CN-122022169-A - Solar resource refined evaluation method and system based on multi-source data fusion

CN122022169ACN 122022169 ACN122022169 ACN 122022169ACN-122022169-A

Abstract

The invention relates to the technical field of renewable energy sources, and discloses a solar resource refined evaluation method and system based on multi-source data fusion. The method comprises the steps of obtaining real-time atmospheric parameters by combining satellite inversion and ground observation data, determining cloud layer aerosol interference intensity, calculating an irradiation component through a radiation transmission equation to obtain a preliminary irradiation estimated value sequence, iteratively adjusting atmospheric transparency parameters to finish correction, constructing a time dimension error accumulation mode, generating time correction coefficient correction data, combining scattering effect parameter iteration optimization to obtain a final dynamic irradiation model, generating a terrain correction irradiation field by combining photovoltaic project regional terrain interaction data, determining a power generation prediction adjustment value, verifying accuracy through a random forest algorithm, and outputting a refined evaluation result. The method breaks through the limitation of weak and error accumulation of the traditional model dynamic interference, accurately captures the influence of the atmosphere and the terrain, improves the evaluation precision, and provides technical support for photovoltaic project site selection and power generation quantity prediction.

Inventors

ZHANG JIN
LIU JIANCHAO
LI HONG
XIE QINGBIAO
WANG JUNXIANG

Assignees

紫金龙净清洁能源有限公司
西藏麻米紫金龙净清洁能源有限公司

Dates

Publication Date: 20260512
Application Date: 20260130

Claims (10)

1. A solar resource refinement evaluation method based on multi-source data fusion is characterized by comprising the following steps: merging satellite inversion information and ground observation data to obtain a real-time atmosphere parameter set, and determining cloud layer aerosol interference intensity of an initial radiation transmission path; Solving an irradiation component of the cloud layer aerosol interference intensity through a radiation transmission equation to obtain a preliminary irradiation estimated value sequence; comparing the preliminary irradiation estimated value sequence with historical observation data, and iteratively adjusting an atmospheric transparency parameter to obtain a corrected irradiation estimated value sequence; constructing a time dimension error accumulation mode for the corrected irradiation estimated value sequence, locking the burst cloud quantity change and generating a time correction coefficient; Correcting the corrected irradiation estimated value sequence by using the time correction coefficient, and obtaining a space homogenization irradiation distribution map through space interpolation processing; Fusing scattering effect parameters to the space homogenized irradiation distribution map, and performing iterative optimization to obtain a final dynamic irradiation model; combining the final dynamic irradiation model with the terrain interaction data of the existing photovoltaic project area to generate a terrain correction irradiation field and determine a power generation prediction adjustment value; and according to the power generation prediction adjustment value, verifying the accuracy of the model by adopting a random forest algorithm, and outputting a refined evaluation result.
2. The method of claim 1, wherein the merging the satellite inversion information with the ground observation data to obtain a real-time set of atmospheric parameters, determining a cloud aerosol interference intensity of the initial radiation transmission path, comprises: Performing space-time registration on the satellite inversion information and ground observation data to generate a space-time registration matrix of high-dimensional meteorological features; Inputting the space-time registration matrix into a preset atmospheric state inversion model to obtain a real-time atmospheric parameter set; Constructing an initial radiation transmission path according to the real-time atmosphere parameter set, and extracting the optical properties of the non-uniform medium on the initial radiation transmission path; and processing the optical properties of the non-uniform medium by adopting a discrete longitudinal standard method to obtain a path attenuation model, quantifying the cloud main scattering contribution rate and the aerosol additional attenuation rate according to the path attenuation model, and determining the cloud layer aerosol interference intensity.
3. The method according to claim 1, wherein said calculating the irradiance component from the irradiance transmission equation for the cloud cover aerosol disturbance intensity, to obtain a preliminary irradiance estimation sequence, comprises: dividing the atmosphere into a plurality of layers of non-uniform medium vertical layered structures from the ground surface to the troposphere top according to the cloud layer aerosol interference intensity distribution data, and distributing corresponding optical characteristics to each layer; Inputting the non-uniform medium vertical layered structure into a radiation transmission equation, and adopting a layered accumulation mode to carry out numerical integration to obtain a diffuse radiation component and a direct radiation component; Coupling and calculating the preset surface reflection characteristic data and the diffuse irradiation component and the direct irradiation component to generate an instantaneous surface irradiation distribution matrix; and extracting the numerical characteristics of the instantaneous surface irradiation distribution matrix, and obtaining a preliminary irradiation estimated value sequence through time sequence mapping.
4. The method of claim 1, wherein iteratively adjusting the atmospheric transparency parameter to the preliminary irradiance estimation sequence, in comparison to the historical observation data, results in a modified irradiance estimation sequence, comprising: Time alignment is carried out on the preliminary irradiation estimated value sequence and the historical observation data to generate a time sequence irradiation deviation matrix; Constructing a root mean square error objective function according to the time sequence irradiation deviation matrix, and solving a gradient vector of the objective function with respect to an atmospheric transparency parameter; If the root mean square error objective function value exceeds a preset error threshold value, calculating a correction step length according to the gradient vector to obtain an optimized atmospheric transparency coefficient; And inputting the optimized atmospheric transparency coefficient into the radiation transmission equation for secondary simulation to obtain a corrected radiation estimated value sequence.
5. The method of claim 1, wherein said constructing a time-dimensional error accumulation pattern for said modified irradiance estimation sequence, locking for sudden cloud cover variations and generating a time correction factor, comprises: calculating the instantaneous deviation gradient of the corrected irradiation estimated value sequence and the historical observation data along the time dimension, and constructing an error accumulation mode; Analyzing the error accumulation mode, and if the accumulated drift amount of the monotonic interval increases sharply and the hysteresis related characteristics break, recording a sudden cloud amount change event; deducing the attenuation oscillation period of the radiation deviation according to the continuous span of the burst cloud quantity change event; and generating a dynamic weighting factor according to the damping oscillation period, and performing reverse compensation operation on the error accumulation mode to obtain a time correction coefficient.
6. The method of claim 1, wherein said correcting said corrected sequence of irradiance estimates with said temporal correction factor, spatially interpolating to obtain a spatially homogenized irradiance distribution map, comprises: multiplying the time correction coefficient with the corrected irradiation estimated value sequence point by point to obtain irradiation data after time domain correction; mapping the irradiation data subjected to time domain correction to a discrete observation point coordinate system, constructing a topological adjacency relation network, calculating a space correlation matrix of the topological adjacency relation network, and analyzing and identifying local microclimate difference characteristics; if the local microclimate difference characteristic exceeds a preset difference threshold value, smoothing the intensity gradient to generate an irradiation gain matrix; And applying the irradiation gain matrix to the irradiation data subjected to time domain correction to generate a space homogenization irradiation distribution map.
7. The method according to claim 1, wherein said iteratively optimizing the spatially homogenized irradiance distribution map, with the scatter effect parameters, to obtain a final dynamic irradiance model, comprises: combining the space homogenization irradiation distribution diagram with the atmospheric light depth data in the real-time atmospheric parameter set to construct a multidimensional radiation transmission equation set; Inputting the multidimensional radiation transmission equation set into a reverse ray tracing engine, and calculating initial scattering effect parameters; Constructing a scattering deviation matrix of the initial scattering effect parameter and actual observation, and if the maximum element of the scattering deviation matrix exceeds a preset deviation threshold value, iteratively adjusting an earth surface albedo variable to obtain a corrected scattering effect parameter; and fusing the corrected scattering effect parameters to the space homogenized irradiation distribution diagram, synthesizing a time-varying three-dimensional irradiation matrix, and analyzing to obtain a final dynamic irradiation model.
8. The method of claim 1, wherein said generating a terrain-modifying irradiation field and determining a power generation predictive adjustment value for said final dynamic irradiation model in combination with existing photovoltaic project area terrain interaction data comprises: matching the final dynamic irradiation model with the geospatial coordinates of the photovoltaic project planning area, and constructing a digital elevation model to obtain a gradient slope factor and an elevation shielding matrix; according to the digital elevation model, fusing the gradient slope factor and the elevation shielding matrix to generate a terrain correction irradiation field; Sampling the terrain correction irradiation field according to the component arrangement vector, generating shadow loss time sequence data, and calculating a theoretical power generation sequence; And if the component temperature rise deviation value corresponding to the theoretical power generation sequence exceeds a preset temperature rise deviation threshold, weighting the theoretical power generation sequence, and determining a power generation prediction adjustment value.
9. The method of claim 1, wherein said verifying model accuracy using a random forest algorithm based on the power generation prediction adjustment value, outputting a refined evaluation result, comprises: Acquiring historical meteorological record data of a target area, constructing a comprehensive meteorological feature data set, sequencing the feature importance of the power generation prediction adjustment value by adopting a random forest algorithm, and extracting key meteorological influence factors; constructing a multidimensional feature space mapping relation based on the key weather influencing factors, and performing deviation analysis on the power generation prediction adjustment value to obtain a deviation analysis result; generating a regional power generation prediction distribution grid fused with the interactive features of the terrain and the weather according to the deviation analysis result; And integrating the accuracy index of the distribution grid and the final dynamic irradiation model, and outputting a refined evaluation result.
10. A solar energy resource refinement evaluation system based on multisource data fusion is characterized by comprising: The data fusion module is used for fusing satellite inversion information and ground observation data to obtain a real-time atmosphere parameter set and determining cloud layer aerosol interference intensity of an initial radiation transmission path; the simulation calculation module is used for calculating an irradiation component according to a radiation transmission equation to the cloud layer aerosol interference intensity to obtain a preliminary irradiation estimated value sequence; The iteration correction module is used for comparing the preliminary irradiation estimated value sequence with historical observation data, and iteratively adjusting the atmospheric transparency parameter to obtain a corrected irradiation estimated value sequence; The time sequence correction module is used for constructing a time dimension error accumulation mode for the corrected irradiation estimated value sequence, locking the burst cloud quantity change and generating a time correction coefficient; The space uniformity module is used for correcting the corrected irradiation estimated value sequence by using the time correction coefficient, and obtaining a space uniformity irradiation distribution map through space interpolation processing; The scattering fusion module is used for homogenizing the irradiation distribution diagram in the space, fusing scattering effect parameters, and performing iterative optimization to obtain a final dynamic irradiation model; The terrain adaptation module is used for generating a terrain correction irradiation field and determining a power generation prediction adjustment value according to the final dynamic irradiation model and the terrain interaction data of the existing photovoltaic project area; and the verification output module is used for verifying the accuracy of the model by adopting a random forest algorithm according to the power generation prediction adjustment value and outputting a refined evaluation result.

Description

Solar resource refined evaluation method and system based on multi-source data fusion Technical Field The invention relates to the technical field of renewable energy sources, in particular to a solar resource refined evaluation method and system based on multi-source data fusion. Background The accuracy of solar energy resource evaluation directly influences the site selection scientificity and the power generation efficiency prediction accuracy of photovoltaic and photo-thermal projects, efficient irradiation simulation and dynamic error correction are keys for guaranteeing the stable operation of new energy projects, and the method has important significance in improving the development and utilization capacity of renewable energy sources by combining a big data processing technology. In the prior art, a static estimation mode based on a single data source or a simplified radiation transmission model is mainly adopted for solar resource estimation, an estimation scheme is formulated through preset fixed parameters, local irradiation calculation is carried out by combining limited observation data, and information such as multi-source data association relation, dynamic atmospheric interference and the like is simply processed. However, because the static estimation mode operates based on fixed parameters, single data source and simplified model feedback are relied on, the irradiation clues in massive multi-source heterogeneous data cannot be deeply mined and associated with analysis by means of big data acquisition and processing technology, the atmospheric condition dynamic change and the hidden interference mode cannot be adapted in real time, and when the dynamic working conditions such as sudden cloud amount change or complex terrain shielding occur in a region, the problems of irradiation estimation deviation, error accumulation or site selection estimation misalignment and the like are easily caused. Therefore, the prior art has a defect of weak dynamic interference coping capability. Disclosure of Invention The invention provides a solar resource refined evaluation method and a system based on multi-source data fusion, which are used for solving the problem of weak dynamic interference coping capacity in the prior art. In a first aspect, the present invention provides a solar resource refinement evaluation method based on multi-source data fusion, including: merging satellite inversion information and ground observation data to obtain a real-time atmosphere parameter set, and determining cloud layer aerosol interference intensity of an initial radiation transmission path; Solving an irradiation component of the cloud layer aerosol interference intensity through a radiation transmission equation to obtain a preliminary irradiation estimated value sequence; comparing the preliminary irradiation estimated value sequence with historical observation data, and iteratively adjusting an atmospheric transparency parameter to obtain a corrected irradiation estimated value sequence; constructing a time dimension error accumulation mode for the corrected irradiation estimated value sequence, locking the burst cloud quantity change and generating a time correction coefficient; Correcting the corrected irradiation estimated value sequence by using the time correction coefficient, and obtaining a space homogenization irradiation distribution map through space interpolation processing; Fusing scattering effect parameters to the space homogenized irradiation distribution map, and performing iterative optimization to obtain a final dynamic irradiation model; combining the final dynamic irradiation model with the terrain interaction data of the existing photovoltaic project area to generate a terrain correction irradiation field and determine a power generation prediction adjustment value; and according to the power generation prediction adjustment value, verifying the accuracy of the model by adopting a random forest algorithm, and outputting a refined evaluation result. In an optional implementation manner, the method for merging satellite inversion information and ground observation data to obtain a real-time atmosphere parameter set and determining cloud layer aerosol interference intensity of an initial radiation transmission path includes: Performing space-time registration on the satellite inversion information and ground observation data to generate a space-time registration matrix of high-dimensional meteorological features; Inputting the space-time registration matrix into a preset atmospheric state inversion model to obtain a real-time atmospheric parameter set; Constructing an initial radiation transmission path according to the real-time atmosphere parameter set, and extracting the optical properties of the non-uniform medium on the initial radiation transmission path; and processing the optical properties of the non-uniform medium by adopting a discrete longitudinal standard method to obtain a path atte