Method for charging station operator to participate in market bidding based on artificial intelligence

Abstract

The invention discloses a method for a charging station operator to participate in market bidding based on artificial intelligence, and the method comprises the following steps: S1, data collection and preprocessing: collecting historical charging data, real-time electricity price information, weather forecast, power grid load information, electric power capacity of a charging station and user charging habit data, cleaning the data, processing missing values, and storing the missing values in a database; and converting a data format so that the data format is suitable for input of a machine learning model. The method has the advantage of low-cost electricity purchase, and solves the problems that in the using process of an existing method that a charging station operator participates in market bidding, it is inconvenient for the charging station operator to purchase electricity at the lowest cost, if the charging station cannot purchase electricity at the lowest cost, the electricity purchase cost is directly increased, and then the overall operation cost is affected; and the charging station needs to increase the charging service price to maintain the profit level due to the relatively high purchase cost, so that the charging cost of the user is indirectly influenced.

Description

Technical Field

[0001] The present invention relates to the technical field of power engineering, and particularly to a method for a charging station operator to participate in market bidding based on artificial intelligence. Background Art

[0002] A method for a charging station operator to participate in market bidding based on artificial intelligence generally refers to using AI technology to optimize the operation efficiency of charging stations, reduce costs, and increase revenues. With the popularization of electric vehicles, the impact of electric vehicle charging loads on the power grid is increasing day by day. Especially during peak load periods, there may be a risk of distribution network overload. In addition, the location of charging stations will affect the route selection and driving time of electric vehicle drivers. As a participant in the power market, a charging station operator needs to purchase electricity through the market according to the predicted charging demand of the previous day and set a reasonable charging price to obtain revenues.

[0003] In the wholesale power market, a charging station operator can use AI to participate in the automated market bidding process. AI algorithms can automatically submit optimal buy and sell offers according to market conditions, so as to obtain the best price while ensuring supply. AI can also be used to monitor the operating status of equipment and predict potential failures, thereby reducing downtime and maintenance costs caused by equipment failures.

[0004] In the process of using the existing method for a charging station operator to participate in market bidding, it is not convenient for the charging station operator to purchase electricity at the lowest cost. If the charging station cannot purchase electricity at the lowest cost, the electricity purchase cost will directly increase, thereby affecting the overall operating cost. Moreover, the higher purchase cost will cause the charging station to increase the charging service price to maintain the profit level, indirectly affecting the charging cost of users. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for a charging station operator to participate in market bidding based on artificial intelligence, which has the advantage of low-cost electricity purchase, and solves the problem that in the process of using the existing method for a charging station operator to participate in market bidding, it is not convenient for the charging station operator to purchase electricity at the lowest cost. If the charging station cannot purchase electricity at the lowest cost, the electricity purchase cost will directly increase, thereby affecting the overall operating cost. Moreover, the higher purchase cost will cause the charging station to increase the charging service price to maintain the profit level, indirectly affecting the charging cost of users.

[0006] To achieve the above purpose, the present invention provides the following technical solution: A method for a charging station operator to participate in market bidding based on artificial intelligence, including the following steps:

[0007] S1. Data collection and preprocessing: Collect historical charging data, real-time electricity prices, weather forecasts, grid load information, the power capacity of charging stations, and user charging habit data. Clean the data, handle missing values, and convert the data format to make it suitable for input into machine learning models.

[0008] S2. Demand forecasting: Select a time series forecasting model, use historical data to train the time series forecasting model, predict the charging demand for a future period, and verify the accuracy of the model through cross-validation methods.

[0009] S3. Dynamic pricing: Build a dynamic pricing model based on reinforcement learning algorithms. The dynamic pricing model adjusts the parameters in the model according to market conditions and tests the effectiveness of the pricing strategy in a simulation environment.

[0010] S4. Energy procurement optimization: Develop an energy procurement plan based on the predicted charging demand and real-time electricity prices.

[0011] S5. Establish a two-layer bidding model: The upper-layer model sets the charging price with the goal of maximizing the revenue of the charging station, and the lower-layer model clears the market with the goal of minimizing the operating cost of the power system.

[0012] S6. Market bidding: Automatically submit the optimal bid according to market dynamics through bidding algorithms and the two-layer bidding model, and integrate the bidding algorithms and the two-layer bidding model into the existing operation system to ensure seamless participation in power market bidding.

[0013] S7. Load management and optimization: Use AI to predict the load situation of the charging station to avoid overload, adjust the working status of the charging piles according to the prediction results to achieve load balancing, and use AI to monitor the operating status of the charging station in real time to ensure system stability.

[0014] Preferably, as a method for a charging station operator based on artificial intelligence to participate in market bidding of the present invention, the time series forecasting model in S2 is one of ARIMA, LSTM, and Prophet.

[0015] Preferably, as a method for a charging station operator based on artificial intelligence to participate in market bidding of the present invention, when verifying the accuracy of the model in S2, use the time series cross-validation method to verify the time series forecasting model. When performing verification, use the time series cross-validation method to train the model within the initial training window, use the trained model to predict one or more future time points, compare the difference between the predicted value and the actual value that occurred, calculate the prediction error, slide the training window forward, add a new observation value, and then repeat the verification. Continue to slide the window until all available historical data has been used for one prediction. Finally, analyze the performance of the model based on the results of cross-validation.

[0016] As a preferred method for charging station operators to participate in market bidding based on artificial intelligence of the present invention, the parameters in the market situation adjustment model in S3 include risk preference parameters and cost function parameters.

[0017] As a preferred method for charging station operators to participate in market bidding based on artificial intelligence of the present invention, when constructing a dynamic pricing model in S3, first prepare a historical data set, then use simulation software to create a simulation environment, which can reflect the operation of charging stations in the real world, and use the selected reinforcement learning algorithm to train the model in the simulation environment. During the training process, the model tries different pricing strategies and learns the optimal strategy based on the reward signal.

[0018] As a preferred method for charging station operators to participate in market bidding based on artificial intelligence of the present invention, when formulating the energy procurement plan in S4, advance procurement strategy, instant procurement strategy and demand response strategy are formulated according to the charging demand forecast and electricity price pattern. The advance procurement strategy is to purchase electricity during the period of lower electricity price and store it in the battery energy storage system for use during peak hours. The instant procurement strategy is to dynamically adjust the procurement volume according to the real-time electricity price, increase the procurement volume when the electricity price is low, and reduce it otherwise. The demand response strategy is to reduce or postpone electricity consumption within a specific time according to the incentive plan provided by the power company, so as to obtain additional economic compensation.

[0019] As a preferred method for charging station operators to participate in market bidding based on artificial intelligence of the present invention, the process of using an optimization algorithm to automatically make purchasing decisions in S4 helps charging stations minimize purchasing costs while meeting demand.

[0020] As a preferred method for charging station operators to participate in market bidding based on artificial intelligence of the present invention, the definition objective function formula of the upper model in S5 is maxp∑t(pt-ct)·dt, where pt is the charging price, ct is the electricity procurement cost, and dt is the charging demand, and the positioning objective function formula of the lower model is ming∑tC(gt), gt is the power generation, and C(gt) is the power generation cost function.

[0021] As a preferred method for charging station operators to participate in market bidding based on artificial intelligence of the present invention, when the optimal bid is automatically submitted in S6, an API interface connected to the power market trading platform is developed to ensure that the system can receive market data in real time and send bids. According to the output of the two-tier bidding model, the optimal bid for each time interval is generated, and a scheduled task is designed so that the system can automatically submit bids at each bidding time point.

[0022] Preferably, as a method for a charging station operator participating in market bidding based on artificial intelligence of the present invention, when performing load balancing in S7, the operating state of the charging pile is dynamically adjusted according to the predicted load situation, and priority charging services are provided for users of different types or urgency levels. When it is predicted that an overload situation is about to occur, measures are taken in advance to reduce the power output of non-critical charging piles.

[0023] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0024] 1. By cleaning noise and outliers, the present invention can improve the quality of data, making subsequent analysis and modeling more reliable. Filling in missing data can avoid analysis biases caused by incomplete data and ensure the accuracy of model training. Converting data into a unified format helps with data consistency and makes the data easier to be processed by machine learning models. By standardizing the data, the influence of dimensions can be eliminated, improving the training efficiency of the model. Through feature engineering, new feature variables can be created or existing variables can be transformed, thereby improving the interpretability and prediction accuracy of the model. By reducing unnecessary features, the complexity of the model can be reduced, avoiding overfitting and accelerating the speed of model training.

[0025] 2. Through the time series cross-validation of the present invention, by means of a sliding window, it is ensured that the data for each training and test are continuous time segments, thus avoiding the overfitting phenomenon. Time series cross-validation can better evaluate the performance of the model on future data because it simulates the rolling prediction process of the model on future data. By making predictions and validations at multiple time points, the performance of the model in different time periods can be more comprehensively evaluated, thereby improving the reliability of the prediction results. As the time window slides, the model can continuously adapt to new data, which helps the model maintain a high prediction accuracy in practical applications.

[0026] 3. Through the reinforcement learning algorithm, the model can automatically learn the optimal pricing strategy under different market conditions, thereby achieving the goal of maximizing revenue or minimizing costs. The reinforcement learning algorithm can dynamically adjust the pricing strategy according to environmental feedback, enabling the model to continuously optimize in a changing market environment. Testing the pricing strategy in a simulated environment can ensure that the model performs well in a variety of possible market situations, improving the robustness of the model. By simulating different market scenarios, the effectiveness of the pricing strategy can be comprehensively evaluated to ensure that the model will not fail due to the failure of a single scenario in practical applications.

[0027] 4. Through an effective energy procurement plan, the present invention can ensure that the charging needs of users are met at any time, improve service quality, dynamically adjust the charging price according to the procurement cost, which can not only attract users but also ensure that the charging station obtains sufficient revenue during peak demand periods. By optimizing the procurement strategy, the charging station can gain a competitive advantage in cost control, improve market competitiveness, and reduce risks brought by market price fluctuations through advance procurement and energy storage management.

[0028] 5. By optimizing the charging price, the present invention can maximize the revenue of the charging station, enabling the charging station to maximize profits while meeting user needs. The optimized charging price can better reflect the electricity procurement cost and market demand, helping to control the overall operating cost. By optimizing the power generation, the operating cost of the power system can be minimized, which not only reduces the cost of the power company but also provides a more economical power supply for the charging station. The double - layer bidding model can help the charging station formulate an optimal price strategy in a dynamic market environment and enhance its competitiveness in the market.

[0029] 6. Through the API interface connected to the power market trading platform, the system can receive market data in real - time and automatically submit the optimal bid according to this data, ensuring that the bid is timely and effective. Automatically submitting the bid reduces the need for manual intervention and improves the efficiency of bidding. Based on the output of the double - layer bidding model, the system can dynamically adjust the bidding strategy to ensure that the optimal bid is submitted in each time interval, thus optimizing resource allocation. By finely managing the bids in each time interval, the charging station can better control costs and increase revenue. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a schematic flow chart of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0031] Please refer to Figure 1 , a method for a charging station operator based on artificial intelligence to participate in market bidding, comprising the following steps:

[0032] S1. Data collection and pre - processing: Collect historical charging data, real - time electricity price information, weather forecasts, grid load information, the power capacity of the charging station, and user charging habit data. Clean the data, process missing values, and convert the data format to make it suitable for the input of machine learning models.

[0033] By cleaning noise and outliers, the quality of the data can be improved, making subsequent analysis and modeling more reliable. Filling in missing data can avoid analysis biases caused by incomplete data and ensure the accuracy of model training. Converting the data into a unified format helps with data consistency and makes the data easier to be processed by machine learning models. By standardizing the data, the influence of the dimension can be eliminated and the training efficiency of the model can be improved. Through feature engineering, new feature variables can be created or existing variables can be transformed, thereby improving the interpretability and prediction accuracy of the model. By reducing unnecessary features, the complexity of the model can be reduced, overfitting can be avoided, and at the same time, the speed of model training can be accelerated.

[0034] High-quality data can improve the accuracy of prediction models, helping charging stations better understand future demand patterns and market dynamics. Based on accurate data analysis, charging stations can formulate more reasonable procurement strategies, pricing strategies, and load management strategies, thereby reducing costs and increasing revenues. By preprocessing real-time data, potential problems can be detected and solved in a timely manner to ensure the stable operation of the system. An anomaly detection mechanism is set up during the preprocessing process to help quickly identify abnormal situations and avoid system failures.

[0035] Moreover, data from different sources, such as historical charging data, real-time electricity price information, weather forecasts, etc., can be integrated during the preprocessing process to form a comprehensive data view to support more complex analysis requirements.

[0036] S2. Demand forecasting: Select a time series forecasting model, use historical data to train the time series forecasting model, and predict the charging demand for a period in the future. Verify the accuracy of the model through the cross-validation method.

[0037] The time series forecasting model is one of ARIMA, LSTM, and Prophet.

[0038] When verifying the accuracy of the model, use the time series cross-validation method to verify the time series forecasting model. When performing the verification, use the time series cross-validation method to train the model within the initial training window, use the trained model to predict one or more future time points, compare the difference between the predicted value and the actual value that occurred, calculate the prediction error, slide the training window forward, add a new observation value, and then repeat the verification. Continue to slide the window until all available historical data has been used for one prediction. Finally, analyze the performance of the model based on the results of the cross-validation.

[0039] Traditional K-fold cross-validation assumes that the data is independent and identically distributed, while time series data has temporal correlation. Time series cross-validation ensures that the data for each training and testing is a continuous time segment through a sliding window approach, thus avoiding overfitting. Time series cross-validation can better evaluate the performance of the model on future data because it simulates the rolling prediction process of the model on future data. By making predictions and validations at multiple time points, the performance of the model in different time periods can be more comprehensively evaluated, thereby improving the reliability of the prediction results. As the time window slides, the model can continuously adapt to new data, which helps the model maintain a high prediction accuracy in practical applications.

[0040] Time series cross-validation allows for model evaluation at different time window lengths and can flexibly adjust the window size according to the actual situation to adapt to different application scenarios. This method is applicable to time series data of different lengths and can effectively verify both short-term and long-term predictions.

[0041] By comparing the differences between the predicted values and the actual values, the prediction effect of the model can be directly observed, which is easy to understand and interpret. The prediction results and the actual data can be plotted in the same chart to visually display the prediction performance of the model. Through multiple validations, the performance differences of the model under different parameter settings can be found, thereby optimizing the model parameters. As time series cross-validation is repeatedly carried out, the model can be continuously adjusted according to the validation results to gradually optimize it. In practical applications, the results of time series cross-validation can be used to adjust the parameters or structure of the model in real time to cope with data changes. By evaluating the performance of the model on future data, potential risk points can be better identified and measures can be taken in advance to prevent them. An accurate prediction model can provide reliable decision-making support for the operation and management of charging stations and reduce the risks caused by incorrect predictions.

[0042] S3. Dynamic pricing, constructing a dynamic pricing model based on reinforcement learning algorithms. The dynamic pricing model adjusts the parameters in the model according to market conditions and tests the effectiveness of the pricing strategy in a simulated environment.

[0043] Adjusting the parameters in the model according to market conditions includes risk preference parameters and cost function parameters.

[0044] When constructing a dynamic pricing model, first prepare the historical data set, and then use simulation software to create a simulated environment that can reflect the operation of charging stations in the real world. Use the selected reinforcement learning algorithm to train the model in the simulated environment. During the training process, the model tries different pricing strategies and learns the optimal strategy based on the reward signal.

[0045] Through the reinforcement learning algorithm, the model can automatically learn the optimal pricing strategy under different market conditions, thus achieving the goal of maximizing revenue or minimizing costs. The reinforcement learning algorithm can dynamically adjust the pricing strategy according to the environmental feedback, enabling the model to continuously optimize in the changing market environment. Testing the pricing strategy in a simulated environment can ensure that the model performs well in a variety of possible market situations, improving the robustness of the model. By simulating different market scenarios, the effectiveness of the pricing strategy can be comprehensively evaluated to ensure that the model will not fail due to the failure of a single scenario in practical applications.

[0046] Conducting experiments in a simulated environment can avoid economic losses caused by incorrect pricing strategies in actual operations. The simulated environment allows for quick adjustment and testing of different strategies, accelerating the iterative and optimization process of the model. Based on the historical dataset and the test results in the simulated environment, more scientific and accurate pricing decisions can be made. By quantitatively evaluating the performance of the model under different strategies, the best pricing strategy can be more objectively selected. By adjusting the parameters in the model, the behavior of the model can be flexibly adjusted according to the actual situation to adapt to different market environments. The dynamic nature of the reinforcement learning algorithm enables the model to self-optimize in the constantly changing market, enhancing flexibility.

[0047] S4. Energy procurement optimization. Develop an energy procurement plan based on the predicted charging demand and real-time electricity price.

[0048] When developing the energy procurement plan, according to the charging demand forecast and electricity price pattern, formulate an advance procurement strategy, an immediate procurement strategy, and a demand response strategy. The advance procurement strategy is to purchase electricity during periods with lower electricity prices and store it in the battery energy storage system for use during peak hours. The immediate procurement strategy is to dynamically adjust the procurement volume according to the real-time electricity price, increasing the procurement volume when the electricity price is low and vice versa. The demand response strategy is to reduce or postpone electricity consumption within a specific time according to the incentive plan provided by the power company to obtain additional economic compensation.

[0049] The process of using an optimization algorithm to automatically make procurement decisions helps the charging station minimize the procurement cost while meeting the demand. By combining the advance procurement, immediate procurement, and demand response strategies, the procurement volume can be increased when the electricity price is low and decreased when the electricity price is high, thus effectively reducing the overall procurement cost. By supplying electricity through energy storage devices or reducing electricity demand during peak electricity price periods, high electricity bills can be avoided.

[0050] Dynamically adjust the procurement volume according to the real-time electricity price, enabling the charging station to obtain the required electricity at the lowest cost at any time. Using an optimization algorithm to automatically make procurement decisions reduces the need for human intervention and improves operational efficiency. Combining advance procurement, immediate procurement, and demand response strategies, the procurement plan can be flexibly adjusted according to market changes, enhancing the ability to cope with market fluctuations. Through the optimization algorithm, the charging station can quickly respond to market changes and timely adjust the procurement strategy to ensure the stability and cost-effectiveness of power supply.

[0051] Through an effective energy procurement plan, it can ensure that the charging needs of users are met at any time and improve service quality. Dynamically adjusting the charging price according to the procurement cost can not only attract users but also ensure that the charging station obtains sufficient revenue during peak demand periods. By optimizing the procurement strategy, the charging station can gain a competitive advantage in cost control and improve its market competitiveness. Through advance procurement and energy storage management, the risks brought about by market price fluctuations can be reduced.

[0052] S5. Establish a two-layer bidding model. The upper-layer model sets the charging price with the goal of maximizing the charging station's revenue, and the lower-layer model clears the market with the goal of minimizing the operating cost of the power system.

[0053] The defined objective function formula of the upper-layer model is maxp∑t(pt - ct)·dt, where pt is the charging price, ct is the power procurement cost, and dt is the charging demand. The defined objective function formula of the lower-layer model is ming∑tC(gt), where gt is the power generation and C(gt) is the power generation cost function.

[0054] By optimizing the charging price, the revenue of the charging station can be maximized, enabling the charging station to maximize profits while meeting user needs. The optimized charging price can better reflect the power procurement cost and market demand, helping to control the overall operating cost. By optimizing the power generation, the operating cost of the power system can be minimized, which not only reduces the cost of the power company but also provides a more economical power supply for the charging station. The two-layer bidding model can help the charging station formulate an optimal price strategy in a dynamic market environment and enhance its competitiveness in the market.

[0055] Based on the optimal price strategy, the charging station can provide more attractive services to attract more users. Through the collaborative optimization of the upper and lower layer models, the best balance of power supply and demand can be achieved, improving the overall efficiency of the power system. The optimized power generation and charging price can guide the rational allocation of resources and avoid resource waste. By optimizing the charging price and power generation, load balancing can be achieved, avoiding power system overload and ensuring the stable operation of the system. During the peak demand period, the optimization model can help the charging station respond quickly by adjusting the price or power generation to cope with sudden demand. The double-layer bidding model can better predict market changes and help the charging station take measures in advance to reduce the risks brought by market price fluctuations.

[0056] S6. Market bidding: Automatically submit the optimal bid according to market dynamics through the bidding algorithm and the double-layer bidding model, and integrate the bidding algorithm and the double-layer bidding model into the existing operation system to ensure seamless participation in the electricity market bidding.

[0057] When automatically submitting the optimal bid, develop an API interface for docking with the electricity market trading platform to ensure that the system can receive market data in real time and send bids. Generate the optimal bid for each time interval according to the output of the double-layer bidding model, and design a timing task to enable the system to automatically submit bids at each bidding time point.

[0058] Through the API interface for docking with the electricity market trading platform, the system can receive market data in real time and automatically submit the optimal bid according to these data to ensure the timeliness and effectiveness of the bid. Automatically submitting bids reduces the need for manual intervention and improves the efficiency of bidding. Based on the output of the double-layer bidding model, the system can dynamically adjust the bidding strategy to ensure that the optimal bid is submitted within each time interval, thereby optimizing resource allocation. By finely managing the bids for each time interval, the charging station can better control costs and increase revenues.

[0059] Automatically submitting bids enables the charging station to respond more quickly to market changes in real-time bidding, improving market competitiveness. Based on real-time market data and optimization algorithms, the charging station can make more scientific decisions, enhancing the market response speed and decision-making quality. Through the automated bidding strategy, the charging station can increase the purchase volume during the period with lower electricity prices, thereby reducing the overall cost.

[0060] S7. Load management and optimization: Use AI to predict the load situation of the charging station to avoid overload, adjust the working state of the charging piles according to the prediction results to achieve load balancing, and use AI to monitor the operation state of the charging station in real time to ensure system stability.

[0061] When performing load balancing, the working state of the charging pile is dynamically adjusted according to the predicted load conditions, and priority charging services are provided for users of different types or emergency levels. By predicting the upcoming overload situation in advance, the working state of the charging pile can be adjusted in a timely manner to avoid the occurrence of overload and ensure the safe and stable operation of the system.

[0062] Dynamic adjustment can reduce equipment failures caused by overload and extend the equipment life. By dynamically adjusting the working state of the charging pile, load balancing is achieved to ensure that there will be no situation where the load is too high or too low in a certain area or time period at any time point. When it is predicted that an overload is about to occur, measures can be taken in a timely manner, such as reducing the power output of non-critical charging piles, thus avoiding the occurrence of emergencies.

[0063] The use of charging piles is reasonably allocated according to the predicted load conditions to avoid resource waste and improve resource utilization. Dynamically adjusting the working state of the charging pile can be flexibly scheduled according to the real-time load conditions. By load balancing, the waiting time of users can be reduced and the user experience can be improved.

[0064] Priority charging services are provided for users of different types or emergency levels to improve the service quality. Measures are taken in advance to reduce the power output of non-critical charging piles.

[0065] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principles of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for charging station operators to participate in market bidding based on artificial intelligence, characterized in that: The following steps are involved: S1. Data collection and preprocessing: collect historical charging data, real-time electricity price information, weather forecast, grid load information, charging station power capacity and user charging habits data, clean the data, process missing values, and convert the data format to make it suitable for the input of the machine learning model; S2. Demand forecasting: select a time series forecasting model, use historical data to train the time series forecasting model, predict charging demand in the future, and verify the accuracy of the model through cross-validation method; S3. Dynamic pricing: Build a dynamic pricing model based on reinforcement learning algorithm. The dynamic pricing model adjusts the parameters in the model according to market conditions and tests the effect of the pricing strategy in a simulation environment. S4, energy procurement optimization, formulate energy procurement plans based on predicted charging demand and real-time electricity prices; S5. Establish a two-tier bidding model. The upper-tier model sets charging prices with the goal of maximizing the revenue of charging stations, and the lower-tier model clears the market with the goal of minimizing the operating cost of the power system. S6, market bidding, automatically submit the best bid according to market dynamics through bidding algorithms and two-tier bidding models, and integrate the bidding algorithms and two-tier bidding models into the existing operating system to ensure seamless participation in power market bidding; S7, load management and optimization, use AI to predict the load conditions of the charging station to avoid overload, adjust the working status of the charging pile according to the prediction results to achieve load balancing, and use AI to monitor the operating status of the charging station in real time to ensure system stability.

2. The method for charging station operators to participate in market bidding based on artificial intelligence according to claim 1, characterized in that: The time series prediction model in S2 is one of ARIMA, LSTM and Prophet.

3. The method for charging station operators to participate in market bidding based on artificial intelligence according to claim 1, characterized in that: When verifying the accuracy of the model in S2, the time series prediction model is verified using the time series cross-validation method. When verifying, the model is trained within the initial training window using the time series cross-validation method, and the trained model is used to predict one or more future time points, and the difference between the predicted value and the actual value is compared. The prediction error is calculated, the training window is slid forward, a new observation is added, and then the verification is repeated. The sliding window is continued until all available historical data are used to make a prediction, and finally the performance of the model is analyzed based on the results of the cross-validation.

4. The method for charging station operators to participate in market bidding based on artificial intelligence according to claim 1, characterized in that: The parameters in the market situation adjustment model in S3 include risk preference parameters and cost function parameters.

5. The method for charging station operators to participate in market bidding based on artificial intelligence according to claim 1, characterized in that: When constructing the dynamic pricing model in S3, the historical data set is first prepared, and then a simulation environment is created using simulation software. The environment can reflect the operation of charging stations in the real world. The model is trained in the simulation environment using the selected reinforcement learning algorithm. During the training process, the model tries different pricing strategies and learns the optimal strategy based on the reward signal.

6. The method for charging station operators to participate in market bidding based on artificial intelligence according to claim 1, characterized in that: When formulating the energy procurement plan in S4, advance procurement strategy, instant procurement strategy and demand response strategy are formulated according to the charging demand forecast and electricity price pattern. The advance procurement strategy is to purchase electricity during the period of lower electricity price and store it in the battery energy storage system for use during peak hours. The instant procurement strategy is to dynamically adjust the procurement quantity according to the real-time electricity price, increase the procurement quantity when the electricity price is low, and reduce it otherwise. The demand response strategy is to reduce or postpone electricity consumption within a specific time according to the incentive plan provided by the power company, so as to obtain additional economic compensation.

7. The method for charging station operators to participate in market bidding based on artificial intelligence according to claim 1, characterized in that: The S4 uses an optimization algorithm to automate the purchasing decision-making process, helping charging stations to meet demand while minimizing purchasing costs.

8. The method for charging station operators to participate in market bidding based on artificial intelligence according to claim 1, characterized in that: The definition objective function formula of the upper model in S5 is maxp∑t(pt-ct)·dt, where pt is the charging price, ct is the electricity procurement cost, and dt is the charging demand. The positioning objective function formula of the lower model is ming∑tC(gt), gt is the power generation, and C(gt) is the power generation cost function.

9. The method for charging station operators to participate in market bidding based on artificial intelligence according to claim 1, characterized in that: When the optimal bid is automatically submitted in S6, an API interface connected to the power market trading platform is developed to ensure that the system can receive market data in real time and send bids. According to the output of the two-tier bidding model, the optimal bid for each time interval is generated, and a scheduled task is designed so that the system can automatically submit bids at each bidding time point.

10. The method for charging station operators to participate in market bidding based on artificial intelligence according to claim 1, characterized in that: When load balancing is performed in S7, the working state of the charging pile is dynamically adjusted according to the predicted load conditions, and priority charging services are provided for users of different types or urgency levels. When an overload condition is predicted to occur, measures are taken in advance to reduce the power output of non-critical charging piles.

Images