How to achieve automatic roof 3d modelling from national open LiDAR data for solar PV shadow analysis?

204 Views Asked by At

this is my first post,

I'm trying to automatize the process of rooftop analysis for the installation of solar photovoltaic (PV) systems. I'm not sure what the input data should be but I've though of the following:

I want to create a 3D model of the rooftops with planar faces (not triangulated irregular networks, (TIN)) to be able to visualize shadows in 3D. The scale intended is municipal. With this resolution of LiDAR, some roof parts like parapets are not clear, satellite images could help with the details.

I've been using TIN files created from LiDAR data (I've tried CloudCompare and LAStools to do that), I've also tried TIN files extracted from Google Maps using RenderDoc. Then, I create a 3D model in Blender based on the TIN, to make a detailed roof. After that, I export the model to a Collada file (.DAE) and import it in Solarius PV software to study shading and automatically place solar panels on any planar surface (inclined or horizontal).

Solarius PV software can import several filetypes, including BIM. Solarius PV can't take 3D local shades into account to calculate the power generated, but I hope it will in the future.

I have two different ideas I would like to test:

  1. Create roof's 3D models based on Artificial Intelligence (AI) (edge detection?).
  2. Create a highly detailed digital elevation model (DEM) of the rooftops based on the input data and use GRASS GIS r.sun tool in QGIS.

I don't know anything about AI and I'm still learning QGIS, my field is electrical engineering, but I've come to a bottleneck in this part of the workflow.

I've tried to find a software that could already implement something like that but I haven't found anything. I know ArcGIS has some 3D tools but I don't have a license to check them out yet. I'm reviewing the literature and I've seen it's an important field of study at the moment.

I would be very grateful if anyone could put me in the right direction.

0

There are 0 best solutions below