Photovoltaic snow covering depth and power loss forecasting method and system based on multi-source data fusion

Abstract

The invention discloses a photovoltaic snow covering depth and power loss forecasting method and system based on multi-source data fusion, and belongs to the technical field of photovoltaic power forecasting. The problems of high-precision accumulated snow influence assessment, short-term risk early warning and power loss quantitative prediction of a photovoltaic system in cold and alpine regions are solved. The method comprises the following steps: selecting three numerical weather forecasting modes, namely a global forecasting system operated by the National Environmental Forecasting Center, a wind energy and solar energy forecasting system of the China Meteorological Administration and a unified mode weather forecast of the British Meteorological Administration, acquiring weather forecasting data, performing unified space-time interpolation, abnormal value correction and unit conversion, and then performing set averaging to obtain a set average value; ensemble forecast data is obtained; based on a temperature-irradiance condition and a friction force mechanism, an accumulated snow covering model is constructed, a photovoltaic panel snow covering range and photovoltaic power loss are obtained, early warning information of snow covering time, duration and power loss is generated through matching of station longitude and latitude and national grid forecasting, and photovoltaic snow covering depth and power loss forecasting based on multi-source data fusion is completed.

Description

Technical Field

[0001] This invention belongs to the field of photovoltaic power prediction technology, specifically involving a method and system for predicting photovoltaic snow cover depth and power loss through multi-source data fusion. Background Technology

[0002] As photovoltaic (PV) power generation continues to account for a growing share of the global energy mix, its power output sensitivity to weather changes is increasingly becoming a significant factor affecting the safe and stable operation of the power grid. Snow accumulation, a common winter weather phenomenon in cold regions, directly shades the surface of PV modules, drastically reducing incident irradiance and causing a significant drop in current and power output. Under conditions of heavy snowfall or sustained low temperatures, snow may even remain on the module surface for extended periods, triggering large-scale load reduction or shutdowns of power plants, posing significant risks to operation management and power generation revenue. Therefore, developing a technological platform capable of integrating multi-source weather forecast data, accurately assessing snow accumulation processes, and quantifying their impact on PV output is of critical engineering significance and an urgent practical need for enhancing the resilience of PV systems and promoting the safe integration of new energy sources.

[0003] In the existing technological system, research on the snow impact assessment of photovoltaic power plants mainly focuses on two technical paths, each with a relatively independent technical framework and model construction approach.

[0004] The first type of technology primarily relies on numerical weather prediction models or ground station observation data to simulate and reproduce surface snow accumulation processes. These methods typically use a single-source meteorological driving field as input, including basic meteorological variables such as snowfall, precipitation phase discrimination, 2m air temperature, 10m wind speed, relative humidity, and net radiation flux. In terms of technical implementation, they often employ empirical relationships (such as temperature index snowmelt models), simplified energy budget equations, or parameterized snow accumulation evolution models to approximate the processes of snow accumulation, compaction, wind-induced redistribution, runoff, and melting. Some studies also introduce fixed compaction coefficients or exponential density variation models to describe the evolution of snow layer structure over time. However, because they mainly rely on a single numerical weather prediction model or ground station observation data to estimate surface snow accumulation processes, their typical approach is to use meteorological driving quantities such as snowfall, air temperature, wind speed, and relative humidity from a single model to simulate the accumulation, compaction, runoff, and melting processes of snow within empirical relationships, energy budget models, or simplified physical parameterization frameworks. However, these methods generally have the following shortcomings: (1) The singularity of meteorological drivers means that model biases cannot be offset by ensemble methods, and differences between different forecast models in terms of physical schemes, surface parameterization, and spatiotemporal resolution can easily introduce systematic errors; (2) Insufficient spatiotemporal continuity makes it difficult to support high-resolution snow cover simulation at the regional scale based on single-point or single-model inputs; (3) Uncertainty is not quantified, and single-source driving results in limited robustness and generalization ability of simulation results under complex weather conditions. Therefore, these methods have certain limitations in terms of prediction accuracy, stability, and operational applicability.

[0005] The second category of technologies primarily focuses on assessing the impact of snow accumulation on the optical characteristics and output performance of photovoltaic modules. These methods typically rely on static snow accumulation characteristics obtained from historical ground observations, UAV aerial photography, or remote sensing, such as snow thickness, snow cover, and surface albedo attenuation coefficients. In terms of model construction, empirical fitting relationships, shading ratio models, or optical attenuation models are often used to estimate the impact of snow accumulation on the module's incident irradiance, effective light-receiving area, and DC output power. However, because they depend on historical observations or remote sensing-derived static snow accumulation characteristics, such as snow thickness, snow cover, and surface albedo attenuation, combined with empirical formulas, they estimate power generation losses. These technologies often have the following characteristics and shortcomings: (1) They lack characterization of dynamic processes, and the processes of snow deposition, compaction, sliding, and melting are often not included in a complete time evolution model; (2) They lack meteorological driving links, and most studies are based on post-event data or qualitative coverage characteristics, which cannot achieve forward-looking predictions; (3) The model coupling is low, and the interaction between snow accumulation processes and irradiance, component temperature, and power plant operating status is not adequately described, making it difficult to meet the needs of actual operation and maintenance decisions for minute- to hour-level predictions. In particular, these technologies have not yet systematically introduced a multi-model ensemble forecasting framework to perform multi-source fusion and probability quantification of meteorological uncertainties. Summary of the Invention

[0006] The problem this invention aims to solve is to meet the requirements of high-precision snow cover impact assessment, short-term risk warning and power loss quantitative prediction for photovoltaic systems in cold and high-altitude regions. It proposes a method and system for predicting photovoltaic snow cover depth and power loss by fusing multi-source data.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A method for predicting photovoltaic snow cover depth and power loss through multi-source data fusion includes the following steps:

[0009] S1. Select three numerical weather prediction models: the Global Forecast System (GFS) operated by the U.S. National Center for Environmental Prediction, the China Meteorological Administration's Wind and Solar Energy Forecast System (CMA-WSP), and the UK Met Office's Unified Model Weather Forecast (UKMO). Collect weather forecast data, perform unified spatiotemporal interpolation, outlier correction, and unit conversion, and then perform ensemble averaging to obtain ensemble forecast data.

[0010] S2. Based on temperature-irradiance conditions and friction mechanism, a snow cover model is constructed. The ensemble forecast data obtained in step S1 is used to determine the process of new snow accumulation, freezing-melting and sliding, so as to realize the physical driving simulation of the snow cover range of photovoltaic array and photovoltaic power loss.

[0011] S3. Based on the snow cover range and photovoltaic power loss of the photovoltaic panels obtained from step S2, the system matches the latitude and longitude of the power station with the national grid forecast to generate early warning information including snow cover time, duration and power loss, thus completing the forecast of photovoltaic snow cover depth and power loss through multi-source data fusion.

[0012] Furthermore, the specific implementation method of step S1 includes the following steps:

[0013] S1.1. Weather forecast data were collected from three numerical weather prediction models: the Global Forecast System (GFS) operated by the U.S. National Center for Environmental Prediction, the China Meteorological Administration's Wind and Solar Energy Forecast System (CMA-WSP), and the UK Met Office's Unified Model Weather Forecast (UKMO).

[0014] S1.2. The spatiotemporal resolution of the three weather forecast data is unified to 0.1° resolution for the next 10 days per hour using spatial nearest neighbor interpolation and temporal interpolation methods, and mapped to a unified target grid. The geographical range of the target grid is latitude 15°N-55°N, the spatial resolution is 0.1°×0.1°, and the number of grid points is 400×700, thus obtaining the three weather forecast data after spatiotemporal interpolation.

[0015] S1.3. Median filtering is used to process outliers in the three types of weather forecast data after spatiotemporal interpolation. NaN values ​​and values ​​that do not conform to physical laws are filled with the values ​​of the surrounding 3×3 spatial grid to obtain the three types of weather forecast data after outlier processing.

[0016] S1.4. Unit conversion is performed on the three types of weather forecast data after outlier processing, with temperature data uniformly converted to... Snow depth converted to cm, snowfall rate converted to Irradiance converted ;

[0017] S1.5. Perform ensemble averaging on the three types of weather forecast data after unit conversion to generate gridded data covering the Chinese region for the next 10 days, hourly, with a spatial resolution of 1°. Output the data in NetCDF file format to obtain ensemble forecast data.

[0018] Furthermore, the specific implementation method of step S2 includes the following steps:

[0019] S2.1. Based on the ensemble forecast data obtained in step S1, first determine whether new snow has been generated. When the snowfall rate variable in the ensemble forecast is greater than 0.1 cm / h and the snow depth threshold is greater than 0.5 cm, it is determined that new snow has been generated, and the snow cover area for that hour is set to 1. If it is determined that no new snow has been generated, the snow cover area for that hour is set to be the same as the previous hour.

[0020] S2.2. Calculate the initial snow cover area by combining the snow depth threshold and system parameters;

[0021] S2.3. The following factors shall be used to control the sliding of snow off the photovoltaic panels:

[0022] Friction factor control: Sliding will occur when the sliding force generated by gravity exceeds the static friction force;

[0023] Control of melting slip factors after freezing: When snow freezes at the interface between the solar panel module and the solar panel module, when the interface temperature rises to 0... As the above begins to melt, the snow will start to slide, and irradiance and temperature are key variables affecting the melting and sliding of snow.

[0024] Obtain the fitting formula The accumulated snow will slide off, where Ta is the temperature and G is the irradiance;

[0025] When snow slides off, the amount of snow that slides off, S, is calculated using the following formula:

[0026]

[0027] in, The coefficient of friction, For the tilt angle of the photovoltaic panel, Let g be the mass of the snow, and g be the acceleration due to gravity.

[0028] Since the mass of snow, m, is difficult to measure directly, the amount of snowfall, S, is calculated using an empirical formula:

[0029] ;

[0030] S2.4. Based on the new snow cover range (the current hourly snow cover range minus the amount of snow that has fallen), calculate the photovoltaic power loss caused by the snow cover range on the photovoltaic panels hourly. Calculate the number of parallel battery strings covered by snow and use their proportion of the total number of parallel battery strings as the DC capacity loss to obtain the photovoltaic power loss caused by the snow cover range. The calculation formula is:

[0031]

[0032] in, This refers to the proportion of the tilt height of the module row covered by snow, i.e., the snow-covered area of ​​the photovoltaic panels. This represents the number of battery strings connected in parallel along the tilt height direction. Let be the number of parallel series covered by snow, where This indicates rounding up to the nearest integer.

[0033] Furthermore, the specific implementation method of step S3 includes the following steps:

[0034] S3.1. Data Extraction: By configuring the specific latitude and longitude of the station, the latitude and longitude of the grid points in the NetCDF format of the national grid forecast product output by the model are matched, and the variable data of the matched grid points are extracted.

[0035] S3.2. Warning Generation: Warning information is stored in a CSV file, including the start time of snow cover, the end time of snow cover, the duration of snow cover, the snow depth and snow coverage during the snow cover process, and the photovoltaic power loss.

[0036] A system for a method of predicting photovoltaic snow cover depth and power loss by multi-source data fusion includes a processor, a memory, and a computer program stored in the memory and run on the processor. When the computer program runs, it implements the steps of the method of predicting photovoltaic snow cover depth and power loss by multi-source data fusion.

[0037] The beneficial effects of this invention are:

[0038] The photovoltaic snow cover depth and power loss forecasting method based on multi-source data fusion described in this invention unifies the processing of data from three independent numerical weather prediction models. This includes spatial interpolation to eliminate resolution differences between forecast models, temporal resampling to achieve timeliness consistency, and ensemble averaging techniques to reduce the uncertainty of individual models and improve the stability and reliability of meteorological driving fields. Based on the fused high-resolution meteorological dataset, the system further constructs a photovoltaic snow cover model based on physically empirical parameters. This model can simulate key processes such as snowfall, snow accumulation, compaction, sliding, and snow melting, identify the degree of snow cover on the module surface, and quantify the DC-side power loss caused by snow accumulation in conjunction with the irradiance transmission mechanism, achieving a dynamic, continuous, and predictable assessment of the impact of snow accumulation on photovoltaic arrays.

[0039] The photovoltaic snow cover depth and power loss prediction method based on multi-source data fusion described in this invention has broad application value. On one hand, it can be used for the daily operation and maintenance of photovoltaic power plants, supporting snow cover monitoring, snow removal decisions, and power generation loss assessment, thus improving the efficiency of power plant operation in winter. On the other hand, its output irradiance and power generation capacity prediction results can provide boundary conditions for new energy output for grid dispatch, assisting in the formulation of more reasonable dispatch strategies and improving the system's adaptability to weather fluctuations. Furthermore, the system can provide reliable forecasting basis for energy trading, reducing the risk of power generation deviations caused by snow disasters. It can also serve as an enhancement to professional forecasting products from meteorological departments, providing data support for the analysis and service of short-term snowfall processes in the region. In summary, this invention has strong engineering practicality and application potential, and can play an important role in several key industries such as photovoltaics, power dispatch, energy trading, and meteorological services. Attached Figure Description

[0040] Figure 1 This is a flowchart of the photovoltaic snow cover depth and power loss prediction method based on multi-source data fusion described in this invention;

[0041] Figure 2 This is a schematic diagram illustrating the spatiotemporal resolution and variable selection for the three numerical weather prediction methods of this invention;

[0042] Figure 3 This is a graph showing the relationship between irradiance, temperature, and whether snow slides off in this invention.

[0043] Figure 4 This is a flowchart of the snow-covered photovoltaic loss algorithm of the present invention;

[0044] Figure 5 This is a comparison chart of the snow cover warning accuracy of this invention with data from ground-based stations. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention; that is, the described specific embodiments are merely a part of the embodiments of the invention, and not all of them. The components of the specific embodiments of the invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations, and the invention may also have other embodiments.

[0046] Therefore, the following detailed description of specific embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected specific embodiments of the invention. All other specific embodiments obtained by those skilled in the art based on these specific embodiments without inventive effort are within the scope of protection of this invention.

[0047] To further understand the invention's content, features, and effects, the following specific embodiments are provided, along with accompanying drawings. Figure 1 -Appendix Figure 5 Detailed explanation is as follows:

[0048] Example 1:

[0049] A method for predicting photovoltaic snow cover depth and power loss through multi-source data fusion includes the following steps:

[0050] S1. Select three numerical weather prediction models: the Global Forecast System (GFS) operated by the U.S. National Center for Environmental Prediction, the China Meteorological Administration's Wind and Solar Energy Forecast System (CMA-WSP), and the UK Met Office's Unified Model Weather Forecast (UKMO). Collect weather forecast data, perform unified spatiotemporal interpolation, outlier correction, and unit conversion, and then perform ensemble averaging to obtain ensemble forecast data.

[0051] Furthermore, the specific implementation method of step S1 includes the following steps:

[0052] S1.1. Weather forecast data were collected from three numerical weather prediction models: the Global Forecast System (GFS) operated by the U.S. National Center for Environmental Prediction, the China Meteorological Administration's Wind and Solar Energy Forecast System (CMA-WSP), and the UK Met Office's Unified Model Weather Forecast (UKMO).

[0053] Snow depth forecasting inherently involves significant uncertainty; different numerical weather prediction models may produce snow depth forecasts for the same time and location with substantial differences. Therefore, ensemble forecasting using multi-source data fusion achieves better forecast results by combining multiple forecast outcomes. Furthermore, multi-source data fusion avoids data gaps in extreme weather conditions, enhances forecast stability and continuity, and reduces forecast risk. This algorithm selects three numerical weather prediction models for multi-source data fusion: the Global Forecast System (GFS) operated by the US National Center for Environmental Prediction, the China Meteorological Administration's Wind and Solar Energy Forecasting System (CMA-WSP), and the UK Met Office Unified Model Weather Forecast (UKMO). The temporal resolution, spatial resolution, and available variables of the three numerical weather prediction products are as follows: Figure 2 As shown.

[0054] S1.2. The spatiotemporal resolution of the three weather forecast data is unified to 0.1° resolution for the next 10 days per hour using spatial nearest neighbor interpolation and temporal interpolation methods, and mapped to a unified target grid. The geographical range of the target grid is latitude 15°N-55°N, the spatial resolution is 0.1°×0.1°, and the number of grid points is 400×700, thus obtaining the three weather forecast data after spatiotemporal interpolation.

[0055] Furthermore, the specific methods for spatial nearest neighbor interpolation and temporal interpolation are as follows:

[0056] S1.2.1. Target Grid Definition: Assume the target spatial grid is a regular latitude and longitude grid:

[0057]

[0058] in, and They represent longitude and latitude respectively. ( ) represents the coordinates of the lower left corner of the study area. The target time series is defined as... That is, from the predicted time Starting from [date], hourly forecasts will be provided for 240 consecutive hours.

[0059] S1.2.2. Spatial Nearest Neighbor Interpolation: Suppose a certain numerical weather forecast product is located in the original spatial grid... The variables are given above. For any grid point in the target grid ( Find the nearest grid point in the original grid. ):

[0060] ( )= ;

[0061] The variable values ​​on the target mesh are defined as follows: ;

[0062] This method can maintain the physical consistency of the original numerical forecast products to the greatest extent without introducing additional smoothing, and is especially suitable for variables with obvious spatial discontinuities, such as cloud cover, precipitation phase, and snow depth.

[0063] S1.2.3. Time Interpolation: Due to differences in time resolution among different products, it is necessary to uniformly interpolate them to an hourly time scale. Let the original time series be { }, The corresponding variable is .

[0064] For the target time Linear time interpolation is used:

[0065]

[0066] in, The original time resolution is set to 1 hour. For products with an original resolution of 1 hour, they are directly mapped to the target time series without interpolation.

[0067] S1.3. Median filtering is used to process outliers in the three types of weather forecast data after spatiotemporal interpolation. NaN values ​​and values ​​that do not conform to physical laws are filled with the values ​​of the surrounding 3×3 spatial grid to obtain the three types of weather forecast data after outlier processing.

[0068] Furthermore, the specific method is as follows: Let the variables after unifying the spatiotemporal resolution be... A value is considered an outlier if it meets the following conditions. Or violate physical constraints ;

[0069] in For the reasonable physical value range corresponding to the variable (such as precipitation) 0, snow depth etc.), for grid points that are judged as abnormal ( At that moment Define its spatial domain window as That is, a 3×3 spatial grid centered at this point, within which all non-outlier values ​​are collected:

[0070]

[0071] If set If the value is not empty, outliers are replaced with the neighborhood median. When the target grid point is located at the boundary of the study area, resulting in an incomplete 3×3 spatial neighborhood, only the available neighborhood grid is used in the calculation.

[0072] S1.4. Unit conversion is performed on the three types of weather forecast data after outlier processing, with temperature data uniformly converted to... Snow depth converted to cm, snowfall rate converted to Irradiance converted ;

[0073] S1.5. Perform ensemble averaging on the three types of weather forecast data after unit conversion to generate gridded data covering the Chinese region for the next 10 days, hourly, with a spatial resolution of 1°. Output the data in NetCDF file format to obtain ensemble forecast data.

[0074] S2. Based on temperature-irradiance conditions and friction mechanism, a snow cover model is constructed. The ensemble forecast data obtained in step S1 is used to determine the process of new snow accumulation, freezing-melting and sliding, so as to realize the physical driving simulation of the snow cover range of photovoltaic array and photovoltaic power loss.

[0075] Furthermore, the specific implementation method of step S2 includes the following steps:

[0076] S2.1. Based on the ensemble forecast data obtained in step S1, first determine whether new snow has been generated. When the snowfall rate variable in the ensemble forecast is greater than 0.1 cm / h and the snow depth threshold is greater than 0.5 cm, it is determined that new snow has been generated, and the snow cover area for that hour is set to 1. If it is determined that no new snow has been generated, the snow cover area for that hour is set to be the same as the previous hour.

[0077] S2.2. Calculate the initial snow cover area by combining the snow depth threshold and system parameters;

[0078] S2.3. The following factors shall be used to control the sliding of snow off the photovoltaic panels:

[0079] Friction factor control: Sliding will occur when the sliding force generated by gravity exceeds the static friction force;

[0080] Control of melting slip factors after freezing: When snow freezes at the interface between the solar panel module and the solar panel module, when the interface temperature rises to 0... As the above begins to melt, the snow will start to slide, and irradiance and temperature are key variables affecting the melting and sliding of snow.

[0081] Obtain the fitting formula The accumulated snow will slide off, where Ta is the temperature and G is the irradiance;

[0082] When snow slides off, the amount of snow that slides off, S, is calculated using the following formula:

[0083]

[0084] in, The coefficient of friction, For the tilt angle of the photovoltaic panel, Let g be the mass of the snow, and g be the acceleration due to gravity.

[0085] Since the mass of snow, m, is difficult to measure directly, the amount of snowfall, S, is calculated using an empirical formula:

[0086] ;

[0087] S2.4. Based on the new snow cover range (the current hourly snow cover range minus the amount of snow that has fallen), calculate the photovoltaic power loss caused by the snow cover range on the photovoltaic panels hourly. If any parallel battery string is partially blocked by snow, the DC power generation capacity of that string is completely lost. Calculate the number of parallel battery strings covered by snow and use their proportion to the total number of parallel battery strings as the DC capacity loss to obtain the photovoltaic power loss caused by the snow cover range. The calculation formula is:

[0088]

[0089] in, This refers to the proportion of the tilt height of the module row covered by snow, i.e., the snow-covered area of ​​the photovoltaic panels. This represents the number of battery strings connected in parallel along the tilt height direction. Let be the number of parallel series covered by snow, where This indicates rounding up to the nearest integer.

[0090] Furthermore, by monitoring six representative photovoltaic systems, irradiance was measured using a Li-COR Li-200 pyranometer instrument with a heater, and air temperature, wind speed, daily snow depth (Campbell SR50A-L ultrasonic rangefinder) and timed digital image recordings of snow cover were obtained.

[0091] Meteorological data collected include: temperature, humidity, wind speed, and irradiance; snow data includes: snow depth at horizontal levels.

[0092] System performance: power generation; visual recording: periodically capturing digital images to record snow cover conditions.

[0093] This model introduces a coverage variable ranging from 0 to 1, representing the extent of snow cover on the photovoltaic panels. The photovoltaic system is connected in strings, and the model assumes that a string does not generate electricity if any module is covered by snow (partial shading triggers a bypass diode). Therefore, energy calculations are performed by accumulating the number of strings capable of generating electricity.

[0094] S3. Based on the snow cover range and photovoltaic power loss of the photovoltaic panels obtained from step S2, the system matches the latitude and longitude of the power station with the national grid forecast to generate early warning information including snow cover time, duration and power loss, thus completing the forecast of photovoltaic snow cover depth and power loss through multi-source data fusion.

[0095] Furthermore, the specific implementation method of step S3 includes the following steps:

[0096] S3.1. Data Extraction: By configuring the specific latitude and longitude of the station, the latitude and longitude of the grid points in the NetCDF format of the national grid forecast product output by the model are matched, and the variable data of the matched grid points are extracted.

[0097] S3.2. Warning Generation: Warning information is stored in a CSV file, including the start time of snow cover, the end time of snow cover, the duration of snow cover, the snow depth and snow coverage during the snow cover process, and the photovoltaic power loss.

[0098] Figure 5 The snow depth variable of this invention is verified at ground-based observation stations. The forecast start time is 12:00 on December 31, 2024. TP, TN, FP and FN represent the results of positive samples being correctly identified, negative samples being correctly identified, false alarms and missed alarms, respectively. Positive samples are cases with snow cover and negative samples are cases without snow cover.

[0099] Example 2:

[0100] A system for a method of predicting photovoltaic snow cover depth and power loss by multi-source data fusion includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the computer program is executed, it implements the steps of the method of predicting photovoltaic snow cover depth and power loss by multi-source data fusion as described in Example 1.

[0101] This embodiment overcomes the shortcomings of existing technologies, such as the instability of single-mode snow cover forecasts, insufficient characterization of snow cover processes, and difficulty in achieving power station-level early warnings, by constructing a technical system that integrates multi-source numerical weather predictions, dynamically simulates snow cover on photovoltaic modules, and provides station-level early warning output. By performing unified spatiotemporal interpolation, outlier correction, and unit conversion on GFS, WSP, and UKMO forecast products and then implementing ensemble averaging, the uncertainty of snow cover forecasts is effectively reduced, and the continuity and accuracy of forecast results are improved. The snow cover model, based on temperature-irradiance conditions and friction mechanisms, can accurately describe the processes of new snow accumulation, freezing-melting, and sliding, enabling physical simulation of the snow cover range and power loss of photovoltaic arrays. By matching the station's latitude and longitude with the national grid forecast, this invention can automatically generate refined early warning information such as snow cover time, duration, and power loss. Therefore, this invention significantly improves the stability, accuracy, and operational applicability of snow cover forecasts, demonstrating a clear technological advancement.

[0102] It should be noted that relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0103] Although this application has been described above with reference to specific embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of this application. In particular, as long as there is no structural conflict, the features in the specific embodiments disclosed in this application can be combined with each other in any way. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, this application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for predicting photovoltaic snow cover depth and power loss through multi-source data fusion, characterized in that, Includes the following steps: S1. Select three numerical weather prediction models: the Global Forecast System (GFS) operated by the U.S. National Center for Environmental Prediction, the China Meteorological Administration's Wind and Solar Energy Forecast System (CMA-WSP), and the UK Met Office's Unified Model Weather Forecast (UKMO). Collect weather forecast data, perform unified spatiotemporal interpolation, outlier correction, and unit conversion, and then perform ensemble averaging to obtain ensemble forecast data. S2. Based on temperature-irradiance conditions and friction mechanism, a snow cover model is constructed. The ensemble forecast data obtained in step S1 is used to determine the process of new snow accumulation, freezing-melting and sliding, so as to realize the physical driving simulation of the snow cover range of photovoltaic array and photovoltaic power loss. S3. Based on the snow cover range and photovoltaic power loss of the photovoltaic panels obtained from step S2, the system matches the latitude and longitude of the power station with the national grid forecast to generate early warning information including snow cover time, duration and power loss, thus completing the forecast of photovoltaic snow cover depth and power loss through multi-source data fusion.

2. The method for predicting photovoltaic snow cover depth and power loss by multi-source data fusion according to claim 1, characterized in that, The specific implementation method of step S1 includes the following steps: S1.

1. Weather forecast data were collected from three numerical weather prediction models: the Global Forecast System (GFS) operated by the U.S. National Center for Environmental Prediction, the China Meteorological Administration's Wind and Solar Energy Forecast System (CMA-WSP), and the UK Met Office's Unified Model Weather Forecast (UKMO). S1.

2. The spatiotemporal resolution of the three weather forecast data sets was unified to an hourly resolution of 0.1° for the next 10 days using spatial nearest neighbor interpolation and temporal interpolation methods, and mapped to a unified target grid. The geographical range of the target grid is latitude 15°N-55°N, and the spatial resolution is [missing information]. With a grid of 400×700, three types of weather forecast data were obtained after spatiotemporal interpolation. S1.

3. Median filtering is used to process outliers in the three types of weather forecast data after spatiotemporal interpolation. NaN values ​​and values ​​that do not conform to physical laws are filled with the values ​​of the surrounding 3×3 spatial grid to obtain the three types of weather forecast data after outlier processing. S1.

4. Unit conversion is performed on the three types of weather forecast data after outlier processing, with temperature data uniformly converted to... Snow depth converted to cm, snowfall rate converted to Irradiance converted ; S1.

5. Perform ensemble averaging on the three types of weather forecast data after unit conversion to generate gridded data covering the Chinese region for the next 10 days, hourly, with a spatial resolution of 1°. Output the data in NetCDF file format to obtain ensemble forecast data.

3. The method for predicting photovoltaic snow cover depth and power loss by multi-source data fusion according to claim 2, characterized in that, The specific implementation method of step S2 includes the following steps: S2.

1. Based on the ensemble forecast data obtained in step S1, first determine whether new snow has been generated. When the snowfall rate variable in the ensemble forecast is greater than 0.1 cm / h and the snow depth threshold is greater than 0.5 cm, it is determined that new snow has been generated, and the snow cover area for that hour is set to 1. If it is determined that no new snow has been generated, the snow cover area for that hour is set to be the same as the previous hour. S2.

2. Calculate the initial snow cover area by combining the snow depth threshold and system parameters; S2.

3. The following factors shall be used to control the sliding of snow off the photovoltaic panels: Friction factor control: Sliding will occur when the sliding force generated by gravity exceeds the static friction force; Control of melting slip factors after freezing: When snow freezes at the interface between the solar panel module and the solar panel module, when the interface temperature rises to 0... As the above begins to melt, the snow will start to slide, and irradiance and temperature are key variables affecting the melting and sliding of snow. Obtain the fitting formula The accumulated snow will slide off, where Ta is the temperature and G is the irradiance; When snow slides off, the amount of snow that slides off, S, is calculated using the following formula: ； in, The coefficient of friction, For the tilt angle of the photovoltaic panel, Let g be the mass of the snow, and g be the acceleration due to gravity. Since the mass of snow, m, is difficult to measure directly, the amount of snowfall, S, is calculated using an empirical formula: ； S2.

4. Based on the new snow cover range (the current hourly snow cover range minus the amount of snow that has fallen), calculate the photovoltaic power loss caused by the snow cover range on the photovoltaic panels hourly. Calculate the number of parallel battery strings covered by snow and use their proportion of the total number of parallel battery strings as the DC capacity loss to obtain the photovoltaic power loss caused by the snow cover range. The calculation formula is: ； in, This refers to the proportion of the tilt height of the module row covered by snow, i.e., the snow-covered area of ​​the photovoltaic panels. This represents the number of battery strings connected in parallel along the tilt height direction. Let be the number of parallel series covered by snow, where This indicates rounding up to the nearest integer.

4. The method for predicting photovoltaic snow cover depth and power loss by multi-source data fusion according to claim 3, characterized in that, The specific implementation method of step S3 includes the following steps: S3.

1. Data Extraction: By configuring the specific latitude and longitude of the station, the latitude and longitude of the grid points in the NetCDF format of the national grid forecast product output by the model are matched, and the variable data of the matched grid points are extracted. S3.

2. Warning Generation: Warning information is stored in a CSV file, including the start time of snow cover, the end time of snow cover, the duration of snow cover, the snow depth and snow coverage during the snow cover process, and the photovoltaic power loss.

5. A system for predicting photovoltaic snow cover depth and power loss using multi-source data fusion, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed, implements the steps of the photovoltaic snow cover depth and power loss prediction method based on multi-source data fusion as described in any one of claims 1-4.

Images