Abstract
Missing or corrupt data is common in real-world datasets; this affects the estimation and operation of analytical models where completeness is assumed or required. Statistical wind power forecasts utilise recent turbine data as model inputs, and must therefore be robust to missing data. We find that wind power data is ‘missing not at random’, with missing patterns also related to the forecast output. Approaches for dealing with this missing data in training and operation are proposed and evaluated through a case study, leading to a suggested forecasting methodology in the presence of missing data. In the training set, missing data was found to have significant negative impact on performance if simply omitted but this can be almost completely mitigated using multiple imputation. Greater increase in forecast errors is seen when input data are missing operationally, and retraining forecast models using the remaining inputs is found to be preferable to imputation.
Original language | English |
---|---|
Article number | 106640 |
Journal | Electric Power Systems Research |
Volume | 189 |
Early online date | 11 Aug 2020 |
DOIs | |
Publication status | Published - 31 Dec 2020 |
Event | XXI Power Systems Computation Conference - Faculty of Engineering, University of Porto, Porto, Portugal Duration: 29 Jun 2020 → 3 Jul 2020 Conference number: 21 https://pscc2020.pt/ |
Keywords
- forecasting
- missing data
- time series
- vector autoregression
- wind power
Fingerprint
Dive into the research topics of 'Missing data in wind farm time series: properties and effect on forecasts'. Together they form a unique fingerprint.Datasets
-
Data for: 'Missing data in wind farm time series: Properties and effect on forecasts'
Tawn, R. (Creator), Browell, J. (Supervisor) & Dinwoodie, I. (Supervisor), University of Strathclyde, 16 May 2022
DOI: 10.15129/eca892b9-7715-4295-9a18-c7f35c57505d
Dataset