We are looking for a student for a final thesis/internship project on the topic of
‘Validation of long-range lidar measurements with machine learning (clustering) techniques’
at ForWind Oldenburg at the Institute of Physics of the Carl von Ossietzky University of Oldenburg.
Start: Flexible
Duration: 3 months
Location: ForWind University Oldenburg (Building W33)
Supervision: Arjun Anantharaman
Wind Energy Systems Research Group
ForWind – University of Oldenburg
Contact: Arjun.anantharaman@uol.de
Description:
The data obtained from Doppler wind lidars is filtered in the processing stage primarily based on the Carrier-to-Noise Ratio (CNR). The method chosen to filter out “invalid” data points can exclude valid data and cause unnecessary data loss (1). Modern filtering techniques use a data density (2) or clustering approach (3) to retain more points in the regions with lower CNR. Newer filtering techniques like OPTICS (4) aim to further automate the methodology with increased computational efficiency and data retention. We have measurements from two long-range lidars, with one being used as a validation source for the other due to its increased power at longer ranges. The tasks involved are: sorting and characterising data from the measurement campaign, apply existing filtering methods to establish a good comparison metric of spatial wind fields and finally testing out and improving the DBSCAN and OPTICS algorithms.
Project goals:
- Learning to process lidar data – flat Plan Position Indicator (PPI) scans and become comfortable in analysis
- Understanding the existing and newer methods of filtering data – also to try and implement new methods and adapt them to the current data set in Matlab/Python
- Compare the effect of the different approaches on velocity statistics and validate the filtering methodologies
Prerequisites: Knowledge of wind energy and physics, experience in Matlab/Python, high motivation and the inclination to conduct independent research.
References:
1. Alcayaga, L. (2020, in review) ‘Filtering of pulsed lidars data using spatial information and a clustering algorithm’. DOI: 10.5194/amt-2019-450.
2. Beck, H. and Kühn, M. (2017) ‘Dynamic data filtering of long-range doppler LiDAR wind speed measurements’, Remote Sensing, 9(6). DOI: 10.3390/rs9060561.
3. Gryning, S.-E. and Floors, R. (2019) ‘Carrier-to-Noise-Threshold Filtering on Off-Shore Wind Lidar Measurements’, Sensors, 19(3), p. 592. DOI: 10.3390/s19030592.
4. Ankerst, Mihael, Markus M. Breunig, Hans-Peter Kriegel, and Jörg Sander. “OPTICS: ordering points to identify the clustering structure.” ACM SIGMOD Record 28, no. 2 (1999): 49-60.
