1-8, characterized in that, include: The anchor price construction module is used to construct a multi-dimensional anchor price based on the anchoring effect, the average historical market transaction price, the comprehensive marginal cost of virtual power plants, and the weighted bids of competitors. The adaptive weighting module is used to dynamically adjust the contribution weight of each component in the anchor price based on market volatility. The three-tiered reporting volume allocation module is used for the profit and loss separation mechanism based on prospect theory. It uses the anchor price as a reference point to determine the profit and loss status and realizes the adaptive allocation of the three-tiered reporting volume. The anchor price update module is used to dynamically adjust the anchor price based on market clearing feedback information using the Markov chain; The two-level optimization model module is used to establish a two-level bidding optimization model with maximizing CVaR risk-adjusted return as the upper-level objective and market clearing and internal scheduling as the lower-level objectives. The solver module is used to solve the two-layer bidding optimization model in collaboration with the frost-ice optimization algorithm and the CPLEX solver, and output the optimal bidding strategy for the virtual power plant.

Images