Method for improving power prediction accuracy for new energy wind power and photovoltaic power

Abstract

The invention discloses a method for improving power prediction accuracy for new energy wind power and photovoltaic power, and particularly relates to the technical field of power regulation and control. By monitoring real-time operation data of a fan, the operation state of the fan is analyzed, and the type of the current state is determined; analyzing the power prediction results of the wind power plant and the photovoltaic power station in the centralized power prediction system, and evaluating the potential influence of the deviation between the power prediction and the actual output on the power grid load; on the basis of the current state category of the fan and the load influence of the power prediction deviation, whether the maintenance capacity needs to be included is determined; when it is confirmed that the maintenance capacity is included, the disturbance degree of cloud shadow dynamic movement on the photovoltaic irradiation dose is evaluated through satellite cloud shadow data with high temporal-spatial resolution; the potential influence of the power prediction deviation on the power grid load and the influence of the cloud shadow disturbance on the photovoltaic irradiation dose are comprehensively analyzed, the accuracy of the power prediction model is evaluated, early warning is carried out on the operation state of the fan, and the stability and economical efficiency of power grid operation can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power regulation and control, and more specifically, to a method for improving power prediction accuracy for new energy wind power and photovoltaic power. Background Art

[0002] With the rapid development of renewable energy power generation, photovoltaic and wind power generation have become important components of the power system. Wind power and photovoltaic power generation are volatile and random, and power output is easily affected by multiple factors such as weather and environmental conditions, making the accuracy of power forecasting a key issue for grid dispatching and safe operation. At present, power forecasting mainly relies on numerical weather forecasts and statistical models. Limited by the chaotic characteristics of the atmospheric state, it is difficult to further improve the prediction accuracy. In addition, with the acceleration of the trend of unmanned operation of new energy sites, the existing prediction models do not adequately consider the dynamic changes of wind turbine operating status and external radiation disturbances, and cannot effectively respond to the needs of deviation correction in complex scenarios.

[0003] In order to solve the above problems, a technical solution is now provided. Summary of the invention

[0004] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present invention provide a method for improving the accuracy of power prediction for new energy wind power and photovoltaic power to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] By monitoring the real-time operating data of wind turbines in new energy stations, the operating status of wind turbines is analyzed and the current status category of wind turbines is determined;

[0007] Analyze the power forecast results of wind farms and photovoltaic power plants in the centralized power forecast system, and evaluate the potential impact of the deviation between power forecast and actual output on grid load;

[0008] Determine whether to add maintenance capacity based on the current status category of the wind turbine and the potential impact of the deviation between power forecast and actual output on the grid load;

[0009] When it is confirmed that maintenance capacity has been added, the disturbance degree of cloud shadow dynamic movement on photovoltaic irradiation is evaluated by analyzing high temporal and spatial resolution satellite cloud shadow data;

[0010] The potential impact of the deviation between power prediction and actual output on grid load and the degree of disturbance of photovoltaic radiation caused by dynamic movement of cloud shadows are comprehensively analyzed to evaluate the accuracy of the power prediction model and provide early warning for wind turbines.

[0011] In a preferred embodiment, by monitoring the real-time operating data of the wind turbine in the new energy station, analyzing the operating status of the wind turbine, and determining the current status category of the wind turbine, specifically:

[0012] Obtain the operating status data of the fan;

[0013] Preprocess the collected raw data;

[0014] Based on historical operating data, use machine learning algorithms to train classification models;

[0015] The real-time collected data is input into the classification model to identify the current status category of the wind turbine, including normal power generation, reduced output operation, environmental standby, maintenance status, manual shutdown, technical standby, fault shutdown, grid failure, and unknown status.

[0016] In a preferred embodiment, the power prediction results of wind farms and photovoltaic power stations in the centralized power prediction system are analyzed to evaluate the potential impact of the deviation between the power prediction and the actual output on the grid load, specifically:

[0017] Obtain power prediction results and corresponding actual output data of wind farms and photovoltaic power plants;

[0018] Align the power prediction results with the actual output data by timestamp to form a unified time series;

[0019] The deviation between the power prediction and the actual output is calculated by the point-by-point difference formula to generate a deviation sequence;

[0020] The deviation sequence is superimposed on the power grid load model to calculate the dynamic impact on load balance: load fluctuation coefficient, the calculation formula is: Wherein, LFC represents the load fluctuation coefficient; N represents the total number of time points; is the corrected grid load balance value at the jth time point; L j Represents the grid load balance value at the jth time point.

[0021] In a preferred embodiment, based on the current status category of the wind turbine and the potential impact of the deviation between the power prediction and the actual output on the grid load, it is determined whether to add maintenance capacity, specifically:

[0022] A load fluctuation coefficient threshold is preset, and the load fluctuation coefficient is compared with the load fluctuation coefficient threshold:

[0023] When the load fluctuation coefficient is greater than or equal to the load fluctuation coefficient threshold, it means that the deviation between the power prediction result and the actual output has reached or exceeded the acceptable range of the power grid;

[0024] When the load fluctuation coefficient is less than the load fluctuation coefficient threshold, it means that the deviation between the power prediction result and the actual output is within the acceptable range of the power grid;

[0025] When the current state category of the wind turbine is normal power generation and the load fluctuation coefficient is less than the load fluctuation coefficient threshold, it is confirmed that there is no need to add maintenance capacity; otherwise, it is confirmed that there is a need to add maintenance capacity.

[0026] In a preferred embodiment, the disturbance degree of the dynamic movement of cloud shadows on the photovoltaic radiation is evaluated by analyzing the satellite cloud shadow data with high temporal and spatial resolution, specifically:

[0027] Collect high temporal and spatial resolution satellite cloud shadow data, extract cloud shadow boundaries and obtain location information;

[0028] Analyze the dynamic trend of cloud shadow time series;

[0029] Calculate cloud shadow movement vectors and predict evolution trajectories;

[0030] Evaluate photovoltaic irradiance disturbance based on solar altitude angle: define the irradiance disturbance coefficient, and the calculation formula is: Where LDC is the radiation disturbance coefficient; G d is the direct radiation; G0 is the theoretical solar radiation in a cloudless state.

[0031] In a preferred embodiment, the potential impact of the deviation between power prediction and actual output on the grid load and the degree of disturbance of the dynamic movement of cloud shadows on the photovoltaic radiation are comprehensively analyzed to evaluate the accuracy of the power prediction model and to give early warning to the wind turbine, specifically:

[0032] The load fluctuation coefficient corresponding to the potential impact of the deviation between power prediction and actual output on the grid load and the irradiation disturbance coefficient corresponding to the disturbance degree of the dynamic movement of cloud shadow on the photovoltaic irradiation are calculated to obtain the accuracy coefficient. The calculation formula is: Among them, ACF is the accuracy coefficient; LFC is the load fluctuation coefficient; LDC is the radiation disturbance coefficient;

[0033] Preset the accuracy coefficient threshold and compare the accuracy coefficient with the accuracy coefficient threshold:

[0034] When the accuracy coefficient is greater than the accuracy coefficient threshold, it indicates that the accuracy of the power prediction model is high and there is no need to issue an early warning for the wind turbine;

[0035] When the accuracy coefficient is less than or equal to the accuracy coefficient threshold, it indicates that the accuracy of the power prediction model is low and an early warning is required for the wind turbine.

[0036] The technical effects and advantages of the method for improving the power prediction accuracy of new energy wind power and photovoltaic power in the present invention are as follows:

[0037] By combining the real-time operation status analysis of wind turbines, power forecast deviation assessment and comprehensive analysis of high-temporal and spatial resolution satellite cloud shadow data, the accuracy of wind and photovoltaic power forecasts has been effectively improved. By real-time monitoring of wind turbine operation status, the current status category can be accurately identified; combined with the potential impact analysis of the power forecast and actual output deviation on the grid load, the maintenance capacity can be confirmed; using high-temporal and spatial resolution satellite cloud shadow data, the degree of disturbance of the dynamic movement of cloud shadows on photovoltaic radiation can be quantified, and the power forecast model of photovoltaic stations can be improved. By comprehensively analyzing the superposition effect of power deviation and environmental disturbance, the accuracy of the forecast model is evaluated, and wind turbine operation warnings are generated, which reduces the economic losses caused by forecast deviations and improves the safety and economic benefits of the operation of new energy stations. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a schematic diagram of a method for improving the power prediction accuracy for new energy wind power and photovoltaic power according to the present invention. DETAILED DESCRIPTION

[0039] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0040] Example 1

[0041] Figure 1 The present invention provides a method for improving the power prediction accuracy of new energy wind power and photovoltaic power, which includes the following steps:

[0042] By monitoring the real-time operating data of wind turbines in new energy stations, the operating status of wind turbines is analyzed and the current status category of wind turbines is determined;

[0043] Analyze the power forecast results of wind farms and photovoltaic power plants in the centralized power forecast system, and evaluate the potential impact of the deviation between power forecast and actual output on grid load;

[0044] Determine whether to add maintenance capacity based on the current status category of the wind turbine and the potential impact of the deviation between power forecast and actual output on the grid load;

[0045] When it is confirmed that maintenance capacity has been added, the disturbance degree of cloud shadow dynamic movement on photovoltaic irradiation is evaluated by analyzing high temporal and spatial resolution satellite cloud shadow data;

[0046] The potential impact of the deviation between power prediction and actual output on grid load and the degree of disturbance of photovoltaic radiation caused by dynamic movement of cloud shadows are comprehensively analyzed to evaluate the accuracy of the power prediction model and provide early warning for wind turbines.

[0047] Specifically, by monitoring the real-time operating data of wind turbines in new energy stations, analyzing the operating status of wind turbines, and determining the current status category of wind turbines, including:

[0048] Obtaining wind turbine operating status data: The operating status of wind turbines in new energy stations is monitored in real time through a variety of sensors. The main data collected include wind turbine power output, blade angle, wind speed, ambient temperature and fault codes.

[0049] Power output: indicates the actual power generated by the wind turbine at present, in kilowatts;

[0050] Blade angle: the rotation angle of the fan blade relative to the horizontal plane, in degrees;

[0051] Wind speed: collect real-time wind speed data in the area where the fan is located, in meters per second;

[0052] Ambient temperature: reflects the temperature conditions of the fan operating environment, in degrees Celsius;

[0053] Fault code: Alarm information during the operation of the fan equipment, used to identify possible fault types.

[0054] The above monitoring data is transmitted to the centralized control center at a fixed frequency (e.g. once per second). The collected raw data is stored in the form of time series, and the data format is a two-dimensional matrix:

[0055] D raw ={(t1,P1,θ1,V1,T1,C1),...,(t n ,P n ,θ n ,V n ,T n ,C n )}; where D raw The original data set representing the real-time operation of the wind turbine contains multi-dimensional operation data at each time point; n Indicates the timestamp of the nth time point; P n represents the power output of the fan at the nth time point; θ n V represents the blade angle of the fan blade at the nth time point; n represents the wind speed at the nth time point; T n Indicates the ambient temperature at the nth time point; C n Indicates the fault code at the nth time point.

[0056] Preprocess the collected raw data: Preprocess the raw data obtained from the sensor. The raw data may be interfered by environmental noise, sensor errors or other factors, and needs to be filtered, denoised, missing value filled, normalized, etc.

[0057] Use appropriate filtering algorithms to remove measurement noise. Common filtering methods include median filtering and Kalman filtering to smooth data. Interpolate missing values, usually using mean interpolation, linear interpolation, or machine learning-based interpolation methods to fill in missing data. The normalization step is used to convert parameters of different dimensions to the same scale to improve the stability and accuracy of model training. For parameters such as wind speed and rotation speed, their data range is converted to between 0 and 1 through maximum and minimum standardization.

[0058] The format of the preprocessed data is:

[0059] D cle ={(t1,P1′,θ1′,V1′,T1′,C′1),...,(t n ,P′ n ,θ′ n ,V′ n ,T′ n ,C′ n )}; where D cle represents the cleaned data set after preprocessing, which contains multidimensional running data at each time point; P′ n Represents the power output of the fan after pretreatment; θ′ n represents the blade angle after pretreatment; V′ n represents the wind speed after preprocessing; T′ n Indicates the ambient temperature after pretreatment; C' n Indicates the fault code after preprocessing.

[0060] Based on historical operation data, use machine learning algorithms to train classification models: extract labeled samples from historical operation data, and the format of the training data set is: tra ={(X1,Y1),...,(X m ,Y m )}; where D tra is the historical running dataset used to train the classification model; X i is the feature vector of the i-th sample, including power output, blade angle, wind speed, ambient temperature and fault code data; Y i is the wind turbine operating status category corresponding to the i-th sample, including normal power generation, reduced output operation, environmental standby, maintenance status, manual shutdown, technical standby, fault shutdown, grid failure, and unknown status.

[0061] The support vector machine algorithm is used to model the training data. The optimization objective function of the support vector machine algorithm is: Among them, w is the weight vector of the support vector machine model; b is the bias term of the support vector machine model; Γ is the regularization parameter, which is used to control the balance between model complexity and misclassification penalty; ||w|| 2 is the L2 regularization term of the weight vector, which is used to prevent the support vector machine model from overfitting.

[0062] Input the real-time collected data into the classification model to identify the current state category of the fan: Extract the feature vector from the preprocessed real-time data set: X re =(P re ,θ re ,V re ,T re ,C re ), where X re is the feature vector collected in real time, representing the multi-dimensional operation data at the current time point; re is the power output data collected in real time; θ re is the blade angle data collected in real time; V re is the wind speed data collected in real time; T re is the ambient temperature data collected in real time; C re It is the fault code data collected in real time.

[0063] Input the feature vector into the support vector machine model and output the current state category of the fan: Y re =sign(wX re +b); where Y re is the real-time wind turbine operating status category output by the support vector machine model, including normal power generation, reduced output operation, environmental standby, maintenance status, manual shutdown, technical standby, fault shutdown, power grid failure, and unknown status; sign is a symbol function. It is used to output the classification results, and the value is +1 or -1, indicating different wind turbine status categories.

[0064] Specifically, the power forecast results of wind farms and photovoltaic power plants in the centralized power forecasting system are analyzed to evaluate the potential impact of the deviation between power forecast and actual output on the grid load, including:

[0065] Obtain the power forecast results and corresponding actual output data of wind farms and photovoltaic power stations: The power forecast results of new energy stations come from the centralized power forecast system, which generates power forecast values ​​at specific future time points based on numerical weather forecasts. These data include power forecast values ​​and corresponding real-time actual output values, which are recorded synchronously based on timestamps.

[0066] The power prediction result data set is defined as: P pred ={(s1,Pp1 ),...,(s N ,P pN )}; where P pred Represents the power prediction result dataset of wind farms and photovoltaic power plants; j Indicates the timestamp of the jth time point, in seconds; P pj It represents the power prediction value at the jth time point in kilowatts; N represents the total number of time points.

[0067] The actual power output of the wind farm and photovoltaic power station is obtained through on-site monitoring equipment, and the recorded data is output at the same time interval in the format of: pr ={(s1,P a1 ),...,(s N ,P aN )};P pr Represents the actual power output data set of wind farms and photovoltaic power plants; P aj Represents the actual power output value at the jth time point.

[0068] Align the power prediction results with the actual output data by timestamp to form a unified time series: In order to analyze the deviation between the power prediction and the actual output, the two data sets need to be aligned using the timestamp as the index to form a unified time series data structure.

[0069] The unified time series is expressed as: P alg ={(s1,P p1 ,P a1 ),...,(s N ,P pN ,P aN )}; where P alg Represents the aligned time series data set, which contains the power prediction value and actual power output value at each time point.

[0070] If the timestamps are inconsistent during the alignment process, for example, the timestamp density of the predicted result is higher than the actual output, the missing actual power output value is calculated using linear interpolation:

[0071] Among them, P aj is the actual power output value at the jth time point calculated by linear interpolation; P ak and P a(k+1) are the actual power output values ​​at the kth time point and the k+1th time point respectively; s k+1 ,s k They are the timestamps of the k+1th time point and the kth time point respectively.

[0072] The deviation between the power prediction and the actual output is calculated by the point-by-point difference formula to generate a deviation sequence: The deviation sequence is generated by calculating the difference between the power prediction value and the actual output value point by point, and the calculation formula is: ΔP j =P pj -P aj ; where ΔP j is the power prediction deviation value at the jth time point.

[0073] The set of deviation sequences is expressed as: ΔP j ={ΔP1, ΔP2, ..., ΔP N}. The deviation sequence is used to reflect the difference between the predicted results and the actual output.

[0074] The deviation sequence is superimposed on the power grid load model to calculate the dynamic impact on load balance: The power grid load model describes the supply and demand balance relationship of the power grid in a specific period of time, which is defined as: L j =G j -D j Among them, L j represents the grid load balance value at the jth time point; G j represents the total power generated at the jth time point; D j represents the total grid demand at the jth time point.

[0075] The power prediction deviation value is added to the grid load model, and the corrected grid load balance value is: in, is the corrected grid load balance value at the jth time point.

[0076] The corrected load balance value is used to analyze the impact of the deviation on the stability of the power grid and calculate the load fluctuation coefficient. The calculation formula is: Wherein, LFC represents the load fluctuation coefficient; N represents the total number of time points; is the corrected grid load balance value at the jth time point; L j Represents the grid load balance value at the jth time point.

[0077] The larger the load fluctuation coefficient, the greater the deviation between the power forecast and actual output of wind farms and photovoltaic power stations, which has a more significant impact on the dynamic balance of grid load. This situation shows that the output volatility of new energy sites is stronger, and the prediction model is insufficient in dealing with real-time changes, which may cause the grid to face higher flexibility requirements during dispatch and even cause the risk of imbalance between supply and demand. When the load fluctuation coefficient is large, the grid needs to mobilize more spare capacity to compensate for the uncertainty of new energy power generation, thereby increasing operating costs. At the same time, it may also aggravate the load fluctuation of grid equipment and put pressure on grid stability.

[0078] Specifically, based on the current status of the wind turbine and the potential impact of the deviation between power forecast and actual output on the grid load, determine whether to add maintenance capacity, including:

[0079] A load fluctuation coefficient threshold is preset, and the load fluctuation coefficient is compared with the load fluctuation coefficient threshold:

[0080] When the load fluctuation coefficient is greater than or equal to the load fluctuation coefficient threshold, it means that the deviation between the power prediction result and the actual output has reached or exceeded the acceptable range of the power grid, which may cause the power grid to face a higher risk of load fluctuation in actual operation. At this time, more active response measures need to be taken, such as increasing the allocation of spare capacity, calling fast-response energy storage devices, or adjusting the load-side management strategy to avoid the impact of supply and demand imbalance on the safe operation of the power grid;

[0081] When the load fluctuation coefficient is less than the load fluctuation coefficient threshold, it means that the deviation between the power prediction result and the actual output is within the acceptable range of the power grid, and the impact on the dynamic balance of the power grid load is small;

[0082] The load fluctuation coefficient threshold is determined comprehensively based on the grid operation characteristics and the grid dispatching reserve capacity configuration capabilities.

[0083] When the current state category of the wind turbine is normal power generation and the load fluctuation coefficient is less than the load fluctuation coefficient threshold, it is confirmed that there is no need to add maintenance capacity; otherwise, it is confirmed that there is a need to add maintenance capacity.

[0084] Specifically, by analyzing satellite cloud shadow data with high temporal and spatial resolution, the disturbance degree of cloud shadow dynamic movement on photovoltaic radiation is evaluated, including:

[0085] Collect high temporal and spatial resolution satellite cloud shadow data, extract cloud shadow boundaries and obtain location information: By connecting to satellite remote sensing data sources, collect satellite cloud shadow data with high temporal resolution and high spatial resolution. High temporal resolution refers to the continuity of data in a short time interval, which is often collected at a frequency of minutes or higher. High spatial resolution refers to the size accuracy of image pixels.

[0086] After completing data collection, it is necessary to extract the boundaries of the cloud shadow area and analyze the location information. The image segmentation algorithm is used to distinguish the cloud shadow area from the non-cloud shadow area in the original remote sensing image.

[0087] For the extracted cloud shadow area, its boundary contour is further calculated, and a coordinate system such as the WGS84 geographic coordinate system is used for positioning to extract the polygonal coordinate point set of the cloud shadow boundary.

[0088] Analyze the dynamic change trend of cloud shadow time series: After extracting the cloud shadow boundary, it is necessary to perform time series analysis on the cloud shadow images at multiple times to study the change trend of cloud shadows over time. First, the cloud shadow boundaries at consecutive time points are aligned to ensure the spatial consistency of images at adjacent times.

[0089] The rate of change of cloud shadow coverage area at adjacent moments is calculated to measure the dynamic characteristics of cloud shadow expansion or contraction. The area change rate can be defined as: Among them, R A (h) is the area change rate; Δh is the time interval between two time points, in seconds; A(h l ) is at time point h l The cloud shadow coverage area at that time is in square meters; A(h l+1 ) is at time point h l+1 The cloud shadow coverage area at that time, in square meters.

[0090] Through analysis, the growth, reduction or stable state of cloud shadows in time series can be identified, providing basic data for subsequent motion vector calculations.

[0091] Calculate the cloud shadow movement vector and predict the evolution trajectory: After clarifying the time series change trend of the cloud shadow, calculate the cloud shadow movement vector. Use the optical flow method to identify the overall movement direction and speed of the cloud shadow by comparing images at consecutive time points.

[0092] The cloud shadow motion vector can be defined as: Among them, v(h l ) indicates that at time point h l The moving velocity vector of the cloud shadow; Δx and Δy represent the horizontal and vertical displacements of the cloud shadow respectively.

[0093] By using the time accumulation of the moving vector, we can predict the future evolution trajectory of the cloud shadow. l The center point of the cloud shadow is (x c,h ,y c,h ), then time point h l+1 The predicted center point is:

[0094] x c,h+1 =x c,h +v x (h l )*Δh,y c,h+1 =y c,h +v y (h l )*Δh; where x c,h+1 At time point h l+1 The cloud shadow center is at the horizontal coordinate position in meters; x c,h At time point h lThe cloud shadow center is located at the horizontal coordinate position in meters; v x (h l ) is at time point h l The horizontal velocity component of the cloud shadow in meters per second; y c,h+1 At time point h l+1 The vertical coordinate position of the cloud shadow center, in meters; y c,h At time point h l The vertical coordinate position of the cloud shadow center, in meters; v y (h l ) is at time point h l The vertical component of the cloud shadow's velocity, measured in meters per second.

[0095] Through calculation, short-term prediction of cloud shadow movement path can be achieved, which is helpful for the analysis of photovoltaic radiation disturbance.

[0096] Evaluation of photovoltaic irradiance disturbance based on solar altitude angle: The movement of cloud shadows directly affects the amount of solar radiation received by the surface, so it is necessary to combine the solar altitude angle for quantitative analysis of photovoltaic irradiance disturbance. The solar altitude angle β is defined as the angle between the sun's rays and the horizon, and its calculation formula is:

[0097] sin(β)=sin(δ)*sin(φ)+cos(δ)*cos(φ)*cos(H); where β is the solar altitude angle; δ is the solar declination angle; φ is the geographic latitude; H is the solar hour angle, which represents the angular difference between the sun and the south direction of the observation point.

[0098] The solar altitude angle is calculated based on the latitude of the observation point, the solar declination angle and the time angle. The solar altitude angle affects the amount of solar radiation received by the surface.

[0099] Based on the calculated solar altitude angle, calculate the direct radiation received per unit area G d : G d =G0*sin(β)*(1-G c ), where G d is the direct radiation received per unit area; G0 is the theoretical solar radiation under cloudless conditions; G c is the cloud shadow occlusion coefficient.

[0100] When the cloud shadow moves and covers the photovoltaic power station area, the irradiance G d The cloud shadow occlusion coefficient G c Increases and decreases, thus affecting the output of the photovoltaic power station.

[0101] The irradiance disturbance coefficient is defined to evaluate the degree of disturbance of cloud shadow on photovoltaic irradiance. The calculation formula is: Where LDC is the radiation disturbance coefficient; G dis the direct radiation; G0 is the theoretical solar radiation in a cloudless state.

[0102] The larger the irradiation disturbance coefficient, the more significant the disturbance of the dynamic movement of cloud shadows on the photovoltaic irradiation, which means that in the area where the photovoltaic power station is located, the rapid changes and frequent shading of clouds will cause large fluctuations in solar irradiation, which in turn affects the stability of the output power of the photovoltaic power station. A larger irradiation disturbance coefficient indicates that the movement of cloud shadows causes a more obvious shading effect of sunlight, which causes a large fluctuation in the effective irradiation received by the photovoltaic modules. Such fluctuations will not only reduce the overall efficiency of photovoltaic power generation, but may also have an adverse impact on the dispatching and power quality of the power grid.

[0103] Specifically, the potential impact of the deviation between power prediction and actual output on grid load and the disturbance degree of cloud shadow dynamic movement on photovoltaic radiation are comprehensively analyzed to evaluate the accuracy of the power prediction model and issue early warning for wind turbines, including:

[0104] The load fluctuation coefficient corresponding to the potential impact of the deviation between power prediction and actual output on the grid load and the irradiation disturbance coefficient corresponding to the disturbance degree of the dynamic movement of cloud shadow on the photovoltaic irradiation are calculated to obtain the accuracy coefficient. The calculation formula is: Among them, ACF is the accuracy coefficient, which is used to evaluate the accuracy of the power prediction model; LFC is the load fluctuation coefficient; LDC is the radiation disturbance coefficient.

[0105] Preset the accuracy coefficient threshold and compare the accuracy coefficient with the accuracy coefficient threshold:

[0106] When the accuracy coefficient is greater than the accuracy coefficient threshold, it indicates that the accuracy of the power prediction model is high and there is no need to issue an early warning for the wind turbine;

[0107] When the accuracy coefficient is less than or equal to the accuracy coefficient threshold, it indicates that the accuracy of the power prediction model is low, and the wind turbine needs to be warned to remind the operation and maintenance personnel to pay attention to the status of the wind turbine. The power prediction model also needs to be corrected or optimized, such as adjusting parameters, introducing more accurate meteorological data, or using more advanced algorithms to improve the model's prediction ability and adaptability.

[0108] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0109] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0110] Those of ordinary skill in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0111] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0112] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0113] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, and may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0114] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0115] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0116] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0117] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for improving the accuracy of power prediction for new energy wind power and photovoltaic power, characterized in that: The steps include: By monitoring the real-time operating data of wind turbines in new energy stations, the operating status of wind turbines is analyzed and the current status category of wind turbines is determined; Analyze the power forecast results of wind farms and photovoltaic power plants in the centralized power forecast system, and evaluate the potential impact of the deviation between power forecast and actual output on grid load; Determine whether to add maintenance capacity based on the current status category of the wind turbine and the potential impact of the deviation between power forecast and actual output on the grid load; When it is confirmed that maintenance capacity has been added, the disturbance degree of cloud shadow dynamic movement on photovoltaic irradiation is evaluated by analyzing high temporal and spatial resolution satellite cloud shadow data; The potential impact of the deviation between power prediction and actual output on grid load and the degree of disturbance of photovoltaic radiation caused by dynamic movement of cloud shadows are comprehensively analyzed to evaluate the accuracy of the power prediction model and provide early warning for wind turbines.

2. The method for improving power prediction accuracy for new energy wind power and photovoltaic power according to claim 1 is characterized in that: By monitoring the real-time operating data of wind turbines in new energy stations, analyzing the operating status of wind turbines, and determining the current status category of wind turbines, specifically: Obtain the operating status data of the fan; Preprocess the collected raw data; Based on historical operating data, use machine learning algorithms to train classification models; The real-time collected data is input into the classification model to identify the current status category of the wind turbine, including normal power generation, reduced output operation, environmental standby, maintenance status, manual shutdown, technical standby, fault shutdown, grid failure, and unknown status.

3. The method for improving power prediction accuracy for new energy wind power and photovoltaic power according to claim 2 is characterized in that: The power forecast results of wind farms and photovoltaic power plants in the centralized power forecasting system are analyzed to evaluate the potential impact of the deviation between power forecast and actual output on the grid load, specifically: Obtain power prediction results and corresponding actual output data of wind farms and photovoltaic power plants; Align the power prediction results with the actual output data by timestamp to form a unified time series; The deviation between the power prediction and the actual output is calculated by the point-by-point difference formula to generate a deviation sequence; The deviation sequence is superimposed on the power grid load model to calculate the dynamic impact on load balance: load fluctuation coefficient, the calculation formula is: Wherein, LFC represents the load fluctuation coefficient; N represents the total number of time points; is the corrected grid load balance value at the jth time point; L j Represents the grid load balance value at the jth time point.

4. The method for improving power prediction accuracy for new energy wind power and photovoltaic power according to claim 3 is characterized in that: Based on the current status category of the wind turbine and the potential impact of the deviation between power forecast and actual output on the grid load, determine whether to add maintenance capacity, specifically: A load fluctuation coefficient threshold is preset, and the load fluctuation coefficient is compared with the load fluctuation coefficient threshold: When the load fluctuation coefficient is greater than or equal to the load fluctuation coefficient threshold, it means that the deviation between the power prediction result and the actual output has reached or exceeded the acceptable range of the power grid; When the load fluctuation coefficient is less than the load fluctuation coefficient threshold, it means that the deviation between the power prediction result and the actual output is within the acceptable range of the power grid; When the current state category of the wind turbine is normal power generation and the load fluctuation coefficient is less than the load fluctuation coefficient threshold, it is confirmed that there is no need to add maintenance capacity; otherwise, it is confirmed that there is a need to add maintenance capacity.

5. The method for improving power prediction accuracy for new energy wind power and photovoltaic power according to claim 4 is characterized in that: By analyzing satellite cloud shadow data with high temporal and spatial resolution, the disturbance degree of cloud shadow dynamic movement on photovoltaic radiation is evaluated, specifically: Collect high temporal and spatial resolution satellite cloud shadow data, extract cloud shadow boundaries and obtain location information; Analyze the dynamic trend of cloud shadow time series; Calculate cloud shadow movement vectors and predict evolution trajectories; Evaluate photovoltaic irradiance disturbance based on solar altitude angle: define the irradiance disturbance coefficient, and the calculation formula is: Where LDC is the radiation disturbance coefficient; G d is the direct radiation; G0 is the theoretical solar radiation in a cloudless state.

6. The method for improving power prediction accuracy for new energy wind power and photovoltaic power according to claim 5, characterized in that: Comprehensively analyze the potential impact of the deviation between power prediction and actual output on grid load and the disturbance degree of cloud shadow dynamic movement on photovoltaic radiation, evaluate the accuracy of power prediction model, and issue early warning for wind turbines. Specifically: The load fluctuation coefficient corresponding to the potential impact of the deviation between power prediction and actual output on the grid load and the irradiation disturbance coefficient corresponding to the disturbance degree of the dynamic movement of cloud shadow on the photovoltaic irradiation are calculated to obtain the accuracy coefficient. The calculation formula is: Among them, ACF is the accuracy coefficient; LFC is the load fluctuation coefficient; LDC is the radiation disturbance coefficient; Preset the accuracy coefficient threshold and compare the accuracy coefficient with the accuracy coefficient threshold: When the accuracy coefficient is greater than the accuracy coefficient threshold, it indicates that the accuracy of the power prediction model is high and there is no need to issue an early warning for the wind turbine; When the accuracy coefficient is less than or equal to the accuracy coefficient threshold, it indicates that the accuracy of the power prediction model is low and an early warning is required for the wind turbine.

Images