Virtual power plant bidding strategy optimization method based on chaotic game optimization algorithm

Abstract

The invention relates to a virtual power plant bidding strategy optimization method based on a chaotic game optimization algorithm. The method comprises the following steps: generating an initial bidding strategy of a virtual power plant through chaotic mapping; each virtual power plant receives feedback information of the market, games by considering factors of cost, income, fluctuation of photovoltaic power generation, scheduling cost and attenuation of an energy storage battery and service life of equipment, and adjusts a bidding strategy of the virtual power plant; punishment income is applied to bidding behaviors violating market ethics or benefit distribution constraints; calculating the income of the power plant according to the adjusted strategy, and updating the bidding strategy of each virtual power plant; and repeating the game process until a convergence condition is reached, and outputting a virtual power plant bidding strategy. Compared with the prior art, the bidding strategy of the virtual power plant is more flexible and adaptive by using the randomness and unpredictability of the chaotic system, and meanwhile, the fairness, sustainability and healthy market development of the system are ensured by introducing ethical constraint and equipment life management.

Description

Technical Field

[0001] The present invention relates to the technical field of virtual power plant bidding strategy optimization, and in particular to a virtual power plant bidding strategy optimization method based on a chaos game optimization algorithm. Background Art

[0002] A virtual power plant (VPP) is a system that aggregates and optimizes distributed energy resources (DERs) such as solar photovoltaics, wind power generation, energy storage systems, dispatchable loads, and micro-generators through advanced information and communication technologies and software systems. It is able to bid and operate as a single entity in the electricity market. The following is the development of technologies related to the virtual power plant bidding system:

[0003] 1. Distributed Energy Resource Management: Technology has enabled monitoring and control of a variety of DERs; through smart grid technology, DERs can respond to grid demands in real time.

[0004] 2. Demand response technology: Demand response (DR) technology allows VPPs to reduce electricity consumption during peak hours or increase consumption during low-demand hours; this technology contributes to the stability of the grid and the efficiency of the electricity market.

[0005] 3. Prediction and optimization algorithms: VPP uses advanced prediction models to predict the power generation and load demand of DERs; optimization algorithms are used to develop optimal power generation and bidding strategies.

[0006] 4. Market participation and bidding strategy: VPPs can participate in bidding in the electricity market and submit bids through an automated bidding system; bidding strategies are usually based on market forecasts, cost analysis and revenue maximization.

[0007] Disadvantages of existing technologies Although the virtual power plant bidding system plays an important role in the electricity market, the existing technologies still have the following disadvantages:

[0008] 1. Forecast accuracy: Existing forecast models are not accurate enough when dealing with uncertainty and extreme weather conditions. Forecast errors may lead to bidding strategy errors and affect the economic benefits of VPP.

[0009] 2. System complexity: VPP involves multiple DERs and complex optimization problems. The system design and operation management are very complex. The existing system may be inefficient when dealing with large-scale DERs.

[0010] 3. Communication and integration: There are still challenges in communication and data integration between DERs, especially between heterogeneous devices and protocols, which leads to high system integration costs and difficult maintenance.

[0011] 4. Market adaptability: The existing bidding system may not be flexible enough to adapt to the rapidly changing electricity market rules and demands. The bidding strategy may be too dependent on historical data and may not respond to market dynamics in a timely manner.

[0012] 5. Economic efficiency: The investment and operating costs of VPP are relatively high, especially in the initial construction and market access stages. Existing technologies have failed to effectively reduce marginal costs, affecting the overall economic efficiency of VPP.

[0013] 6. Security: With the application of information and communication technology in VPP, the network security issues of the system are becoming increasingly prominent, and the existing system may lack sufficient security measures to resist network attacks.

[0014] Virtual power plant bidding systems play an important role in integrating distributed energy resources, improving power market efficiency, and promoting renewable energy consumption. However, existing technologies still face challenges in prediction accuracy, system complexity, communication integration, market adaptability, economy, and safety. Summary of the invention

[0015] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and provide a virtual power plant bidding strategy optimization method based on chaos game optimization algorithm.

[0016] The purpose of the present invention can be achieved by the following technical solutions:

[0017] A virtual power plant bidding strategy optimization method based on a chaotic game optimization algorithm, comprising the following steps:

[0018] Generate the initial bidding strategy of virtual power plant through chaos mapping;

[0019] Each virtual power plant receives feedback from the market, including market dynamics and bidding behavior of other power plants;

[0020] Each virtual power plant adjusts its bidding strategy based on market feedback, taking into account costs and benefits, the volatility of photovoltaic power generation, the dispatch cost and attenuation of energy storage batteries, and the life of equipment.

[0021] Determine whether the bidding behavior of virtual power plants violates market ethics or benefit distribution constraints. If so, impose penalty benefits.

[0022] Calculate the power plant's revenue based on the adjusted strategy and update the bidding strategy of each virtual power plant;

[0023] The above game process is repeated until the convergence condition or the maximum number of iterations is reached, and the virtual power plant bidding strategy is output.

[0024] As a preferred technical solution, the virtual power plant adjusts its bidding strategy based on an objective function that considers costs and benefits, the volatility of photovoltaic power generation, the scheduling cost and attenuation of energy storage batteries, and the life of the equipment, which can be expressed as a comprehensive economic optimization problem:

[0025]

[0026] In the formula, C i (P i ,S i ,E i ) represents the total cost function of the i-th virtual power plant, including the photovoltaic power generation P i , Energy storage usage S i and the amount of energy storage dispatched E i ; B i (P i ) represents the profit function of the ith virtual power plant, which represents the economic benefits obtained through photovoltaic power generation, energy storage discharge or market regulation; f battery (E i ) represents the attenuation function of the energy storage battery, taking into account the recycling and attenuation cost of the energy storage battery; f lifetime (D i ) represents the loss cost function related to the equipment life, taking into account the aging and loss of the equipment during long-term operation; α and β are weight coefficients used to balance the relationship between different costs and benefits.

[0027] As a preferred technical solution, the constraints set for the objective function of the bidding strategy adjustment include: photovoltaic power generation constraints, energy storage capacity constraints, benefit distribution constraints, ethical constraints, load constraints, equipment life constraints and energy storage battery capacity attenuation constraints.

[0028] As a preferred technical solution, the process of adjusting the bidding strategy of the virtual power plant based on the objective function and the constraint conditions is as follows:

[0029] Each virtual power plant will be based on the current bidding strategy Calculate market feedback, including the power plant's market revenue, competitors' bidding strategies, and changes in market prices;

[0030] The virtual power plant calculates the gradient of the objective function based on the current market feedback, and uses the gradient descent method to adjust the bidding strategy in the direction of increasing revenue based on the gradient of the objective function;

[0031] When updating the strategy, the VPP ensures that the new strategy satisfies all constraints:

[0032] Project the updated bidding strategy into the constraint space. If the updated bidding strategy does not satisfy the constraint g(s i ), then adjust its value by projection method:

[0033]

[0034] Add a penalty term to the objective function to punish violations of the constraint g(s i ) strategy, then add a penalty term to the objective function:

[0035] f(s i )=f(s i )+λ·g(s i )

[0036] Among them, λ is the penalty coefficient.

[0037] As a preferred technical solution, the above-mentioned benefit distribution constraint is used to balance the benefit distribution of each participant:

[0038] B i (P i )≤γ·B total

[0039] Where: B total Represents the total revenue; γ is used to adjust the maximum benefit share of each participant in the overall market to prevent monopoly;

[0040] In the profit distribution plan, the fairness of the market can be controlled by adjusting the value of γ to ensure that the profits are reasonably distributed among different participants.

[0041] As a preferred technical solution, the ethical constraints are used to ensure that market monopoly or manipulation will not occur due to unfair competition:

[0042]

[0043] When the market revenue concentration exceeds the threshold θ, the system adjusts the bidding strategy to reduce the revenue concentration to meet the requirements of ethical constraints.

[0044] As a preferred technical solution, the load constraint uses historical load data and a deep learning model to perform load forecasting.

[0045] Collect historical load data, weather data, holiday information, and market electricity price data, perform feature engineering after data cleaning, and extract time features, weather features, and historical load features;

[0046] A load value predicted using a deep learning model comprising a long short-term memory network or a gated recurrent unit;

[0047] The architecture of the deep learning model is as follows: input layer: receiving processed feature vectors; hidden layer: multiple LSTM / GRU layers, each layer contains multiple neurons, used to capture long-term dependencies in time series data; output layer: output predicted load values;

[0048] The total load of the virtual power plant is constrained to meet the system balance based on the predicted load value.

[0049] As a preferred technical solution, the equipment life constraint and the energy storage battery capacity attenuation constraint are specifically implemented as follows:

[0050] The current bidding strategy is obtained based on the objective function optimization solution;

[0051] The number of battery charge and discharge cycles is calculated based on the current bidding strategy. The remaining life is calculated using a battery capacity attenuation model based on the number of battery charge and discharge cycles and deep discharge to determine whether the equipment life constraints and energy storage battery capacity attenuation constraints are met:

[0052]

[0053] Where: D i (t) represents the operating loss of the i-th device at time t; Indicates the maximum allowable loss or life of the equipment;

[0054]

[0055] Where: E i (t) represents the remaining capacity of the i-th energy storage device at time t; represents the maximum capacity of the i-th energy storage device; n i Indicates the number of times the energy storage device is used; δ indicates the capacity decay rate of the battery after each cycle;

[0056] The charging and discharging strategies are dynamically adjusted based on the actual status of the battery and the calculated remaining life. When the battery life approaches the threshold, maintenance or replacement is performed in advance.

[0057] As a preferred technical solution, if the behavior of a virtual power plant violates market ethical constraints or benefit distribution constraints, the penalty mechanism will impose a fine on its revenue:

[0058] B i (P i )-λ·Penalty(P i )

[0059] Where: λ is the penalty coefficient; Penalty(·) represents the penalty function when market fairness is violated.

[0060] As a preferred technical solution, the optimization method introduces disturbance after each round of game iteration to avoid local optimality, specifically:

[0061] After each round of game iteration, when all virtual power plants adjust their strategies based on market feedback, or apply a specific perturbation function to modify the current strategy after the output of the strategy update;

[0062] The range of the perturbation should match the range of the current bidding strategy; a stronger perturbation is used in the early stage of iteration, and a weaker perturbation is used in the later stage of convergence;

[0063] After applying the disturbance, the strategy is evaluated through market feedback; if the disturbance makes the strategy closer to the global optimal solution, the disturbance is retained, otherwise the reverse adjustment can be made.

[0064] Compared with the prior art, the present invention has the following beneficial effects:

[0065] 1) This invention introduces the Chaos Game Optimization (CGO) algorithm for the first time, which can simulate the complex competitive relationship between market participants and explore a wider solution space through chaotic mapping to avoid falling into the local optimum. Compared with existing linear optimization or heuristic algorithms, Chaos Game Optimization can better cope with the complexity and uncertainty of bidding in the power market. By utilizing the randomness and unpredictability of chaotic systems, the bidding strategy of virtual power plants is more flexible and adaptable, and can find better bidding solutions in the fiercely competitive market.

[0066] 2) The present invention introduces ethical constraints in the bidding process. By setting market fairness indicators (such as market share restrictions, price fairness, etc.), it ensures that the behavior of market participants complies with the principle of fair competition. After the profit calculation, penalties are imposed to suppress violations and avoid the occurrence of monopoly and unfair competition. It ensures that virtual power plants can not only improve their own benefits in bidding, but also maintain the healthy and sustainable development of the power market.

[0067] 3) The present invention specifically considers the battery capacity attenuation and equipment life issues of energy storage equipment in long-term use. By introducing these physical constraints in the optimization process, the virtual power plant can more reasonably dispatch the energy storage system, thereby extending the service life of the equipment and reducing long-term maintenance costs. This makes the operation of the virtual power plant more sustainable and avoids the frequent replacement and high maintenance costs caused by ignoring the aging of energy storage equipment in the prior art.

[0068] 4) This patent combines deep learning technology with virtual power plant bidding, using historical load data for accurate load forecasting, thereby improving the accuracy of bidding strategies. Compared with traditional methods that rely only on simple time series analysis or rule-based forecasting models, deep learning models can automatically extract features from large-scale data and adapt to complex load fluctuations and diverse market conditions. This enables the system to have higher load forecasting accuracy, thereby optimizing power dispatch and energy storage dispatch strategies and reducing the negative impact of uncertainty on virtual power plant bidding. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 This is a flow chart of a virtual power plant bidding strategy optimization method based on a chaotic game optimization algorithm of the present invention;

[0070] Figure 2 This is a comparison chart of annual returns before and after a virtual power plant implements a virtual power plant bidding strategy optimization method based on a chaos game optimization algorithm in one embodiment of the present invention. DETAILED DESCRIPTION

[0071] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0072] Example 1

[0073] 3. Gaming strategy

[0074] When considering the game between multiple virtual power plants, a chaotic game optimization algorithm is used to simulate the bidding behavior of each power plant. The goal of the game is to maximize the revenue of each power plant while satisfying various constraints. The specific chaotic game optimization process is as follows:

[0075] S1. Initialization strategy generation: Generate the initial bidding strategy of the virtual power plant through chaos mapping.

[0076] S2. Whether the maximum number of iterations has been reached: Check whether the preset maximum number of iterations has been reached. If not, proceed to the next round of game.

[0077] S3. Market feedback: Receive feedback from the market, understand market trends and the bidding behavior of other power plants.

[0078] S4. Adjust bidding strategy: Based on market feedback, the virtual power plant adjusts its bidding strategy.

[0079] S5. Whether it violates market rules: Determine whether the power plant's behavior violates market ethics or profit distribution constraints.

[0080] 5.1. If violated, proceed to the penalty step to reduce the benefits.

[0081] 5.2. If there is no violation, proceed with the profit calculation.

[0082] S6. Calculate profit: Calculate the profit of the power plant based on the adjusted strategy.

[0083] S7. Update bidding strategy: Update the bidding strategy of the power plant to respond to market changes.

[0084] S8. Introduce disturbances to avoid local optimality: Add disturbances during the update process to prevent the power plant's strategy from falling into the local optimal solution.

[0085] S9: Check whether the strategy has converged. If not, return to the market feedback link and continue iterating.

[0086] S10, End: When the maximum number of iterations or convergence condition is reached, the game ends.

[0087] The specific implementation of each major step is as follows:

[0088] In step S1, the initial strategy is generated by chaotic mapping: the initial bidding strategy of each virtual power plant is generated by chaotic mapping. The mapping function used in this embodiment is Logistic Map:

[0089] x_{t+1}=rx_t(1-x_t)

[0090] This mapping can be used to generate chaotic sequences as input to the initial strategy.

[0091] The specific steps of adjusting the bidding strategy according to market feedback in step S4 are as follows: In each iteration, the virtual power plant adjusts the bidding strategy according to market feedback and the behavior of other participants. The adjustment is based on the objective function and constraints, and its goal is to maximize the benefits of the virtual power plant while satisfying market rules and constraints. The process of adjusting the strategy can be achieved through an optimization algorithm.

[0092] The virtual power plant adjusts its bidding strategy based on the following objective function, which does not consider costs and benefits, but considers factors such as the volatility of photovoltaic power generation, the scheduling cost and attenuation of energy storage batteries, and the life of equipment. This objective function is used to guide the virtual power plant on how to formulate the optimal bidding strategy, which is expressed as a comprehensive economic optimization problem, as follows:

[0093]

[0094] In the formula, C i (P i ,S i ,Ei ) represents the total cost function of the i-th virtual power plant, including the photovoltaic power generation P i , Energy storage usage S i and the amount of energy storage dispatched E i ; B i (P i ) represents the profit function of the ith virtual power plant, which represents the economic benefits obtained through photovoltaic power generation, energy storage discharge or market regulation; f battery (E i ) represents the attenuation function of the energy storage battery, taking into account the recycling and attenuation cost of the energy storage battery; f 1ifetime (D i ) represents the loss cost function related to the equipment life, taking into account the aging and loss of the equipment during long-term operation; α and β are weight coefficients used to balance the relationship between different costs and benefits.

[0095] Specifically, the objective function takes into account the following factors:

[0096] Cost and Benefit Balancing: C in the Objective Function i (P i ,S i ,E i ) represents the total cost function of the ith virtual power plant, including the photovoltaic power generation P i , Energy storage usage S i and the amount of energy storage dispatched E i . B i (P i ) represents the revenue function of the ith virtual power plant, which represents the economic benefits obtained through photovoltaic power generation, energy storage discharge or market regulation. By minimizing the difference between cost and revenue, the virtual power plant can maximize profits.

[0097] Energy storage battery efficiency and maintenance: battery (E i ) represents the efficiency function of the energy storage battery, which reflects the cycle use and decay cost of the energy storage battery. This function encourages the virtual power plant to minimize the loss of the battery when charging to increase its service life.

[0098] Equipment life: lifetime (D i ) represents the loss cost function related to the equipment life. This function encourages virtual power plants to consider the long-term operation of equipment when formulating strategies to avoid premature damage caused by overuse.

[0099] Weight coefficients α and β: These two coefficients are used to adjust the importance of different factors in the objective function. For example, when α is larger, it means that the cost factor is more important; when β is larger, it means that the benefit factor is more critical.

[0100] For the process of adjusting the bidding strategy, set the following constraints:

[0101] Photovoltaic power generation constraint: Photovoltaic power generation P pv Affected by weather and time, the variability of photovoltaic power generation needs to be considered in the bidding of virtual power plants:

[0102]

[0103] Where: P pv,i (t) represents the photovoltaic power generation of the i-th virtual power plant at time t; Indicates the maximum output of photovoltaic power generation affected by factors such as sunshine and weather.

[0104] Energy storage capacity constraints: The charging and discharging of energy storage equipment must meet the limitations of its capacity and charging and discharging rate:

[0105]

[0106] Where: S i (t) represents the state of energy storage capacity of the i-th virtual power plant at time t; Indicates the maximum capacity of the energy storage device.

[0107] At the same time, the charging and discharging rate of energy storage is also limited:

[0108]

[0109] Where: is the maximum charge and discharge rate.

[0110] Benefit distribution constraint: In order to ensure fair benefit distribution, a constraint is set to balance the benefit distribution of each participant:

[0111] B i (P i )≤γ·B total

[0112] Where: B total represents the total revenue; γ is used to adjust the maximum benefit share of each participant in the overall market to prevent monopoly.

[0113] The effects of benefit allocation constraints are as follows:

[0114] Fairness guarantee: Ensure that all market participants can obtain reasonable benefits from the market and prevent some participants from occupying too much market share due to excessive market power.

[0115] Preventing market monopoly: By limiting the maximum profit share of each participant, it is possible to prevent individual participants from monopolizing the market and maintain the competitiveness of the market.

[0116] Incentive balance: Promote the balance of cooperation and competition among all participants, so that small and medium-sized participants also have the opportunity to gain benefits in the market, thereby motivating them to continue to participate in market activities.

[0117] Specific applications:

[0118] In the optimization algorithm, the profit distribution constraint will serve as a restriction to ensure that the profit of each participant will not exceed the set maximum share γ when solving the optimal bidding strategy.

[0119] In the profit distribution plan, the fairness of the market can be controlled by adjusting the value of γ to ensure that the profits are reasonably distributed among different participants.

[0120] Ethical constraints: Market competition needs to introduce fairness constraints to ensure that market monopoly or manipulation will not occur due to unfair competition:

[0121]

[0122] This formula is used to calculate the concentration of income in the market, similar to the Gini coefficient. If it exceeds the threshold θ, it indicates that the market is unfair.

[0123] Ethical constraints are as follows:

[0124] Market fairness: By calculating the market revenue concentration, we ensure that market competition is fair and there is no manipulation of market prices or monopolistic behavior.

[0125] Social and ethical responsibility: Ensure that the bidding strategy of virtual power plants not only pursues economic benefits, but also takes into account social responsibility and ethics to maintain the healthy development of the market.

[0126] Preventing market failures: By limiting the concentration of market gains, market failures such as price manipulation or excessive control of the market by a single player can be prevented.

[0127] Specific applications:

[0128] In the process of optimizing bidding strategies, ethical constraints will serve as an additional restriction to ensure that the optimization results do not lead to excessive concentration of market returns.

[0129] When the market revenue concentration exceeds the threshold θ, the system may adjust the bidding strategy to reduce the revenue concentration to meet the requirements of ethical constraints.

[0130] By monitoring and evaluating the distribution of market returns, potential market unfairness can be discovered in a timely manner and corresponding measures can be taken.

[0131] Load constraints: The total load of the virtual power plant needs to meet the system balance constraints:

[0132]

[0133] Where: L(t) represents the total load demand of the system.

[0134] The present invention uses historical load data to make accurate load forecasts and perform load constraints. The load forecasting scheme is as follows:

[0135] Data preprocessing: Data collection: collect historical load data, weather data, holiday information, market electricity prices, etc.; Data cleaning: remove outliers, process missing values, and standardize data; Feature engineering: extract time features (such as hours, weeks, seasons), weather features (such as temperature, humidity, wind speed), historical load features, etc.;

[0136] Load forecasting model: Model selection: Use deep learning models such as long short-term memory network (LSTM) or gated recurrent unit (GRU).

[0137] Model architecture: Input layer: receives the processed feature vector. Hidden layer: multiple LSTM / GRU layers, each layer contains multiple neurons to capture long-term dependencies in time series data. Output layer: outputs the predicted load value.

[0138] Model training: Loss function: Use mean square error (MSE) or mean absolute error (MAE) as the loss function. Optimizer: Use optimization algorithms such as Adam or RMSprop.

[0139] Equipment life constraints: Equipment life constraints are related to its use in long-term operation. In particular, the number of cycles of energy storage equipment will affect its life:

[0140]

[0141] Where: D i (t) represents the operating loss of the i-th device at time t; Indicates the maximum allowable loss or life of the equipment.

[0142] Energy storage battery capacity decay constraint: The capacity of the energy storage battery will gradually decay over time, affecting its actual energy storage capacity:

[0143]

[0144] Where: E i (t) represents the remaining capacity of the i-th energy storage device at time t; n i Indicates the number of times the energy storage device is used; δ indicates the capacity attenuation rate of the battery after each cycle.

[0145] The energy storage system equipment life and battery capacity attenuation constraint scheme is as follows:

[0146] Building an Energy Storage System Model

[0147] Battery model: Use an equivalent circuit model of the battery (such as the Thevenin model or the PNGV model) to simulate the charge and discharge characteristics of the battery.

[0148] Life model: Based on the battery cycle life test data, a battery capacity attenuation model is established, usually using empirical formulas or physics-based models.

[0149] Battery degradation model selection:

[0150] Use empirical models based on battery charge and discharge cycles, depth of discharge (DOD), and other factors, such as the Cox model or NCA model.

[0151] Cox Model:

[0152] Q trmainimg =Q nominal ×(1-α×N) β

[0153] In the formula, Q remaining is the remaining capacity; Q nominal is the nominal capacity; N is the number of charging cycles; α, β are model parameters.

[0154] NCA Model:

[0155] Q remaining =Q nemin ×e -(×DODNδ)×N

[0156] Where DOD stands for depth of discharge; γ and δ are model parameters.

[0157] Lifetime constraint introduction

[0158] Life target: Set the target life of the energy storage system (e.g. 10 years).

[0159] Constraints: During the optimization process, the remaining life of the battery is taken as one of the constraints.

[0160] Optimization strategy implementation

[0161] Step 1: Define the optimization problem

[0162]

[0163] Step 2: Optimize the algorithm

[0164] initial_bid_strategy = {...} # Initial strategy

[0165] result = minimize(objective_function, initial_bid_strategy, method = 'SLSQP', constraints = constraints) # Solve using optimization algorithm

[0166] optimal_bid_strategy=result.x#output optimal strategy

[0167] Step 3: Battery Life Calculation

[0168]

[0169]

[0170] Lifespan management strategy

[0171] Dynamic adjustment: Dynamically adjust the charging and discharging strategy based on the actual state and predicted life of the battery.

[0172] Preventive maintenance: Perform maintenance or replacement in advance when the battery life is approaching the threshold.

[0173] Result analysis and verification

[0174] Simulation test: Verify the effectiveness of the life management strategy by simulating different bidding strategies and battery usage scenarios.

[0175] Comparison with actual data: Compare simulation results with actual operating data to further optimize the model.

[0176] Through the above solution, virtual power plants can fully consider the long-term use of energy storage equipment and battery capacity degradation while optimizing bidding strategies, thereby extending the service life of equipment and reducing operating costs.

[0177] The process of adjusting the bidding strategy based on the objective function and constraints is as follows:

[0178] After each iteration, each virtual power plant will adjust its bidding strategy according to the following steps:

[0179] Calculate the market feedback of the current strategy

[0180] Each virtual power plant will be based on the current bidding strategy Calculate market feedback. Market feedback can include the market revenue of the power plant, the bidding strategies of competitors, and changes in market prices.

[0181] Calculate the gradient of the objective function

[0182] Based on the current market feedback, the virtual power plant will calculate the gradient of the objective function. The gradient reflects the changing trend of the current strategy on the objective function, which can help the power plant adjust its strategy in the direction of increasing revenue:

[0183]

[0184] in, is the gradient of the objective function with respect to the strategy, which represents the impact of market feedback on the power plant strategy.

[0185] Strategy Update

[0186] The power plant updates the strategy based on the gradient of the objective function, usually using the gradient descent method for adjustment. The update formula is as follows:

[0187]

[0188] Among them, α is the learning rate, which determines the step size of the policy adjustment. Through this formula, the virtual power plant updates the policy according to the gradient direction of the objective function.

[0189] Handling of constraints

[0190] When updating the strategy, the virtual power plant must ensure that the new strategy meets all constraints. Common constraint handling methods are:

[0191] Project the updated strategy into the constraint space. Assume that the constraint is g(s i )<0, if the updated strategy does not meet the constraints, adjust its value through the projection method:

[0192]

[0193] Add a penalty term to the objective function to penalize strategies that violate constraints. For example, if the constraint g(s i )>0, a penalty term can be added to the objective function:

[0194] f(s i )=f(s i )+λ·g(s i )

[0195] Among them, λ is the penalty coefficient.

[0196] Market feedback and adjustments

[0197] After the strategy is updated, the power plant will receive feedback from the market, such as price fluctuations, demand changes, competitor bids, etc. The power plant will adjust its strategy again based on this feedback.

[0198] In step S5, if the market rules are violated, the penalties imposed are as follows:

[0199] In order to ensure that virtual power plants comply with market rules, such as market ethical constraints, price constraints and other constraints, penalties are imposed after the profit calculation to curb violations and ensure the fairness of the game process and the effectiveness of the rules.

[0200] If a power plant's behavior violates market ethics constraints or benefit distribution constraints, the penalty mechanism will impose a fine on its revenue:

[0201] B i (P i )-λ·Penalty(P i )

[0202] Where: λ is the penalty coefficient; Penalty(·) represents the penalty function when market fairness is violated, which can be linked to constraints such as market share and revenue fairness.

[0203] Step S8 introduces disturbances to avoid local optimality. Specifically, after each round of iteration, all virtual power plants adjust their strategies based on market feedback and apply disturbances before the next strategy update. The disturbances can be applied directly when adjusting the strategy or after the output of the strategy update.

[0204] The scope of the perturbation should match the scope of the current bidding strategy. For example, if the value of the bidding strategy is a price range, the scope of the perturbation should also be limited to this range. The strength of the perturbation determines the impact of the perturbation on the strategy. Stronger perturbations may cause drastic changes in the strategy, while weaker perturbations will produce small adjustments. The perturbation strength can be adjusted dynamically, such as using stronger perturbations in the early stages of the iteration and weaker perturbations in the later stages when converging.

[0205] Modify the current policy through a specific perturbation function. The perturbation can be random or based on certain rules. The perturbation is usually based on a chaotic mapping function, and the perturbation value can be determined by the parameters in the chaotic mapping, for example, the control parameters in the logistic mapping.

[0206] After applying the perturbation, the strategy is evaluated through market feedback. If the perturbation makes the strategy closer to the global optimal solution, the perturbation is retained, otherwise the reverse adjustment can be made.

[0207] Example 2

[0208] As one of the specific implementation examples of the present invention, this embodiment uses photovoltaic power generation and energy storage to participate in the bidding strategy of the electricity market for two virtual power plants in a certain city, and compares their economic benefits before and after the implementation of this patent.

[0209] Virtual power plant 1: 50MW photovoltaic capacity, 30MWh energy storage capacity; Virtual power plant 2: 40MW photovoltaic capacity, 25MWh energy storage capacity;

[0210] Weather data: 70% sunny, 20% cloudy, 10% rainy; Photovoltaic power generation efficiency: 100% sunny, 50% cloudy, 20% rainy; Market electricity price: 0.5 yuan / kWh during peak hours, 0.2 yuan / kWh during trough hours; Energy storage charging and discharging efficiency: 90%; Energy storage battery attenuation rate: 0.5% / each cycle.

[0211] Profits before implementation: Before the implementation of this patent, the strategy of the virtual power plant is to simply participate in the bidding for photovoltaic power generation, without considering energy storage scheduling and market price fluctuations.

[0212] Benefits after implementation: Through the chaos game optimization algorithm proposed in this patent, the virtual power plant combines the scheduling strategy of energy storage, discharges during peak electricity price periods and charges during low electricity price periods, and takes into account the volatility of photovoltaic power generation.

[0213] Virtual Power Plant 1:

[0214] Energy storage strategy: discharge during peak hours and charge during off-peak hours. The estimated annual energy storage benefits are:

[0215] B S,1 =30×365×(0.5-0.2)×90%=29565 yuan

[0216] The total revenue is:

[0217]

[0218] Virtual Power Plant 2:

[0219] Energy storage strategy: Discharge during peak hours and charge during off-peak hours. The estimated annual energy storage benefits are:

[0220] B s,2 =25×365×(0.5-0.2)×90%=24638 yuan

[0221] The total revenue is:

[0222]

[0223] Comparison of benefits before and after implementation

[0224]

[0225]

[0226] The results show that the solution of the present invention has the following technical advantages:

[0227] 1. Significantly improve the benefits of bidding strategies:

[0228] By introducing the chaos game optimization algorithm, the virtual power plant can find a better bidding solution under complex market conditions and improve the success rate of bidding. Compared with the traditional method, the annual income of the virtual power plant has increased by 20-30%. As shown in the above example, the annual income of virtual power plant 1 and virtual power plant 2 increased by 29,665 yuan and 24,638 yuan respectively. This beneficial effect significantly improves the overall economic benefits of the virtual power plant and provides strong support for the virtual power plant to participate in the power market bidding.

[0229] 2. Higher load forecasting accuracy:

[0230] Using deep learning models for load forecasting enables virtual power plants to more accurately predict future load demand, thereby formulating more accurate power generation and energy storage scheduling plans. Compared with traditional load forecasting methods, the accuracy is improved by about 10-15%, reducing scheduling errors caused by inaccurate load forecasts. This means that virtual power plants can more accurately grasp market fluctuations, reduce unnecessary costs caused by power fluctuations, and thus improve the stability of the power grid.

[0231] 3. Improvement of market fairness and stability:

[0232] By introducing ethical constraints, monopoly and unfair competition in the market are avoided, making the competition environment in the power market more fair and transparent. Compared with existing technologies, this solution not only optimizes the benefits of individual virtual power plants, but also maintains the healthy operation of the entire market and improves the trust and willingness of market participants to participate. This effectively avoids market failures caused by monopoly behavior and makes the power market more robust and sustainable.

[0233] 4. Reduce the operating cost of energy storage system and extend the service life of equipment:

[0234] By taking into account the battery capacity decay and equipment life constraints of energy storage equipment, this patent can perform reasonable energy storage scheduling with minimal impact of battery capacity decay. Compared with traditional energy storage scheduling strategies, the service life of the equipment is extended by 5-10%, and the decay cost caused by excessive use of batteries is reduced. This advantage greatly reduces the long-term operation and maintenance costs of virtual power plants and improves the return on investment of equipment.

[0235] 5. Flexibility of coordinated dispatch of energy storage and photovoltaics:

[0236] This patent combines chaos game optimization with energy storage scheduling strategies to flexibly respond to the uncertainty of photovoltaic power generation, especially to make full use of the energy storage system to obtain maximum benefits during peak electricity price periods, and to charge during low electricity price periods to give full play to the regulatory role of energy storage. Compared with traditional scheduling methods, the energy storage utilization rate has increased by 15-20%. This flexibility effectively improves the dynamic adaptability of the bidding strategy of the virtual power plant to adapt to the changing market environment.

[0237] Through the above improvements, compared with the existing technology, the present invention not only improves the benefits and efficiency of virtual power plant bidding, but also ensures the fairness, sustainability and healthy market development of the system by introducing ethical constraints and equipment life management.

[0238] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A virtual power plant bidding strategy optimization method based on chaos game optimization algorithm, characterized in that the steps include: Generate the initial bidding strategy of virtual power plant through chaos mapping; Each virtual power plant receives feedback from the market, including market dynamics and bidding behavior of other power plants; Each virtual power plant adjusts its bidding strategy based on market feedback, taking into account costs and benefits, the volatility of photovoltaic power generation, the dispatch cost and attenuation of energy storage batteries, and the life of equipment. Determine whether the bidding behavior of virtual power plants violates market ethics or benefit distribution constraints. If so, impose penalty benefits. Calculate the power plant's revenue based on the adjusted strategy and update the bidding strategy of each virtual power plant; The above game process is repeated until the convergence condition or the maximum number of iterations is reached, and the virtual power plant bidding strategy is output.

2. According to claim 1, a virtual power plant bidding strategy optimization method based on chaos game optimization algorithm is characterized in that: The virtual power plant adjusts its bidding strategy based on the objective function that considers costs and benefits, the volatility of photovoltaic power generation, the scheduling cost and attenuation of energy storage batteries, and the life of the equipment, which can be expressed as a comprehensive economic optimization problem: In the formula, C i (P i ,S i ,E i ) represents the total cost function of the i-th virtual power plant, including the photovoltaic power generation P i , energy storage usage S i and the amount of energy storage dispatched E i ; B i (P i ) represents the profit function of the ith virtual power plant, which represents the economic benefits obtained through photovoltaic power generation, energy storage discharge or market regulation; f battery (E i ) represents the attenuation function of the energy storage battery, taking into account the recycling and attenuation cost of the energy storage battery; f lifetime (D i ) represents the loss cost function related to the equipment life, taking into account the aging and loss of the equipment during long-term operation; α and β are weight coefficients used to balance the relationship between different costs and benefits.

3. The method for optimizing bidding strategy of virtual power plant based on chaos game optimization algorithm according to claim 2 is characterized in that: The constraint conditions set for the objective function of adjusting the bidding strategy include: photovoltaic power generation constraint, energy storage capacity constraint, benefit distribution constraint, ethical constraint, load constraint, equipment life constraint and energy storage battery capacity attenuation constraint.

4. The method for optimizing bidding strategy of virtual power plant based on chaos game optimization algorithm according to claim 3 is characterized in that: The process of adjusting the bidding strategy of the virtual power plant based on the objective function and the constraints is as follows: Each virtual power plant will be based on the current bidding strategy Calculate market feedback, including the power plant's market revenue, competitors' bidding strategies, and changes in market prices; The virtual power plant calculates the gradient of the objective function based on the current market feedback, and uses the gradient descent method to adjust the bidding strategy in the direction of increasing revenue based on the gradient of the objective function; When updating the strategy, the VPP ensures that the new strategy satisfies all constraints: Project the updated bidding strategy into the constraint space. If the updated bidding strategy does not satisfy the constraint g(s i ), then adjust its value by projection method: Add a penalty term to the objective function to punish violations of the constraint g(s i ) strategy, then add a penalty term to the objective function: f(s i )=f(s i )+λ·g(s i ) Among them, λ is the penalty coefficient.

5. The method for optimizing bidding strategy of virtual power plant based on chaos game optimization algorithm according to claim 4 is characterized in that: The above-mentioned interest distribution constraints are used to balance the interest distribution of each participant: B i (P i )≤γ·B total Where: B total Represents the total revenue; γ is used to adjust the maximum benefit share of each participant in the overall market to prevent monopoly; In the profit distribution plan, the fairness of the market can be controlled by adjusting the value of γ to ensure that the profits are reasonably distributed among different participants.

6. The method for optimizing bidding strategy of virtual power plant based on chaos game optimization algorithm according to claim 4 is characterized in that: The ethical constraints described above are used to ensure that market monopoly or manipulation is not caused by unfair competition: When the market revenue concentration exceeds the threshold θ, the system adjusts the bidding strategy to reduce the revenue concentration to meet the requirements of ethical constraints.

7. The method for optimizing bidding strategy of virtual power plant based on chaos game optimization algorithm according to claim 4 is characterized in that: The load constraint described above utilizes historical load data and uses a deep learning model to perform load forecasting. Collect historical load data, weather data, holiday information, and market electricity price data, perform feature engineering after data cleaning, and extract time features, weather features, and historical load features; A load value predicted using a deep learning model comprising a long short-term memory network or a gated recurrent unit; The architecture of the deep learning model is as follows: Input layer: receives the processed feature vector; Hidden layer: multiple LSTM / GRU layers, each layer contains multiple neurons to capture long-term dependencies in time series data; Output layer: output predicted load value; The total load of the virtual power plant is constrained to meet the system balance based on the predicted load value.

8. The method for optimizing bidding strategy of virtual power plant based on chaos game optimization algorithm according to claim 4 is characterized in that: The equipment life constraint and the energy storage battery capacity attenuation constraint are specifically implemented as follows: The current bidding strategy is obtained based on the objective function optimization solution; The number of battery charge and discharge cycles is calculated based on the current bidding strategy. The remaining life is calculated using a battery capacity attenuation model based on the number of battery charge and discharge cycles and deep discharge to determine whether the equipment life constraints and energy storage battery capacity attenuation constraints are met: Where: D i (t) represents the operating loss of the i-th device at time t; Indicates the maximum allowable loss or life of the equipment; Where: E i (t) represents the remaining capacity of the i-th energy storage device at time t; represents the maximum capacity of the i-th energy storage device; n i Indicates the number of times the energy storage device is used; δ indicates the capacity decay rate of the battery after each cycle; The charging and discharging strategies are dynamically adjusted based on the actual status of the battery and the calculated remaining life. When the battery life approaches the threshold, maintenance or replacement is performed in advance.

9. The method for optimizing bidding strategy of virtual power plant based on chaos game optimization algorithm according to claim 5 is characterized in that: If the behavior of a virtual power plant violates market ethical constraints or benefit distribution constraints, the penalty mechanism will impose a fine on its revenue: B i (P i )-λ·Penalty(P i ) Where: λ is the penalty coefficient; Penalty(·) represents the penalty function when market fairness is violated.

10. The method for optimizing bidding strategy of virtual power plant based on chaos game optimization algorithm according to claim 1 is characterized in that: The optimization method introduces disturbance after each round of game iteration to avoid local optimality, specifically: After each round of game iteration, when all virtual power plants adjust their strategies based on market feedback, or apply a specific perturbation function to modify the current strategy after the output of the strategy update; The range of the perturbation should match the range of the current bidding strategy; a stronger perturbation is used in the early stage of iteration, and a weaker perturbation is used in the later stage of convergence; After applying the disturbance, the strategy is evaluated through market feedback; if the disturbance makes the strategy closer to the global optimal solution, the disturbance is retained, otherwise the reverse adjustment can be made.

Images