Energy management method and system for a photovoltaic power plant

Abstract

The application relates to an energy management method and system of a photovoltaic power station. The energy management method of the photovoltaic power station comprises the following steps: acquiring power generation power prediction deviation information, environmental change indication information and photovoltaic array efficiency deviation information of the photovoltaic power station; evaluating the explanation capability of the environmental change indication information and wide-spectrum irradiance of a corresponding region to the prediction deviation and the efficiency deviation according to the power generation power prediction deviation information, the environmental change indication information and the photovoltaic array efficiency deviation information; generating a hidden risk signal when the explanation capability is lower than a preset risk threshold and the efficiency deviation continuously exists; quantifying an uncertainty index of power generation power prediction generated according to the power generation power prediction deviation information, the environmental change indication information and the hidden risk signal, and determining a risk preference level of the photovoltaic power station based on the uncertainty index; obtaining a storage battery charging and discharging correction strategy and a power grid grid-connected power correction strategy based on the risk preference level, and generating and issuing a control instruction.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of energy management, in particular to an energy management method and system for a photovoltaic power station. BACKGROUND

[0002] The energy management system of a large ground photovoltaic power station relies on a power generation power short-term prediction function to formulate daily energy output and grid-connected strategy. In the early stage of construction, the function is based on long-term historical operation data and fine regional meteorological model, and the prediction is accurate, which can effectively guide energy scheduling.

[0003] However, as the operation time of the power station increases, the prediction deviation problem gradually emerges. On the one hand, the geographical environment around the power station continues to change slightly: natural vegetation growth causes moving shadows on the photovoltaic array under certain seasons and sunlight angles, changes the local irradiance; hydrological environment fluctuations affect the convective heat dissipation on the surface of the components; the construction of small buildings around or the widening of roads changes the local wind field, affecting the heat dissipation and dust accumulation of the components. These micro-changes have local and subtle nature, which are difficult to capture by conventional macro-weather station sensors, but gradually deviate the basic assumptions of the prediction model from the actual situation.

[0004] On the other hand, the performance of photovoltaic components degrades after long-term operation: the surface anti-reflection coating is degraded due to ultraviolet radiation, wind and rain erosion, polymer structure decomposition or micro-wear, reducing light transmittance; the glass panel is scratched or the internal structure changes, reducing the light transmittance. This uniform degradation and microclimate influence superimpose, resulting in persistent and cumulative deviations in the prediction function at specific times such as early morning and evening, and conventional parameter adjustment cannot eliminate this systematic error.

[0005] Prediction deviation directly leads to energy scheduling errors: the charging and discharging plan of energy storage batteries and the grid-connected power strategy cannot match the actual operation. If the predicted power generation is lower than the actual power generation, the excess power may be wasted due to full storage or limited grid connection; if the predicted power generation is higher than the actual power generation, it may over-discharge and accelerate battery wear, or fail to meet the grid connection commitment and affect the stability of the grid.

[0006] To cope with the scheduling problem, the operation and maintenance team needs to frequently intervene manually, compare data and adjust parameters, which not only increases labor intensity and cost, but also makes it difficult to achieve real-time optimization due to decision lag and dependence on experience, and may even exacerbate scheduling efficiency loss.

[0007] In addition, the grid-connected power frequently deviates from the plan, resulting in the stability of the power grid being affected, and the power grid dispatching center frequently issuing warnings. According to the examination clauses of the grid-connected protocol, the power station gradually accumulates the penalty for power deviation out of the range, which becomes a heavy operating cost, and the problem is upgraded from internal efficiency optimization to an urgent task that affects economic benefits and market reputation. At the same time, the energy storage battery ages faster than expected due to non-optimal charging and discharging cycles, such as frequent shallow charging and discharging, deep discharging, or non-ideal temperature operation. The battery health state estimation model does not fully consider the influence of non-optimal working conditions, and the estimation accuracy decreases, making it difficult for the power station to grasp the true condition of the battery and increasing the long-term operation risk.

[0008] In summary, long-term running photovoltaic power stations urgently need a new energy management method to dynamically adapt to prediction uncertainty, real-time optimization scheduling, and decision-making combined with device health status to ensure long-term stable operation. SUMMARY

[0009] The present application provides an energy management method and system for a photovoltaic power station to at least solve the problem of power generation prediction deviation of large ground photovoltaic power stations caused by the surrounding environment and component aging, as well as a series of operation problems such as energy management system scheduling mismatch, high cost of manual intervention, abnormal power grid warning and penalty, etc.

[0010] In a first aspect, the present application provides an energy management method for a photovoltaic power station, comprising the following steps:

[0011] Obtain the power generation prediction deviation information, environmental change indication information and photovoltaic array efficiency deviation information of the photovoltaic power station, wherein the power generation prediction deviation information includes the prediction deviation between the actual power generation and the predicted power generation of the photovoltaic power station, and the trend of the prediction deviation; the environmental change indication information includes macro weather forecast information and local environmental change information collected by the microclimate sensor inside the photovoltaic power station; the photovoltaic array efficiency deviation information includes the efficiency deviation between the instantaneous photoelectric conversion efficiency of the photovoltaic array and the pre-constructed dynamic expected efficiency benchmark, and the trend of the efficiency deviation;

[0012] According to the power generation prediction deviation information, the environmental change indication information and the photovoltaic array efficiency deviation information, evaluate the explanation ability of the environmental change indication information and the wide spectrum irradiance of the corresponding area to the prediction deviation and the efficiency deviation;

[0013] When the explanation ability is lower than a preset risk threshold and the efficiency deviation persists, generate a hidden risk signal;

[0014] According to the power generation prediction deviation information, the environmental change indication information, and the implicit risk signal, an uncertainty index of the power generation prediction is quantitatively generated, and based on the uncertainty index, a risk preference level of the photovoltaic power station is determined;

[0015] Based on the risk preference level, a backup capacity reservation ratio and a charge-discharge rate of the energy storage battery are adjusted to obtain a charge-discharge correction strategy of the energy storage battery;

[0016] According to the risk preference level, an upper limit of grid-connected power is set to obtain a grid-connected power correction strategy;

[0017] According to the charge-discharge correction strategy of the energy storage battery and the grid-connected power correction strategy, a control instruction is generated and delivered to an energy storage battery management system and a photovoltaic inverter.

[0018] Optionally, the obtaining of the photovoltaic array efficiency deviation information comprises:

[0019] Based on the continuously collected actual output power of the photovoltaic array, the wide spectrum irradiance of the corresponding area, and the surface temperature of the photovoltaic component, an instantaneous photoelectric conversion efficiency of the photovoltaic array is calculated and obtained;

[0020] A dynamic expected efficiency benchmark is established;

[0021] The instantaneous photoelectric conversion efficiency is compared with the dynamic expected efficiency benchmark to obtain an efficiency deviation;

[0022] The persistence of the efficiency deviation is tracked to obtain the photovoltaic array efficiency deviation information.

[0023] Optionally, when the interpretation capability is lower than a preset risk threshold and the efficiency deviation persists, an implicit risk signal is generated, comprising:

[0024] The mode of the efficiency deviation is identified and classified, and the mode is associated with potential implicit risk sources;

[0025] The contribution degree of each of the potential implicit risk sources to the efficiency deviation is evaluated, and a comprehensive risk degree is comprehensively determined according to the contribution degrees;

[0026] According to the comprehensive risk degree, an implicit risk signal reflecting the comprehensive risk degree is generated.

[0027] Optionally, the determination of the risk preference level of the photovoltaic power station based on the uncertainty index comprises:

[0028] The current value of the uncertainty index and the change trend of the uncertainty index in a preset time window are monitored;

[0029] start a timer when the uncertainty index approaches a switching threshold of the risk preference level, wherein a delay time is pre-configured in the timer;

[0030] perform switching of the risk preference level when the uncertainty index is stable in a new level region within the delay time, and adjust a hysteresis interval of the level switching according to a fluctuation amplitude of the uncertainty index.

[0031] Optionally, the adjusting the reserve capacity reservation ratio and the charging and discharging rate of the energy storage battery based on the risk preference level to obtain a charging and discharging correction strategy of the energy storage battery comprises:

[0032] obtain operation data of each battery module, and evaluate a health state of each battery module according to the operation data;

[0033] differentially adjust a reserve capacity reservation ratio of each battery module according to the risk preference level and the health state;

[0034] differentially adjust a charging and discharging rate of each battery module according to the risk preference level and the health state.

[0035] Optionally, the setting an upper limit of grid-connected power according to the risk preference level to obtain a grid-connected power correction strategy comprises:

[0036] obtain operation data of a photovoltaic inverter and a combiner box, wherein the operation data comprises output power, operating temperature, and fault indication of the inverter, and branch current and insulation state of the combiner box;

[0037] evaluate a health state and a current maximum output capability of the photovoltaic inverter and the combiner box according to the operation data;

[0038] set an upper limit of grid-connected power according to the risk preference level, the predicted power generation, and the health state and the current maximum output capability of the photovoltaic inverter and the combiner box, wherein the upper limit of grid-connected power is not higher than the current maximum output capability of the photovoltaic inverter and the combiner box.

[0039] Optionally, the evaluating an explanation capability of the environmental change indication information and wide spectrum irradiance of a corresponding region to the prediction deviation and the efficiency deviation comprises:

[0040] calculate a correlation degree between the environmental change indication information and the wide spectrum irradiance and the prediction deviation and the efficiency deviation in different preset time windows, respectively;

[0041] According to the correlation degrees calculated in each time window, an optimal time scale with the strongest explanatory ability is identified, and a sampling period and a data smoothing window for the explanatory ability evaluation are adjusted according to the optimal time scale;

[0042] Under the adjusted sampling period and data smoothing window, the environmental change indication information and the wide-spectrum irradiance of the corresponding area are evaluated for the explanatory ability to the prediction deviation and the efficiency deviation.

[0043] Optionally, the quantification of the uncertainty index of the generated power prediction according to the generated power prediction deviation information, the environmental change indication information, and the implicit risk signal comprises:

[0044] The fluctuation characteristics and the time sequence correlation among the generated power prediction deviation information, the environmental change indication information, and the implicit risk signal on different time scales are calculated.

[0045] According to the fluctuation characteristics and the time sequence correlation, a contribution mode of the generated power prediction deviation information, the environmental change indication information, and the implicit risk signal to the uncertainty of the generated power prediction is identified.

[0046] According to the contribution mode, the weight and the nonlinear mapping relationship of the generated power prediction deviation information, the environmental change indication information, and the implicit risk signal in quantifying the uncertainty index are dynamically adjusted, wherein when a multiple collinearity relationship is identified, the weight of the information with strong correlation is reduced.

[0047] According to the adjusted weight and the nonlinear mapping relationship, the generated power prediction deviation information, the environmental change indication information, and the implicit risk signal are fused to generate the uncertainty index.

[0048] Optionally, the setting of the upper limit of the grid-connected power according to the risk preference level to obtain the grid-connected power correction strategy further comprises:

[0049] Detailed information of all power deviation penalty clauses in the grid-connected protocol is obtained, wherein the detailed information includes the trigger condition, the penalty calculation method, and the penalty degree of each penalty clause.

[0050] According to the risk preference level, the risk of each power deviation penalty clause is evaluated in combination with the uncertainty range of the predicted generated power, and the probability and the expected penalty cost of triggering each power deviation penalty clause under different upper limits of the grid-connected power are quantified.

[0051] The probability and the expected penalty cost are integrated, and the grid-connected power upper limit is iteratively adjusted to find an optimal grid-connected power upper limit value, wherein the optimal grid-connected power upper limit value minimizes the integrated expected penalty cost of all the power deviation penalty terms under the premise of meeting the risk preference level requirement;

[0052] In the iterative adjustment process, when the adjusted grid-connected power upper limit causes the expected penalty cost of a certain power deviation penalty term to significantly increase, the grid-connected power upper limit is preferentially adjusted to reduce the expected penalty cost of the power deviation penalty term.

[0053] In a second aspect, the present application provides an energy management system of a photovoltaic power plant, the system comprising:

[0054] An information acquisition module is configured to acquire power generation prediction deviation information, environmental change indication information, and photovoltaic array efficiency deviation information of the photovoltaic power plant, wherein the power generation prediction deviation information comprises a prediction deviation between an actual power generation and a predicted power generation of the photovoltaic power plant, and a change trend of the prediction deviation; the environmental change indication information comprises macro weather forecast information and local environmental change information collected by a microclimate sensor inside the photovoltaic power plant; and the photovoltaic array efficiency deviation information comprises an efficiency deviation between an instantaneous photoelectric conversion efficiency of a photovoltaic array and a pre-constructed dynamic expected efficiency benchmark, and a change trend of the efficiency deviation;

[0055] An explanation capability evaluation module is configured to evaluate an explanation capability of the environmental change indication information and wide-spectrum irradiance of a corresponding region on the prediction deviation and the efficiency deviation according to the power generation prediction deviation information, the environmental change indication information, and the photovoltaic array efficiency deviation information;

[0056] A hidden risk signal generation module is configured to generate a hidden risk signal when the explanation capability is lower than a preset risk threshold and the efficiency deviation persists;

[0057] A risk preference determination module is configured to quantify an uncertainty index of power generation prediction according to the power generation prediction deviation information, the environmental change indication information, and the hidden risk signal, and determine a risk preference level of the photovoltaic power plant based on the uncertainty index;

[0058] An energy storage battery strategy correction module is configured to adjust a backup capacity reservation ratio and a charge-discharge rate of an energy storage battery based on the risk preference level to obtain a charge-discharge correction strategy of the energy storage battery;

[0059] A grid-connected power correction module is configured to set a grid-connected power upper limit according to the risk preference level to obtain a grid-connected power correction strategy;

[0060] The control instruction issuing module is configured to generate and issue control instructions to the energy storage battery management system and the photovoltaic inverter according to the energy storage battery charging and discharging correction strategy and the grid-connected power correction strategy.

[0061] Compared with the related art, the energy management method and system of the photovoltaic power station provided by the present application at least has the following technical effects:

[0062] By obtaining the power generation prediction deviation information, the environmental change indication information and the photovoltaic array efficiency deviation information, various factors affecting the power generation performance of the photovoltaic power station can be comprehensively and meticulously captured. Among them, the power generation prediction deviation information provides a basis for identifying systematic errors; the environmental change indication information can accurately reflect the actual operating environment of the power station; and the photovoltaic array efficiency deviation information can accurately identify the efficiency decay caused by hidden problems such as component aging. Subsequently, by evaluating the explanatory ability of the environmental change indication information and the wide spectrum irradiance on the power generation prediction deviation and the efficiency deviation, the contribution degree of these factors to the prediction accuracy and the efficiency decay is analyzed in depth. When the explanatory ability is lower than the preset threshold and the efficiency deviation persists, a hidden risk signal is generated, solving the problem that it is difficult to identify and quantify the gradual progressive microclimate change and the hidden risks such as component aging in the prior art. Further, the power generation prediction uncertainty index is quantitatively generated by comprehensively considering the power generation prediction deviation information, the environmental change indication information and the hidden risk signal, and the risk preference level of the power station is determined according to the index. This mechanism enables the energy management system to dynamically adjust its risk tolerance according to the actual risk level faced by the power station. Finally, according to the determined risk preference level, the reserve capacity reservation ratio and the charging and discharging rate of the energy storage battery are adjusted, and the upper limit of the grid-connected power is set, to obtain the adjusted energy storage battery charging and discharging strategy and the grid-connected power strategy. The dynamically adjusted strategy enables the energy scheduling of the photovoltaic power station to more accurately match the actual power generation and the grid demand, significantly reducing manual intervention and operation and maintenance costs; at the same time, the grid-connected power strategy is optimized, effectively avoiding the penalty clauses in the grid-connected power protocol, reducing the risk of fines, and improving the economic benefits of the photovoltaic power station.

[0063] In summary, through multi-dimensional information acquisition, hidden risk identification, uncertainty quantification and risk preference level determination, the present application realizes the intelligentization and refinement of the energy management of the photovoltaic power station, effectively solves a series of problems such as prediction deviation, scheduling mismatch, high operation and maintenance cost and fine risk in the prior art caused by microclimate change and component aging, and improves the operation efficiency, economic benefits and grid stability of the photovoltaic power station.

[0064] The details of one or more embodiments of the present application are set forth in the following drawings and description, so that other features, objects and advantages of the present application are more apparent. BRIEF DESCRIPTION OF DRAWINGS

[0065] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0066] Figure 1 This is a flowchart illustrating an energy management method for a photovoltaic power plant according to an exemplary embodiment.

[0067] Figure 2 This is a flowchart illustrating step S3 according to an exemplary embodiment.

[0068] Figure 3 This is a partial flowchart illustrating step S4 according to an exemplary embodiment.

[0069] Figure 4 This is a flowchart illustrating step S5 according to an exemplary embodiment.

[0070] Figure 5 This is a flowchart illustrating step S6 according to an exemplary embodiment.

[0071] Figure 6 This is a flowchart illustrating step S2 according to an exemplary embodiment.

[0072] Figure 7 This is a block diagram illustrating an energy management system for a photovoltaic power plant according to an exemplary embodiment. Detailed Implementation

[0073] To make the objectives, technical solutions, and advantages of this application clearer, the application is described and illustrated below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application. All other embodiments obtained by those skilled in the art based on the embodiments provided in this application without inventive effort are within the scope of protection of this application.

[0074] Obviously, the accompanying drawings described below are merely some examples or embodiments of this application. Those skilled in the art can apply this application to other similar scenarios based on these drawings without any inventive effort. Furthermore, it is understood that although the efforts made in this development process may be complex and lengthy, for those skilled in the art related to the content disclosed in this application, any changes to design, manufacturing, or production based on the technical content disclosed in this application are merely conventional technical means and should not be construed as insufficient disclosure of the content of this application.

[0075] In this application, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that is mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0076] Unless otherwise defined, the technical or scientific terms used in this application shall have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms “a,” “an,” “an,” “the,” and similar words used in this application do not indicate quantity limitation and may indicate singular or plural. The terms “comprising,” “including,” “having,” and any variations thereof used in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units not listed, or may include other steps or units inherent to these processes, methods, products, or devices. The terms “connected,” “linked,” “coupled,” and similar words used in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. “Multiple” used in this application refers to two or more. “And / or” describes the relationship between related objects, indicating that three relationships may exist; for example, “A and / or B” can represent: A alone, A and B simultaneously, and B alone. The character " / " generally indicates that the preceding and following objects are in an "or" relationship. The terms "first," "second," and "third" used in this application are merely to distinguish similar objects and do not represent a specific ordering of the objects.

[0077] In related technologies, as power plants age, prediction biases become increasingly prominent. On one hand, the surrounding geographical environment undergoes subtle but continuous changes: natural vegetation growth causes shifting shadows on photovoltaic arrays under specific seasons and solar angles, altering local irradiance; fluctuations in the hydrological environment affect convective heat dissipation from the module surface; and the construction of small buildings or road widening alters the local wind field, impacting module heat dissipation and dust accumulation. These micro-changes are localized and subtle, difficult for conventional macro-meteorological sensors to capture, yet they cause the fundamental assumptions of prediction models to gradually deviate from reality.

[0078] On the other hand, photovoltaic modules show performance degradation after long-term operation: the surface anti-reflection coating is degraded due to ultraviolet radiation, wind and sand and rain erosion, polymer structure decomposition or micro-abrasion, reducing light transmittance; the glass panel is scratched or the internal structure changes, and the light transmittance decreases. This uniform degradation is superimposed with microclimate effects, resulting in persistent and cumulative deviations in the prediction function at specific times such as early morning and evening, and regular parameter adjustments cannot eliminate these systematic errors.

[0079] Prediction errors directly lead to inaccurate energy scheduling: the charging and discharging plan of energy storage batteries and the grid-connected power strategy cannot match the actual operation. If the predicted power generation is lower than the actual power generation, the excess power may be wasted due to full storage or limited grid connection; if the predicted power generation is higher than the actual power generation, the battery may be over-discharged to accelerate battery wear or may not meet the grid connection commitment to affect the stability of the power grid.

[0080] To deal with scheduling problems, the operation and maintenance team needs to frequently intervene manually, compare data and adjust parameters, which not only increases labor intensity and cost, but also makes it difficult to achieve real-time optimization due to decision lag and reliance on experience, and may even exacerbate scheduling efficiency loss.

[0081] In addition, frequent deviations of grid-connected power from the plan affect the stability of the power grid, and the power grid dispatching center frequently issues warnings. According to the examination clauses of the grid connection agreement, the power plant will be fined for power deviation beyond the range, and the fine will gradually accumulate, becoming a heavy operating cost. The problem has escalated from internal efficiency optimization to an urgent task that affects economic benefits and market reputation.

[0082] At the same time, the energy storage battery ages faster than expected due to non-optimal charging and discharging cycles, such as frequent shallow charging and discharging, deep discharging, or non-ideal temperature operation. The battery health state estimation model does not fully consider the impact of non-optimal working conditions, resulting in decreased estimation accuracy, making it difficult for the power plant to grasp the true state of the battery and increasing the risk of long-term operation.

[0083] Based on the above, the embodiments of the present application provide a photovoltaic power plant energy management method and system, which are described in detail below in conjunction with specific embodiments and drawings.

[0084] Embodiment 1

[0085] The embodiments of the present application provide a photovoltaic power plant energy management method. Figure 1 is a flow chart of a photovoltaic power plant energy management method according to an exemplary embodiment. As shown in Figure 1 , the method comprises the following steps:

[0086] S1, obtain power generation prediction deviation information, environment change indication information and photovoltaic array efficiency deviation information of the photovoltaic power station; wherein the power generation prediction deviation information includes a prediction deviation between the actual power generation and the predicted power generation of the photovoltaic power station, and a change trend of the prediction deviation; the environment change indication information includes macro weather forecast information and local environment change information collected by microclimate sensors inside the photovoltaic power station; and the photovoltaic array efficiency deviation information includes an efficiency deviation between the instantaneous photoelectric conversion efficiency of the photovoltaic array and the pre-constructed dynamic expected efficiency benchmark, and a change trend of the efficiency deviation.

[0087] In the embodiment, the power generation prediction deviation information is the difference between the actual power generation and the predicted power generation of the photovoltaic power station, which not only contains the instantaneous prediction deviation, but also covers the change trend of the prediction deviation over time, which helps to capture the dynamic mismatch between the prediction model and the actual operation. Specifically, the instantaneous deviation is obtained by monitoring the actual power generation of the photovoltaic power station in real time and comparing it with the pre-set predicted power generation. For example, the actual power generation data is collected every 5 minutes and compared with the predicted data to calculate the instantaneous deviation. At the same time, the change trend of the instantaneous deviation can be obtained by time series analysis of these instantaneous deviation data, for example, using moving average or exponential smoothing method.

[0088] The environment change indication information is a comprehensive concept, which not only includes traditional macro weather forecast data, but also emphasizes the local environment change data provided by the local microclimate sensors inside the power station, such as local temperature, humidity, wind speed, etc., which have a significant impact on the performance of the photovoltaic array. Specifically, it is obtained from two aspects: the macro weather forecast information can be obtained by accessing the data interface of the third-party weather service platform or the weather bureau, providing regional weather, temperature, humidity, etc. data; and the local microclimate sensors inside the power station, such as temperature sensors, wind speed sensors, humidity sensors installed in different areas of the photovoltaic array, can collect local environmental data in real time to reflect the microclimate changes inside the power station.

[0089] The photovoltaic array efficiency deviation information is obtained by monitoring the output power, irradiance and component temperature of the photovoltaic array in real time, and comparing with the dynamic expected efficiency benchmark, aiming to reflect the deviation degree and persistence of the actual photoelectric conversion efficiency of the photovoltaic array. Specifically, the steps of obtaining the photovoltaic array efficiency deviation information include:

[0090] S101, based on the continuously collected actual output power of the photovoltaic array, the wide spectrum irradiance of the corresponding area and the surface temperature of the photovoltaic component, the instantaneous photoelectric conversion efficiency of the photovoltaic array is calculated and obtained;

[0091] In this embodiment, continuous acquisition is achieved by installing power sensors, wide-spectrum irradiance meters, and temperature sensors on each photovoltaic string or subarray. Based on these data, the instantaneous photoelectric conversion efficiency of the photovoltaic array can be calculated.

[0092] S102, establishing a dynamic expected efficiency benchmark;

[0093] In this embodiment, historical data and machine learning models can be used to consider factors such as component aging and seasonal changes to establish a dynamic expected efficiency benchmark to predict the theoretical best efficiency under current irradiance and temperature conditions.

[0094] S103, comparing the instantaneous photoelectric conversion efficiency with the dynamic expected efficiency benchmark to obtain the efficiency deviation, tracking the persistence of the efficiency deviation, and obtaining the photovoltaic array efficiency deviation information;

[0095] In this embodiment, the instantaneous photoelectric conversion efficiency is compared with the dynamic expected efficiency benchmark to obtain the efficiency deviation. Finally, by continuously tracking these efficiency deviations, it can be determined whether they have persistence, thereby identifying potential systemic problems.

[0096] S2, according to the power generation prediction deviation information, the environmental change indication information, and the photovoltaic array efficiency deviation information, evaluating the explanation ability of the environmental change indication information and the wide-spectrum irradiance of the corresponding area to the prediction deviation and the efficiency deviation;

[0097] In this embodiment, the explanation ability is used to understand to what extent external environmental factors can explain the current prediction deviation and efficiency problem. For example, through statistical methods such as correlation analysis or regression analysis, the degree of correlation between macro weather forecast information, local microclimate information, and wide-spectrum irradiance and power generation prediction deviation and efficiency deviation is quantified. If it is found that the prediction deviation and efficiency deviation are highly correlated with a certain environmental factor (such as a local temperature anomaly) in a certain period of time, it indicates that the environmental factor has strong explanation ability.

[0098] S3, generating a hidden risk signal when the explanation ability is lower than the preset risk threshold and the efficiency deviation persists;

[0099] In this embodiment, the hidden risk signal is a warning signal generated by the system when the explanation ability of environmental changes and irradiance to prediction deviation and efficiency deviation is insufficient and the efficiency deviation persists, indicating that there may be deep problems that are not captured by conventional models. When conventional environmental factors are insufficient to explain the current problem, there may be deeper and more difficult to detect risks. For example, if the persistent efficiency deviation cannot be explained by known irradiance or temperature changes, it may indicate that there are hidden defects in the photovoltaic components or internal connection problems in the array.

[0100] S4, according to the power generation power prediction deviation information, the environmental change indication information and the implicit risk signal, quantitatively generate the uncertainty index of the power generation power prediction, and determine the risk preference level of the photovoltaic power station based on the uncertainty index;

[0101] In this embodiment, the power generation power prediction uncertainty index is a comprehensive quantitative index that comprehensively considers the above-mentioned various information, and is used to evaluate the reliability of future power generation power prediction. It not only considers the direct deviation of the prediction, but also integrates the complexity of the environmental change and the potential implicit risk. For example, fuzzy logic or neural network model can be used to fuse these multi-source information, and output an uncertainty index between 0 and 1, wherein a higher index represents a higher prediction uncertainty.

[0102] The risk preference level is the risk bearing capacity of the photovoltaic power station determined according to the uncertainty index. For example, the uncertainty index can be divided into several intervals, and each interval corresponds to a risk preference level, such as "conservative", "moderate" or "aggressive". When the uncertainty index is low, the photovoltaic power station can adopt a more aggressive strategy to maximize the benefit; and when the uncertainty index is high, a more conservative strategy should be adopted to avoid risks.

[0103] S5, based on the risk preference level, adjusting the reserve capacity reservation ratio and the charging and discharging rate of the energy storage battery to obtain a charging and discharging correction strategy of the energy storage battery;

[0104] In this embodiment, the reserve capacity reservation ratio and the charging and discharging rate of the energy storage battery are key parameters of the operation of the energy storage system. By adjusting these key parameters, the response capability and economic benefit of the energy storage system can be optimized. For example, when the risk preference level is "conservative", the reserve capacity reservation ratio of the energy storage battery can be increased to cope with sudden power generation power fluctuations, and the charging and discharging rate can be reduced to prolong the battery life and improve the system stability. Conversely, when the risk preference level is "aggressive", the reserve capacity reservation ratio can be appropriately reduced, and the charging and discharging rate can be increased to maximize the utilization efficiency and economic benefit of the energy storage system.

[0105] S6, according to the risk preference level, setting the upper limit of the grid-connected power to obtain a grid-connected power correction strategy;

[0106] In this embodiment, the upper limit of the grid-connected power is the maximum limit of the power transmitted by the photovoltaic power station to the grid, and its setting needs to consider the stability of the grid, the income of the power station and the potential risk of penalty. For example, when the risk preference level is "conservative", a lower upper limit of the grid-connected power can be set to ensure the stability of the grid-connected power and avoid triggering the penalty clause of the grid due to power fluctuations. When the risk preference level is "aggressive", a higher upper limit of the grid-connected power can be set to maximize the grid-connected income of the power station under the premise of ensuring the safety of the grid.

[0107] S7、According to the energy storage battery charge and discharge correction strategy and the grid-connected power correction strategy, control instructions are generated and issued to the energy storage battery management system and the photovoltaic inverter;

[0108] In this embodiment, the control instructions will guide the energy storage battery system to perform charge and discharge operations according to the new strategy, and control the photovoltaic inverter to adjust the grid-connected power, so as to finally realize fine management of the energy output of the photovoltaic power station.

[0109] The technical solutions of the above embodiments introduce multi-dimensional information perception, not only obtain power generation power prediction deviation information, but also obtain environmental change indication information and photovoltaic array efficiency deviation information, so that the system can more comprehensively and deeply understand the source of power generation power prediction uncertainty. Subsequently, a hidden risk signal generation mechanism is introduced. When the environmental change and the wide spectrum irradiance have a lower explanation ability for the power generation power prediction deviation and the efficiency deviation than a preset threshold and the efficiency deviation persists, the system generates a hidden risk signal. This breaks through the limitation of the prior art that only relies on explicit data for analysis, can early warn possible deep problems that are not captured by conventional models, such as component hidden defects or system failures, and thus realizes earlier risk identification and intervention. Then, by comprehensively quantifying the power generation power prediction deviation information, the environmental change indication information, and the hidden risk signal, a power generation power prediction uncertainty index is generated, and the risk preference level of the power station is determined according to the index. This enables the energy management strategy to be dynamically adjusted according to the risk bearing capacity of the power station. For example, when the uncertainty is high, the system will adopt a more conservative energy storage battery charge and discharge strategy and grid-connected power strategy to avoid risks; and when the uncertainty is low, a more aggressive strategy can be adopted to maximize economic benefits. This dynamic strategy adjustment based on the risk preference level enables the energy management system to more flexibly and intelligently adapt to the actual operating conditions of the power station and external environmental changes.

[0110] In summary, the above embodiments of the present application effectively solve the problems of persistent prediction deviation, high cost of manual intervention, and impact on grid stability in the prior art through the innovative combination of multi-dimensional information perception, intelligent risk assessment, and dynamic strategy adjustment. Not only does it improve the accuracy and efficiency of photovoltaic power station energy management, but also reduces operating costs and potential penalty risks, providing technical support for the long-term stable operation and economic benefit improvement of photovoltaic power stations.

[0111] In one possible design, Figure 2 is a flowchart of step S3 according to an example embodiment. Referring to the accompanying drawings, Figure 2 Step S3 includes:

[0112] S301, identify and classify the mode of efficiency deviation, and associate the mode with potential hidden risk sources;

[0113] In this embodiment, the observed efficiency deviation is classified into different types or patterns by deep analysis of the efficiency deviation information of the photovoltaic array, using time series analysis, clustering algorithm or pattern recognition techniques. These patterns can include but are not limited to: persistent small decline, periodic fluctuation, sudden large decline, local regional deviation, etc. Subsequently, for each type of efficiency deviation pattern identified, a mapping relationship between the pattern and the potential latent risk source that can cause the deviation is established. For example, persistent small decline can be related to long-term aging degradation of the photovoltaic component; periodic fluctuation can be related to local shading (such as bird droppings, dust accumulation) or sensor drift; sudden large decline can be directed to physical damage (such as cracks, hot spots) of the component or inverter failure. In practical applications, this association can be established through expert experience, historical failure data analysis, machine learning model training, etc.

[0114] S302, evaluate the contribution of each potential latent risk source to the efficiency deviation, and determine the comprehensive risk level according to the contribution;

[0115] In this embodiment, for each type of efficiency deviation pattern identified, the contribution of each potential latent risk source associated with the pattern to the formation or persistence of the pattern is quantified. For example, the relative influence weight of different risk sources (such as component aging, local shading, inverter efficiency decline, etc.) on the overall efficiency deviation can be calculated by statistical analysis, causal inference model or simulation based on physical model. After quantifying the contribution of each risk source, a comprehensive risk level indicator is calculated according to pre-set rules or algorithms. For example, if the contribution of multiple risk sources is high and the interaction is complex, a weighted average or fuzzy logic based comprehensive risk level can be calculated.

[0116] S303, generating a latent risk signal reflecting the comprehensive risk level according to the comprehensive risk level;

[0117] In this embodiment, the above-mentioned comprehensive risk level is converted into a specific, identifiable and processable latent risk signal by the system. The latent risk signal not only indicates the existence of latent risk, but further contains the nature of the risk (such as "component aging risk", "local shading risk") or its severity (such as "moderate risk", "high risk"). For example, a structured signal containing risk type code and risk level value is generated.

[0118] The technical solutions of the above embodiments can specifically convert the fuzzy "efficiency deviation" into understandable abnormal forms through mode recognition and classification of the efficiency deviation. On this basis, by associating these modes with potential implicit risk sources and further evaluating the contribution degrees of the risk sources, the system can deeply mine the root causes behind the efficiency deviation from the appearance. Thus, not only whether there is an implicit risk can be identified, but also the type, source and severity of the risk can be determined, so as to convert a general "implicit risk signal" into a signal with diagnostic and guiding significance. This refined risk identification mechanism provides more accurate and insightful input for the subsequent quantification of the power generation power prediction uncertainty index, effectively making up for the limitations of generating a general signal only by the interpretability and efficiency deviation.

[0119] In one example, it is assumed that the monitoring system of the photovoltaic power station continuously collects the operation data of the photovoltaic array and calculates the efficiency deviation. When it is found by evaluation that the environmental change indication information and the wide spectrum irradiance have an interpretability lower than a preset threshold to the efficiency deviation, and the efficiency deviation persists, the system will start the implicit risk signal generation process.

[0120] Specifically, the system first analyzes the historical efficiency deviation data and identifies a "persistent slight decline" efficiency deviation mode. Through a pre-established knowledge base or machine learning model, the system associates this mode with multiple potential implicit risk sources, such as long-term degradation of photovoltaic components, poor contact of junction boxes, and local slight dust accumulation.

[0121] Subsequently, the system will evaluate the contribution degrees of these potential risk sources to the current "persistent slight decline" mode. For example, by analyzing the operation life of the components and the historical degradation curve, the contribution degree of the long-term degradation of the components is calculated to be 60%; by analyzing the temperature data and current fluctuation of the junction box, the contribution degree of the poor contact of the junction box is calculated to be 20%; and by analyzing the dust accumulation indicated by the local microclimate sensor, the contribution degree of the local slight dust accumulation is calculated to be 20%.

[0122] According to these contribution degrees, the system determines that the "long-term degradation of photovoltaic components" is the dominant risk source. Finally, the system generates an implicit risk signal, which not only indicates the existence of an implicit risk, but also explicitly contains information such as "dominant risk source: long-term degradation of photovoltaic components" and "risk degree: moderate".

[0123] In one possible design, Figure 3 is a partial flowchart of step S4 according to an example embodiment. Referring to the accompanying drawings, Figure 3 In step S4, based on the uncertainty index, the risk preference level of the photovoltaic power station is determined, including:

[0124] S401, monitor the current value of the uncertainty index and the change trend of the uncertainty index in a preset time window;

[0125] In this embodiment, the historical data of the uncertainty index of power generation prediction is continuously collected and analyzed to obtain the instantaneous state and dynamic change rule in a period of time. The preset time window can be understood as a configurable time length, for example, 5 minutes, 10 minutes or longer, which aims to smooth short-term fluctuations and more accurately reflect the long-term trend of the index.

[0126] S402, when the uncertainty index approaches the switching threshold of the risk preference level, start a timer, wherein the delay time is pre-configured in the timer;

[0127] In this embodiment, when the monitored index value enters the preset threshold interval (for example, the range of ±X% from the switching threshold), the system automatically activates a timing module. The purpose of the timer is to introduce a time delay to avoid immediate triggering of level switching due to transient fluctuations.

[0128] S403, when the uncertainty index is stable in the new level area within the delay time, execute the switching of the risk preference level, and adjust the hysteresis interval of the level switching according to the fluctuation amplitude of the uncertainty index;

[0129] In this embodiment, during the running of the timer, the system does not immediately execute the level switching, but continuously monitors the index value. If the index value is always stable in the new level area to be switched to within the delay time, it indicates that the change has a certain persistence. When the uncertainty index continuously meets the conditions of the new level in the entire preset delay time, the system finally confirms and executes the switching of the risk preference level, ensuring that the decision of level switching is based on stable trends rather than short-term fluctuations. Finally, the system dynamically adjusts the hysteresis interval of the level switching according to the historical fluctuation characteristics of the index. For example, when the index fluctuates greatly, the hysteresis interval can be appropriately increased to improve the robustness of switching; when the index fluctuates less, the hysteresis interval can be reduced to improve the response speed. The purpose of adjusting the hysteresis interval is to further optimize the sensitivity and stability of level switching, and prevent "shock" switching when the index frequently crosses the threshold.

[0130] The technical solution described above, by introducing a multi-confirmation mechanism, effectively addresses the problem of frequent risk preference level switching that may result from frequent fluctuations in the power generation forecast uncertainty index near the level switching threshold. First, by monitoring the current value and trend of the power generation forecast uncertainty index, the system can comprehensively grasp its dynamic characteristics. Second, when the index approaches the switching threshold, a timer is started and a delay time is set. This mechanism prevents the system from immediately responding to instantaneous fluctuations, instead waiting for the index to stabilize within the new level range for a period of time. It is precisely this delayed confirmation that makes the level switching decision more prudent and reliable. Finally, the hysteresis range of the level switching is dynamically adjusted according to the fluctuation amplitude of the index, further enhancing the system's adaptability and ensuring stable and accurate level switching under different fluctuation environments, thereby avoiding the uncertainty of energy management strategies and operational risks caused by frequent switching.

[0131] In one example, assume a photovoltaic power station has three risk preference levels: "low risk," "medium risk," and "high risk." The switching threshold from "low risk" to "medium risk" is 0.3, and the switching threshold from "medium risk" to "high risk" is 0.6. The system is set with a 5-minute timer delay and an initial hysteresis interval of ±0.05.

[0132] When the power generation forecast uncertainty index rises from 0.25 (low-risk zone) to 0.32, the system will start a 5-minute timer because it is approaching and exceeding the switching threshold of 0.3. During these 5 minutes, the system will continuously monitor the index. If the index remains above 0.30 (i.e., medium-risk zone) during these 5 minutes, for example, stabilizing between 0.31 and 0.35, then after the 5 minutes, the system will switch the risk preference level from "low risk" to "medium risk".

[0133] However, if during the timer's operation, the index rises to 0.32 and then quickly falls back to 0.28 (low-risk zone), and fails to stabilize in the medium-risk zone within 5 minutes, then the system will not perform a level switch after the timer ends, and the risk preference level will remain "low-risk".

[0134] If the system detects a significant fluctuation in the power generation forecast uncertainty index over the past hour, such as a standard deviation exceeding 0.1, the system can adjust the hysteresis range from ±0.05 to ±0.08 based on this fluctuation. This means that a level switch will only be triggered when the index stabilizes outside the switching threshold of ±0.08 and remains there for the delay period, thereby further enhancing stability in volatile environments. In this way, the proposed solution effectively avoids misjudgments and frequent switching of risk preference levels caused by short-term, non-persistent fluctuations.

[0135] In one possible design, Figure 4 This is a flowchart illustrating step S5 according to an exemplary embodiment. (Refer to the attached document.) Figure 4 Step S5 includes:

[0136] S501. Obtain the operating data of each battery module and evaluate the health status of each battery module based on the operating data.

[0137] In this embodiment, real-time operating parameters of each independent battery module in the energy storage system are continuously collected. This operating data may include, but is not limited to, battery module voltage, current, temperature, internal resistance, charge-discharge cycle count, and historical charge-discharge curves. These parameters are then monitored and recorded in real-time by sensors in the Battery Management System (BMS) and transmitted to the Energy Management System for further processing. Subsequently, using the collected operating data, the health status of each battery module is quantitatively assessed through specific algorithms and models. For example, based on indicators such as changes in internal resistance, capacity decay rate, self-discharge rate, and temperature anomalies, combined with machine learning algorithms or expert systems, the State of Health (SOH) or Remaining Useful Life (RUL) of each battery module can be calculated. The assessment result can be a percentage value representing the health level of the battery module relative to its initial state, or a rating system such as "Good," "Average," or "Poor."

[0138] S502. Adjust the reserve capacity ratio of each battery module according to risk preference level and health status.

[0139] In this embodiment, after determining the overall reserve capacity requirement of the energy storage system, different reserve capacity reservation ratios are allocated to each battery module based on its health status and the power station's risk preference level. For example, for battery modules in poor health, their reserve capacity reservation ratio can be appropriately increased to reduce deep charge-discharge cycles and slow down their degradation rate; while for battery modules in good health, their reserve capacity reservation ratio can be appropriately decreased to improve their energy utilization rate. The risk preference level serves as a macro-control factor; for example, when the risk preference is high, the overall reserve capacity reservation ratio may be low, but internal differentiated allocations will still be made based on the module's health status.

[0140] S503. Adjust the charging and discharging rates of each battery module differently based on risk preference level and health status.

[0141] In this embodiment, the allowable charging and discharging current of each battery module is dynamically adjusted based on its health status and the power station's risk preference level. For example, for battery modules in poor health, their maximum charging and discharging rate is limited to a lower level to avoid further damage from overcurrent; while for battery modules in good health, they can be allowed to charge and discharge at a higher rate to respond to the grid's rapid power demand. The risk preference level also affects the overall charging and discharging strategy. For example, in risk-averse mode, the charging and discharging rates of all modules may be conservatively limited, but fine-tuning will still be made internally based on their health status.

[0142] The technical solutions described above can significantly improve the operational reliability, safety, and economy of photovoltaic power plant energy storage systems. Specifically, by assessing the health status of each battery module and implementing differentiated adjustments, it is possible to effectively avoid overall system performance degradation or failure caused by premature degradation of some battery modules, thereby extending the overall service life of the energy storage system. Furthermore, this refined management approach helps balance the load of each battery module, preventing localized overload or overheating, and significantly improving the operational safety of the energy storage system. Simultaneously, by fully utilizing the performance of battery modules in good health and protecting those in poor health, energy throughput efficiency can be optimized, operation and maintenance costs reduced, and the overall economic benefits of the photovoltaic power plant improved. Compared to a scheme that only makes uniform adjustments based on the overall risk appetite level, the differentiated adjustment strategy of this application can better adapt to the heterogeneity within the energy storage battery pack, achieving more efficient energy management.

[0143] In one example, suppose a photovoltaic power plant's energy storage system consists of 100 independent battery modules. By continuously collecting operational data such as voltage, current, temperature, and internal resistance of each battery module, and using a State of Health (SOH) assessment algorithm, the system evaluates that 80 battery modules have an SOH above 90% (good), 15 battery modules have an SOH between 80% and 90% (average), and 5 battery modules have an SOH below 80% (poor). Simultaneously, based on the power generation prediction uncertainty index, the power plant is classified as having a "medium risk appetite."

[0144] In this scenario, the energy management system will make differentiated adjustments based on the power plant's "medium risk appetite" level and the health status of each battery module:

[0145] For 80 good battery modules with a state of harmlessness (SOH) of over 90%, the reserve capacity ratio can be set to 10%, and the charge / discharge rate is allowed to reach 90% of its rated maximum rate.

[0146] For 15 general battery modules with a state of equilibrium (SOH) between 80% and 90%, the reserve capacity ratio can be appropriately increased to 15%, and the charge / discharge rate is limited to 70% of its rated maximum rate.

[0147] For the five poorer battery modules with a State of Harm (SOH) of less than 80%, the reserve capacity ratio will be further increased to 25%, and the charge / discharge rate will be strictly limited to 50% of its rated maximum rate to protect these modules to the greatest extent and delay their further degradation.

[0148] By making differentiated adjustments, the entire energy storage system can meet the requirements of a medium risk appetite in power plants while maximizing the lifespan of all battery modules, thus avoiding the impact on the performance and safety of the entire system due to the premature failure of a few weaker modules.

[0149] In one possible design, Figure 5 This is a flowchart illustrating step S6 according to an exemplary embodiment. (Refer to the attached document.) Figure 5 Step S6 includes:

[0150] S601. Obtain the operating data of the photovoltaic inverter and combiner box, including the inverter's output power, operating temperature, fault indication, and the combiner box's branch current and insulation status.

[0151] In this embodiment, key operating parameters of the photovoltaic inverter and combiner box are collected in real time or periodically through sensors, communication interfaces, etc. Among these, the inverter's output power reflects its current power generation capacity; operating temperature is an important indicator for assessing the inverter's heat dissipation performance and potential overheating risk; and fault indication directly indicates whether the equipment has any abnormalities or faults. For the combiner box, branch current is used to monitor the operating status of each photovoltaic string and whether there is current imbalance; insulation status reflects the safety of the electrical system and whether there is a risk of leakage.

[0152] S602. Based on the operating data, assess the health status and current maximum output capacity of the photovoltaic inverter and combiner box;

[0153] In this embodiment, the collected operational data is analyzed and processed. For example, data trend analysis, anomaly detection algorithms, and machine learning models can be used to comprehensively assess the wear and tear, performance degradation, and potential failure risks of the equipment, thereby determining the equipment's health status. Simultaneously, by combining the equipment's rated parameters, historical operational data, and current environmental conditions, the maximum power that the equipment can safely and stably output under the current conditions is calculated, i.e., the current maximum output capacity.

[0154] S603. Based on the risk preference level, predicted power generation, and the health status and current maximum output capacity of the photovoltaic inverter and combiner box, set an upper limit for grid-connected power, wherein the upper limit for grid-connected power shall not exceed the current maximum output capacity of the photovoltaic inverter and combiner box.

[0155] In this embodiment, a multi-objective optimization algorithm or rule-based decision-making logic is employed. Considering the power plant's risk preference level (e.g., a risk-averse power plant might set a lower grid-connected power limit to ensure safety, while a risk-seeking power plant might set a higher limit to maximize profits) and the predicted power generation (as the theoretical maximum power output), the health status and current maximum output capacity of the photovoltaic inverter and combiner box are used as hard constraints. This ensures that the set grid-connected power limit meets both the power plant's risk management needs and the actual operating capacity of the equipment, avoiding overload operation or forced high power output in an unhealthy state. Specifically, the grid-connected power limit is strictly limited to not exceeding the current maximum output capacity of the photovoltaic inverter and combiner box to prevent equipment damage or safety accidents caused by the strategy settings.

[0156] The technical solutions described above enable more precise and secure setting of the grid-connected power limit. They not only consider the overall risk appetite and power generation forecasts of the power plant, but also incorporate the actual operating conditions and health levels of key equipment such as photovoltaic inverters and combiner boxes into the decision-making process, effectively avoiding grid connection risks caused by equipment performance degradation or potential failures. By ensuring that the set grid-connected power limit never exceeds the current maximum output capacity of the equipment, the reliability and safety of the photovoltaic power plant operation are improved, the service life of the equipment is extended, and unplanned downtime and maintenance costs are reduced. Furthermore, this refined strategy adjustment helps optimize the overall operating efficiency of the power plant, maximizing the effective utilization of power generation and grid connection benefits while ensuring safety.

[0157] In one possible design, Figure 6 This is a flowchart illustrating step S2 according to an exemplary embodiment. (Refer to the attached document.) Figure 6 Step S2 includes:

[0158] S201. Calculate the correlation between environmental change indication information and broadband irradiance with prediction deviation and efficiency deviation within different preset time windows.

[0159] In this embodiment, the preset time window can include, but is not limited to, various time granularities such as minutes, hours, and days, aiming to capture the potential correlation between environmental factors and power generation performance deviations at different time scales. The correlation degree can be calculated using various statistical methods, such as Pearson correlation coefficient, Granger causality test, and mutual information, to quantify the impact of environmental change indicators and broad-spectrum irradiance on power generation prediction deviations and efficiency deviations.

[0160] S202. Based on the correlation calculated within each time window, identify the dominant time scale with the strongest explanatory power, and adjust the sampling period and data smoothing window for the explanatory power assessment according to the dominant time scale.

[0161] In this embodiment, the time scale with the highest correlation is selected as the optimal evaluation benchmark. For example, if the correlation within an hourly time window is significantly higher than that within a minute or day, then the hourly time scale is identified as the dominant time scale. Based on this, the sampling period and data smoothing window for the explanatory power assessment are adjusted according to the identified dominant time scale. For example, if the dominant time scale is the hourly time scale, the sampling period can be set to 15 minutes or 30 minutes, and the data smoothing window can be set to 1 hour or 2 hours to ensure that the evaluation process can fully reflect the data characteristics at this time scale. The sampling period refers to the time interval for data collection; the data smoothing window refers to the time range used to eliminate short-term fluctuations and reveal long-term trends during data analysis.

[0162] S203. Under the adjusted sampling period and data smoothing window, evaluate the explanatory power of environmental change indication information and the corresponding broadband irradiance of the region on prediction bias and efficiency bias.

[0163] In this embodiment, under the adjusted sampling period and data smoothing window, the explanatory power of environmental change indication information and broadband irradiance on power generation prediction deviation and efficiency deviation is finally evaluated. The evaluation results will be used for subsequent risk assessment and strategy adjustment.

[0164] The technical solutions described above overcome the limitations of traditional single-timescale assessments, enabling a more comprehensive and accurate evaluation of the explanatory power of environmental change indicators and broad-spectrum irradiance on power generation prediction and efficiency deviations. Furthermore, by dynamically identifying advantageous timescales and adjusting assessment parameters, the reliability and effectiveness of the assessment results are enhanced, thereby helping to more accurately identify potential risks and optimize power plant operation strategies.

[0165] In one possible design, in step S4, based on the power generation prediction deviation information, environmental change indication information, and implicit risk signals, an uncertainty index for power generation prediction is quantified and generated, including:

[0166] S411. Calculate the fluctuation characteristics and temporal correlations between power generation prediction deviation information, environmental change indication information, and hidden risk signals at different time scales.

[0167] In this embodiment, multi-scale analysis is performed on the input power generation prediction deviation information, environmental change indication information, and implicit risk signals. For example, wavelet transform, Fourier transform, or empirical mode decomposition can be used to extract the fluctuation amplitude, frequency, phase, and other features under different frequency components. Simultaneously, techniques such as cross-correlation functions, Granger causality tests, or dynamic time warping are used to analyze the interdependence and influence intensity of these information under different time lags.

[0168] S412. Based on fluctuation characteristics and time series correlation, identify the contribution patterns of power generation prediction deviation information, environmental change indication information, and implicit risk signals to the uncertainty of power generation prediction.

[0169] In this embodiment, machine learning algorithms, such as support vector machines, random forests, or neural networks, are used to perform pattern recognition and classification on the extracted fluctuation features and time-series correlations. The contribution patterns can reflect which information sources contribute more significantly to prediction uncertainty under specific operating conditions, and how they work synergistically or independently.

[0170] S413. Based on the contribution model, dynamically adjust the weights and nonlinear mapping relationships of power generation prediction deviation information, environmental change indication information, and implicit risk signals in the quantification uncertainty index. Among them, when multicollinearity is identified, the weight of highly correlated information is reduced.

[0171] In this embodiment, based on the identified contribution patterns, the importance (weight) of each information source in the uncertainty index calculation and their nonlinear relationship with the uncertainty index are updated in real time. For example, when environmental change indication information is identified as having a dominant influence on prediction bias within a certain time period, its weight is increased accordingly. In particular, when there is a high correlation (i.e., multicollinearity) among multiple information sources, to avoid redundant calculations and model overfitting, the weight of the more correlated information is actively reduced to ensure that the independent contribution of each information source is accurately evaluated.

[0172] S414. Based on the adjusted weights and nonlinear mapping relationship, integrate power generation prediction deviation information, environmental change indication information, and implicit risk signals to generate an uncertainty index;

[0173] In this embodiment, the information sources, after dynamic weight adjustment and nonlinear mapping, are comprehensively calculated through a fusion model (e.g., a fusion model based on fuzzy logic, Bayesian network, or deep learning) to finally output a quantitative index that can comprehensively reflect the uncertainty level of the current power generation forecast.

[0174] The technical solution of the above embodiments firstly analyzes the multi-scale fluctuation characteristics and temporal correlation of power generation prediction deviation information, environmental change indication information, and implicit risk signals to deeply explore the dynamic behavior and mutual influence mechanisms of these information sources at different time dimensions. Based on this, by dynamically adjusting the weights and nonlinear mapping relationships of each information source according to the identified contribution patterns, and specifically considering multicollinearity to reduce the weight of highly correlated information, the independent contribution of each information source is accurately evaluated when quantifying the uncertainty index of power generation prediction, and the model has stronger adaptability to the complexity and dynamism of the input data.

[0175] In another possible design, step S6 also includes:

[0176] S611. Obtain detailed information on all power deviation penalty clauses in the grid connection protocol, including the triggering conditions, penalty calculation method, and penalty intensity for each penalty clause.

[0177] In this embodiment, the system needs to parse the grid connection protocol text, identify and extract all penalty clauses related to power deviation. This detailed information includes, but is not limited to, the specific conditions that trigger the penalty (e.g., power deviation exceeding a certain percentage or duration), the calculation method of the penalty (e.g., linear penalty based on deviation or tiered penalty), and the severity of the penalty (e.g., the penalty amount per megawatt-hour of deviation).

[0178] S612. Based on the risk appetite level and the uncertainty range of the predicted power generation, conduct a risk assessment for each power deviation penalty clause, and quantify the probability and expected penalty cost of triggering each power deviation penalty clause under different grid-connected power caps.

[0179] In this embodiment, the system utilizes the uncertainty information of predicted power generation, combined with the photovoltaic power plant's own risk preference (e.g., whether it tends to operate conservatively to avoid penalties or aggressively to maximize power generation revenue), to simulate and analyze each potential penalty clause. Through Monte Carlo simulation or other probabilistic statistical methods, the probability of each penalty clause being triggered and the resulting expected economic loss can be calculated under different grid-connected power cap settings.

[0180] S613, taking into account the probability and expected penalty cost, and iteratively adjusting the grid-connected power limit to find an optimal grid-connected power limit value, wherein the optimal grid-connected power limit value minimizes the comprehensive expected penalty cost of all power deviation penalty clauses while meeting the risk preference level requirements.

[0181] In this embodiment, the process of finding an optimal grid-connected power limit can employ optimization algorithms, such as genetic algorithms, particle swarm optimization, or linear programming, to search within a preset grid-connected power limit range. Each iteration calculates the comprehensive expected penalty cost based on the current grid-connected power limit and adjusts the grid-connected power limit according to the optimization objective, until a grid-connected power limit that minimizes the total expected penalty cost while satisfying the risk preference level is found.

[0182] S614. During the iterative adjustment process, when the adjusted grid-connected power limit causes a significant increase in the expected penalty cost of a certain power deviation penalty clause, the grid-connected power limit shall be adjusted first to reduce the expected penalty cost of the power deviation penalty clause.

[0183] In this embodiment, this step is an intelligent optimization strategy designed to improve adjustment efficiency and effectiveness. For example, if a small change in a grid-connected power limit leads to a significant increase in the probability of triggering a high-penalty clause, the system will immediately adjust that clause instead of blindly performing a global search.

[0184] The technical solution described above addresses the problem of traditional methods potentially ignoring economic risks when setting grid-connected power limits by introducing detailed consideration of power deviation penalty clauses in grid connection protocols. First, acquiring and parsing detailed information about the penalty clauses lays the foundation for subsequent quantitative risk assessment. Second, combining the power plant's risk appetite level and the uncertainty of predicted power generation, a refined risk assessment is conducted for each penalty clause, quantifying the trigger probability and expected penalty cost under different grid-connected power limits. This transforms risk from a vague concept into a calculable value. Finally, the system iteratively optimizes and dynamically adjusts the grid-connected power limit to minimize the overall expected penalty cost of all penalty clauses while satisfying the risk appetite level, achieving more economical energy management.

[0185] In summary, the energy management method for photovoltaic power plants provided in this invention, by acquiring power generation prediction deviation information, environmental change indication information, and photovoltaic array efficiency deviation information, can comprehensively and meticulously capture various factors affecting the power generation performance of photovoltaic power plants. Specifically, power generation prediction deviation information provides a basis for identifying systematic errors; environmental change indication information accurately reflects the actual operating environment of the power plant; and photovoltaic array efficiency deviation information accurately identifies efficiency degradation caused by latent problems such as component aging. Subsequently, by evaluating the explanatory power of environmental change indication information and broadband irradiance on power generation prediction deviation and efficiency deviation, the contribution of these factors to prediction accuracy and efficiency degradation is analyzed in depth. When the explanatory power is below a preset threshold and efficiency deviation persists, a latent risk signal is generated, solving the problem in existing technologies of difficulty in identifying and quantifying latent risks such as gradual microclimate changes and component aging. Furthermore, by integrating power generation prediction deviation information, environmental change indication information, and latent risk signals, a power generation prediction uncertainty index is quantified, and the risk preference level of the power plant is determined based on this index. This mechanism enables the energy management system to dynamically adjust its risk tolerance capacity according to the actual risk level faced by the power plant. Finally, based on the determined risk appetite level, the reserve capacity ratio and charge / discharge rate of the energy storage battery are adjusted, and a grid-connected power limit is set, resulting in the adjusted energy storage battery charge / discharge strategy and grid-connected power strategy. This dynamically adjusted strategy allows for more precise matching of photovoltaic power plant energy dispatch with actual power generation and grid demand, significantly reducing manual intervention and lowering operation and maintenance costs. Simultaneously, the optimized grid-connected power strategy effectively avoids penalty clauses in grid connection agreements, reduces the risk of fines, and improves the economic benefits of photovoltaic power plants.

[0186] Example 2

[0187] Embodiment 2 of the present invention provides an energy management system for a photovoltaic power station. Figure 7 This is a block diagram illustrating an energy management system for a photovoltaic power plant according to an exemplary embodiment. (See attached diagram.) Figure 7 The system includes:

[0188] The information acquisition module 01 is used to acquire power generation prediction deviation information, environmental change indication information, and photovoltaic array efficiency deviation information of the photovoltaic power station. Among them, the power generation prediction deviation information includes the prediction deviation between the actual power generation of the photovoltaic power station and the predicted power generation, as well as the trend of the prediction deviation; the environmental change indication information includes macro-meteorological forecast information and local environmental change information collected by microclimate sensors inside the photovoltaic power station; the photovoltaic array efficiency deviation information includes the efficiency deviation between the instantaneous photoelectric conversion efficiency of the photovoltaic array and the pre-constructed dynamic expected efficiency benchmark, as well as the trend of the efficiency deviation.

[0189] The explanatory power assessment module 02 is used to assess the explanatory power of the environmental change indication information and the corresponding region's broadband irradiance on the prediction deviation and efficiency deviation based on the power generation prediction deviation information, environmental change indication information, and photovoltaic array efficiency deviation information.

[0190] The latent risk signal generation module 03 is used to generate latent risk signals when the explanatory power is lower than the preset risk threshold and the efficiency deviation persists.

[0191] The risk preference determination module 04 is used to quantify and generate an uncertainty index for power generation prediction based on power generation prediction deviation information, environmental change indication information, and implicit risk signals, and to determine the risk preference level of the photovoltaic power station based on the uncertainty index.

[0192] The energy storage battery strategy correction module 05 is used to adjust the reserve capacity ratio and charge / discharge rate of the energy storage battery based on the risk preference level, so as to obtain the energy storage battery charge / discharge correction strategy.

[0193] The grid-connected power correction module 06 is used to set the upper limit of grid-connected power according to the risk preference level and obtain the grid-connected power correction strategy.

[0194] The control command issuing module 07 is used to generate and issue control commands to the energy storage battery management system and the photovoltaic inverter based on the energy storage battery charge and discharge correction strategy and the grid-connected power correction strategy.

[0195] In summary, the energy management method and system for photovoltaic power plants provided by the embodiments of the present invention achieve intelligent and refined energy management of photovoltaic power plants through multi-dimensional information acquisition, hidden risk identification, uncertainty quantification, and risk preference level determination. It effectively solves a series of problems in the prior art, such as prediction deviation, scheduling mismatch, high operation and maintenance costs, and penalty risks caused by microclimate changes and component aging, thereby improving the operating efficiency, economic benefits, and grid stability of photovoltaic power plants.

[0196] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0197] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. An energy management method for a photovoltaic power station, characterized in that, Includes the following steps: The system acquires power generation prediction deviation information, environmental change indication information, and photovoltaic array efficiency deviation information for the photovoltaic power station. The power generation prediction deviation information includes the prediction deviation between the actual power generation and the predicted power generation of the photovoltaic power station, as well as the trend of the prediction deviation. The environmental change indication information includes macro-meteorological forecast information and local environmental change information collected by microclimate sensors inside the photovoltaic power station. The photovoltaic array efficiency deviation information includes the efficiency deviation between the instantaneous photoelectric conversion efficiency of the photovoltaic array and a pre-constructed dynamic expected efficiency benchmark, as well as the trend of the efficiency deviation. Based on the power generation prediction deviation information, the environmental change indication information, and the photovoltaic array efficiency deviation information, evaluate the explanatory power of the environmental change indication information and the broadband irradiance of the corresponding region on the prediction deviation and the efficiency deviation; When the explanatory power is lower than a preset risk threshold and the efficiency deviation persists, a latent risk signal is generated. Based on the power generation prediction deviation information, the environmental change indication information, and the implicit risk signal, an uncertainty index for power generation prediction is quantified and generated, and based on the uncertainty index, the risk preference level of the photovoltaic power station is determined. Based on the risk preference level, the reserve capacity ratio and charge / discharge rate of the energy storage battery are adjusted to obtain the energy storage battery charge / discharge correction strategy. Based on the risk preference level, a grid-connected power limit is set to obtain a grid-connected power correction strategy. Based on the energy storage battery charge / discharge correction strategy and the grid-connected power correction strategy, control commands are generated and sent to the energy storage battery management system and the photovoltaic inverter.

2. The method according to claim 1, characterized in that, The steps for obtaining the photovoltaic array efficiency deviation information include: The instantaneous photoelectric conversion efficiency of the photovoltaic array is calculated based on the actual output power of the photovoltaic array, the broadband irradiance of the corresponding region, and the surface temperature of the photovoltaic module, which are continuously collected. Establish dynamic expected efficiency benchmarks; The efficiency deviation is obtained by comparing the instantaneous photoelectric conversion efficiency with the dynamic expected efficiency benchmark. By tracking the persistence of the efficiency deviation, the efficiency deviation information of the photovoltaic array is obtained.

3. The method according to claim 1, characterized in that, When the explanatory power is lower than a preset risk threshold and the efficiency deviation persists, a latent risk signal is generated, including: Identify and classify the patterns of the efficiency deviations, and associate the patterns with potential hidden risk sources; Assess the contribution of each potential hidden risk source to the efficiency deviation, and determine the overall risk level based on the contribution. Based on the overall risk level, a latent risk signal reflecting the overall risk level is generated.

4. The method according to claim 1, characterized in that, The determination of the risk appetite level of the photovoltaic power station based on the uncertainty index includes: Monitor the current value of the uncertainty index and its changing trend within a preset time window; When the uncertainty index approaches the switching threshold of the risk preference level, a timer is started, wherein a delay time is pre-configured within the timer; When the uncertainty index stabilizes in the new level range within the specified delay time, a risk preference level switch is performed, and the lag range for the level switch is adjusted according to the fluctuation range of the uncertainty index.

5. The energy management method for a photovoltaic power station according to claim 1, characterized in that, The step of adjusting the reserve capacity ratio and charge / discharge rate of the energy storage battery based on the risk preference level to obtain the energy storage battery charge / discharge correction strategy includes: Obtain the operating data of each battery module, and evaluate the health status of each battery module based on the operating data; Based on the risk preference level and the health status, the reserve ratio of the backup capacity of each battery module is adjusted in a differentiated manner. The charging and discharging rates of each battery module are adjusted differently based on the risk preference level and the health status.

6. The method according to claim 1, characterized in that, The step of setting an upper limit for grid-connected power based on the risk preference level to obtain a grid-connected power correction strategy includes: The operation data of the photovoltaic inverter and combiner box are obtained, including the inverter's output power, operating temperature, fault indication, and the combiner box's branch current and insulation status. Based on the operational data, assess the health status and current maximum output capacity of the photovoltaic inverter and combiner box; Based on the risk preference level, predicted power generation, and the health status and current maximum output capacity of the photovoltaic inverter and combiner box, a grid-connected power limit is set, wherein the grid-connected power limit is not higher than the current maximum output capacity of the photovoltaic inverter and combiner box.

7. The method according to claim 1, characterized in that, The step of evaluating the explanatory power of the environmental change indication information and the corresponding region's broadband irradiance for the prediction deviation and the efficiency deviation based on the power generation prediction deviation information, the environmental change indication information, and the photovoltaic array efficiency deviation information includes: Within different preset time windows, the correlation between the environmental change indication information and the broadband irradiance with the prediction deviation and the efficiency deviation is calculated respectively. Based on the correlation calculated within each time window, the dominant time scale with the strongest explanatory power is identified, and the sampling period and data smoothing window for the explanatory power assessment are adjusted according to the dominant time scale. Under the adjusted sampling period and data smoothing window, evaluate the explanatory power of the environmental change indication information and the corresponding broadband irradiance of the region on the prediction bias and the efficiency bias.

8. The method according to claim 1, characterized in that, The step of quantifying and generating an uncertainty index for power generation forecasting based on the power generation forecasting deviation information, the environmental change indication information, and the implicit risk signal includes: Calculate the fluctuation characteristics and temporal correlations between the power generation prediction deviation information, the environmental change indication information, and the latent risk signal at different time scales; Based on the fluctuation characteristics and the time-series correlation, identify the contribution patterns of the power generation prediction deviation information, the environmental change indication information, and the latent risk signal to the uncertainty of power generation prediction; Based on the contribution mode, the weights and nonlinear mapping relationships of the power generation prediction deviation information, the environmental change indication information, and the implicit risk signal in quantifying the uncertainty index are dynamically adjusted. When multicollinearity is identified, the weights of highly correlated information are reduced. Based on the adjusted weights and the nonlinear mapping relationship, the uncertainty index is generated by fusing the power generation prediction deviation information, the environmental change indication information, and the implicit risk signal.

9. The method according to claim 1, characterized in that, The step of setting an upper limit for grid-connected power based on the risk preference level to obtain a grid-connected power correction strategy further includes: Obtain detailed information on all power deviation penalty clauses in the grid connection protocol, including the triggering conditions, penalty calculation method, and penalty intensity for each penalty clause; Based on the risk preference level and the uncertainty range of the predicted power generation, a risk assessment is conducted on each of the power deviation penalty clauses to quantify the probability and expected penalty cost of triggering each of the power deviation penalty clauses under different grid-connected power caps. By combining the probabilities and the expected penalty costs, and iteratively adjusting the grid-connected power limit, an optimal grid-connected power limit value is found, wherein the optimal grid-connected power limit value minimizes the overall expected penalty cost of all the power deviation penalty clauses while meeting the risk preference level requirements. During the iterative adjustment process, when the adjusted grid-connected power limit leads to a significant increase in the expected penalty cost of a certain power deviation penalty clause, the grid-connected power limit is adjusted first to reduce the expected penalty cost of the power deviation penalty clause.

10. An energy management system for a photovoltaic power station, characterized in that, The system includes: The information acquisition module is used to acquire power generation prediction deviation information, environmental change indication information, and photovoltaic array efficiency deviation information of the photovoltaic power station. The power generation prediction deviation information includes the prediction deviation between the actual power generation of the photovoltaic power station and the predicted power generation, as well as the trend of the prediction deviation. The environmental change indication information includes macro-meteorological forecast information and local environmental change information collected by microclimate sensors inside the photovoltaic power station. The photovoltaic array efficiency deviation information includes the efficiency deviation between the instantaneous photoelectric conversion efficiency of the photovoltaic array and a pre-constructed dynamic expected efficiency benchmark, as well as the trend of the efficiency deviation. The explanatory power assessment module is used to assess the explanatory power of the environmental change indication information and the broadband irradiance of the corresponding region on the prediction deviation and the efficiency deviation, based on the power generation prediction deviation information, the environmental change indication information, and the photovoltaic array efficiency deviation information. The latent risk signal generation module is used to generate a latent risk signal when the explanatory power is lower than a preset risk threshold and the efficiency deviation persists. The risk preference determination module is used to quantify and generate an uncertainty index for power generation prediction based on the power generation prediction deviation information, the environmental change indication information, and the implicit risk signal, and to determine the risk preference level of the photovoltaic power station based on the uncertainty index. The energy storage battery strategy correction module is used to adjust the reserve capacity ratio and charge / discharge rate of the energy storage battery based on the risk preference level, so as to obtain the energy storage battery charge / discharge correction strategy. The grid-connected power correction module is used to set the upper limit of grid-connected power according to the risk preference level, and obtain the grid-connected power correction strategy. The control command issuing module is used to generate and issue control commands to the energy storage battery management system and the photovoltaic inverter based on the energy storage battery charge and discharge correction strategy and the grid-connected power correction strategy.

Images