A photovoltaic power prediction optimization method and system for a photovoltaic power station

Abstract

The present application relates to the technical field of prediction optimization, and discloses a photovoltaic power prediction optimization method and system for a photovoltaic power station, which comprises the following steps: step 1, normalizing photovoltaic probability prediction into quantile form, and determining a default probability budget and a standby trigger probability budget; step 2, comparing a confidence level prediction value with a maximum available output to obtain a baseline output; step 3, determining an upward adjustment standby capacity and a downward adjustment standby capacity, and setting a power range constraint; step 4, determining a bid curve segmentation, a standby quota and an energy storage plan; step 5, calculating a real-time deviation, and determining an error cost threshold; step 6, determining energy storage supplement charging capacity, discharging capacity and necessary limiting generation capacity, and solving a minimum total cost of rescheduling; and step 7, counting a default rate and a standby trigger rate, updating unified risk budget parameters, and generating a default and standby trigger probability budget. The present application realizes the efficiency and stability of power grid dispatching, and improves the power generation utilization rate of the photovoltaic power station.

Description

Technical Field

[0001] This invention belongs to the field of prediction and optimization technology, specifically relating to a method and system for predicting and optimizing photovoltaic power in photovoltaic power plants. Background Technology

[0002] With the rapid development of renewable energy, especially photovoltaic (PV) power generation, its large-scale integration into the electricity market has brought profound changes and challenges to traditional operating models. PV power generation is significantly affected by weather conditions, exhibiting high uncertainty and volatility. In the electricity market environment, this uncertainty directly translates into economic risks for power generators: if the predicted power output exceeds the actual generating capacity, it may lead to market contract defaults and incur high assessment fees; if the bid is too conservative, potential electricity sales revenue will be lost. Furthermore, to mitigate volatility and fulfill contracts, PV power plants typically need to purchase or reserve standby capacity services, which constitutes their main ancillary service costs. How to formulate the optimal bidding strategy and standby purchase plan based on uncertain forecasts and market rules to maximize profits and control risks has become a core decision-making challenge for PV power plant operators.

[0003] Meanwhile, energy storage systems, as important flexibility assets, provide photovoltaic power plants with new regulatory tools for market participation. Energy storage can not only participate in the energy market to achieve arbitrage opportunities during periods of low energy consumption and high energy generation, but also provide ancillary services such as backup power. However, its dispatching decisions are highly complex; charging and discharging strategies directly affect equipment lifespan and must be coordinated with the power plant's market application strategy. Traditional decision-making methods often consider power forecasting, market application, and energy storage operation plans separately, lacking a unified risk management and economic optimization framework. This makes it difficult to achieve the optimal balance between market revenue, backup costs, deviation assessments, and energy storage losses, thus affecting the overall market competitiveness and return on investment of the power plant. Summary of the Invention

[0004] This invention provides a photovoltaic power prediction and optimization method and system for photovoltaic power plants, which solves the technical problem in related technologies where photovoltaic power plants lack a unified risk quantification and economic optimization framework based on uncertain power prediction in the power market environment. This leads to fragmented market bidding, reserve purchase and energy storage trading strategies, making it difficult to coordinate and achieve maximum returns and controllable risks.

[0005] This invention provides a method for predicting and optimizing photovoltaic power in photovoltaic power plants, comprising the following steps:

[0006] Step 1: Obtain unified risk budget parameters, normalize the photovoltaic probability forecast into quantile form, and determine the default probability budget and reserve trigger probability budget for electricity market bidding and reserve dispatch based on the baseline risk coefficient and reserve risk coefficient.

[0007] Step 2: Select the corresponding confidence level forecast value in quantile form based on the default probability budget, and compare it with the maximum available output to obtain the baseline output for electricity market application;

[0008] Step 3: Determine whether to increase or decrease the reserve capacity based on the reserve trigger probability budget, and set a power range constraint consistent with the energy storage discharge power and energy storage charging power;

[0009] Step 4: Optimize baseline output, increase reserve capacity, decrease reserve capacity and energy storage charging and discharging power within a rolling time window, and output segmented pricing curves, reserve quotas and energy storage operation plans;

[0010] Step 5: Obtain the actual power output and calculate the real-time deviation, and determine the error cost threshold based on the principle of cost minimization to decide whether to trigger the rescheduling process;

[0011] Step 6: In the rescheduling process, determine the additional discharge amount, additional power supply, and necessary power restriction amount of the energy storage, solve for the adjustment result that minimizes the total rescheduling cost for the current period, and execute it.

[0012] Step 7: Calculate the default rate and the standby trigger rate, update the unified risk budget parameters according to the preset step size and the preset target level, and generate the default probability budget and the standby trigger probability budget accordingly.

[0013] This invention provides a photovoltaic power prediction and optimization system for photovoltaic power plants, comprising:

[0014] The risk budget and forecast module is used to obtain unified risk budget parameters, normalize the photovoltaic probability forecast into quantile form, and determine the default probability budget and reserve trigger probability budget for electricity market bidding and reserve dispatch based on the baseline risk coefficient and reserve risk coefficient.

[0015] The baseline output optimization module is used to select the corresponding confidence level prediction value in quantile form based on the default probability budget, and compare it with the maximum available output to obtain the baseline output for electricity market declaration;

[0016] The reserve capacity and constraint module is used to determine whether to increase or decrease the reserve capacity based on the reserve trigger probability budget, and to set power range constraints consistent with the energy storage discharge power and energy storage charging power.

[0017] The scheduling optimization and resource allocation module is used to optimize baseline output, increase or decrease reserve capacity and energy storage charging and discharging power in a rolling time window, and output price curve segments, reserve quotas and energy storage operation plans.

[0018] The deviation and error management module is used to obtain the actual power output and calculate the real-time deviation, determine the error cost threshold, and determine the switching conditions between the scheduling optimization and resource allocation module and rescheduling.

[0019] The rescheduling and adjustment module is used to determine the additional discharge amount, additional power amount, and necessary power restriction amount of energy storage during rescheduling, and to solve for the adjustment result that minimizes the total rescheduling cost for the current period and execute it.

[0020] The risk update and optimization module is used to calculate the default rate and the standby trigger rate, update the unified risk budget parameters according to the preset step size and the preset target level, and generate the default probability budget and the standby trigger probability budget accordingly.

[0021] The beneficial effects of this invention are as follows: By optimizing the power prediction and dispatch strategies of photovoltaic (PV) power plants and combining them with risk budget management, this invention achieves efficient coordination between grid dispatch and PV power generation. Specifically, by calculating the probability of default and the probability of reserve triggering, the output of PV power plants and the charging and discharging strategies of energy storage systems can be dynamically adjusted, ensuring grid stability in response to fluctuations in PV power generation. Based on error cost thresholds and energy storage dispatch optimization, dispatch fluctuations caused by prediction errors are reduced, effectively mitigating over-allocation of reserve capacity and unnecessary charging and discharging of energy storage systems, thereby improving the economy and flexibility of grid dispatch. Through rolling time window optimization, PV power plants can maximize power generation efficiency while ensuring the safe operation of the grid, reducing unnecessary energy waste. Furthermore, the risk budget optimization method of this invention provides a more intelligent and efficient dispatching means for the power system, enhancing the market participation capability of PV power plants and promoting the smooth integration and efficient utilization of renewable energy in the grid. Attached Figure Description

[0022] Figure 1 This is a flowchart of a photovoltaic power prediction and optimization method for photovoltaic power plants according to the present invention. Detailed Implementation

[0023] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0024] like Figure 1 As shown, a photovoltaic power prediction and optimization method for photovoltaic power plants includes the following steps:

[0025] Step 1: Obtain unified risk budget parameters, normalize the photovoltaic probability forecast into quantile form, and determine the default probability budget and reserve trigger probability budget for electricity market bidding and reserve dispatch based on the baseline risk coefficient and reserve risk coefficient.

[0026] Step 2: Select the corresponding confidence level forecast value in quantile form based on the default probability budget, and compare it with the maximum available output to obtain the baseline output for electricity market application;

[0027] Step 3: Determine whether to increase or decrease the reserve capacity based on the reserve trigger probability budget, and set a power range constraint consistent with the energy storage discharge power and energy storage charging power;

[0028] Step 4: Optimize baseline output, increase reserve capacity, decrease reserve capacity and energy storage charging and discharging power within a rolling time window, and output segmented pricing curves, reserve quotas and energy storage operation plans;

[0029] Step 5: Obtain the actual power output and calculate the real-time deviation, and determine the error cost threshold based on the principle of cost minimization to decide whether to trigger the rescheduling process;

[0030] Step 6: In the rescheduling process, determine the additional discharge amount, additional power supply, and necessary power restriction amount of the energy storage, solve for the adjustment result that minimizes the total rescheduling cost for the current period, and execute it.

[0031] Step 7: Calculate the default rate and the standby trigger rate, update the unified risk budget parameters according to the preset step size and the preset target level, and generate the default probability budget and the standby trigger probability budget accordingly.

[0032] In one embodiment of the present invention, a unified risk budget parameter is obtained, photovoltaic probability prediction is normalized into a quantile form, and default probability budget and reserve trigger probability budget for electricity market bidding and reserve dispatch are determined based on the baseline risk coefficient and reserve risk coefficient, including:

[0033] Step 11: Obtain unified risk budget parameters, baseline risk coefficient, reserve risk coefficient, rolling time window and maximum available output, fill missing moments with linear interpolation, and unify time granularity and power unit.

[0034] Among them, the unified risk budget parameter is dynamically determined by comprehensively analyzing historical operating data and prediction errors, combined with factors such as the design capacity and geographical location of the photovoltaic power station, and is used as an overall parameter to characterize the uncertainty risk that the entire photovoltaic power station system can withstand; the baseline risk coefficient is used to reflect the deviation between the baseline power and the actual power generation output, indicating the degree of risk that the photovoltaic power station's baseline power fails to reach the expected output; the reserve risk coefficient indicates the degree of risk that reserve capacity needs to be dispatched to cope with the error in the photovoltaic power station's power prediction; the maximum available output indicates the upper limit of the physical output that the photovoltaic power station can achieve under the current operating conditions; during the data acquisition process, if the photovoltaic power prediction or related parameters are missing at some times, linear interpolation is used to fill in the gaps, and the time granularity and power unit are unified for all data to ensure consistency in time scale and dimension between different data sources.

[0035] Step 12: The photovoltaic probability prediction is normalized into quantile form, and the upper and lower bounds are truncated according to the maximum available output. The data is then sorted from smallest to largest confidence level to eliminate non-monotonic data and form quantile data. Photovoltaic probability prediction is usually given in the form of probability distribution or multiple scenarios. Directly using it for scheduling decisions has the problems of structural complexity and insufficient interpretability. In this embodiment, the photovoltaic probability prediction is extracted according to different confidence levels, and each prediction value is limited to between zero and the maximum available output to ensure that the prediction results meet the physical operation constraints of the photovoltaic power station. Furthermore, non-monotonic cases that may occur during the sorting process are corrected to eliminate quantile crossover problems caused by prediction noise or modeling errors, and finally, monotonic and ordered quantile data is formed.

[0036] Step 13: The product of the unified risk budget parameter and the baseline risk coefficient is determined as the default probability budget, which is used to constrain the probability level of output below the baseline when selecting the baseline power of the photovoltaic power plant. The product of the unified risk budget parameter and the reserve risk coefficient is determined as the reserve trigger probability budget, which is used to constrain the probability level of the photovoltaic power plant triggering an increase or decrease in reserve during operation. Values ​​of the two probability budgets that exceed zero to one are replaced by boundary values ​​to ensure that the default probability budget and the reserve trigger probability budget are always within a reasonable range.

[0037] This embodiment constructs quantile-based data, enabling photovoltaic power prediction results to participate in grid dispatch decisions in a standardized and adjustable manner, thereby improving the availability and stability of the prediction results. By unifying risk budget parameters and their derived default probability budgets and reserve trigger probability budgets, this invention achieves effective risk control of uncertainties in photovoltaic power generation during grid dispatch, coordinating baseline default risk and reserve trigger risk. Through this optimization process, this invention can improve the power generation dispatch efficiency of photovoltaic power plants while ensuring the safe operation of the grid, reducing unnecessary reserve capacity configuration and power curtailment, thereby enhancing the dispatch flexibility and overall efficiency of the entire power system.

[0038] In one embodiment of the present invention, selecting a corresponding confidence level prediction value based on the default probability budget in quantile form and comparing it with the maximum available output to obtain the baseline output for electricity market declaration includes:

[0039] The confidence level is obtained by subtracting the default probability budget from 1. The confidence level ranges from 0 to 1 and represents the reliability of the predicted value. Based on this confidence level, the corresponding predicted value is selected from the data that has been normalized to quantile form. Photovoltaic power prediction data is usually given in quantile form, meaning that each confidence level corresponds to a specific power value. If the selected confidence level falls exactly at a discrete point, the predicted value corresponding to that point is directly taken; if the confidence level is between discrete points, linear interpolation is used to perform interpolation calculations between two adjacent prediction points to obtain the predicted value at that confidence level.

[0040] The selected confidence level prediction value is compared with the maximum available output of the photovoltaic power station. If the confidence level prediction value is less than zero, it means that the predicted photovoltaic power is negative. In this case, it should be set to zero, indicating that the photovoltaic power station is not generating electricity. If the confidence level prediction value is greater than zero, the smaller value between it and the maximum available output is taken as the baseline output, which represents the expected power generation of photovoltaic power without dispatch intervention.

[0041] Through the above steps, this invention provides a baseline output for the power grid dispatch system that conforms to the characteristics of photovoltaic power generation while also taking into account risk control, based on the default probability budget and photovoltaic power quantile prediction data, combined with the actual power generation capacity of the photovoltaic power plant. This process, by optimizing power prediction and risk budget management, effectively improves the coordination between the photovoltaic power plant and the power grid dispatch system, reduces dispatch instability caused by power prediction errors, thereby enhancing the power generation utilization efficiency and flexibility of the photovoltaic power plant, ensuring the stable operation of the power system in a highly uncertain power generation environment, and ultimately guaranteeing the safety and stability of the power grid.

[0042] In one embodiment of the present invention, determining whether to increase or decrease the reserve capacity based on the reserve trigger probability budget, and setting a power range constraint consistent with the energy storage discharge power and energy storage charging power, includes:

[0043] Step 21: Based on the standby trigger probability budget, determine two confidence levels in the quantile-form data for calculating standby capacity. Specifically, subtract the standby trigger probability budget from one confidence level to obtain the first confidence level, and use the standby trigger probability budget as the second confidence level. These two confidence levels correspond to the high and low ends of the photovoltaic power prediction distribution, respectively, and are used to characterize the potential upward or downward skewness of photovoltaic output under large fluctuations. Subsequently, select the corresponding predicted values ​​in the quantile-form data according to the two confidence levels.

[0044] Step 22: Compare the predicted value (with confidence level 1 minus the reserve trigger probability budget) with the baseline output. If the predicted value is greater than the baseline output, the difference is used to adjust the reserve capacity upwards; if the predicted value is not greater than the baseline output, the adjusted reserve capacity is zero. Correspondingly, compare the baseline output with the predicted value (with confidence level 1) of the reserve trigger probability budget. If the baseline output is greater than the predicted value, the difference is used to adjust the reserve capacity downwards; if the baseline output is not greater than the predicted value, the adjusted reserve capacity is zero. Adjusting the reserve capacity upwards covers situations where the photovoltaic output may be lower than the baseline output, while adjusting the reserve capacity downwards covers situations where the photovoltaic output may be higher than the baseline output, thus symmetrically quantifying the photovoltaic output deviation in a probabilistic sense.

[0045] Step 23: After determining whether to increase or decrease the reserve capacity, to ensure its effective implementation in actual operation, it is necessary to ensure consistency with the operational constraints of the energy storage system configured in the photovoltaic power plant. Specifically, the sum of the energy storage charging power and the decreased reserve capacity is limited to the maximum charging power range of the energy storage, and the sum of the energy storage discharging power and the increased reserve capacity is limited to the maximum discharging power range of the energy storage. This power range constraint ensures that when the reserve is triggered, the energy storage system has sufficient charging and discharging capacity to support the corresponding reserve adjustment needs, avoiding situations where the reserve capacity exceeds the physical capacity of the energy storage system, thereby improving the feasibility and operational safety of the reserve configuration.

[0046] Through the above implementation methods, this embodiment fully utilizes the quantile form data of photovoltaic power prediction, enabling the determination of reserve capacity to have a clear probabilistic meaning and risk control basis. By coordinating with the power range constraints of the energy storage system, the feasibility of reserve capacity in actual grid dispatch is ensured. Furthermore, through the application of this scheme, this invention can enhance the adaptability of photovoltaic power plants to power fluctuations while ensuring the safe operation of the power grid, reducing dispatch risks and system impacts caused by prediction errors. This method not only improves the output regulation capability of photovoltaic power plants but also further optimizes grid load dispatch, improving the overall operating efficiency and dispatch management effectiveness of the power system.

[0047] In one embodiment of the present invention, a rolling time window optimizes baseline output, increases reserve capacity, decreases reserve capacity and energy storage charging and discharging power, and outputs segmented pricing curves, reserve quotas and energy storage operation plans, including:

[0048] Step 31: Set a rolling time window. Within this window, power dispatch, reserve capacity configuration, and energy storage management are optimized to minimize the total cost. In each time period, the reserve cost coefficient is multiplied by the increase and decrease in reserve capacity, and then summed to obtain the reserve cost. The reserve cost coefficient is set based on the grid's dispatch requirements and market prices. The reserve cost is used to assess the economic cost of reserving reserve capacity to cope with the uncertainty of photovoltaic power plant power forecasts. The absolute difference between the baseline output and the predicted value for the corresponding time period in the quantile data is multiplied by the deviation cost coefficient to obtain the deviation cost. The deviation cost coefficient is set based on grid dispatch requirements and the volatility of photovoltaic power generation. The deviation cost is used to measure the cost resulting from the difference between the actual power generation of the photovoltaic power plant and the baseline output. The energy storage degradation cost coefficient is multiplied by the sum of the energy storage charging power and the energy storage discharging power to obtain the energy storage degradation cost. The energy storage degradation cost coefficient is obtained through equipment life cycle analysis. The energy storage degradation cost refers to the energy loss caused by the charging and discharging cycle and the long-term maintenance and replacement costs caused by the degradation of energy storage equipment during the charging and discharging process of the energy storage system. By summing these three costs periodically within a rolling time window, the total cost for that period is obtained. Ultimately, by minimizing the total cost, the optimized baseline output, increased reserve capacity, decreased reserve capacity, and energy storage charging / discharging power are derived. This process ensures that the photovoltaic power plant maximizes cost-effectiveness while guaranteeing stable grid operation and reduces unnecessary waste during system operation.

[0049] Step 32: The baseline output is limited to a range that is not less than zero and does not exceed the maximum available output, ensuring that the power output of the photovoltaic power station complies with physical limitations. The energy storage charging power is limited to between zero and the maximum charging power, and the energy storage discharging power is limited to between zero and the maximum discharging power, preventing the energy storage system from exceeding its charging and discharging capabilities. For more refined dispatching, the baseline output is divided into several segments, each with a corresponding segment price, and these prices are made to increase or decrease monotonically to conform to the electricity market's pricing rules and dispatching requirements. This ensures that dispatching decisions between the photovoltaic power station and the grid reflect the economic efficiency of the electricity market while optimizing dispatching.

[0050] Step 33: Combining the aforementioned optimization results, the increased and decreased reserve capacity for each time period are summarized to obtain the reserve quota; the energy storage charging power, energy storage discharging power, and energy storage state of charge for each time period are summarized to generate an energy storage operation plan. By combining this information with the needs of grid dispatch, the bidding curve segments, reserve quota, and energy storage operation plan are output so that the grid dispatch system can execute these optimization results. This output not only meets the needs of grid dispatch but also ensures the stable operation of photovoltaic power plants and the efficient utilization of energy storage resources.

[0051] This invention effectively improves the accuracy and efficiency of grid dispatch by comprehensively considering the baseline output, reserve capacity, and energy storage charging and discharging power of photovoltaic power plants within a rolling time window and optimizing their cost structure. Through the setting of segmented pricing and the generation of reserve quotas and energy storage operation plans, this invention improves the power utilization efficiency of photovoltaic power plants while reducing the grid's demand for reserve capacity, thus optimizing the trading and dispatching modes of the electricity market. Furthermore, the synergistic optimization of energy storage and reserve capacity further enhances the system's flexibility and adjustability, thereby improving the overall efficiency of the power system while ensuring the safe and stable operation of the grid.

[0052] In one embodiment of the present invention, the actual generated power is obtained and the real-time deviation is calculated. An error cost threshold is determined based on the principle of cost minimization to decide whether to trigger a rescheduling process, including:

[0053] Step 41: Obtain the actual power generation and baseline output of the photovoltaic power station, and ensure that they are aligned at the same time granularity and power unit. Actual power generation refers to the actual power generated by the photovoltaic power station in a certain period of time. The real-time deviation is obtained by calculating the difference between the actual power generation and the baseline output. The absolute value of the real-time deviation is taken as the absolute deviation, which serves as the basis for subsequent error cost threshold determination and rescheduling judgment.

[0054] Step 42: Set candidate error cost thresholds as the tolerance bandwidth for the absolute value of real-time deviation. For each candidate error cost threshold, calculate the total threshold cost under that threshold. Specifically, the difference between the absolute value of real-time deviation and the candidate error cost threshold is taken as the over-threshold deviation, and the product of the over-threshold deviation and the deviation cost is taken as the deviation cost item, reflecting the difference between power generation and the expected plan, usually the economic loss incurred when the photovoltaic power station fails to reach the expected power generation capacity. When the over-threshold deviation is greater than zero, calculate the difference between the maximum charging power of energy storage and the energy storage charging power, and the difference between the maximum discharging power of energy storage and the energy storage discharging power. Take the smaller of the two differences as the available adjustment amount of energy storage. The energy storage degradation cost and the available adjustment amount of energy storage are then combined. Multiplying these terms yields the energy storage degradation cost, which reflects the equipment degradation and energy efficiency loss caused by the charging and discharging process of the energy storage system, i.e., the long-term operating cost, which is usually related to the usage frequency of the energy storage system. The smaller value between increasing and decreasing the reserve capacity is taken as the reserve available adjustment amount, and the reserve cost is multiplied by the reserve available adjustment amount to obtain the reserve cost item. The above three items are added together to obtain the total threshold cost corresponding to the candidate error cost threshold. The threshold value corresponding to the minimum total threshold cost is selected as the error cost threshold. The error cost threshold reflects the maximum tolerable range of power generation deviation between the photovoltaic power plant and the grid, and provides a basis for subsequent dispatch decisions.

[0055] Step 43: Compare the absolute value of the real-time deviation with the error cost threshold. If the absolute value of the real-time deviation is not greater than the error cost threshold, it means that the deviation is within the acceptable range. At this time, the normal scheduling decision is executed, and the photovoltaic power plant power scheduling strategy in step 4 is continued. If the absolute value of the real-time deviation is greater than the error cost threshold, it means that the deviation exceeds the acceptable range. The system will automatically trigger the rescheduling process to recalculate the power scheduling and reserve capacity configuration of the photovoltaic power plant to ensure that the grid can operate smoothly and cope with large power fluctuations.

[0056] By calculating deviations in real time and determining error cost thresholds, this invention can effectively manage power fluctuations in photovoltaic power plants, ensuring the efficiency and flexibility of grid dispatching decisions. Specifically, by setting a reasonable error cost threshold, the grid can maintain its original dispatching strategy when photovoltaic power generation fluctuations are small. When the deviation exceeds the set threshold, the system can automatically trigger re-dispatch to reduce the dispatching risks caused by the uncertainty of photovoltaic power generation, thereby improving the stability and economy of the power system.

[0057] Furthermore, this invention optimizes the configuration of energy storage systems and reserve capacity, reducing over-provisioning of reserve capacity and overcharging / discharging of energy storage systems caused by prediction errors. This lowers grid operating costs and improves the power generation efficiency and flexibility of photovoltaic power plants. This process not only enhances the accuracy of grid dispatch but also reduces the potential risks to the power system caused by fluctuations in photovoltaic power generation.

[0058] In one embodiment of the present invention, the process of determining the additional discharge amount, additional charge amount, and necessary power rationing amount of energy storage during rescheduling, solving for and executing the adjustment result that minimizes the total rescheduling cost for the current time period, includes:

[0059] Step 51: Based on the operating status of the photovoltaic power station, calculate the charging power margin by subtracting the current charging power from the maximum charging power of the energy storage system. This charging power margin represents the maximum charging power that the energy storage system can still accept. Calculate the discharging power margin by subtracting the current discharging power from the maximum discharging power of the energy storage system. This discharging power margin represents the maximum discharging power that the energy storage system can still provide. Based on the upper and lower limits of the energy storage's state of charge (SOC), charging efficiency, and discharging efficiency, determine the allowable range of charging and discharging amounts for the current time period. The upper and lower limits of SOC refer to the minimum and maximum charge levels that the batteries of the energy storage devices in the energy storage system can withstand.

[0060] Step 52: Subtract the sum of the additional energy storage capacity, the additional energy storage discharge capacity, and the necessary power restriction capacity from the absolute deviation. If the difference is greater than zero, it is taken as the pricing deviation; if the difference is not greater than zero, the pricing deviation is zero; otherwise, the pricing deviation is the difference. Add the product of deviation cost and pricing deviation, the product of the sum of additional energy storage capacity and additional energy storage discharge capacity and the energy storage degradation cost coefficient, and the product of the necessary power restriction cost coefficient and the necessary power restriction capacity to obtain the total rescheduling cost. Minimize the total rescheduling cost under comprehensive feasibility constraints to obtain the additional energy storage discharge capacity, additional energy storage capacity, and necessary power restriction capacity. Among them, the necessary power restriction cost coefficient is determined according to the grid dispatch demand and the electricity market price. The necessary power restriction capacity refers to the amount of power generation that the photovoltaic power station must reduce due to grid dispatch or other external factors within a certain period of time. The first product reflects the economic impact of deviation on grid dispatch, the second product reflects the degradation cost of energy storage equipment during operation, and the third product reflects the additional cost of photovoltaic power station due to dispatch constraints.

[0061] The comprehensive feasibility constraints include: the supplementary energy storage capacity is not greater than the charging power margin, the supplementary energy storage discharge capacity is not greater than the discharge power margin, the necessary power limitation is not greater than the baseline output, the supplementary energy storage discharge capacity is set to zero when the real-time deviation is greater than zero, and the supplementary energy storage capacity and the necessary power limitation are set to zero when the real-time deviation is less than zero.

[0062] The additional charging power required by the energy storage system during the current period represents the additional charging power needed by the system. The charging power margin is the difference between the current charging power and the maximum charging power of the energy storage system. The charging power margin limits the maximum charging capacity of the energy storage system, ensuring that the charging power does not exceed its designed maximum power. The additional discharging power required by the energy storage system during the current period represents the additional discharging power needed by the system. The discharging power margin is the difference between the current discharging power and the maximum discharging power of the energy storage system, limiting the maximum discharging capacity of the energy storage system.

[0063] Step 53: After completing the rescheduling optimization, update the charging power and discharging power of the energy storage based on the calculated additional energy storage charge and additional energy storage discharge. The updated energy storage charging power is the sum of the energy storage charging power and the additional energy storage charge; the updated energy storage discharging power is the sum of the energy storage discharging power and the additional energy storage discharge.

[0064] Based on the updated energy storage charging and discharging power and the operating status of the energy storage system, corresponding scheduling adjustments are made to ensure the coordinated operation of the photovoltaic power station and the energy storage system, and to achieve optimal grid scheduling.

[0065] This embodiment effectively balances the difference between photovoltaic power plant generation and grid load by introducing optimized calculations of energy storage supplementary discharge, energy storage supplementary power, and necessary power curtailment during the rescheduling process. By optimizing the configuration of energy storage charging and discharging power and reserve capacity, it ensures that grid dispatch can flexibly adjust according to the real-time situation of photovoltaic power generation, thereby improving the power generation utilization rate of photovoltaic power plants. Through reasonable rescheduling optimization, it reduces the over-configuration of reserve capacity and lowers grid dispatch costs. By rationally configuring energy storage and reserve capacity, this invention ensures the stable operation of the grid in the face of uncertainties in photovoltaic power generation and reduces fluctuations and instability in the grid during dispatching.

[0066] In one embodiment of the present invention, the default rate and the standby trigger rate are statistically analyzed, and a unified risk budget parameter is updated according to a preset step size and a preset target level, thereby generating a default probability budget and a standby trigger probability budget, including:

[0067] Step 61: Align the actual generated power with the baseline output within the rolling time window. The proportion of periods where the actual generated power is less than the baseline output is used as the default rate. The default rate reflects the deviation between the actual and expected power generation of the photovoltaic power plant, and is used to assess the predictability of photovoltaic power generation and the risk of grid dispatch. Calculate the reserve capacity triggering status of the photovoltaic power plant, and count the periods where reserve capacity is increased to a value greater than zero and the periods where reserve capacity is decreased to a value greater than zero. Combine these two types of periods and count their proportion of the total periods as the reserve trigger rate. The reserve trigger rate indicates the frequency with which the reserve capacity of the photovoltaic power plant is called up in actual operation, reflecting the utilization of the grid's reserve capacity.

[0068] Step 62: Calculate the difference between the default rate and the preset target level and the difference between the standby trigger rate and the preset target level. Multiply the two differences by the preset step size and add them to the unified risk budget parameter to obtain the temporary parameter. Limit the temporary parameter to the preset range to obtain the updated unified risk budget parameter.

[0069] Step 63: The product of the updated unified risk budget parameter and the baseline risk coefficient is determined as the default probability budget, which represents the probability level that the photovoltaic power plant will output less than the baseline under given conditions; the product of the updated unified risk budget parameter and the reserve risk coefficient is determined as the reserve trigger probability budget, which controls the probability level of reserve capacity triggering; the default probability budget and the reserve trigger probability budget are limited to between zero and one, and marked as effective for the next rolling time window.

[0070] This embodiment updates the risk budget based on real-time default rates and reserve trigger rates, enabling the power grid to more precisely adjust the output of photovoltaic power plants and the scheduling of energy storage systems, ensuring a balance between grid load and generation capacity. By dynamically updating unified risk budget parameters, the power grid can better manage risks, avoiding grid instability caused by photovoltaic power plant generation errors; enabling the power grid to flexibly respond to the uncertainties of photovoltaic power plant generation, ensuring the safe, stable, and efficient operation of the power grid.

[0071] This invention provides a photovoltaic power prediction and optimization system for photovoltaic power plants, comprising:

[0072] The risk budget and forecast module is used to obtain unified risk budget parameters, normalize the photovoltaic probability forecast into quantile form, and determine the default probability budget and reserve trigger probability budget for electricity market bidding and reserve dispatch based on the baseline risk coefficient and reserve risk coefficient.

[0073] The baseline output optimization module is used to select the corresponding confidence level prediction value in quantile form based on the default probability budget, and compare it with the maximum available output to obtain the baseline output for electricity market declaration;

[0074] The reserve capacity and constraint module is used to determine whether to increase or decrease the reserve capacity based on the reserve trigger probability budget, and to set power range constraints consistent with the energy storage discharge power and energy storage charging power.

[0075] The scheduling optimization and resource allocation module is used to optimize baseline output, increase or decrease reserve capacity and energy storage charging and discharging power in a rolling time window, and output price curve segments, reserve quotas and energy storage operation plans.

[0076] The deviation and error management module is used to obtain the actual power output and calculate the real-time deviation, determine the error cost threshold, and determine the switching conditions between the scheduling optimization and resource allocation module and rescheduling.

[0077] The rescheduling and adjustment module is used to determine the additional discharge amount, additional power amount, and necessary power restriction amount of energy storage during rescheduling, and to solve for the adjustment result that minimizes the total rescheduling cost for the current period and execute it.

[0078] The risk update and optimization module is used to calculate the default rate and the standby trigger rate, update the unified risk budget parameters according to the preset step size and the preset target level, and generate the default probability budget and the standby trigger probability budget accordingly.

[0079] It should be noted that the interval and threshold sizes are set for ease of comparison. The size of the threshold depends on the amount of sample data and the base number set by those skilled in the art for each set of sample data, as long as it does not affect the proportional relationship between the parameter and the quantized value. Furthermore, the above formulas are all dimensionless calculations, and the formulas are derived from software simulations using a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0080] The embodiments of the present invention have been described above, but the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of the present embodiments, all of which are within the protection scope of the present embodiments.

Claims

1. A photovoltaic power prediction and optimization method for photovoltaic power plants, characterized in that, Includes the following steps: Step 1: Obtain unified risk budget parameters, normalize the photovoltaic probability forecast into quantile form, and determine the default probability budget and reserve trigger probability budget for electricity market bidding and reserve dispatch based on the baseline risk coefficient and reserve risk coefficient. Step 2: Select the corresponding confidence level forecast value in quantile form based on the default probability budget, and compare it with the maximum available output to obtain the baseline output for electricity market application; Step 3: Determine whether to increase or decrease the reserve capacity based on the reserve trigger probability budget, and set a power range constraint consistent with the energy storage discharge power and energy storage charging power; Step 4: Optimize baseline output, increase reserve capacity, decrease reserve capacity and energy storage charging and discharging power within a rolling time window, and output segmented pricing curves, reserve quotas and energy storage operation plans; Step 5: Obtain the actual power output and calculate the real-time deviation, and determine the error cost threshold based on the principle of cost minimization to decide whether to trigger the rescheduling process; Step 6: In the rescheduling process, determine the additional discharge amount, additional power supply, and necessary power restriction amount of the energy storage, solve for the adjustment result that minimizes the total rescheduling cost for the current period, and execute it. Step 7: Calculate the default rate and the standby trigger rate, update the unified risk budget parameters according to the preset step size and the preset target level, and generate the default probability budget and the standby trigger probability budget accordingly.

2. The photovoltaic power prediction and optimization method for photovoltaic power plants according to claim 1, characterized in that, Obtain unified risk budget parameters, normalize photovoltaic probability forecasts into quantile form, and determine default probability budgets and reserve trigger probability budgets for electricity market bidding and reserve dispatch based on baseline risk coefficients and reserve risk coefficients, including: Step 11: Obtain unified risk budget parameters, baseline risk coefficient, reserve risk coefficient, rolling time window and maximum available output, fill missing moments with linear interpolation, and unify time granularity and power unit. Step 12: Normalize the photovoltaic probability prediction into quantile form, truncate the upper and lower bounds according to the maximum available power output, sort according to the confidence level from small to large, eliminate non-monotonic data, and form quantile form data; Step 13: The product of the unified risk budget parameter and the baseline risk coefficient is determined as the default probability budget, and the product of the unified risk budget parameter and the standby risk coefficient is determined as the standby trigger probability budget. Values ​​of the two probability budgets that exceed zero to one are replaced by boundary values.

3. The photovoltaic power prediction and optimization method for photovoltaic power plants according to claim 1, characterized in that, Based on the default probability budget, the corresponding confidence level forecast value is selected in quantile form, and compared with the maximum available output to obtain the baseline output for electricity market declaration, including: The confidence level is obtained by subtracting the default probability budget from 1. Based on the confidence level, the predicted value corresponding to the confidence level is selected from the quantile data as the confidence level prediction value. If the confidence level does not fall within the discrete point set, the linear interpolation method is used to calculate the confidence level prediction value. The predicted value is compared with the maximum available output. If the predicted value is less than zero, it is set to zero; otherwise, the smaller value between the predicted value and the maximum available output is taken as the baseline output.

4. The photovoltaic power prediction and optimization method for photovoltaic power plants according to claim 1, characterized in that, Determine whether to increase or decrease reserve capacity, and set power range constraints, including: Step 21: Based on the backup trigger probability budget, determine two confidence levels in the quantile form data, namely, one minus the backup trigger probability budget and the backup trigger probability budget, and select the corresponding predicted values ​​in the quantile form data according to the two. Step 22: Subtract the baseline output from the predicted value of the standby trigger probability budget with a confidence level of 1. If the difference is positive, the standby capacity is increased; otherwise, the standby capacity is increased to zero. Subtract the predicted value of the standby trigger probability budget with a confidence level of 1 from the baseline output. If the difference is positive, the standby capacity is decreased; otherwise, the standby capacity is decreased to zero. Step 23: Limit the sum of the energy storage charging power and the reduced reserve capacity to the maximum energy storage charging power, and limit the sum of the energy storage discharging power and the increased reserve capacity to the maximum energy storage discharging power.

5. The photovoltaic power prediction and optimization method for photovoltaic power plants according to claim 1, characterized in that, The rolling time window optimizes baseline output, increases reserve capacity, decreases reserve capacity and energy storage charging and discharging power, and outputs segmented pricing curves, reserve quotas and energy storage operation plans, including: Step 31: Set a rolling time window, multiply the reserve cost coefficient by the upward and downward reserve capacity respectively and add them together to get the reserve cost, multiply the deviation cost coefficient by the absolute difference between the baseline output and the predicted value of the quantile data in the corresponding period to get the deviation cost, multiply the energy storage degradation cost by the sum of the energy storage charging power and the energy storage discharging power to get the energy storage degradation cost, add the above three items in each period within the rolling time window to get the total cost, and take the minimum total cost as the objective. Step 32: Limit the baseline output to a range greater than or equal to zero and not exceeding the maximum available output; limit the energy storage charging power to a range between zero and the maximum energy storage charging power; limit the energy storage discharging power to a range between zero and the maximum energy storage discharging power; divide the baseline output into several segments; assign a corresponding segment price to each segment output; and make the segment prices monotonic. Step 33: Summarize the increased and decreased reserve capacity for each time period into reserve quota, and summarize the energy storage charging power, energy storage discharging power and energy storage state of charge for each time period into energy storage operation plan, and output the price quotation curve segments, reserve quota and energy storage operation plan.

6. The photovoltaic power prediction and optimization method for photovoltaic power plants according to claim 1, characterized in that, The system acquires the actual generated power and calculates the real-time deviation, then determines the error cost threshold based on the principle of cost minimization to decide whether to trigger the rescheduling process, including: Step 41: Obtain the actual power output, align it with the baseline output in terms of time granularity and power unit, take the difference between the actual power output and the baseline output as the real-time deviation, and take the absolute value of the real-time deviation as the absolute deviation. Step 42: Use the candidate error cost threshold as the tolerance bandwidth of the absolute value of the real-time deviation, and calculate the total threshold cost for each candidate error cost threshold. Specifically, the difference between the absolute value of the real-time deviation and the candidate error cost threshold is taken as the over-threshold deviation, and the product of the over-threshold deviation and the deviation cost is taken as the deviation cost item. When the over-threshold deviation is greater than zero, the difference between the maximum energy storage charging power and the energy storage charging power, and the difference between the maximum energy storage discharging power and the energy storage discharging power are calculated. The smaller of the two differences is taken as the energy storage available adjustment amount, and the energy storage degradation cost is multiplied by the energy storage available adjustment amount to obtain the energy storage degradation cost item. The smaller of the increase and decrease of the reserve capacity is taken as the reserve available adjustment amount, and the reserve cost is multiplied by the reserve available adjustment amount to obtain the reserve cost item. The above three items are added together to obtain the total threshold cost corresponding to the candidate error cost threshold, and the threshold value corresponding to the minimum total threshold cost is selected as the error cost threshold. Step 43: Compare the absolute value of the real-time deviation with the error cost threshold. If the absolute value of the real-time deviation is not greater than the error cost threshold, proceed to step 4. If the absolute value of the real-time deviation is greater than the error cost threshold, proceed to rescheduling.

7. The photovoltaic power prediction and optimization method for photovoltaic power plants according to claim 1, characterized in that, In the rescheduling process, the additional discharge amount, additional charge amount, and necessary power rationing amount of energy storage are determined. The adjustment result that minimizes the total rescheduling cost for the current time period is calculated and executed, including: Step 51: Subtract the energy storage charging power from the maximum energy storage charging power to obtain the charging power margin, and subtract the energy storage discharging power from the maximum energy storage discharging power to obtain the discharging power margin. Step 52: Subtract the sum of the additional energy storage capacity, the additional energy storage discharge capacity, and the necessary power restriction capacity from the absolute deviation. If the difference is greater than zero, it is taken as the pricing deviation; if the difference is not greater than zero, the pricing deviation is zero. Add the product of the deviation cost coefficient and the pricing deviation, the product of the sum of the additional energy storage capacity and the additional energy storage discharge capacity and the energy storage degradation cost coefficient, and the product of the necessary power restriction cost and the necessary power restriction capacity to obtain the total rescheduling cost. Minimize the total rescheduling cost under the comprehensive feasibility constraint to obtain the additional energy storage discharge capacity, the additional energy storage capacity, and the necessary power restriction capacity. Step 53: Add the energy storage charging power to the energy storage supplementary power to obtain the updated energy storage charging power, and add the energy storage discharging power to the energy storage supplementary discharge power to obtain the updated energy storage discharging power.

8. The photovoltaic power prediction and optimization method for photovoltaic power plants according to claim 7, characterized in that, The comprehensive feasibility constraints include: the supplementary energy storage capacity is not greater than the charging power margin, the supplementary energy storage discharge capacity is not greater than the discharge power margin, the necessary power limitation is not greater than the baseline output, the supplementary energy storage discharge capacity is set to zero when the real-time deviation is greater than zero, and the supplementary energy storage capacity and the necessary power limitation are set to zero when the real-time deviation is less than zero.

9. The photovoltaic power prediction and optimization method for photovoltaic power plants according to claim 1, characterized in that, The default rate and standby trigger rate are statistically analyzed, and the unified risk budget parameters are updated according to a preset step size and a preset target level. Based on this, default probability budget and standby trigger probability budget are generated, including: Step 61: Align the actual power output with the baseline output within the rolling time window. Use the percentage of periods when the actual power output is less than the baseline output as the default rate. Combine the periods when the reserve capacity is increased to a value greater than zero with the periods when the reserve capacity is decreased to a value greater than zero, and use the percentage of these periods as the reserve trigger rate. Step 62: Calculate the difference between the default rate and the preset target level and the difference between the standby trigger rate and the preset target level. Multiply the two differences by the preset step size and add them to the unified risk budget parameter to obtain the temporary parameter. Limit the temporary parameter to the preset range to obtain the updated unified risk budget parameter. Step 63: The product of the updated unified risk budget parameter and the baseline risk coefficient is determined as the default probability budget, and the product of the updated unified risk budget parameter and the standby risk coefficient is determined as the standby trigger probability budget. Both are limited to between zero and one, and marked as effective for the next rolling time window.

10. A photovoltaic power prediction and optimization system for photovoltaic power plants, characterized in that, The photovoltaic power prediction and optimization method for photovoltaic power plants as described in any one of claims 1-9 includes: The risk budget and forecast module is used to obtain unified risk budget parameters, normalize the photovoltaic probability forecast into quantile form, and determine the default probability budget and reserve trigger probability budget for electricity market bidding and reserve dispatch based on the baseline risk coefficient and reserve risk coefficient. The baseline output optimization module is used to select the corresponding confidence level prediction value in quantile form based on the default probability budget, and compare it with the maximum available output to obtain the baseline output for electricity market declaration; The reserve capacity and constraint module is used to determine whether to increase or decrease the reserve capacity based on the reserve trigger probability budget, and to set power range constraints consistent with the energy storage discharge power and energy storage charging power. The scheduling optimization and resource allocation module is used to optimize baseline output, increase or decrease reserve capacity and energy storage charging and discharging power in a rolling time window, and output price curve segments, reserve quotas and energy storage operation plans. The deviation and error management module is used to obtain the actual power output and calculate the real-time deviation, determine the error cost threshold, and determine the switching conditions between the scheduling optimization and resource allocation module and rescheduling. The rescheduling and adjustment module is used to determine the additional discharge amount, additional power amount, and necessary power restriction amount of energy storage during rescheduling, and to solve for the adjustment result that minimizes the total rescheduling cost for the current period and execute it. The risk update and optimization module is used to calculate the default rate and the standby trigger rate, update the unified risk budget parameters according to the preset step size and the preset target level, and generate the default probability budget and the standby trigger probability budget accordingly.

Images