Dual-track coordinated regulation method of optical storage charging station for power spot and ancillary service market

Abstract

This invention discloses a dual-track coordinated control method for photovoltaic-storage charging stations targeting both the electricity spot market and ancillary services markets. The method first constructs a day-ahead dual-track coordinated bidding optimization model embedded with conditional risk value, simultaneously deciding on electricity volume in the spot market and capacity in the ancillary services market. It also employs a hierarchical structure combining upper-level instruction decomposition and lower-level model predictive control to achieve precise tracking of grid instructions and smooth handling of internal fluctuations. Through day-ahead risk optimization and real-time multi-objective coordinated control in a dual-track closed-loop control system, this invention significantly improves the overall profitability and operational stability of photovoltaic-storage charging stations participating in the electricity market, while simultaneously ensuring charging service quality and optimizing equipment lifespan.

Description

Technical Field

[0001] This invention relates to the field of electricity market and energy aggregation regulation technology, and in particular to a dual-track coordinated regulation method for photovoltaic-storage-charging stations targeting the electricity spot market and ancillary services market. Background Technology

[0002] Today's electricity spot market integrates photovoltaic power generation, energy storage systems, and electric vehicle charging loads. Photovoltaic-storage charging stations are no longer just electricity consumers, but have the potential to become active market participants and generate revenue by providing multiple types of services. Existing technologies have focused on optimizing bidding strategies for photovoltaic-storage systems or charging stations in a single market, or on using energy storage to smooth out photovoltaic fluctuations and achieve peak-valley arbitrage and other single-site-level management goals. At the same time, for the adjustability of charging load, some technologies have also attempted to participate in demand response through price signals or direct control.

[0003] Most existing methods employ sequential or independent decision-making models, meaning they first determine the strategy for participating in one market and then use the outcome as a boundary condition for participating in another. This fails to fully consider the deep synergistic relationship between the two markets in terms of temporal coupling, resource competition, and risk interlocking, resulting in the overall revenue potential not being fully realized. Furthermore, at the real-time operational level, existing control strategies often focus on tracking day-ahead plans or satisfying single services, lacking a unified framework to simultaneously and in real-time respond to highly volatile AGC commands, absorb photovoltaic forecast deviations, and dynamically guarantee diversified charging service demands. This can easily lead to market performance penalties or decreased user satisfaction. Moreover, most bidding models do not adequately consider the operational risks arising from market price volatility and the uncertainty of ancillary service calls, or only impose simple constraints, lacking refined risk quantification and management mechanisms, making the strategies less robust in real-world market environments.

[0004] Therefore, in response to the problems mentioned above, this invention proposes a dual-track coordinated control method for photovoltaic-storage-charging stations oriented towards the electricity spot market and ancillary services market. Summary of the Invention

[0005] To overcome the problems of insufficient coordination, lack of real-time control capabilities, and inadequate risk consideration in existing technologies when photovoltaic-storage charging stations participate in the dual-track market, this invention proposes a dual-track coordinated control method for photovoltaic-storage charging stations oriented towards the electricity spot market and ancillary services market. By constructing a risk-embedded joint optimization model and a hierarchical real-time control architecture, this method aims to maximize the overall revenue and ensure safe and stable operation of photovoltaic-storage charging stations in both the electricity spot market and ancillary services market.

[0006] The technical solution of this invention is: a dual-track coordinated control method for photovoltaic-storage-charging stations targeting the electricity spot market and ancillary services market, comprising the following steps: S1, before the daily market declaration deadline, generates day-ahead forecast curves for photovoltaic power generation and base charging load based on high-precision numerical weather forecasts, historical load data and electric vehicle user reservation information; S2. Construct a dual-track bidding optimization model that takes into account risks. The model aims to maximize the overall operating revenue of photovoltaic-storage charging stations. The first track decision variable is the day-ahead electricity bidding curve in the electricity spot market, and the second track decision variable is the combination of frequency regulation capacity and frequency regulation mileage bidding in the ancillary service market. S3, for the dual-track bidding optimization model, takes into account the prediction curve, historical spot market prices, statistical characteristics of ancillary service market clearing prices and call-up probabilities, energy storage system charging and discharging efficiency and cycle life depreciation costs, and the constraint of uninterrupted charging services, and solves to obtain the optimal day-ahead bidding plan, including time-of-use electricity sales / purchase volume, energy storage benchmark power point, and declared frequency regulation capacity. S4, during the real-time operation phase, receives AGC instructions from the power grid dispatching agency every 15 minutes. These instructions include the frequency regulation direction and regulation power requirements. S5, based on real-time photovoltaic output, charging load, energy storage state of charge and received AGC commands, constructs a real-time collaborative control model. The goal of this model is to minimize the deviation between real-time operation and the day-ahead bidding plan while meeting the accuracy and speed of grid command response, and to prioritize the use of photovoltaic fluctuations for internal balance. S6 solves the real-time collaborative control model, generates dynamic power limit instructions for charging pile groups, real-time charging and discharging power instructions for energy storage systems, and power adjustment instructions for settlement with the spot market, thereby realizing the collaborative control of photovoltaic, energy storage, charging load and grid interaction power. The method for generating dynamic power limit instructions for charging pile groups is as follows: Adjustable power is dynamically allocated based on the charging priority of electric vehicles, the current state of battery charge, and the user's selected participation demand response mode; for vehicles participating in frequency regulation services, their power can be adjusted within a preset range; for vehicles with priority protection, their power is not lower than the protection threshold. S7, after daily market settlement, calculates the actual revenue, forecast deviation, and instruction tracking performance indicators of each market, and uses historical data to continuously update the parameters of the forecast model, the statistical distribution of the probability of calling the ancillary service market, and the risk cost coefficient, thereby optimizing the bidding strategy for subsequent days.

[0007] Preferably, the risk-considered dual-track bidding optimization model constructed in step S2 has the objective function: Maximize(R1 + R2 - C1 - C2), where: R1 represents the expected revenue from electricity in the spot market, which is the difference between electricity sales revenue and electricity purchase cost; R2 represents the expected revenue from the ancillary services market, calculated based on the declared frequency regulation capacity, mileage unit price, and historical dispatch probability; C1 represents the estimated cost of assessment penalties due to forecast deviation; C2 represents the risk cost of revenue volatility in the value at risk, which quantifies the tail risk of market price and dispatch uncertainty through the conditional value at risk method. The dual-track bidding optimization model includes four constraints, specifically: The power balance constraint is that at any given moment, the sum of photovoltaic output, energy storage discharge power, and grid power purchase is equal to the sum of charging load, energy storage charging power, and grid power sales. Energy storage operation constraints include upper and lower limits of state of charge, limits of charging and discharging power, and constraints on the equality of the initial and final state of charge considering intraday cycles; Frequency regulation capability constraints: the declared upward frequency regulation capacity shall not exceed the sum of the current dischargeable power of energy storage and the power that photovoltaics can reduce; the declared downward frequency regulation capacity shall not exceed the sum of the current chargeable power of energy storage and the adjustable margin of charging load. The calculation method for the adjustable margin of charging load in this constraint is as follows: subtract the minimum average power required by all vehicles to ensure that they reach the target state of charge at the expected departure time from the real-time total charging demand power. This minimum average power is updated online based on the vehicle battery capacity, current state of charge, target state of charge and expected dwell time. Charging service constraints ensure that the total charging power provided to all connected electric vehicles at any given time is not less than their minimum demand threshold, which is dynamically determined based on user contracts and battery status.

[0008] As a preferred option, the real-time collaborative control model adopts a hierarchical structure, specifically as follows: The upper layer is the instruction decomposition layer, which decomposes the AGC adjustment demand issued by the power grid into energy storage power adjustment instructions, charging load aggregation adjustment instructions, and power deviation instructions that balance with the spot market in a 5-minute cycle. The lower layer is a fast tracking layer that operates on a second-level cycle. It receives instructions from the upper layer and uses a model predictive control method to perform rolling optimization based on the instantaneous fluctuations of photovoltaic and charging loads. This generates the final dynamic power limit for each charging pile and the millisecond-level charge and discharge setpoint for energy storage, ensuring that the total output power accurately tracks the combined value of the AGC instructions and the day-ahead power plan. The method for generating the dynamic power limit instruction for the charging pile group in step S6 is as follows: dynamically allocate adjustable power according to the charging priority of electric vehicles, the current state of battery charge, and the participation demand response mode selected by the user; for vehicles participating in frequency regulation services, their power can be adjusted within a preset range; for vehicles with priority protection, their power is not lower than the protection threshold.

[0009] Preferably, in the model solving of steps S3 and S5, for mixed integer programming problems containing integer variables, the branch and bound method is used; for convex optimization problems with continuous variables, the interior point method or quadratic programming algorithm is used, and it is guaranteed that a feasible solution is obtained within their respective decision time scales.

[0010] The beneficial effects of this invention are: 1. This invention constructs a day-ahead dual-track collaborative bidding optimization model, placing the electricity trading in the spot market and the capacity bidding in the ancillary services market within the same decision-making framework for synchronous joint optimization. This breaks the traditional sequential or independent decision-making model and realizes the optimal allocation of resources such as photovoltaic, energy storage and adjustable charging loads across markets in the spatiotemporal dimension. Thus, while ensuring charging services, it significantly improves the overall benefits of photovoltaic-storage charging stations participating in the electricity market.

[0011] 2. This invention embeds Conditional Value at Risk (CVaR) into the objective function of the dual-track bidding optimization model, quantifying market price fluctuations and frequency regulation uncertainty into manageable risk costs. This enables the bidding strategy to not only maximize expected returns but also proactively avoid extreme market risks, significantly enhancing the robustness and sustainability of the commercial operation of photovoltaic-storage charging stations. Experimental data show that this method can reduce the probability of loss in extreme scenarios from over 15% to less than 2%.

[0012] 3. The hierarchical collaborative control architecture of day-ahead planning plus real-time commands designed in this invention, especially the second-level fast tracking layer based on model predictive control at the lower layer, achieves millisecond-level accurate response to grid AGC commands and smooth processing of internal photovoltaic and load fluctuations. It solves the problems of poor command tracking accuracy and severe resource conflicts in traditional methods, reduces the average tracking error of frequency regulation commands by more than 50%, and significantly reduces the impact of the regulation process on the user's charging experience.

[0013] 4. This invention clearly proposes the concept and calculation method of minimum guaranteed charging power based on vehicle battery status and user reservations, and embeds it as a constraint into the bidding and real-time control model. This allows the adjustable margin of charging load to be accurately quantified and safely utilized. Thus, while deeply participating in grid frequency regulation ancillary services, it also ensures the core charging needs of electric vehicle users, achieving a combination of market benefits and charging service quality. Attached Figure Description

[0014] Figure 1 The diagram shown illustrates the overall workflow of this invention. Figure 2 The diagram shown illustrates the principle of the dual-track collaborative bidding optimization model of this invention. Figure 3 The diagram shown is a hierarchical structure diagram of the real-time collaborative control model of the present invention. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] Please see Figure 1 and Figure 2 This invention provides an embodiment of a dual-track coordinated control method for photovoltaic-storage-charging stations targeting the electricity spot market and ancillary services market: (1) Collect various data before the daily market reporting deadline, including: 1) For meteorological data, receive high-resolution numerical weather forecasts for the receiving area, including total horizontal radiation, direct radiation and ambient temperature for the next 72 hours, with a time resolution of no less than 15 minutes.

[0017] 2) For market data, obtain historical day-ahead price data and forecast curves for the electricity spot market, and historical capacity clearing prices, mileage clearing prices, and historical call probability distribution statistics for AGC instructions in the ancillary services market.

[0018] 3) For site data, monitor in real time the output of photovoltaic array, state of charge (SOC) of energy storage system, charging and discharging power and health status, connection status of each interface of charging pile, charging power, and vehicle battery information (current SOC, capacity, estimated dwell time, target SOC, etc.).

[0019] 4) For user reservation data, the charging station operation platform is used to obtain users' reservation charging information, including estimated arrival time, estimated departure time, required power or target SOC. This data is used to improve the accuracy of load forecasting and to formulate a reliable charging plan.

[0020] Based on the above data, a hybrid forecasting method combining physical models and data-driven approaches is adopted. Photovoltaic power generation forecasting first calculates baseline values ​​based on physical photovoltaic module models and weather data, then uses a neural network trained with historical error data for error correction. Baseline charging load forecasting combines historical load curves for the same period, real-time charging load, reservation information, and date type (weekday, holiday), generating a day-ahead forecast curve through a gradient boosting decision tree model. Before real-time operation, a Kalman filter-based regression model is used to continuously refresh photovoltaic output and charging load for the next 15 minutes to 1 hour, providing more accurate boundary conditions for real-time control.

[0021] (2) Construct a risk-considered dual-track bidding optimization model. Its objective function is to maximize the expected net revenue of the photovoltaic-storage charging station while minimizing the risk caused by market uncertainty. Specifically, the objective function is expressed as Maximize(R1 + R2 - C1 - C2), where: R1 represents the expected revenue from electricity sales in the spot market, which is the difference between the planned revenue from selling electricity in the spot market and the cost of purchasing electricity. Electricity sales originate from surplus photovoltaic output and energy storage discharge, while electricity purchases are used to supplement charging needs that photovoltaic and energy storage cannot meet, or to charge energy storage. R2 represents the anticipated revenue from the ancillary services market, primarily considering frequency modulation (FM) service revenue, which consists of two parts: capacity revenue and mileage revenue. This revenue is calculated based on the declared FM capacity, the published mileage price per unit in the market, and the expected probability of actual AGC command calls derived from historical statistics. C1 represents the estimated cost of penalties due to forecast deviations. It anticipates potential penalties for spot market imbalances or substandard frequency regulation performance if actual operating power deviates from the day-ahead bidding plan due to forecast deviations in photovoltaic and load data. This cost is quantified by introducing the probability distribution of forecast errors. C2 represents the cost of return volatility risk measured by Value at Risk (VaR). It quantifies the tail risk of market price and call uncertainty using the Conditional Value at Risk (VaR) method. VaR measures the average tail loss at a given confidence level. C2 is simply the VaR multiplied by a risk aversion coefficient. The main sources of risk include spot market price fluctuations, FM mileage price fluctuations, and the uncertainty of the actual call rate of AGC commands. The model characterizes these uncertainties by generating a large amount of historically statistically based market prices and call scenarios, and embeds VaR constraints or cost terms into the objective function, enabling the decision to pursue high returns while avoiding huge losses in extreme market conditions.

[0022] The model is solved under the following physical and market constraints: 1) For power balance constraints, it is necessary to ensure the instantaneous balance of power generation, power consumption and energy storage within the power station. That is, the sum of photovoltaic power, energy storage discharge power and power purchased from the grid must be equal to the sum of total charging load, energy storage charging power and power sold to the grid.

[0023] 2) For the operation constraints of energy storage systems, there are upper and lower limits of energy storage SOC, maximum charging and discharging power constraints and mutual exclusion constraints of charging and discharging states. It is usually required that the SOC at the beginning and end of a scheduling cycle (24 hours) is equal to ensure the sustainable recycling of energy storage. At the same time, the energy storage charging and discharging cycle losses can be converted into costs and included in the objective function.

[0024] 3) Regarding the constraints on frequency modulation capability applications, specifically: During any reporting period, the up-regulation capacity that can be reported (which requires increasing output or reducing power consumption) cannot exceed the sum of the current available discharge power margin of energy storage and the power that can be reduced from the predicted output of photovoltaic power.

[0025] The declared down-regulation capacity (which requires reducing output or increasing power consumption) cannot exceed the sum of the current available charging power margin of the energy storage and the adjustable charging load margin.

[0026] The adjustable charging load margin is the difference between the real-time total charging demand and the minimum guaranteed charging power. The minimum guaranteed charging power is calculated for all connected vehicles to ensure they reach the target State of Charge (SOC) upon expected departure. This power is dynamically calculated online and aggregated based on each vehicle's battery capacity, current SOC, target SOC, and remaining dwell time. This ensures that frequency regulation capability claims are made without compromising core charging service commitments.

[0027] 4) Regarding the charging service guarantee constraint, the constraint requires that at any time, the total power actually allocated to all charging vehicles shall not be lower than the minimum guaranteed charging power threshold calculated above, so as to guarantee the basic charging needs of users.

[0028] This model is a typical mixed-integer stochastic programming or robust optimization problem. In practical engineering applications, the sample averaging approximation method can be used to transform the stochastic programming problem into a large-scale deterministic mixed-integer linear programming problem, which can then be solved efficiently using a commercial optimization solver. The solution results are the spot market power purchase and sale plan every 15 minutes or hour for the next 24 hours, the baseline charging and discharging power plan of the energy storage system, and the time-of-use frequency regulation capacity curve submitted to the ancillary service market.

[0029] Please see Figure 3 (3) During the real-time operation phase, it is necessary to simultaneously respond to the power grid AGC commands, track the day-ahead plan, and handle internal fluctuations. This invention adopts a hierarchical real-time collaborative control architecture, specifically: 1) For the upper command decomposition layer, the basic operating cycle is 5 minutes. Its main task is to coordinate the received power grid AGC commands with the latest ultra-short-term forecast information and day-ahead plans.

[0030] Its inputs include the total AGC regulation demand issued by the grid for the next few minutes, real-time photovoltaic and ultra-short-term load forecasts, real-time SOC of energy storage, and the current adjustable margin of charging load. During processing, this layer runs a fast real-time rolling optimization model. The goal of this model is to minimize the deviation between the real-time operating total power and the combined value of "day-ahead planned power and AGC regulation demand," while optimizing internal resource allocation. Its outputs are the decomposed energy storage power regulation setpoint for the next 5-minute period, the aggregated power regulation target value for charging load, and the power deviation correction command required for settlement with the spot market.

[0031] 2) For the lower fast tracking layer, this layer receives the power adjustment target sent down from the upper layer and is responsible for executing it in a very short time.

[0032] This layer employs a model predictive control framework, performing rolling optimization with a 2-second cycle. It establishes a simplified linear or nonlinear predictive model encompassing photovoltaic, energy storage, and charging load aggregation. Its optimization objective is to ensure that the total exchange power between the charging station and the grid connection point accurately tracks the power reference trajectory issued from the upper layer within a short timeframe. Simultaneously, while maintaining tracking accuracy, it smooths energy storage power operations and dynamically allocates adjustable power limits among charging piles according to a predetermined priority strategy. Finally, the optimization results are converted into specific equipment instructions, including: millisecond-level charge / discharge power setpoints sent to the energy storage converter; and dynamic power limit instructions sent to the charging pile management system. For vehicles not participating in adjustment or with the highest priority, their charging power is unrestricted or only a minimum guaranteed power is set.

[0033] (4) After the daily market settlement is completed, this invention automatically collects the actual market clearing price, AGC call details, actual photovoltaic output, charging load, operating status of each device, and the final revenue settlement statement. Then, it calculates key performance indicators, including total revenue, frequency regulation command tracking performance indicators, day-ahead forecast mean absolute error, and planned deviation penalty costs. Then, it uses the collected new data to continuously update the parameters of the forecasting model.

[0034] Comparative Example 1 provided by the present invention: This simulation environment is based on the actual electricity market rules and meteorological data of a certain province in China. The photovoltaic-storage charging station is set up with the following configuration: photovoltaic installed capacity of 500kwp, energy storage system installed capacity of 0.5MW / 1MWh, total charging pile power of 1200kW, including 10 DC fast charging piles.

[0035] This example compares the present invention with Comparative Example 1. Comparative Example 1 employs sequential decision-making, initially focusing on energy storage and photovoltaic participation in peak-valley arbitrage in the spot market to formulate a charging and discharging plan. Then, within the remaining energy storage and load regulation capacity, it requests ancillary services, tracks only the day-ahead plan in real-time, and reserves a fixed proportion of energy storage capacity to respond to AGC. The simulated results for one quarter are shown in the table below.

[0036] Table 1 Comparison of Operational Results for One Quarter

[0037] As shown in Table 1, this invention, through collaborative optimization, slightly sacrifices some spot arbitrage profits (-8%), but allocates energy storage and load regulation resources more precisely and in larger quantities to the higher-priced ancillary services market, resulting in a significant 52.7% increase in ancillary service revenue and a final 20% increase in total revenue. Simultaneously, because real-time tiered control enables more precise resource allocation, the frequency regulation command completion rate is significantly improved, and because the strategy optimizes energy storage actions, its average daily cycle count is reduced, which is beneficial for extending equipment lifespan.

[0038] Comparative Example 2 is provided in this invention: This example uses the same simulation environment as Example 1, where the experimental group incorporates a Conditional Value at Risk (CVaR) cost term into the objective function of this invention, with a confidence level of 95%. Comparative Example 2 removes the Conditional Value at Risk (CVaR) cost term, focusing solely on maximizing expected return. This example simulates a week of extreme price volatility in the market, and the results are shown in the table below.

[0039] Table 2 Comparison of Market Performance During Weeks with Extreme Price Fluctuations

[0040] As shown in Table 2, under normal or favorable market conditions, the risk-insensitive strategy (Comparative Example 2) could potentially achieve higher average returns (75,000 vs. 58,000 this week) due to its more aggressive bidding. However, the method of this invention, through CVaR management, significantly smooths out return volatility (standard deviation decreased from 3.8 to 1.2). Most importantly, in extremely unfavorable market scenarios (tail risk), the method of this invention effectively avoids losses, with an average return of positive value (40,000 RMB) even in the worst-case 5% scenario, while Comparative Example 2 suffers an average loss of 25,000 RMB, with a loss probability as high as 15%. This demonstrates the significant value of the method of this invention in ensuring operational stability.

[0041] Comparative Example 3 is provided in this invention: The purpose of this example is to test the effect of the hierarchical real-time collaborative control architecture of the present invention. The comparative example 3 of this example adopts the traditional proportional allocation. Its real-time layer adopts a simple proportional allocation strategy, specifically, the adjustment demand is allocated proportionally according to the current power of each charging pile, and the energy storage responds to the remaining commands in a fixed proportion.

[0042] This example simulates a drastic up-frequency modulation command process lasting 10 minutes to observe the effects of both methods. Key performance indicators are shown in the table below.

[0043] Table 3 Comparison of the effects of hierarchical collaborative architecture and proportional allocation

[0044] As shown in Table 3, the hierarchical real-time collaborative control architecture adopted in this invention can proactively optimize resource allocation. When responding to frequency modulation commands, it can more intelligently determine when to prioritize adjusting the charging load and when to activate energy storage. This allows for a smoother reduction in charging power while maintaining tracking accuracy, minimizing the impact of adjustment actions on users and reducing the proportion of vehicles with delayed charging progress from 25% to 8%. Simultaneously, MPC optimizes the energy storage action sequence, avoiding unnecessary drastic fluctuations and benefiting equipment lifespan. In contrast, the simple proportional law in ratio 3 results in a rigid response, poor tracking performance, and a significant impact on user experience.

[0045] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A dual-track coordinated control method for photovoltaic-storage-charging stations targeting the electricity spot market and ancillary services market, characterized in that: Includes the following steps: S1, before the daily market declaration deadline, generates day-ahead forecast curves for photovoltaic power generation and base charging load based on high-precision numerical weather forecasts, historical load data and electric vehicle user reservation information; S2. Construct a dual-track bidding optimization model that takes into account risks. The model aims to maximize the overall operating revenue of photovoltaic-storage charging stations. The first track decision variable is the day-ahead electricity bidding curve in the electricity spot market, and the second track decision variable is the combination of frequency regulation capacity and frequency regulation mileage bidding in the ancillary service market. S3, for the dual-track bidding optimization model, takes into account the prediction curve, historical spot market prices, statistical characteristics of ancillary service market clearing prices and call-up probabilities, energy storage system charging and discharging efficiency and cycle life depreciation costs, and the constraint of uninterrupted charging services, and solves to obtain the optimal day-ahead bidding plan. S4, during the real-time operation phase, receives AGC instructions from the power grid dispatching agency every 15 minutes. These instructions include the frequency regulation direction and regulation power requirements. S5, based on real-time photovoltaic output, charging load, energy storage state of charge and received AGC commands, constructs a real-time collaborative control model. The goal of this model is to minimize the deviation between real-time operation and the day-ahead bidding plan while meeting the accuracy and speed of grid command response, and to prioritize the use of photovoltaic fluctuations for internal balance. S6 solves the real-time collaborative control model, generating dynamic power limit instructions for charging pile groups, real-time charging and discharging power instructions for energy storage systems, and power adjustment instructions settled with the spot market, thereby achieving collaborative control of the interactive power of photovoltaic, energy storage, charging loads, and grid.

2. The dual-track coordinated control method for photovoltaic-storage-charging stations oriented towards the electricity spot market and ancillary services market as described in claim 1, characterized in that, The risk-considered dual-track bidding optimization model constructed in step S2 has the objective function: Maximize(R1 + R2 - C1 - C2), where: R1 represents the expected revenue from electricity in the spot market, which is the difference between electricity sales revenue and electricity purchase cost. R2 represents the expected revenue from the ancillary services market, calculated based on the declared frequency regulation capacity, mileage unit price, and historical call probability. C1 is the estimated cost of performance penalties caused by prediction bias; C2 represents the cost of return volatility risk in value at risk, which quantifies the tail risk of market price and invoking uncertainty through the conditional value at risk method.

3. The dual-track coordinated control method for photovoltaic-storage-charging stations oriented towards the electricity spot market and ancillary services market as described in claim 2, is characterized in that... The dual-track bidding optimization model in step S2 includes the following constraints: The power balance constraint is that at any given moment, the sum of photovoltaic output, energy storage discharge power, and grid power purchase is equal to the sum of charging load, energy storage charging power, and grid power sales. Energy storage operation constraints include upper and lower limits of state of charge, limits of charging and discharging power, and constraints on the equality of the initial and final state of charge considering intraday cycles; Frequency regulation capacity constraints: the declared upward frequency regulation capacity shall not exceed the sum of the current dischargeable power of energy storage and the power that can be reduced by photovoltaic power; the declared downward frequency regulation capacity shall not exceed the sum of the current chargeable power of energy storage and the adjustable margin of charging load. Charging service constraints ensure that the total charging power provided to all connected electric vehicles at any given time is not less than their minimum demand threshold, which is dynamically determined based on user contracts and battery status.

4. The dual-track coordinated control method for photovoltaic-storage-charging stations oriented towards the electricity spot market and ancillary services market as described in claim 3, characterized in that: The calculation method for the adjustable margin of charging load in the frequency regulation capability constraint is as follows: subtract the minimum average power required by all vehicles to ensure that they reach the target state of charge at the expected departure time from the real-time total charging demand power. This minimum average power is updated online based on the vehicle battery capacity, current state of charge, target state of charge, and expected dwell time.

5. The dual-track coordinated control method for photovoltaic-storage-charging stations oriented towards the electricity spot market and ancillary services market as described in claim 4, characterized in that: The real-time collaborative control model in step S5 adopts a hierarchical structure.

6. The dual-track coordinated control method for photovoltaic-storage-charging stations oriented towards the electricity spot market and ancillary services market as described in claim 5, characterized in that, The hierarchical structure is specifically as follows: The upper layer is the instruction decomposition layer, which decomposes the AGC instructions issued by the power grid into energy storage power adjustment instructions, charging load aggregation adjustment instructions, and power deviation instructions that balance with the spot market, with a 5-minute cycle. The lower layer is a fast tracking layer that operates on a second-level cycle. It receives instructions from the upper layer and uses a model predictive control method to perform rolling optimization based on the instantaneous fluctuations of photovoltaic and charging loads. This generates the final dynamic power limit for each charging pile and the millisecond-level charge and discharge setpoint for energy storage, ensuring that the total output power accurately tracks the combined value of the AGC instructions and the day-ahead power plan.

7. The dual-track coordinated control method for photovoltaic-storage-charging stations oriented towards the electricity spot market and ancillary services market as described in claim 6, characterized in that: The optimal day-ahead bidding plan in step S3 includes the electricity sales / purchase volume in different time periods, the energy storage benchmark power point, and the declared frequency regulation capacity.

8. The dual-track coordinated control method for photovoltaic-storage-charging stations oriented towards the electricity spot market and ancillary services market as described in claim 7, characterized in that: The method for generating the dynamic power limit instruction for the charging pile group in step S6 is as follows: dynamically allocate adjustable power according to the electric vehicle charging priority, the current battery state of charge, and the user-selected participation demand response mode. For vehicles participating in frequency modulation services, their power can be adjusted within a preset range; for vehicles given priority, their power shall not be lower than the guarantee threshold.

9. The dual-track coordinated control method for photovoltaic-storage-charging stations oriented towards the electricity spot market and ancillary services market as described in claim 8, characterized in that: In the model solving process of steps S3 and S5, the branch and bound method is used to solve mixed integer programming problems containing integer variables; the interior point method or quadratic programming algorithm is used to solve convex optimization problems with continuous variables, and it is guaranteed that a feasible solution is obtained within their respective decision time scales.

10. The dual-track coordinated control method for photovoltaic-storage-charging stations oriented towards the electricity spot market and ancillary services market as described in claim 9, characterized in that: The method calculates the actual revenue, prediction deviation, and instruction tracking performance indicators for each market after daily market settlement. It then uses historical data to continuously update the parameters of the prediction model, the statistical distribution of the probability of calling ancillary service markets, and the risk cost coefficient, thereby optimizing the bidding strategy for subsequent days.

Images