Real-time compensation control system and method for photovoltaic power station output fluctuation

Abstract

The present application relates to photovoltaic power station technical field, disclose photovoltaic power station output fluctuation real-time compensation control system and method, the system includes: trend correction module obtains the real-time power sequence and output fluctuation mark of target photovoltaic power station, obtains trend correction feature;Working condition switching module utilizes output fluctuation mark to determine fluctuation interval, obtains fluctuation interval feature, and according to the data difference between adjacent fluctuation interval data determines the working condition switching mark corresponding to real-time power sequence;Fusion processing module determines the compensation weight of real-time power sequence according to working condition switching mark, carries out fusion to trend correction feature and fluctuation interval feature to obtain fusion feature, and carries out collaborative mechanism processing to obtain time sequence collaborative feature;Compensation control module deduces initial compensation feature to time sequence collaborative feature, utilizes fluctuation interval feature to correct processing fluctuation interval in initial compensation feature, realizes the accurate generation and smooth transition of output fluctuation compensation control instruction.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic power plant technology, and more specifically, to a real-time compensation control system and method for output fluctuations in photovoltaic power plants. Background Technology

[0002] Photovoltaic power plant output fluctuation compensation refers to the technical process of smoothing out intermittent changes in photovoltaic output through power regulation to ensure the quality of grid-connected power. It is widely used in centralized and distributed photovoltaic power plants. Existing compensation technologies mostly adopt fixed thresholds or simple moving average filtering schemes, which directly generate compensation commands based on the deviation between real-time power and planned values, and achieve output smoothing through energy storage or inverter regulation.

[0003] However, existing technologies fail to distinguish between trend components and fluctuation components in the power sequence, resulting in trend changes and random fluctuations being treated equally, leading to insufficient compensation accuracy. In addition, there is a lack of a mechanism to identify the switching of operating conditions between adjacent fluctuation intervals. When irradiation conditions change abruptly, the compensation weight cannot be adaptively adjusted, which can easily cause overcompensation or undercompensation. At the same time, no collaborative correction mechanism for adjacent time windows has been established, and compensation commands are prone to jump at the window boundaries, affecting the stability of inverter execution and equipment lifespan. Summary of the Invention

[0004] To address the technical problems of low compensation accuracy caused by the lack of differentiation between trend and fluctuation components and the absence of adaptive operation condition switching in existing photovoltaic power output fluctuation compensation systems, this application aims to provide a real-time compensation control system and method for photovoltaic power plant output fluctuations. By extracting multi-scale features and correcting trends, combined with operation condition switching identification and adjacent window collaborative correction, the system can achieve accurate generation and smooth transition of output fluctuation compensation control commands.

[0005] To achieve the above objectives, the present invention provides a real-time compensation control system for output fluctuations in photovoltaic power plants, comprising: The trend correction module is used to acquire the real-time power sequence and corresponding output fluctuation identifier of the target photovoltaic power station, and to perform feature extraction, time series estimation and trend correction on the real-time power sequence and the output fluctuation identifier to obtain trend correction features; The operating condition switching module is used to determine the fluctuation range in the real-time power sequence using the output fluctuation identifier, extract features from the fluctuation range data to obtain fluctuation range features, and determine the operating condition switching identifier corresponding to the real-time power sequence based on the data difference between adjacent fluctuation range data; wherein, the operating condition switching identifier indicates whether there is an operating condition switch or not. The fusion processing module is used to determine the compensation weight of the real-time power sequence according to the operating condition switching identifier, fuse the trend correction feature and the fluctuation range feature according to the compensation weight to obtain the fusion feature, and perform a collaborative mechanism processing on the fusion features of adjacent time windows to obtain the time-series collaborative feature. The compensation control module is used to deduce the initial compensation features from the time-series coordinated features, correct the fluctuation range in the initial compensation features using the fluctuation range features, and generate output fluctuation compensation control commands using the corrected initial compensation features.

[0006] Furthermore, the trend correction module is used for: The real-time power sequence is divided into multi-scale sliding windows, and the power mean, power variance and power range in each window are extracted as primary power features. Based on the power output fluctuation identifier, each primary power feature is subjected to fluctuation-sensitive weighting to obtain a weighted power feature set. Trend components are extracted from the weighted power feature set to obtain a power trend baseline, and the trend deviation between the weighted power feature set and the power trend baseline is calculated. The trend deviation is smoothed and corrected to obtain the trend correction feature.

[0007] Furthermore, the operating condition switching module is used for: The start and end points of the fluctuation intervals in the real-time power sequence are located based on the fluctuation marker bits in the output fluctuation identifier, and the duration of each fluctuation interval is determined. The power change rate, power fluctuation amplitude, and power fluctuation frequency within each fluctuation range are extracted as the characteristics of the fluctuation range.

[0008] Furthermore, the operating condition switching module is used for: Calculate the power mean difference and the power fluctuation amplitude ratio between adjacent fluctuation intervals. In response to determining that the power mean difference is greater than a preset operating condition switching threshold and the power fluctuation amplitude ratio is within a preset switching ratio range, determine that the operating condition switching flag indicates that an operating condition switching exists; otherwise, determine that the operating condition switching flag indicates that an operating condition switching does not exist.

[0009] Furthermore, the fusion processing module is used for: In response to determining that the working condition switching identifier indicates that a working condition switching has occurred, the compensation weight is set to a first compensation weight value, which is greater than the benchmark weight. In response to determining that the operating condition switching identifier does not exist, the compensation weight is set to a second compensation weight value, which is equal to or less than the baseline weight; or the baseline weight is linearly interpolated and adjusted according to the power mean difference to obtain the compensation weight.

[0010] Furthermore, the fusion processing module is used for: The compensation weight and the trend correction feature are weighted to obtain the weighted trend feature, and the weighted trend feature and the fluctuation range feature are vector-concatenated to obtain the fusion feature; Obtain the feature difference vector between the fusion feature of the current time window and the fusion feature of the previous adjacent time window. In response to determining that the magnitude of the feature difference vector is greater than a preset cooperative mutation threshold, perform a gradual correction on the fusion feature of the current time window in the direction of the fusion feature of the previous adjacent time window, and use the corrected fusion feature as the temporal cooperative feature.

[0011] Furthermore, the compensation control module is used for: The temporal collaborative features are input into the compensation inference network; The power deviation trend features are extracted by the time-series mapping layer in the compensation inference network, and the power compensation demand at a preset time step is inferred by the compensation amount generation layer based on the power deviation trend features to obtain the initial compensation features; wherein, the compensation inference network is trained based on the mapping relationship between historical power fluctuation data and corresponding historical compensation amounts.

[0012] Furthermore, the compensation control module is used for: Identify the compensation segment to be corrected in the initial compensation feature that corresponds to the fluctuation range, and determine the compensation gain coefficient based on the power fluctuation amplitude in the fluctuation range feature; Based on the power fluctuation frequency, a compensation response delay constraint is determined. According to the compensation gain coefficient and the compensation response delay constraint, the gain adjustment and amplitude limiting processing of the compensation segment to be corrected are performed to obtain the corrected initial compensation characteristics.

[0013] Furthermore, the compensation control module is used for: The corrected initial compensation characteristics are converted into inverter power regulation values, and the output fluctuation compensation control command is generated and sent to the photovoltaic power plant inverter for execution.

[0014] To achieve the above objectives, the present invention also provides a real-time compensation and control method for output fluctuations in photovoltaic power plants, comprising: The real-time power sequence and corresponding output fluctuation identifier of the target photovoltaic power station are obtained, and the real-time power sequence and the output fluctuation identifier are subjected to feature extraction, time series estimation and trend correction to obtain trend correction features. The fluctuation range in the real-time power sequence is determined by the output fluctuation identifier, the fluctuation range data is feature extracted to obtain the fluctuation range features, and the operating condition switching identifier corresponding to the real-time power sequence is determined based on the data difference between adjacent fluctuation range data; wherein, the operating condition switching identifier indicates whether there is an operating condition switch or not. The compensation weight of the real-time power sequence is determined according to the operating condition switching identifier. The trend correction feature and the fluctuation range feature are fused according to the compensation weight to obtain the fused feature. The fused features of adjacent time windows are processed by a collaborative mechanism to obtain the time-series collaborative feature. The timing coordination features are used to deduce the initial compensation features. The fluctuation range in the initial compensation features is corrected using the fluctuation range features. The corrected initial compensation features are then used to generate the output fluctuation compensation control command.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention discloses a real-time compensation control system and method for power output fluctuations in photovoltaic power plants. The system includes: a trend correction module that acquires the real-time power sequence and power output fluctuation identifier of the target photovoltaic power plant to obtain trend correction features; a mode switching module that uses the power output fluctuation identifier to determine the fluctuation range to obtain fluctuation range features, and determines the mode switching identifier corresponding to the real-time power sequence based on the data differences between adjacent fluctuation ranges; a fusion processing module that determines the compensation weight of the real-time power sequence based on the mode switching identifier, fuses the trend correction features and fluctuation range features to obtain fusion features, and performs coordination mechanism processing to obtain time-series coordination features; and a compensation control module that deduces the time-series coordination features to obtain initial compensation features, and uses the fluctuation range features to correct the fluctuation range in the initial compensation features to achieve accurate generation and smooth transition of power output fluctuation compensation control commands. Attached Figure Description

[0016] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 This invention illustrates a schematic diagram of the structure of a real-time compensation control system for power output fluctuations in a photovoltaic power plant, as shown in an embodiment of the present invention. Figure 2A flowchart illustrating the real-time compensation control method for power output fluctuations in a photovoltaic power plant according to an embodiment of the present invention is shown. Detailed Implementation

[0017] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0018] In the description of this application, it should be understood that the terms "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0019] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0020] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0021] The following is a description of preferred embodiments of the present invention in conjunction with the accompanying drawings.

[0022] like Figure 1 As shown, embodiments of the present invention disclose a real-time compensation control system for power output fluctuations in photovoltaic power plants, comprising: The trend correction module 210 is used to acquire the real-time power sequence and corresponding output fluctuation identifier of the target photovoltaic power station, and to perform feature extraction, time series estimation and trend correction on the real-time power sequence and the output fluctuation identifier to obtain trend correction features; Specifically, the real-time power sequence is acquired through a three-phase power transmitter deployed at the grid connection point of the photovoltaic power plant. The output fluctuation indicator is generated by the power fluctuation detection module. This module performs sliding window analysis on the real-time power sequence. When the difference between the maximum and minimum power values ​​within the window exceeds 0.5 MW and this difference persists for more than 3 seconds, the corresponding time period is marked as having output fluctuation, and an output fluctuation indicator is generated. The indicator value is 1 to indicate the presence of fluctuation and 0 to indicate stability.

[0023] In some embodiments of this application, the trend correction module is used for: The real-time power sequence is divided into multi-scale sliding windows, and the power mean, power variance and power range in each window are extracted as primary power features. Based on the power output fluctuation identifier, each primary power feature is subjected to fluctuation-sensitive weighting to obtain a weighted power feature set. Trend components are extracted from the weighted power feature set to obtain a power trend baseline, and the trend deviation between the weighted power feature set and the power trend baseline is calculated. The trend deviation is smoothed and corrected to obtain the trend correction feature.

[0024] Specifically, the real-time power sequence is segmented using a multi-scale sliding window: three window sizes are set: a small window with a width of 10 sampling points (corresponding to 1 second) to capture rapid power fluctuations; a medium window with a width of 50 sampling points (corresponding to 5 seconds) to capture typical power fluctuations caused by cloud drift; and a large window with a width of 300 sampling points (corresponding to 30 seconds) to capture gradual trends in irradiance conditions. Primary power features within each window are extracted: power mean, power variance, and power range (maximum value minus minimum value within the window, in megawatts). Fluctuation sensitivity weighting is applied to each primary power feature based on the power fluctuation indicator: when the power fluctuation indicator is 1, the weight of the small window feature is set to 0.5, the medium window to 0.3, and the large window to 0.2, enhancing sensitivity to rapid fluctuations; when the power fluctuation indicator is 0, the weight of the small window is set to 0.1, the medium window to 0.3, and the large window to 0.6, enhancing the ability to track long-term trends. After obtaining the weighted power feature set, trend component extraction is performed: a moving average method is used to smooth the large-window weighted power features, with a window width of 600 sampling points (corresponding to 1 minute), to obtain the power trend baseline (unit: megawatt). The trend deviation between the weighted power feature set and the power trend baseline is calculated: at each sampling time, the power trend baseline value is subtracted from the weighted power feature value to obtain the deviation (unit: megawatt). The trend deviation is then smoothed and corrected: a sliding window mid-range filter is used, with a window width of 5 sampling points, to filter out instantaneous spikes caused by sensor noise, resulting in the trend correction feature (unit: megawatt).

[0025] This application's solution uses multi-scale window segmentation and fluctuation-sensitive weighting to enable trend correction features to focus on rapid changes during volatile periods and track long-term trends during stable periods. This solves the problem of trend and fluctuation aliasing caused by traditional fixed-window filtering and improves the targeting of subsequent compensation control.

[0026] The operating condition switching module 220 is used to determine the fluctuation range in the real-time power sequence using the output fluctuation identifier, extract features from the fluctuation range data to obtain fluctuation range features, and determine the operating condition switching identifier corresponding to the real-time power sequence based on the data difference between adjacent fluctuation range data; wherein, the operating condition switching identifier indicates whether there is an operating condition switch or not. In some embodiments of this application, the operating condition switching module is used for: The start and end points of the fluctuation intervals in the real-time power sequence are located based on the fluctuation marker bits in the output fluctuation identifier, and the duration of each fluctuation interval is determined. The power change rate, power fluctuation amplitude, and power fluctuation frequency within each fluctuation range are extracted as the characteristics of the fluctuation range.

[0027] In some embodiments of this application, the operating condition switching module is used for: Calculate the power mean difference and the power fluctuation amplitude ratio between adjacent fluctuation intervals. In response to determining that the power mean difference is greater than a preset operating condition switching threshold and the power fluctuation amplitude ratio is within a preset switching ratio range, determine that the operating condition switching flag indicates that an operating condition switching exists; otherwise, determine that the operating condition switching flag indicates that an operating condition switching does not exist.

[0028] Specifically, the start and end points of fluctuation intervals in the real-time power sequence are located based on the fluctuation marker bits in the power fluctuation identifier: when the power fluctuation identifier changes from 0 to 1, the sampling point is marked as the start point of the fluctuation interval; when the power fluctuation identifier changes from 1 to 0 and remains at 0 for more than 5 seconds, the previous sampling point is marked as the end point of the fluctuation interval. The duration of each fluctuation interval is determined (the time difference between the end point and the start point). The power change rate (the difference between the maximum and minimum power values ​​within the interval divided by the duration), power fluctuation amplitude (the difference between the maximum and minimum power values ​​within the interval), and power fluctuation frequency (the number of times the power crosses the mean within the interval divided by the duration) are extracted as fluctuation interval features. The power mean difference between adjacent fluctuation intervals (the power mean of the later interval minus the power mean of the earlier interval) and the power fluctuation amplitude ratio (the fluctuation amplitude of the later interval divided by the fluctuation amplitude of the earlier interval) are calculated. In response to a situation where the difference in average power is greater than the preset switching threshold of 0.8 MW (this threshold is determined based on the typical inverter string capacity of a photovoltaic power plant; 0.8 MW corresponds to approximately 40% of the output change of a 2 MW inverter string, representing a significant abrupt change in irradiance conditions) and the power fluctuation amplitude ratio is within the preset switching ratio range of 0.3 to 3.0 (this range excludes extreme cases where the fluctuation amplitude in the latter range is too small or too large, ensuring comparability between the two ranges), the switching condition is identified as present, with a value of 1; otherwise, the switching condition is identified as absent, with a value of 0.

[0029] This application's solution identifies operating condition switching events caused by sudden irradiation by locating fluctuation markers and comparing data differences between adjacent intervals. This solves the problem of mismatch in compensation strategies caused by the failure of traditional compensation to distinguish between continuous fluctuations and sudden changes in operating conditions, enabling the compensation weights to be adaptively adjusted according to whether the operating conditions are switched.

[0030] The fusion processing module 230 is used to determine the compensation weight of the real-time power sequence according to the operating condition switching identifier, fuse the trend correction feature and the fluctuation interval feature according to the compensation weight to obtain the fusion feature, and perform a collaborative mechanism processing on the fusion features of adjacent time windows to obtain the time-series collaborative feature. In some embodiments of this application, the fusion processing module is used for: In response to determining that the working condition switching identifier indicates that a working condition switching has occurred, the compensation weight is set to a first compensation weight value, which is greater than the benchmark weight. In response to determining that the operating condition switching identifier does not exist, the compensation weight is set to a second compensation weight value, which is equal to or less than the baseline weight; or the baseline weight is linearly interpolated and adjusted according to the power mean difference to obtain the compensation weight.

[0031] Specifically, a baseline weight of 0.5 (dimensionless) is set. This value is based on the conventional compensation intensity of the energy storage system and represents the baseline contribution ratio of trend correction characteristics and fluctuation range characteristics during integration. In response to the determination that a switching condition has occurred, the compensation weight is set to a first compensation weight value of 0.8 (dimensionless), which is greater than the baseline weight of 0.5. This represents an increase in the weight of trend correction characteristics during the switching condition, enabling a rapid response to power trend changes under the new condition. In response to the determination that a switching condition has not occurred, the compensation weight is set to a second compensation weight value of 0.4 (dimensionless), which is less than the baseline weight. This represents a fine-tuning effect of reducing trend weight and enhancing fluctuation range characteristics under stable operating conditions. Alternatively, the baseline weight can be linearly interpolated based on the power mean difference: when the power mean difference is between 0 and 0.8 MW, the compensation weight equals the baseline weight plus the power mean difference divided by 0.8 MW and multiplied by 0.3, resulting in a continuous weight value between 0.5 and 0.8, achieving a smooth transition in the degree of switching condition.

[0032] This application's solution uses dynamic setting of compensation weights driven by operating condition switching. When operating conditions change abruptly, it enhances trend tracking to quickly adapt to the new steady state, and enhances fine-tuning of fluctuations to suppress random disturbances when operating conditions are stable. This solves the contradiction that fixed weights cannot balance response speed and fine-tuning.

[0033] In some embodiments of this application, the fusion processing module is used for: The compensation weight and the trend correction feature are weighted to obtain the weighted trend feature, and the weighted trend feature and the fluctuation range feature are vector-concatenated to obtain the fusion feature; Obtain the feature difference vector between the fusion feature of the current time window and the fusion feature of the previous adjacent time window. In response to determining that the magnitude of the feature difference vector is greater than a preset cooperative mutation threshold, perform a gradual correction on the fusion feature of the current time window in the direction of the fusion feature of the previous adjacent time window, and use the corrected fusion feature as the temporal cooperative feature.

[0034] Specifically, the compensation weight and trend correction feature are weighted: the weighted trend feature equals the compensation weight multiplied by the trend correction feature. When the compensation weight is 0.8, the contribution of the trend correction feature is enhanced; when it is 0.4, the contribution is weakened. The weighted trend feature is then concatenated with the fluctuation range feature (a three-dimensional vector composed of power change rate, power fluctuation amplitude, and power fluctuation frequency) to obtain the fused feature. The feature difference vector between the fused feature of the current time window and the fused feature of the previous adjacent time window is obtained: at the current sampling time, the fused feature vector of the current window is subtracted from the fused feature vector of the previous window to obtain the difference vector. The modulus of the feature difference vector (the square root of the sum of squares of each component) is calculated. In response to a modulus greater than a preset collaborative mutation threshold of 0.5 (this threshold is determined based on the normalized historical statistical standard deviation of each component of the fused feature; twice the standard deviation is used as the mutation judgment boundary to prevent feature jumps at the window boundary), the fused feature of the current time window is gradually corrected in the direction of the fused feature of the previous adjacent time window: the corrected fused feature is equal to the current fused feature multiplied by a correction coefficient of 0.7 plus the previous window fused feature multiplied by 0.3, achieving a smooth transition of features between the two windows. The corrected fused feature is then used as the temporal collaborative feature.

[0035] This application's solution suppresses jumps in compensation control commands at window boundaries by using differential detection and gradual correction of adjacent window features. This solves the problem of sudden changes in inverter power commands caused by traditional independent processing of separate windows, thereby improving the stability of compensation execution and extending equipment lifespan.

[0036] The compensation control module 240 is used to deduce the timing coordination features to obtain the initial compensation features, use the fluctuation range features to correct the fluctuation range in the initial compensation features, and use the corrected initial compensation features to generate output fluctuation compensation control commands.

[0037] In some embodiments of this application, the compensation control module is used for: The temporal collaborative features are input into the compensation inference network; The power deviation trend features are extracted by the time-series mapping layer in the compensation inference network, and the power compensation demand at a preset time step is inferred by the compensation amount generation layer based on the power deviation trend features to obtain the initial compensation features; wherein, the compensation inference network is trained based on the mapping relationship between historical power fluctuation data and corresponding historical compensation amounts.

[0038] Specifically, the temporal collaborative features are input into the compensation projection network, which is trained based on the mapping relationship between historical power fluctuation data and corresponding historical compensation amounts. The training data selects the output fluctuation event records of photovoltaic power plants in the past 6 months, including the power sequence from 30 seconds before the fluctuation to 60 seconds after the fluctuation and the manually calibrated optimal compensation amount sequence. The compensation projection network includes a temporal mapping layer and a compensation amount generation layer: the temporal mapping layer adopts a bidirectional long short-term memory network with a hidden layer dimension of 64. The input is the temporal collaborative feature sequence of the most recent 10 sampling points (corresponding to 1 second), extracting the power deviation trend feature (128-dimensional vector) to represent the direction and acceleration of the power deviation change; the compensation amount generation layer adopts a fully connected network, inputs the 128-dimensional power deviation trend feature, and outputs the power compensation demand sequence for the next preset time step (the next 2 seconds, corresponding to 20 sampling points). Each time step outputs a compensation amount (unit: megawatt), obtaining the initial compensation feature (20-dimensional vector).

[0039] This application's solution uses a compensation inference network trained on historical data to map temporal collaborative features into future multi-step compensation requirements, enabling forward-looking prediction of compensation control. This solves the problem of lag in traditional compensation commands generated solely based on current deviations and improves the ability to suppress rapid fluctuations in advance.

[0040] In some embodiments of this application, the compensation control module is used for: Identify the compensation segment to be corrected in the initial compensation feature that corresponds to the fluctuation range, and determine the compensation gain coefficient based on the power fluctuation amplitude in the fluctuation range feature; Based on the power fluctuation frequency, a compensation response delay constraint is determined. According to the compensation gain coefficient and the compensation response delay constraint, the gain adjustment and amplitude limiting processing of the compensation segment to be corrected are performed to obtain the corrected initial compensation characteristics.

[0041] In some embodiments of this application, the compensation control module is used for: The corrected initial compensation characteristics are converted into inverter power regulation values, and the output fluctuation compensation control command is generated and sent to the photovoltaic power plant inverter for execution.

[0042] Specifically, the compensation segments to be corrected corresponding to the fluctuation range in the initial compensation characteristics are identified: the time series of the initial compensation characteristics is aligned with the start and end points of the fluctuation range, and the compensation amount located within the fluctuation range is extracted as the compensation segments to be corrected. The compensation gain coefficient is determined based on the power fluctuation amplitude in the fluctuation range characteristics: when the power fluctuation amplitude is greater than 1.0 MW, the gain coefficient is set to 1.3; when the amplitude is between 0.5 and 1.0 MW, it is set to 1.1; and when the amplitude is less than 0.5 MW, it is set to 0.9. This coefficient setting is based on the principle that large fluctuations require increased compensation to quickly smooth them out, while small fluctuations require reduced gain to avoid overcompensation. The compensation response delay constraint is determined based on the power fluctuation frequency: when the power fluctuation frequency is greater than 0.5 Hz, the delay constraint is set to 20 milliseconds (requiring fast compensation response); when the frequency is between 0.2 and 0.5 Hz, it is set to 50 milliseconds; and when the frequency is less than 0.2 Hz, it is set to 100 milliseconds. This constraint represents the maximum allowable response delay when the compensation command is issued to the inverter. Based on the compensation gain coefficient and compensation response delay constraint, the gain adjustment and limiting processing of the compensation segment to be corrected is performed: the compensation amount of each time step in the compensation segment to be corrected is multiplied by the compensation gain coefficient, and then limited to the rated regulation range of the inverter (for example, the maximum regulation amount of a single inverter is 10% of the rated power, i.e., 0.2 MW; if there are 10 inverters in the power station, the total limit is 2.0 MW), thus obtaining the corrected initial compensation characteristics. The corrected initial compensation characteristics are converted into inverter power regulation amounts: the compensation amount is evenly distributed according to the number of inverters, generating a power regulation command for each inverter (unit: MW, accuracy: 0.001 MW), which is sent to each inverter in the photovoltaic power station for execution via fiber optic Ethernet at a frequency of 10 times per second.

[0043] This application's solution uses gain adjustment driven by fluctuation amplitude and delay constraint driven by fluctuation frequency to enhance the strength and speed of compensation commands during large and rapid fluctuations, and conservatively adjust them during small and slow fluctuations. This solves the problems of overcompensation or response lag caused by uniform compensation intensity, and improves the adaptability of compensation control and the safety of inverter execution.

[0044] To further illustrate the technical concept of this invention, the technical solution of this invention will now be described in conjunction with specific application scenarios.

[0045] Correspondingly, such as Figure 2 As shown, this application also provides a real-time compensation and control method for output fluctuations in photovoltaic power plants, including: S110: Obtain the real-time power sequence and corresponding output fluctuation identifier of the target photovoltaic power station, and perform feature extraction, time series estimation and trend correction on the real-time power sequence and the output fluctuation identifier to obtain trend correction features; S120: The fluctuation range in the real-time power sequence is determined by the output fluctuation identifier, the fluctuation range data is subjected to feature extraction to obtain the fluctuation range features, and the operating condition switching identifier corresponding to the real-time power sequence is determined according to the data difference between adjacent fluctuation range data; wherein, the operating condition switching identifier indicates whether there is an operating condition switch or not. S130: Determine the compensation weight of the real-time power sequence according to the operating condition switching identifier, fuse the trend correction feature and the fluctuation interval feature according to the compensation weight to obtain the fused feature, and process the fused features of adjacent time windows through a collaborative mechanism to obtain the time-series collaborative feature. S140: The timing coordination features are deduced to obtain the initial compensation features. The fluctuation range in the initial compensation features is corrected using the fluctuation range features. The corrected initial compensation features are then used to generate the output fluctuation compensation control command.

[0046] In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0047] Although the invention has been described above with reference to embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of the invention. In particular, as long as there is no structural conflict, the features in the embodiments disclosed in this invention can be combined with each other in any way. The fact that not all of these combinations are described in this specification is merely for the sake of brevity and resource conservation.

[0048] It will be understood by those skilled in the art that the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A real-time compensation control system for output fluctuations in a photovoltaic power plant, characterized in that, include: The trend correction module is used to acquire the real-time power sequence and corresponding output fluctuation identifier of the target photovoltaic power station, and to perform feature extraction, time series estimation and trend correction on the real-time power sequence and the output fluctuation identifier to obtain trend correction features; The operating condition switching module is used to determine the fluctuation range in the real-time power sequence using the output fluctuation identifier, extract features from the fluctuation range data to obtain fluctuation range features, and determine the operating condition switching identifier corresponding to the real-time power sequence based on the data difference between adjacent fluctuation range data; wherein, the operating condition switching identifier indicates whether there is an operating condition switch or not. The fusion processing module is used to determine the compensation weight of the real-time power sequence according to the operating condition switching identifier, fuse the trend correction feature and the fluctuation range feature according to the compensation weight to obtain the fusion feature, and perform a collaborative mechanism processing on the fusion features of adjacent time windows to obtain the time-series collaborative feature. The compensation control module is used to deduce the initial compensation features from the time-series coordinated features, correct the fluctuation range in the initial compensation features using the fluctuation range features, and generate output fluctuation compensation control commands using the corrected initial compensation features.

2. The real-time compensation control system for output fluctuations in a photovoltaic power station according to claim 1, characterized in that, The trend correction module is used for: The real-time power sequence is divided into multi-scale sliding windows, and the power mean, power variance and power range in each window are extracted as primary power features. Based on the power output fluctuation identifier, each primary power feature is subjected to fluctuation-sensitive weighting to obtain a weighted power feature set. Trend components are extracted from the weighted power feature set to obtain a power trend baseline, and the trend deviation between the weighted power feature set and the power trend baseline is calculated. The trend deviation is smoothed and corrected to obtain the trend correction feature.

3. The real-time compensation control system for output fluctuations in a photovoltaic power plant according to claim 1, characterized in that, The operating condition switching module is used for: The start and end points of the fluctuation intervals in the real-time power sequence are located based on the fluctuation marker bits in the output fluctuation identifier, and the duration of each fluctuation interval is determined. The power change rate, power fluctuation amplitude, and power fluctuation frequency within each fluctuation range are extracted as the characteristics of the fluctuation range.

4. The real-time compensation control system for output fluctuations in a photovoltaic power plant according to claim 3, characterized in that, The operating condition switching module is used for: Calculate the power mean difference and the power fluctuation amplitude ratio between adjacent fluctuation intervals. In response to determining that the power mean difference is greater than a preset operating condition switching threshold and the power fluctuation amplitude ratio is within a preset switching ratio range, determine that the operating condition switching flag indicates that an operating condition switching exists; otherwise, determine that the operating condition switching flag indicates that an operating condition switching does not exist.

5. The real-time compensation control system for output fluctuations in a photovoltaic power station according to claim 1, characterized in that, The fusion processing module is used for: In response to determining that the working condition switching identifier indicates that a working condition switching has occurred, the compensation weight is set to a first compensation weight value, which is greater than the benchmark weight. In response to determining that the working condition switching identifier does not exist, the compensation weight is set to a second compensation weight value, which is equal to or less than the baseline weight. Alternatively, the compensation weight can be obtained by linearly interpolating the baseline weight based on the difference in the average power value.

6. The real-time compensation control system for output fluctuations in a photovoltaic power station according to claim 1, characterized in that, The fusion processing module is used for: The compensation weight and the trend correction feature are weighted to obtain the weighted trend feature, and the weighted trend feature and the fluctuation range feature are vector-concatenated to obtain the fusion feature; Obtain the feature difference vector between the fusion feature of the current time window and the fusion feature of the previous adjacent time window. In response to determining that the magnitude of the feature difference vector is greater than a preset cooperative mutation threshold, perform a gradual correction on the fusion feature of the current time window in the direction of the fusion feature of the previous adjacent time window, and use the corrected fusion feature as the temporal cooperative feature.

7. The real-time compensation control system for output fluctuations in a photovoltaic power plant according to claim 1, characterized in that, The compensation control module is used for: The temporal collaborative features are input into the compensation inference network; The power deviation trend features are extracted by the time-series mapping layer in the compensation inference network, and the power compensation demand at a preset time step is inferred by the compensation amount generation layer based on the power deviation trend features to obtain the initial compensation features; wherein, the compensation inference network is trained based on the mapping relationship between historical power fluctuation data and corresponding historical compensation amounts.

8. The real-time compensation control system for output fluctuations in a photovoltaic power plant according to claim 1, characterized in that, The compensation control module is used for: Identify the compensation segment to be corrected in the initial compensation feature that corresponds to the fluctuation range, and determine the compensation gain coefficient based on the power fluctuation amplitude in the fluctuation range feature; Based on the power fluctuation frequency, a compensation response delay constraint is determined. According to the compensation gain coefficient and the compensation response delay constraint, the gain adjustment and amplitude limiting processing of the compensation segment to be corrected are performed to obtain the corrected initial compensation characteristics.

9. The real-time compensation control system for output fluctuations in a photovoltaic power station according to claim 8, characterized in that, The compensation control module is used for: The corrected initial compensation characteristics are converted into inverter power regulation values, and the output fluctuation compensation control command is generated and sent to the photovoltaic power plant inverter for execution.

10. A method for real-time compensation and control of output fluctuations in a photovoltaic power plant, applied to the real-time compensation and control system for output fluctuations in a photovoltaic power plant as described in any one of claims 1-9, characterized in that, include: The real-time power sequence and corresponding output fluctuation identifier of the target photovoltaic power station are obtained, and the real-time power sequence and the output fluctuation identifier are subjected to feature extraction, time series estimation and trend correction to obtain trend correction features. The fluctuation range in the real-time power sequence is determined by the output fluctuation identifier, the fluctuation range features are extracted from the fluctuation range data, and the operating condition switching identifier corresponding to the real-time power sequence is determined based on the data difference between adjacent fluctuation range data; wherein, the operating condition switching identifier indicates whether there is an operating condition switch or not. The compensation weight of the real-time power sequence is determined according to the operating condition switching identifier. The trend correction feature and the fluctuation range feature are fused according to the compensation weight to obtain the fused feature. The fused features of adjacent time windows are processed by a collaborative mechanism to obtain the time-series collaborative feature. The timing coordination features are used to deduce the initial compensation features. The fluctuation range in the initial compensation features is corrected using the fluctuation range features. The corrected initial compensation features are then used to generate the output fluctuation compensation control command.

Images