SMU Data Science Review


In this paper, modeling techniques for the forecasting of wind speed using historical values observed by Light Detection and Ranging (LIDAR) sensors in an offshore context are described. Both univariate time series and multivariate time series modeling techniques leveraging meteorological data collected simultaneously with the LIDAR data are evaluated for potential contributions to predictive ability. Accurate and timely ability to predict wind values is essential to the effective integration of wind power into existing power grid systems. It allows for both the management of rapid ramp-up / down of base production capacity due to highly variable wind power inputs and integration of wind power into regional and national energy trading markets. Modeling successfully indicates that Autoregressive Integrated Moving Average (ARIMA) models, given data histories of one day at one minute intervals, provide the most useful forecasts, even when compared to more advanced modeling techniques such as Long Short Term Memory (LSTM) neural networks. These findings demonstrate the continued utility of long-standing autoregressive techniques and their more rapid time to train as an advantage over more complex machine learning techniques. To drive the operational utility of the analysis for users familiar with the data set provided by Pacific Northwest National Labs, a prototype web-based wind data exploration dashboard is also provided to allow users to conduct "on the fly" Exploratory Data Analysis (EDA) and identify best fit models.

Creative Commons License

Creative Commons Attribution-Noncommercial 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License