CN-122018044-A - Photovoltaic snow covering depth and power loss forecasting method and system based on multi-source data fusion

CN122018044ACN 122018044 ACN122018044 ACN 122018044ACN-122018044-A

Abstract

A photovoltaic snow covering depth and power loss forecasting method and system for multi-source data fusion belong to the technical field of photovoltaic power forecasting. The method aims to solve the problems of high-precision snow impact assessment, short-term risk early warning and power loss quantitative prediction of the photovoltaic system in cold and alpine regions. The invention selects three numerical weather forecast modes of a global forecast system operated by a national environment forecast center in the United states, a wind energy and solar energy forecast system of a Chinese weather bureau and a unified weather forecast of a British weather bureau, collects weather forecast data, performs unified space-time interpolation, outlier correction and unit conversion, then performs collection and average to obtain collection forecast data, builds a snow coverage model based on temperature-irradiance conditions and a friction mechanism to obtain a snow coverage range and photovoltaic power loss of a photovoltaic panel, and generates early warning information of snow coverage time, duration and power loss through matching of longitude and latitude of a station with national grid forecast to finish photovoltaic snow coverage depth and power loss forecast of multi-source data fusion.

Inventors

YANG DAZHI
MA YUHANG
ZHANG HAO
LIU BAI

Assignees

哈尔滨工业大学

Dates

Publication Date: 20260512
Application Date: 20260127

Claims (5)

1. A photovoltaic snow covering depth and power loss forecasting method based on multi-source data fusion is characterized by comprising the following steps: s1, selecting three numerical weather forecast modes of a global forecast system GFS operated by a national environmental forecast center, a Chinese weather bureau wind energy solar energy forecast system CMA-WSP and a unified weather forecast UKMO of a British weather bureau, collecting weather forecast data, carrying out unified space-time interpolation, outlier correction and unit conversion, and then carrying out aggregate average to obtain aggregate forecast data; S2, constructing a snow cover model based on temperature-irradiance conditions and a friction mechanism, judging the processes of new snow accumulation, freezing-melting and sliding by utilizing the set forecast data obtained in the step S1, and realizing the physical driving simulation of the snow cover range and the photovoltaic power loss of the photovoltaic array; And S3, matching the snow coverage range and the photovoltaic power loss of the photovoltaic panel obtained by simulation in the step S2 with national grid forecast through the longitude and latitude of the station, generating early warning information comprising snow coverage time, duration and power loss, and finishing the forecast of the depth and the power loss of the photovoltaic snow coverage fused by the multi-source data.
2. The method for forecasting the depth and the power loss of the photovoltaic snow covering integrated with the multi-source data according to claim 1, wherein the specific implementation method of the step S1 comprises the following steps: S1.1, collecting weather forecast data by selecting three numerical weather forecast modes, namely a global forecast system GFS operated by a national environmental forecast center, a China weather bureau wind energy and solar energy forecast system CMA-WSP and a unified weather forecast UKMO of the United kingdom weather bureau; S1.2, unifying the space-time resolution of three weather forecast data into the resolution of 0.1 degree from hour to hour in the future 10 days by a spatial nearest neighbor interpolation and time interpolation method, and mapping the three weather forecast data to a unified target grid, wherein the geographic range of the target grid is 15-55 degrees N of latitude, and the spatial resolution is The grid point number is 400 multiplied by 700, and three weather forecast data after space-time interpolation are obtained; s1.3, performing outlier processing on the three weather forecast data after time interpolation by adopting a median filtering method, and filling NaN values and values which do not accord with a physical rule with grid values in the surrounding 3X 3 space domain to obtain three weather forecast data after outlier processing; s1.4, carrying out unit conversion on three weather forecast data after abnormal value processing, and uniformly converting the temperature into The depth of snow is converted into cm, and the snowfall rate is converted into Conversion of irradiance into ; S1.5, carrying out aggregate average on three weather forecast data after unit conversion to generate gridding data which covers the China area with the spatial resolution of 1 DEG every hour for 10 days, and obtaining aggregate forecast data by using the output data format as NetCDF file format.
3. The method for forecasting the depth and the power loss of the photovoltaic snow covering integrated with the multi-source data according to claim 2, wherein the specific implementation method of the step S2 comprises the following steps: S2.1, based on the set forecast data obtained in the step S1, firstly judging whether new snow is generated, when the snowfall rate variable in the set forecast is larger than 0.1cm/h and the snow depth threshold value is larger than 0.5cm, setting the snow coverage area of the hour as 1, judging that no new snow is generated, and setting the snow coverage area of the hour as the same as that of the last hour; S2.2, calculating an initial snow coverage range by combining a snow depth threshold value and system parameters; s2.3, setting the sliding condition of snow on the photovoltaic panel to be controlled by the following factors: friction factor control, namely sliding occurs when sliding force generated by gravity exceeds static friction force; controlling the factors of melting and sliding after freezing, namely when the interface temperature of the snow and the photovoltaic panel module is raised to 0 after the freezing The above starts to slide when melting starts, wherein irradiance and temperature variables are key variables affecting snow melting and sliding; Obtaining a fitting formula Snow can slide off during the process, wherein Ta is the temperature, and G is the irradiance; When snow falls off, the falling off amount S is calculated, and the formula is as follows: ; Wherein, the In order to provide a coefficient of friction, For the inclination angle of the photovoltaic panel, The weight g is gravity acceleration; since the snow mass m is difficult to directly measure, the slip amount S is calculated according to an empirical formula, which is: ; S2.4, calculating photovoltaic power loss influenced by the snow covering range of the photovoltaic panel from time to time according to the new snow covering range which is the current hour snow covering range minus the sliding amount of snow, calculating the number of parallel battery strings covered by snow, taking the proportion of the number of the parallel battery strings as the direct current capacity loss, and obtaining the photovoltaic power loss influenced by the snow covering range of the photovoltaic panel The calculation formula is as follows: ; Wherein, the The component rows covered by snow cover are inclined in height proportion, namely the snow covering range of the photovoltaic panel, To connect the number of battery strings in parallel in the oblique height direction, For the number of parallel strings covered by snow, where Representing an upward rounding.
4. The method for forecasting the depth and the power loss of the photovoltaic snow covering with the multi-source data fusion according to claim 3, wherein the specific implementation method of the step S3 comprises the following steps: S3.1, extracting data, namely matching the specific longitude and latitude positions of the station with the longitude and latitude of grid points in the form of a national grid point forecast product NetCDF output by a model, and extracting matched grid point variable data; And S3.2. Early warning generation, wherein early warning information is stored in a csv file and comprises snow covering start time, snow covering end time, snow covering duration time, snow depth and snow coverage rate in the snow covering process and photovoltaic power loss.
5. A system of a multi-source data fusion photovoltaic snow depth and power loss forecasting method, characterized by comprising a processor, a memory and a computer program stored in the memory and executable on the processor, which computer program when run implements the steps of the multi-source data fusion photovoltaic snow depth and power loss forecasting method according to any one of claims 1-4.

Description

Photovoltaic snow covering depth and power loss forecasting method and system based on multi-source data fusion Technical Field The invention belongs to the technical field of photovoltaic power prediction, and particularly relates to a photovoltaic snow covering depth and power loss prediction method and system for multi-source data fusion. Background As the duty ratio of photovoltaic power generation in a global energy structure is continuously increased, the sensitivity of the output power to weather changes is becoming an important factor affecting the safe and stable operation of a power grid. The snow cover is used as a common weather phenomenon in winter in cold areas, and can directly cover the surface of the photovoltaic component, so that the incident irradiance is suddenly reduced, and the current and power output are obviously reduced. Under certain strong snowfall or continuous low-temperature conditions, snow can even adhere to the surfaces of components for a long time, so that large-area load shedding or outage of a power station is caused, and significant risks are brought to operation management and power generation benefits. Therefore, a technical platform which can fuse multisource weather forecast data, accurately evaluate the snow accumulation process and quantify the influence on the photovoltaic output is developed, and the technology platform has key engineering significance and urgent practical requirements for improving the risk resistance of a photovoltaic system and promoting the safe absorption of new energy. In the prior art system, researches on evaluation of snow impact of a photovoltaic power station are mainly focused on two types of technical paths, and each technical path has a relatively independent technical frame and model construction thought. The first technology mainly relies on a numerical weather forecast mode or ground site observation data to simulate and reproduce the surface snow accumulation process. The method generally takes a single-source meteorological driving field as input, and comprises basic meteorological variables such as snowfall, rainfall phase state discrimination, 2m air temperature, 10m wind speed, relative humidity, net radiation flux and the like. In technical implementation, empirical relation (such as a temperature index snow melting model), simplified energy balance equation or parameterized snow evolution model are mostly adopted to approximately solve the processes of accumulation, compaction, wind-induced redistribution, sliding, melting and the like of snow. There are also some studies describing the evolution of the snowy layer structure over time by introducing a fixed compaction factor or exponential density change model. But the estimation of the surface snow accumulation process is mainly based on a single numerical weather forecast mode or ground station observation data. The typical practice is to simulate accumulation, compaction, sliding and melting processes of accumulated snow under an empirical relationship, an energy balance model or a simplified physical parameterization framework by utilizing meteorological driving amounts such as snowfall, air temperature, air speed, relative humidity and the like from a single mode. However, the method has the defects that (1) mode deviation cannot be counteracted by an aggregation means due to the singleness of meteorological driving, systematic errors are easily introduced due to the difference of different forecasting modes in terms of physical scheme, surface parameterization and space resolution, (2) space-time continuity is insufficient, high-resolution snow simulation of a region scale is difficult to support based on single-point or single-mode input, uncertainty is not quantized, and robustness and generalization capability of a simulation result are limited under complex weather conditions due to single-source driving. Therefore, such methods have limitations in terms of prediction accuracy, stability, and business applicability. The second technology mainly focuses on evaluation of the influence of snow on the optical characteristics and output performance of a photovoltaic module. The method is generally based on static snow characteristic quantities obtained through historical ground observation, unmanned aerial vehicle aerial photography or remote sensing inversion, such as snow thickness, snow coverage rate, surface albedo attenuation coefficient and other indexes. In the aspect of model construction, an empirical fitting relation, a shielding proportion model or an optical attenuation model is mostly adopted to estimate the influence of snow on the incident radiation quantity, the effective light receiving area and the direct current output power of the component. However, because static snow features such as snow thickness, snow coverage, surface albedo attenuation and the like which depend on historical observation or remote sensing inversion are needed, the power g