Solar power generation adaptive transmission method and system based on prediction model

Abstract

The invention discloses a solar power generation adaptive transmission method and system based on a prediction model, and relates to the technical field of data processing, and the method comprises the steps: obtaining a to-be-managed transmission time period, obtaining a historical power generation transmission data sequence, and obtaining an average power generation transmission change rate; selecting a fluctuation threshold according to the to-be-managed conveying time period, and obtaining an over-limit coverage rate; if the overrun coverage rate does not exceed the preset threshold value, obtaining a first target transmission data sequence of the to-be-managed transmission time period based on a prediction model and the average power generation transmission change rate; and if the overrun coverage rate exceeds a preset threshold value, obtaining a correction proportion according to the historical power generation transmission data sequence, obtaining a corrected power generation transmission change rate according to the average power generation transmission change rate and the correction proportion, and obtaining a second target transmission data sequence of the to-be-managed transmission time period based on the prediction model and the corrected power generation transmission change rate. The method has the advantages of being high in prediction efficiency, reliable in prediction and good in conveying management effect.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and particularly relates to a solar power generation adaptive delivery method and system based on a prediction model. BACKGROUND

[0002] In a solar power generation system, due to the significant intermittency, volatility and uncontrollability of solar energy resources, the power generation output frequently fluctuates with weather conditions (such as cloud cover, changes in solar intensity) and environmental factors (such as temperature, seasonal changes), and this instability makes the power generation delivery process uncontrollable and unable to achieve adaptive adjustment, which brings serious challenges to power delivery management.

[0003] Specifically, the existing power delivery method uses fixed threshold control, which cannot adapt to seasonal and weather pattern changes, and does not fully consider the inherent stability defects of historical data sequences, especially when the historical data fluctuates frequently and greatly, directly applying the average rate will amplify the prediction error, because the average rate only reflects the overall trend, ignoring the influence of local mutation areas (such as sudden increase or sudden decrease points), resulting in that the prediction result deviates from the actual power generation capacity, which will further cause the mismatch between the delivery plan and the actual power generation, causing energy waste, power grid frequency oscillation, power imbalance and even equipment overload risk. At the same time, the existing method lacks a quantitative evaluation mechanism for data volatility, cannot distinguish between data stable states, and has no adaptive adjustment strategy to cope with unstable scenarios, so it cannot provide reliable delivery sequences in periods of strong volatility, ultimately reducing the overall efficiency, renewable energy utilization rate and economic benefits of the system. SUMMARY

[0004] The technical problem of the present application can be summarized as how to adaptively identify the data stable state and intelligently adjust the power generation delivery strategy to ensure that a relatively reliable power generation delivery sequence can be generated under any fluctuation condition. In view of the above technical problem, the present application provides a solar power generation adaptive delivery method and system based on a prediction model.

[0005] The solar power generation adaptive delivery method based on a prediction model comprises: obtaining a to-be-managed delivery period, obtaining a historical power generation delivery data sequence of a preset time period before the to-be-managed delivery period, and obtaining an average power generation delivery change rate according to the historical power generation delivery data sequence; selecting a fluctuation threshold according to the to-be-managed delivery period, obtaining a proportion of the historical power generation delivery data sequence exceeding the fluctuation threshold as an overrun coverage rate; if the overrun coverage rate does not exceed a preset threshold, obtaining a first target delivery data sequence of the to-be-managed delivery period based on a prediction model and the average power generation delivery change rate; if the overrun coverage rate exceeds the preset threshold, obtaining a correction ratio according to the historical power generation delivery data sequence, obtaining a corrected power generation delivery change rate according to the average power generation delivery change rate and the correction ratio, and obtaining a second target delivery data sequence of the to-be-managed delivery period based on the prediction model and the corrected power generation delivery change rate.

[0006] Optionally, obtaining the average power generation delivery change rate according to the historical power generation delivery data sequence comprises: obtaining a first and last power generation delivery change amount of the historical power generation delivery data sequence and a historical delivery time length; and obtaining the average power generation delivery change rate according to a ratio of the first and last power generation delivery change amount to the historical delivery time length.

[0007] Optionally, obtaining the first target delivery data sequence of the to-be-managed delivery period based on the prediction model and the average power generation delivery change rate comprises: obtaining a plurality of key time points of the to-be-managed delivery period according to the prediction model; taking power generation delivery data at the end of the historical power generation delivery data sequence as starting power generation delivery data of the to-be-managed delivery period, and obtaining predicted power generation delivery data corresponding to the plurality of key time points based on the prediction model, the average power generation delivery change rate and the starting power generation delivery data; and forming a first pre-delivery data sequence of the to-be-managed delivery period according to the predicted power generation delivery data corresponding to the plurality of key time points.

[0008] Optionally, selecting the fluctuation threshold according to the to-be-managed delivery period comprises: obtaining a plurality of numerical intervals in sequence, each numerical interval corresponding to a different fluctuation threshold, wherein the larger the numerical value, the smaller the fluctuation threshold corresponding to the numerical interval; selecting a numerical interval to which the to-be-managed delivery period belongs according to the size of the to-be-managed delivery period, and obtaining the fluctuation threshold corresponding to the numerical interval.

[0009] Optionally, obtaining the proportion of the historical power generation delivery data sequence exceeding the fluctuation threshold as the overrun coverage rate comprises: obtaining a difference value of each adjacent power generation delivery data in the historical power generation delivery data sequence; obtaining a number of difference values exceeding the fluctuation threshold as an overrun number, and obtaining a proportion of the overrun number in the number of difference values in the historical power generation delivery data sequence as the overrun coverage rate.

[0010] Optionally, the acquiring the correction ratio according to the historical power generation delivery data sequence comprises: acquiring total trend variables of rising trend and falling trend respectively according to the historical power generation delivery data sequence, acquiring a total trend variable same as the trend of the average power generation delivery change rate as a target variable, and acquiring a total trend variable opposite to the trend of the average power generation delivery change rate as a reference variable; and acquiring the correction ratio according to the target variable and the reference variable.

[0011] Optionally, the acquiring the correction ratio according to the target variable and the reference variable is represented as: ; wherein, is the correction ratio, is the target variable, is the reference variable.

[0012] A solar power generation adaptive delivery system based on a prediction model is also provided, the system comprising: a first data processing module configured to acquire a to-be-managed delivery period, acquire a historical power generation delivery data sequence of a preset time period before the to-be-managed delivery period, and acquire an average power generation delivery change rate according to the historical power generation delivery data sequence; a second data processing module configured to select a fluctuation threshold according to the to-be-managed delivery period, acquire a proportion of the historical power generation delivery data sequence exceeding the fluctuation threshold as an overrun coverage rate; a first prediction delivery module configured to, when the overrun coverage rate does not exceed a preset threshold, acquire a first target delivery data sequence of the to-be-managed delivery period based on a prediction model and the average power generation delivery change rate; and a second prediction delivery module configured to, when the overrun coverage rate exceeds the preset threshold, acquire a correction ratio according to the historical power generation delivery data sequence, acquire a corrected power generation delivery change rate according to the average power generation delivery change rate and the correction ratio, and acquire a second target delivery data sequence of the to-be-managed delivery period based on the prediction model and the corrected power generation delivery change rate.

[0013] Optionally, the first prediction delivery module is further configured to: acquire a plurality of key time points of the to-be-managed delivery period according to the prediction model; take power generation delivery data at an end of the historical power generation delivery data sequence as starting power generation delivery data of the to-be-managed delivery period, and acquire predicted power generation delivery data corresponding to the plurality of key time points based on the prediction model, the average power generation delivery change rate, and the starting power generation delivery data; and form a first delivery data sequence of the to-be-managed delivery period according to the predicted power generation delivery data corresponding to the plurality of key time points.

[0014] Optionally, the second prediction delivery module is further configured to: acquire total trend variables of rising trend and falling trend respectively according to the historical power generation delivery data sequence, acquire a total trend variable same as the trend of the average power generation delivery change rate as a target variable, and acquire a total trend variable opposite to the trend of the average power generation delivery change rate as a reference variable; and acquire the correction ratio according to the target variable and the reference variable.

[0015] The beneficial effects of the present application are embodied in: In the entire prediction model-based solar power generation adaptive delivery method, firstly, based on the average change rate calculated by the head and tail values of the historical sequence, the overall trend characteristics are quickly captured to lay the foundation for initial prediction; secondly, according to the design of intelligently matching the fluctuation threshold value according to the management period, the defects of the fixed threshold value are overcome, and the dynamic balance of short-term noise tolerance and long-term sensitivity is realized; by calculating the overrun coverage rate of the difference value of adjacent data points, the fluctuation intensity of the historical data is accurately quantified to provide an objective standard for stable state judgment, so that linear extrapolation prediction is directly used in the stable scene to ensure efficiency, and in the fluctuation scene, depth trend conflict analysis is started to separate and quantify the confrontation intensity of the same direction trend and the reverse fluctuation, and according to this, the proportion correction is generated to more accurately reflect the true trend: when the trend is dominant, the sudden interference is weakened in the positive direction, and when the fluctuation is dominant, the optimistic prediction deviation is suppressed in the negative direction, and finally the smooth delivery sequence after the fluctuation noise is eliminated is output. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the specific embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the specific embodiments or prior art description will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, each element or part is not necessarily drawn according to the actual proportion.

[0017] Figure 1 The steps of the prediction model-based solar power generation adaptive delivery method of the present application in one embodiment are shown in the following schematic diagram. Figure 2 A part of the steps of S1 in the prediction model-based solar power generation adaptive delivery method of the present application are shown in the following schematic diagram. Figure 3 A part of the steps of S3 in the prediction model-based solar power generation adaptive delivery method of the present application are shown in the following schematic diagram. Figure 4 A part of the steps of S2 in the prediction model-based solar power generation adaptive delivery method of the present application are shown in the following schematic diagram. Figure 5 Another part of the steps of S2 in the prediction model-based solar power generation adaptive delivery method of the present application are shown in the following schematic diagram. Figure 6 A part of the steps of S4 in the prediction model-based solar power generation adaptive delivery method of the present application are shown in the following schematic diagram. DETAILED DESCRIPTION

[0018] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some but not all of the embodiments of the present application. The components of the embodiments of the present application described and shown in the drawings can be arranged and designed in various different configurations.

[0019] Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed application, but only represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work are within the scope of protection of the present application.

[0020] It should be noted that: similar reference numbers and letters represent similar items in the following drawings, therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. In addition, the terms "first", "second" and the like are only used to distinguish description, and cannot be understood as indicating or implying relative importance.

[0021] As shown in Figure 1 a solar power generation adaptive delivery method based on a prediction model is provided, in one embodiment, the method comprises: S1, obtaining a to-be-managed delivery period, and obtaining a historical power generation delivery data sequence of a preset time period before the to-be-managed delivery period, and obtaining an average power generation delivery change rate according to the historical power generation delivery data sequence; S2, selecting a fluctuation threshold according to the to-be-managed delivery period, and obtaining a proportion exceeding the fluctuation threshold in the historical power generation delivery data sequence as an overrun coverage rate; S3, if the overrun coverage rate does not exceed a preset threshold, obtaining a first target delivery data sequence of the to-be-managed delivery period based on the prediction model and the average power generation delivery change rate; S4, if the overrun coverage rate exceeds the preset threshold, obtaining a correction ratio according to the historical power generation delivery data sequence, and obtaining a corrected power generation delivery change rate according to the average power generation delivery change rate and the correction ratio, and obtaining a second target delivery data sequence of the to-be-managed delivery period based on the prediction model and the corrected power generation delivery change rate.

[0022] In this embodiment, it is necessary to note that in S1, the target period for which the delivery management needs to be performed is first determined as the object of subsequent prediction and control. To this end, a historical power generation delivery data sequence of a preset length before the target period is obtained, which contains the record of continuous change of past power generation and reflects the behavior pattern of actual power generation output under the influence of weather and environmental factors. Based on this historical sequence, the average power generation delivery change rate is calculated, which represents the overall trend change level of power generation in the overall period, and the core is to simplify the representation by the difference between the first and last values of the sequence divided by the historical duration. This method avoids cumbersome point-to-point calculation, but implicitly regards the historical data as linear change, thereby providing basic input for the prediction model to support the subsequent decision whether to directly apply the rate.

[0023] Further, for example, it is assumed that on a cloudy spring day, the solar power generation needs to manage the delivery period from 14:00 to 16:00. The historical data sequence before the target period is extracted, for example, the power generation change record of the previous 5 hours, and the starting point of the sequence corresponds to the power generation level at 9:00 am, and the ending point corresponds to the power generation level at 13:00. First, the difference between the first value (9:00 am data) and the last value (13:00 data) of the sequence is evaluated, and then divided by the duration of 5 hours to obtain the average rate; this is equivalent to simplifying the frequent cloud cover caused by the up and down fluctuations in the sequence into a whole upward or downward trend value. The rate value generated in this way ignores internal mutations, but lays an initial reference benchmark for subsequent adaptive judgment of stable state.

[0024] In S2, first, the fluctuation threshold is dynamically adapted to cope with the characteristics of different management periods: a plurality of consecutive time length intervals (such as 1 h below, 2 h-3 h, 3 h above period) are preset, and differentiated fluctuation judgment thresholds are assigned to each interval, and the design logic follows the core principle that the longer the management period, the smaller the threshold. This is because longer management periods are more sensitive to fluctuations (for example, a summer all-day delivery plan), and slight changes may accumulate into significant deviations, so a smaller threshold is needed to capture subtle fluctuations; on the contrary, shorter periods (such as 2 hours in the afternoon) allow larger fluctuations, so a larger threshold is used. By matching the actual delivery period to be managed to the corresponding interval, the adaptive fluctuation threshold is automatically selected, thereby solving the problem that a fixed threshold cannot adapt to seasonal, weather, or period changes.

[0025] Further, based on the dynamic threshold, the fluctuation of the historical sequence is quantitatively evaluated: the difference of the power generation change between adjacent time points in the sequence (such as the power generation increase or decrease amplitude every minute or every quarter of an hour) is calculated in sequence, the proportion of the number of times that exceeds the fluctuation threshold in all difference values is counted, and the coverage ratio is formed. The coverage ratio directly reflects the inherent fluctuation intensity of the historical data - for example, in the morning data sequence of a certain day with cloudy and frequent flashes, the cloud movement causes frequent mutations of power generation between adjacent minutes, a large number of difference values exceed the threshold, and the coverage ratio increases significantly; while in the stable afternoon sequence, most of the difference values are within the threshold, and the coverage ratio is relatively low. This mechanism can objectively distinguish between stable state (low coverage ratio) and fluctuation state (high coverage ratio), and provide quantitative basis for subsequent adaptive decision-making.

[0026] In S3, when the coverage ratio does not exceed the preset threshold (indicating that the historical power generation transmission data sequence is in a stable state), the prediction is directly based on the average power generation transmission change rate: first, the split duration is defined by the prediction model, and then the to-be-managed transmission period is split into multiple key time nodes with interval split duration (for example, the split duration is defined as 30 minutes, and a certain 3-hour period can be divided into 6 key points every 30 minutes). These nodes represent the output targets of the prediction model, and the distribution logic is determined according to the total length of the historical power generation transmission data sequence and the total length of the to-be-managed transmission period: for example, for a period of stable and decaying sunlight in the afternoon, the key points may be reduced, and the split duration will be increased; for example, the split duration does not exceed the total length of the historical power generation transmission data sequence.

[0027] Further, a linear prediction sequence is constructed using the average rate: the last end of the historical sequence is taken as the starting reference value of the to-be-managed transmission period, and the theoretical power generation of each key time point is calculated based on this origin and the average rate. The calculation method is as follows for each key point: multiply the average rate by the time span of the point from the starting time, and superimpose the starting power generation data. Finally, all the predicted values of the key points are connected in time sequence to form a smooth and continuous first target transmission data sequence. For example, in the scenario of a stable sunny day in spring, if the historical sequence shows that the power generation decreases steadily every hour, the sequence generated by the prediction model will show a uniform decrease, and the slope strictly matches the average rate, ensuring that the transmission plan matches the power generation capacity and avoiding energy scheduling deviation. In summary, relying only on rough trend extrapolation can not only simplify the calculation complexity, but also ensure the generation of a high-reliability transmission scheme in a low-fluctuation scenario, thereby improving the stability of the power grid and the energy utilization rate.

[0028] In S4, when the over-limit coverage exceeds the preset threshold (indicating that the historical data is in an unstable fluctuation state), firstly, the conflict trend of the historical sequence is deeply analyzed: based on the power generation change difference of adjacent time points, the trend force in the same direction as the average rate (referred to as the target variable) and the fluctuation force in the opposite direction (referred to as the reference variable) are calculated respectively. For example, under strong convective weather, a certain period of historical sequence may contain both the sudden increase of power generation caused by the dispersion of cloud layer (positive trend) and the sudden drop caused by the cloud layer blocking (opposite fluctuation), and all positive and negative difference values are accumulated to form two quantitative forces to reflect the intensity of the trend and fluctuation.

[0029] Further, a dynamic correction ratio is generated according to the strength of the two forces: if the trend force represented by the target variable is dominant (such as the overall power generation still showing an upward trend when the weather changes from sunny to cloudy), the correction ratio is a positive adjustment range, which weakens the influence of sudden fluctuations on the average rate; on the contrary, if the fluctuation force represented by the reference variable is stronger (such as continuous rain leading to a cliff-like drop), the correction ratio is a negative adjustment, which significantly inhibits the optimistic prediction tendency of the average rate. Finally, the correction ratio is multiplied by the average power transmission change rate to form a corrected rate that eliminates the disturbance of fluctuations. Based on the average power transmission change rate, the same prediction model in S3 is used to construct the power generation sequence at the key time point (with the historical end data as the starting point, the time span as the horizontal axis, and the correction rate as the slope to generate a linear prediction value), forming the second target transmission sequence.

[0030] For example, on a strong and variable wind day in the summer and autumn, even if the overall trend of the historical data is slowly rising, the negative correction ratio will be amplified by the minute-level power generation shock caused by frequent gusts, and the slope of the generated second sequence will be gentler than the original average rate, avoiding the problem of power imbalance caused by blind transmission. This mechanism ensures that in a high fluctuation scenario, the game between trend and disturbance can still be automatically weighed, and a controllable and feasible solution can be output.

[0031] In summary, throughout the entire prediction model-based solar power generation adaptive delivery method, first, based on the average change rate calculated based on the head and tail values of the historical sequence, the overall trend characteristics are quickly captured, laying the foundation for the initial prediction; second, according to the design of intelligently matching the fluctuation threshold value in the management period, the defects of the fixed threshold value are overcome, and the dynamic balance of short-term noise tolerance and long-term sensitivity is realized; by calculating the overrun coverage rate of the difference value of adjacent data points, the fluctuation intensity of the historical data is accurately quantified, providing an objective standard for stable state judgment, so that in the stable scene, linear extrapolation prediction is directly used to ensure efficiency, and in the fluctuation scene, depth trend conflict analysis is started to separate and quantify the confrontation intensity of the same trend and the reverse fluctuation, and according to this, the proportion correction is generated to more accurately reflect the true trend: when the trend is dominant, the sudden interference is weakened in the positive direction, and when the fluctuation is dominant, the optimistic prediction deviation is suppressed in the negative direction, and finally the smooth delivery sequence after the fluctuation noise is eliminated is output.

[0032] In one embodiment, the average power generation delivery change rate is obtained according to the historical power generation delivery data sequence in S1, which includes: S11, obtaining the head and tail power generation delivery change amount and the historical delivery time length of the historical power generation delivery data sequence; S12, obtaining the average power generation delivery change rate according to the ratio of the head and tail power generation delivery change amount and the historical delivery time length.

[0033] In this embodiment, it should be noted that in S11, the historical power generation delivery data sequence is first extracted, the key values at the beginning and end of the sequence are accurately located, and the algebraic difference amount between the two is calculated. This amount directly reflects the net change amplitude of the power generation capacity in the preset time period (which may be positive growth or negative decay). At the same time, the total span of the continuous historical delivery time covered by the sequence is strictly counted, and this time length is used as the denominator for rate calculation to ensure that the result has unified time dimension comparability (such as hourly rate or minute rate).

[0034] In S12, the head and tail power generation delivery change amount in S11 is divided by the historical delivery time length to obtain a single average power generation delivery change rate. This operation is essentially to compress all complex fluctuations in the historical sequence (including instantaneous drops caused by sudden cloud cover or sudden rises caused by clear skies) into a linear expression. For example, in a historical sequence at a certain winter afternoon, even if there are several snowmelt-induced sawtooth fluctuations in power generation, the calculation result will still abstract it into a single slope of continuous and gentle rise or fall. This simplified calculation based on the head and tail value framework significantly reduces the computational complexity while providing a quantifiable and comparable trend benchmark value for subsequent stable state judgment, avoiding the high computational load and real-time bottleneck caused by point-by-point analysis.

[0035] In one embodiment, the step of obtaining the first target delivery data sequence of the to-be-managed delivery period based on the prediction model and the average power generation delivery change rate in S3 comprises: S31, obtaining a plurality of key time points of the to-be-managed delivery period according to the prediction model; S32, taking the power generation delivery data at the end of the historical power generation delivery data sequence as the starting power generation delivery data of the to-be-managed delivery period, and obtaining the predicted power generation delivery data corresponding to the plurality of key time points based on the prediction model, the average power generation delivery change rate, and the starting power generation delivery data; S33, forming the first delivery data sequence of the to-be-managed delivery period according to the predicted power generation delivery data corresponding to the plurality of key time points.

[0036] In the present embodiment, it should be noted that in S31, the to-be-managed delivery period is finely divided into a time grid by the prediction model. According to the length characteristics of the historical power generation delivery data sequence and the span of the to-be-managed delivery period, the distribution positions of the key time nodes are determined: for example, in the noon period with strong regularity of light, the node interval can be appropriately relaxed to simplify the calculation; while in the sunset period with frequent weather changes, the nodes are automatically encrypted to capture nonlinear changes. These nodes constitute the time coordinate framework of the prediction, and the distribution logic ensures that both typical power generation patterns (such as seasonal light rising and falling curves) and management needs of different lengths (such as the difference between short-term regulation and long-term planning) can be reflected.

[0037] In S32, the latest power generation data at the end of the historical sequence is taken as the dynamic prediction reference origin, which is taken as the starting power generation of the to-be-managed period. Based on the reference value and the average power generation delivery change rate generated in S1, linear trend extrapolation calculation is performed for each key time node: the average rate is multiplied by the time span of the node from the reference origin, and the reference power generation is superimposed to obtain the theoretical power generation prediction value of the node. For example, in the stable scenario of autumn with high and cool weather, if the historical end data shows that the power generation is at a high level and the average rate is slowly decreasing, the prediction value will decrease uniformly with the passage of time, forming a linear evolution path that strictly follows the overall trend.

[0038] In S33, the prediction values of all key nodes are connected in time sequence to construct a continuous and smooth target delivery sequence. The sequence fills in the transition values between nodes by linear interpolation, forming a smooth curve without jumps; further, the overall slope completely inherits the linear characteristics of the average rate; further, the sequence starting value is forced to align with the current latest power generation state. For example, for the winter continuous weak light scenario, the generated sequence represents a delivery plan that starts from the current power generation and continuously and gently decreases according to the historical trend, ensuring that the power generation capacity is strictly synchronized with the grid dispatching plan under stable conditions.

[0039] In one embodiment, the selecting the fluctuation threshold value according to the to-be-managed delivery period in S2 comprises: S21, a plurality of value intervals are obtained in sequence, each value interval corresponding to a different fluctuation threshold value, wherein the larger the value of the value interval, the smaller the fluctuation threshold value corresponding to the value interval; S22, selecting the value interval to which the to-be-managed delivery period belongs according to the size of the to-be-managed delivery period, and obtaining the fluctuation threshold value corresponding to the value interval.

[0040] In the present embodiment, it should be noted that in S21, a preset dynamic threshold mapping rule is constructed: a set of continuous and non-overlapping management period length intervals (such as 0-1 hour, 1-3 hours, and more than 3 hours) are constructed, and a differentiated fluctuation determination threshold value is configured for each interval. The core principle is that the larger the upper limit of the interval length, the smaller the corresponding threshold value. This is because a longer management period has lower tolerance to fluctuations - subtle changes can significantly affect overall delivery stability through cumulative effects (for example, an 8-hour cross-day delivery plan), so a smaller threshold value is required to capture minor fluctuations; otherwise, the threshold value can be relaxed for short periods (such as a 30-minute peak shaving period) to avoid excessive sensitivity to triggering the correction mechanism.

[0041] In S22, by comparing the actual to-be-managed delivery period with the preset interval library, the interval category to which it belongs is determined, and the fluctuation threshold value corresponding to the interval is automatically extracted. For example, a certain 4-hour seasonal delivery plan will match the long period interval of more than 3 hours, and the corresponding smaller fluctuation threshold value will be enabled; while the 1-hour lunch peak regulation will be classified into the short period interval of 0-1 hour, and the corresponding larger fluctuation threshold value will be used. This mechanism enables dynamic adaptation to the sensitivity differences of different seasons, weather patterns, and power grid load scenarios.

[0042] In one embodiment, the obtaining the proportion of the fluctuation threshold value exceeded in the historical power generation delivery data sequence in S2 and taking the proportion as the overrun coverage rate comprises: S23, obtaining the difference value of each adjacent power generation delivery data in the historical power generation delivery data sequence; S24, obtaining the number of difference values exceeding the fluctuation threshold value as the overrun number, and obtaining the proportion of the overrun number in the number of difference values in the historical power generation delivery data sequence as the overrun coverage rate.

[0043] In the present embodiment, it should be noted that in S23, the fluctuation trajectory sequence is formed by calculating the power generation change difference value of adjacent time points in the sequence (such as the change amplitude between data points every second, every minute, or every 5 minutes). The sequence essentially reflects the instantaneous change intensity of the power generation capacity in the historical period, for example, the adjacent minute difference value may present a sawtooth-like sharp oscillation under cloudy weather, and a smooth gradual change under sunny weather. This operation generates a set of fluctuation quantification basic data, which provides direct input for subsequent threshold determination.

[0044] In S24, the number of differences exceeding the selected threshold in the fluctuation trajectory sequence in S22 is counted, and the proportion of the number in the total number of differences (i.e., the overrun coverage) is calculated. This index directly quantifies the density of mutation points within the historical data (e.g., 70% of the differences exceed the threshold during a strong convective weather period); this index reflects the duration of the fluctuation intensity (e.g., a high coverage state lasting for 30 minutes). For example, during the passage of a sandstorm, when minute-level differences are detected to continuously exceed the threshold and the coverage breaks through the preset threshold, it can be determined as an extremely unstable state, providing a decision basis for adaptive correction.

[0045] In one embodiment, the correction ratio obtained in S4 according to the historical power generation and transmission data sequence includes: S41, obtain total trend variables of rising trend and falling trend respectively according to the historical power generation and transmission data sequence, and obtain a total trend variable with the same trend as the average power generation and transmission change rate as a target variable, and obtain a total trend variable with the opposite trend as the average power generation and transmission change rate as a reference variable; S42, obtain the correction ratio according to the target variable and the reference variable.

[0046] In this embodiment, it should be noted that in S41, based on the power generation change difference between adjacent time points (e.g., the power generation increase or decrease value from a certain time to the next time), all differences are decomposed into rising total trend variables (the sum of all positive differences) and falling total trend variables (the sum of absolute values of all negative differences). Then, taking the average rate direction generated in S1 as the benchmark: if the average rate is positive (overall rising trend), the rising trend combined force is defined as the target variable supporting the trend, and the falling trend combined force is defined as the reference variable of the reverse interference; conversely, if the average rate is negative (overall falling trend), the falling trend combined force becomes the target variable, and the rising trend combined force becomes the reference variable. For example, during a period of frequent summer thunderstorms, the minute-level power generation surge caused by sunlight exposure and the cliff-like drop caused by rainstorm attack form an antagonistic relationship, and the total sum of both forces can be quantified by classified accumulation.

[0047] In S42, a correction ratio is generated according to the confrontation energy ratio of the target variable and the reference variable. When the target variable is dominant (e.g., the total amount of positive fluctuation is greater in the overall upward trend), the relative proportion of the target variable exceeding the reference variable is calculated as a positive correction value, and the reliability of the original trend is amplified. Conversely, when the reference variable is stronger (e.g., the total amount of negative fluctuation exceeds the positive fluctuation due to sudden hail), the proportion of the reference variable exceeding the target variable is calculated and taken as a negative value, and the prediction deviation of the original trend is suppressed. For example, on the day a typhoon passes, although the average rate shows a slow upward trend, the instantaneous power generation drop caused by frequent squall lines causes the reference variable to surge, and a negative correction ratio is generated accordingly, so that the final corrected prediction trend is closer to the actual fluctuation risk. This conflict quantization mechanism ensures that the trend dominance and the intensity of sudden disturbances can be intelligently balanced in a high-noise scenario.

[0048] It should be further noted that the corrected power generation and transmission change rate is obtained by multiplying the correction ratio by the average power generation and transmission change rate, and then the first target transmission data sequence of the to-be-managed transmission period can be obtained based on the prediction model and the average power generation and transmission change rate in S3, that is, the second target transmission data sequence of the to-be-managed transmission period is obtained based on the prediction model and the corrected power generation and transmission change rate.

[0049] In one embodiment, the correction ratio in S42 is obtained according to the target variable and the reference variable, and is expressed as: ; wherein, is the correction ratio, is the target variable, is the reference variable.

[0050] In this embodiment, it should be noted that the entire expression essentially builds a dynamic trend credibility evaluation, which solves the problem of insensitivity of the average rate to local mutations in the solar power generation fluctuation scenario. Specifically, when , that is, the trend is dominant, the expression calculates the net advantage proportion of the total trend variable after removing the interference; if the total amount of fluctuation in the same direction as the average rate in the historical sequence is significantly greater than the total amount of fluctuation in the opposite direction , it indicates that although there are local mutations, the overall trend is still reliable, and at this time the correction ratio is positive, indicating that only the average rate needs to be moderately weakened (the numerator is the net trend force, divided by normalized to the proportion), so as to retain the trend direction while reflecting the weakening effect of sudden disturbances. For similar sunny-to-cloudy scenarios (overall upward trend but occasional cloud cover), it avoids excessive correction due to a few sudden drops and loses the main trend, thereby improving prediction stability.

[0051] Further, when Time (i.e. volatility dominates): Expression The relative strength of the reverse volatility exceeding the trend is calculated. If the total amount of reverse volatility exceeds or equals the total amount of the trend , it means that the sudden disturbance has seriously distorted the overall trend. At this time, the correction ratio is negative, and by (the numerator is the net disturbance strength, divided by normalization) to generate a strong inhibition coefficient, directly adjust the average power transmission change rate in the opposite direction, so that the subsequent predicted trend changes from rising to falling, or from falling to rising, while the strong inhibition coefficient weakens the absolute value of the change rate, reflecting the influence of another trend.

[0052] Further, regardless of whether the target or reference variable is dominant, the total amount of itself is used as the denominator, or to ensure that the correction ratio is always in the interval [-1, 1]. When , is in (0, 1], the larger or the smaller, the closer the correction ratio is to 1, avoiding the stable trend of the same average power transmission change rate from being overcorrected. When , is in [-1, 0), the stronger the disturbance, that is, the larger or the smaller, the closer the negative ratio is to -1, ensuring that the corrected power transmission change rate is directly opposite to the original average power transmission change rate trend, and the stable trend, avoiding the value from being overcorrected.

[0053] Also provided is a solar power generation adaptive transmission system based on a prediction model, the system comprising: a first data processing module configured to obtain a to-be-managed transmission period, obtain a historical power generation transmission data sequence of a preset time period before the to-be-managed transmission period, and obtain an average power generation transmission change rate according to the historical power generation transmission data sequence; a second data processing module configured to select a volatility threshold according to the to-be-managed transmission period, and obtain a proportion of historical power generation transmission data sequence exceeding the volatility threshold as an overrun coverage rate; a first prediction transmission module configured to, when the overrun coverage rate does not exceed a preset threshold, obtain a first target transmission data sequence of the to-be-managed transmission period based on a prediction model and the average power generation transmission change rate; The second prediction delivery module is configured to obtain a correction ratio according to the historical power generation delivery data sequence when the over-limit coverage exceeds the preset threshold, obtain a corrected power generation delivery change rate according to the average power generation delivery change rate and the correction ratio, and obtain a second target delivery data sequence of the to-be-managed delivery period based on the prediction model and the corrected power generation delivery change rate.

[0054] In one embodiment, the first prediction delivery module is further configured to obtain a plurality of key time points of the to-be-managed delivery period according to the prediction model, take the power generation delivery data at the end of the historical power generation delivery data sequence as the initial power generation delivery data of the to-be-managed delivery period, and obtain predicted power generation delivery data corresponding to the plurality of key time points based on the prediction model, the average power generation delivery change rate and the initial power generation delivery data; and form a first delivery data sequence of the to-be-managed delivery period according to the predicted power generation delivery data corresponding to the plurality of key time points.

[0055] In one embodiment, the second prediction delivery module is further configured to obtain total trend variables of the upward trend and the downward trend respectively according to the historical power generation delivery data sequence, obtain a total trend variable with the same trend as the average power generation delivery change rate as a target variable, and obtain a total trend variable with the opposite trend as the average power generation delivery change rate as a reference variable, and obtain the correction ratio according to the target variable and the reference variable.

[0056] In the present embodiment, it should be noted that the specific manner of performing operations of the above-mentioned solar power generation adaptive delivery system based on a prediction model has been described in detail in the embodiments of the method for solar power generation adaptive delivery based on a prediction model, and will not be described in detail here.

[0057] The preferred embodiments of the present disclosure are described in detail above in combination with the drawings, but the present disclosure is not limited to the specific details in the above-described embodiments. Within the technical concept of the present disclosure, various simple modifications can be made to the technical solutions of the present disclosure, and these simple modifications all belong to the protection scope of the present disclosure.

[0058] In addition, it should be noted that the various specific technical features described in the above-described specific embodiments can be combined in any appropriate manner without contradiction. In order to avoid unnecessary repetition, various possible combination manners are not described again in the present disclosure.

[0059] In addition, the various different embodiments of the present disclosure can also be combined in any appropriate manner, as long as they do not deviate from the idea of the present disclosure, and they should also be considered as disclosed by the present disclosure.

[0060] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit the present application; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the technical solutions recorded in the foregoing embodiments can still be modified, or some or all of the technical features can be replaced by equivalent replacements; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application, and they should be covered in the scope of the claims and the description of the present application.

Claims

1. A method for adaptive delivery of solar power generation based on a predictive model, the method comprising: receiving a plurality of inputs from a plurality of sources; processing the plurality of inputs to generate a plurality of outputs; and generating a predictive model based on the plurality of outputs. The system comprises: acquire a to-be-managed delivery period, acquire a historical power generation delivery data sequence of a preset time period before the to-be-managed delivery period, and acquire an average power generation delivery change rate according to the historical power generation delivery data sequence; select a fluctuation threshold according to the to-be-managed delivery period, acquire a proportion of the historical power generation delivery data sequence that exceeds the fluctuation threshold as an overrun coverage rate; if the overrun coverage rate does not exceed a preset threshold, acquire a first target delivery data sequence of the to-be-managed delivery period based on a prediction model and the average power generation delivery change rate; if the overrun coverage rate exceeds the preset threshold, acquire a correction ratio according to the historical power generation delivery data sequence, acquire a corrected power generation delivery change rate according to the average power generation delivery change rate and the correction ratio, and acquire a second target delivery data sequence of the to-be-managed delivery period based on the prediction model and the corrected power generation delivery change rate.

2. The method of claim 1, wherein, The acquiring of the average power generation delivery change rate according to the historical power generation delivery data sequence comprises: acquiring a first and last power generation delivery change amount of the historical power generation delivery data sequence and a historical delivery time length; acquiring the average power generation delivery change rate according to a ratio of the first and last power generation delivery change amount to the historical delivery time length.

3. The method of claim 1, wherein, The acquiring of the first target delivery data sequence of the to-be-managed delivery period based on the prediction model and the average power generation delivery change rate comprises: acquiring a plurality of key time points of the to-be-managed delivery period according to the prediction model; taking power generation delivery data at the end of the historical power generation delivery data sequence as starting power generation delivery data of the to-be-managed delivery period, and acquiring predicted power generation delivery data corresponding to the plurality of key time points based on the prediction model, the average power generation delivery change rate, and the starting power generation delivery data; forming a first pre-delivery data sequence of the to-be-managed delivery period according to the predicted power generation delivery data corresponding to the plurality of key time points.

4. The method of claim 1, wherein, The selecting of the fluctuation threshold according to the to-be-managed delivery period comprises: acquiring a plurality of numerical intervals that are continuous in sequence, each numerical interval corresponding to a different fluctuation threshold, wherein the larger the numerical value, the smaller the fluctuation threshold corresponding to the numerical interval; selecting a numerical interval to which the to-be-managed delivery period belongs according to the size of the to-be-managed delivery period, and acquiring a fluctuation threshold corresponding to the numerical interval.

5. The method of claim 1, wherein, The acquiring of the proportion of the historical power generation delivery data sequence that exceeds the fluctuation threshold as the overrun coverage rate comprises: acquiring a difference value of each adjacent power generation delivery data in the historical power generation delivery data sequence; acquiring a number of difference values that exceed the fluctuation threshold as an overrun number, and acquiring a proportion of the overrun number in the number of difference values in the historical power generation delivery data sequence as the overrun coverage rate.

6. The method of claim 1, wherein, The acquiring of the correction ratio according to the historical power generation delivery data sequence comprises: acquiring total trend variables of rising trends and falling trends respectively according to the historical power generation delivery data sequence, acquiring a total trend variable with the same trend as the average power generation delivery change rate as a target variable, and acquiring a total trend variable with a trend opposite to the average power generation delivery change rate as a reference variable; acquiring the correction ratio according to the target variable and the reference variable.

7. The method of claim 6, wherein, The acquiring of the correction ratio according to the target variable and the reference variable is expressed as: ; wherein, is a correction ratio, is a target variable, is a reference variable.

8. A solar power generation adaptive delivery system based on a predictive model, characterized by, The system comprises: The first data processing module is configured to obtain a to-be-managed delivery period, obtain a historical power generation delivery data sequence of a preset time period before the to-be-managed delivery period, and obtain an average power generation delivery change rate according to the historical power generation delivery data sequence; The second data processing module is configured to select a fluctuation threshold according to the to-be-managed delivery period, obtain a proportion of historical power generation delivery data in the historical power generation delivery data sequence that exceeds the fluctuation threshold as an overrun coverage rate; The first prediction delivery module is configured to, when the overrun coverage rate does not exceed a preset threshold, obtain a first target delivery data sequence of the to-be-managed delivery period based on a prediction model and the average power generation delivery change rate; The second prediction delivery module is configured to, when the overrun coverage rate exceeds the preset threshold, obtain a correction ratio according to the historical power generation delivery data sequence, obtain a corrected power generation delivery change rate according to the average power generation delivery change rate and the correction ratio, and obtain a second target delivery data sequence of the to-be-managed delivery period based on a prediction model and the corrected power generation delivery change rate.

9. The prediction model based solar power adaptive delivery system of claim 8, wherein, The first prediction delivery module is further configured to: obtain a plurality of key time points of the to-be-managed delivery period according to the prediction model; take power generation delivery data at an end of the historical power generation delivery data sequence as starting power generation delivery data of the to-be-managed delivery period, and obtain predicted power generation delivery data corresponding to the plurality of key time points based on the prediction model, the average power generation delivery change rate, and the starting power generation delivery data; form a first predicted delivery data sequence of the to-be-managed delivery period according to the predicted power generation delivery data corresponding to the plurality of key time points.

10. The prediction model based solar power adaptive delivery system of claim 9, wherein, The second prediction delivery module is further configured to: obtain total trend variables of rising trends and falling trends respectively according to the historical power generation delivery data sequence, obtain a total trend variable with a same trend as the average power generation delivery change rate as a target variable, and obtain a total trend variable with an opposite trend to the average power generation delivery change rate as a reference variable; obtain the correction ratio according to the target variable and the reference variable.

Images