Search

CN-121997075-A - Intelligent recommendation method, system, equipment and medium for mountain photovoltaic ecological restoration scheme

CN121997075ACN 121997075 ACN121997075 ACN 121997075ACN-121997075-A

Abstract

The application provides an intelligent recommendation method, system, equipment and medium for a mountain photovoltaic ecological restoration scheme, wherein the method comprises the steps of obtaining geographic data of a target area, wherein the geographic data comprise remote sensing data and a digital elevation model, performing target recognition based on deep learning based on the geographic data to obtain photovoltaic panel geographic data, performing space intelligent deduction and cloud pixel filling on the remote sensing data through a visual large model to obtain reconstruction time sequence reflectivity data, calculating a normalized vegetation index based on the reconstruction time sequence reflectivity data, performing ecological influence assessment on the target area based on the photovoltaic panel geographic data and the normalized vegetation index to obtain ecological influence data, and performing ecological restoration scheme matching based on multi-rule constraint according to the ecological influence data to obtain the ecological restoration recommendation scheme. By adopting the method, the intelligent conversion from the mountain photovoltaic ecological assessment result to the executable scheme can be realized.

Inventors

  • JIANG HOU

Assignees

  • 中国科学院地理科学与资源研究所

Dates

Publication Date
20260508
Application Date
20260407

Claims (8)

  1. 1. An intelligent recommendation method for a mountain photovoltaic ecological restoration scheme is characterized by comprising the following steps: obtaining geographic data of a target area, wherein the geographic data comprises remote sensing data and a digital elevation model; Performing target recognition based on deep learning based on the geographic data to obtain photovoltaic panel geographic data, wherein the photovoltaic panel geographic data comprises mountain photovoltaic pattern spots; Performing space intelligent deduction and cloud pixel filling on the remote sensing data through a visual large model to obtain reconstruction time sequence reflectivity data, and calculating a normalized vegetation index based on the reconstruction time sequence reflectivity data; Carrying out ecological influence assessment on the target area based on the photovoltaic panel geographic data and the normalized vegetation index to obtain ecological influence data; And carrying out ecological restoration scheme matching based on multi-rule constraint according to the ecological influence data to obtain an ecological restoration recommended scheme, wherein the ecological restoration recommended scheme is used for indicating restoration technical measures to be adopted by each mountain photovoltaic pattern spot.
  2. 2. The method of claim 1, wherein performing deep learning-based object recognition based on the geographic data to obtain photovoltaic panel geographic data comprises: Simulating the topographic shadow distribution of the target area through an illumination model based on the digital elevation model and the solar azimuth angle and the elevation angle at the imaging moment of the remote sensing data to generate a topographic shadow mask map; Carrying out space registration on the remote sensing data and the topographic shadow mask map, and carrying out self-adaptive spectrum enhancement on pixels falling in a shadow area in the remote sensing data to obtain a topographic calibration image, wherein the self-adaptive spectrum enhancement is used for weakening the influence of topographic shadow on the spectral characteristics of ground objects; Inputting the terrain calibration image into an improved U-Net deep learning model that introduces a mechanism of attention at the encoder section to focus the photovoltaic panel features in the mountain environment; And carrying out pixel-level semantic segmentation on the terrain calibration image through the improved U-Net deep learning model, outputting a binarization photovoltaic panel distribution map, and vectorizing the binarization photovoltaic panel distribution map to obtain the mountain photovoltaic pattern spots in a vector format.
  3. 3. The method of claim 1, wherein the performing spatial intelligent deduction and cloud pixel filling on the remote sensing data through the visual large model to obtain reconstructed time sequence reflectivity data comprises: Acquiring a multi-temporal satellite image sequence of a target area in a preset time span, and carrying out cloud and cloud shadow pixel identification on each scene image in the multi-temporal satellite image sequence to generate a corresponding binary cloud mask, wherein in the binary cloud mask, a first numerical representation corresponding to the pixel position is covered by cloud or cloud shadow, and a second numerical representation corresponding to the pixel position is clearly available; inputting the multi-temporal satellite image sequence and the binary cloud mask into a pre-trained visual large model, and extracting space-time joint features of the multi-temporal satellite image sequence through an encoder of the visual large model, wherein the encoder captures long-range dependency relations among pixels at any space-time position through a self-attention mechanism so as to extract the space-time joint features; Based on the space-time joint characteristics, carrying out conditional probability reasoning on all missing pixels marked by the first numerical value in the multi-temporal satellite image sequence through a decoder of the visual large model, reconstructing pixel by pixel to obtain interpolation reflectivity values of the missing pixels, and generating interpolation pixels; And combining the pixel identified by the second numerical value and the interpolation pixel based on the binary cloud mask to obtain the reconstruction time sequence reflectivity data.
  4. 4. The method of claim 1, wherein the performing ecological impact assessment on the target area based on the photovoltaic panel geographic data and the normalized vegetation index to obtain ecological impact data comprises: Acquiring auxiliary geographic environment data of the target area, wherein the auxiliary geographic environment data comprise soil type space distribution data, gradient slope data and meteorological time sequence data; Inputting the photovoltaic panel geographic data, the normalized vegetation index and the auxiliary geographic environment parameters into an ecological service function evaluation model, and respectively simulating and calculating ecological service function amounts of the target area in a photovoltaic pre-construction scene and a photovoltaic post-construction scene to generate pre-construction ecological space data and post-construction ecological space data; and carrying out space superposition analysis and pixel level difference operation on the pre-construction ecological service function space data and the post-construction ecological space data to obtain the ecological influence data.
  5. 5. The method according to claim 1, wherein the performing ecological restoration scheme matching based on multi-rule constraint according to the ecological influence data, to obtain an ecological restoration recommendation scheme, includes: constructing a repair measure knowledge rule base, wherein each repair technical measure rule is associated with a group of geographic constraint conditions and applicable conditions, and the geographic constraint conditions comprise a gradient range, a soil thickness threshold value and an ecological influence level; Carrying out space superposition analysis on the ecological influence data, gradient data obtained based on the digital elevation model and soil thickness data to generate an ecological restoration evaluation unit vector image layer with multiple attribute fields, wherein the attribute fields of each ecological restoration evaluation unit comprise a space range, an average gradient value, an average soil thickness value and an ecological influence level; performing traversal matching on the attribute fields of each ecological restoration evaluation unit and rules in the restoration measure knowledge rule base, and screening out all technically feasible restoration technical measure combinations through a rule engine; And sequencing the repair technical measure combinations based on a preset economic optimization target, determining the comprehensive optimal repair technical measure combination for each ecological repair evaluation unit, and summarizing to generate the spatial ecological repair recommended scheme.
  6. 6. An intelligent recommendation system for a mountain photovoltaic ecological restoration scheme for implementing the method as set forth in any one of claims 1 to 5, characterized in that the system includes: the data acquisition module is used for acquiring geographic data of a target area, wherein the geographic data comprises remote sensing data and a digital elevation model; The photovoltaic panel identification module is used for carrying out target identification based on deep learning based on the geographic data to obtain photovoltaic panel geographic data, wherein the photovoltaic panel geographic data comprises mountain photovoltaic pattern spots; The cloud removing reconstruction module is used for carrying out space intelligent deduction and cloud pixel filling on the remote sensing data through a visual large model to obtain reconstruction time sequence reflectivity data, and calculating a normalized vegetation index based on the reconstruction time sequence reflectivity data; The ecological assessment module is used for carrying out ecological influence assessment on the target area based on the photovoltaic panel geographic data and the normalized vegetation index to obtain ecological influence data; and the ecological restoration recommendation module is used for carrying out ecological restoration scheme matching based on multi-rule constraint according to the ecological influence data to obtain an ecological restoration recommendation scheme, wherein the ecological restoration recommendation scheme is used for indicating restoration technical measures to be adopted by each mountain photovoltaic pattern spot.
  7. 7. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the method of any one of claims 1 to 5 when executing the computer program.
  8. 8. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the method of any one of claims 1 to 5.

Description

Intelligent recommendation method, system, equipment and medium for mountain photovoltaic ecological restoration scheme Technical Field The invention belongs to the technical field of photovoltaic ecological restoration, and particularly relates to an intelligent recommendation method, system, equipment and medium for a mountain photovoltaic ecological restoration scheme. Background With the development of the remote sensing technology in the ecological monitoring field, the satellite image-based photovoltaic identification and vegetation monitoring method is widely applied to the ecological restoration technology, has the characteristic of periodically acquiring surface information in a large range, and further forms a mountain photovoltaic ecological management mode which mainly relies on the combination of traditional remote sensing monitoring and artificial experience judgment at present. In the current working practice in industry, the boundary of a photovoltaic panel is generally extracted by using a common optical satellite image, vegetation growth and the like are monitored by means of time sequence remote sensing data, and then ecological influence is evaluated and a restoration scheme is selected by combining expert experience, however, the current mode has obvious limitations that the photovoltaic recognition precision and the continuity of vegetation time sequence monitoring are insufficient due to the restriction of complex terrain and climate conditions, and meanwhile, scientific quantitative support related to the depth of a geographic environment is lacking between ecological evaluation and restoration decisions, so that the overall management efficiency is low and the reproducibility of decisions is weak. Disclosure of Invention Based on the above, it is necessary to provide an intelligent recommendation method, system, device and medium for a mountain photovoltaic ecological restoration scheme, which can break through the bottleneck of disjointing evaluation and restoration measures, and realize the intelligent conversion of the evaluation result to an executable scheme. In a first aspect, the application provides an intelligent recommendation method for a mountain photovoltaic ecological restoration scheme, which comprises the following steps: Obtaining geographic data of a target area, wherein the geographic data comprises remote sensing data and a digital elevation model; Performing target recognition based on deep learning based on geographic data to obtain photovoltaic panel geographic data, wherein the photovoltaic panel geographic data comprises mountain photovoltaic pattern spots; performing space intelligent deduction and cloud pixel filling on remote sensing data through a visual large model to obtain reconstruction time sequence reflectivity data, and calculating a normalized vegetation index based on the reconstruction time sequence reflectivity data; Carrying out ecological influence assessment on the target area based on the photovoltaic panel geographic data and the normalized vegetation index to obtain ecological influence data; And carrying out ecological restoration scheme matching based on multi-rule constraint according to the ecological influence data to obtain an ecological restoration recommended scheme, wherein the ecological restoration recommended scheme is used for indicating restoration technical measures to be adopted by each mountain photovoltaic pattern spot. In one embodiment, performing deep learning-based target recognition based on geographic data to obtain photovoltaic panel geographic data includes: Simulating the topographic shadow distribution of the target area through the illumination model based on the digital elevation model and the solar azimuth angle and the altitude angle at the imaging moment of the remote sensing data to generate a topographic shadow mask map; Carrying out space registration on the remote sensing data and a topographic shadow mask map, and carrying out self-adaptive spectrum enhancement on pixels falling in a shadow area in the remote sensing data to obtain a topographic calibration image, wherein the self-adaptive spectrum enhancement is used for weakening the influence of topographic shadow on the spectral characteristics of ground objects; Inputting the terrain calibration image into an improved U-Net deep learning model, wherein the improved U-Net deep learning model introduces a focusing mechanism in an encoder part to focus the characteristics of the photovoltaic panel in the mountain environment; And carrying out pixel-level semantic segmentation on the terrain calibration image by improving a U-Net deep learning model, outputting a binarization photovoltaic panel distribution map, and vectorizing the binarization photovoltaic panel distribution map to obtain mountain photovoltaic image spots in a vector format. In one embodiment, performing spatial intelligent deduction and cloud pixel filling on remote sensing data t