Bias correction of wind power forecasts with SCADA data and continuous learning

Stefan Jonas,Kevin Winter,Bernhard Brodbeck,Angela Meyer

2024-02-22

Abstract:Wind energy plays a critical role in the transition towards renewable energy sources. However, the uncertainty and variability of wind can impede its full potential and the necessary growth of wind power capacity. To mitigate these challenges, wind power forecasting methods are employed for applications in power management, energy trading, or maintenance scheduling. In this work, we present, evaluate, and compare four machine learning-based wind power forecasting models. Our models correct and improve 48-hour forecasts extracted from a numerical weather prediction (NWP) model. The models are evaluated on datasets from a wind park comprising 65 wind turbines. The best improvement in forecasting error and mean bias was achieved by a convolutional neural network, reducing the average NRMSE down to 22%, coupled with a significant reduction in mean bias, compared to a NRMSE of 35% from the strongly biased baseline model using uncorrected NWP forecasts. Our findings further indicate that changes to neural network architectures play a minor role in affecting the forecasting performance, and that future research should rather investigate changes in the model pipeline. Moreover, we introduce a continuous learning strategy, which is shown to achieve the highest forecasting performance improvements when new data is made available.

Machine Learning

What problem does this paper attempt to address?

### What problem does this paper attempt to solve? This paper aims to address the issues of uncertainty and variability in wind power forecasting to enhance the potential of wind energy and the growth of wind power capacity. Specifically, the paper achieves this goal through the following points: 1. **Bias Correction**: Proposes and evaluates four machine learning-based wind power forecasting models that can correct and improve the 48-hour forecasts extracted from numerical weather prediction (NWP) models. 2. **Performance Improvement**: Validates the effectiveness of these models in reducing forecast errors and mean bias through comparative experiments, and finds that convolutional neural networks (CNN) perform best in reducing the normalized root mean square error (NRMSE), lowering the NRMSE from 35% in the uncorrected model to 22%. 3. **Continuous Learning Strategy**: Introduces a continuous learning strategy that can achieve the highest forecast performance improvement when new data becomes available. In summary, the core issue of this paper is to improve the accuracy of wind power forecasting through machine learning methods, thereby better supporting applications such as power management, energy trading, and maintenance scheduling.

Bias correction of wind power forecasts with SCADA data and continuous learning

Machine Learning Approaches for Wind Speed Forecasting Using Long-Term Monitoring Data: a Comparative Study

Wind power prediction based on high-frequency SCADA data along with isolation forest and deep learning neural networks

A Hybrid Model for Long-Term Wind Power Forecasting Utilizing NWP Subsequence Correction and Multi-Scale Deep Learning Regression Methods

Short-term regional wind power forecasting for small datasets with input data correction, hybrid neural network, and error analysis

Power prediction using high-resolution SCADA data with a farm-wide deep neural network approach

Forecasting model for short-term wind speed using robust local mean decomposition, deep neural networks, intelligent algorithm, and error correction

Adaptive machine learning for forecasting in wind energy: A dynamic, multi-algorithmic approach for short and long-term predictions

Short-Term Wind Speed Hybrid Forecasting Model Based on Bias Correcting Study and Its Application

Wind power forecasting of an offshore wind turbine based on high-frequency SCADA data and deep learning neural network

A Long-Term Wind Speed Ensemble Forecasting System With Weather Adapted Correction

Wind turbine short-term power forecasting method based on hybrid probabilistic neural network

Machine learning forecasts of Scandinavian numerical weather prediction wind model residuals with control theory for wind energy

Wind power predictions from nowcasts to 4-hour forecasts: A learning approach with variable selection

A novel approach to generate bias-corrected regional wind infeed timeseries based on reanalysis data

SCADA system dataset exploration and machine learning based forecast for wind turbines

Machine Learning–Adjusted WRF Forecasts to Support Wind Energy Needs in Black Start Operations

A novel wind power prediction model improved with feature enhancement and autoregressive error compensation

Wind power error compensation prediction model based on CEEMD-SE-ELM-TCN

Short-term wind power combination forecasting method based on wind speed correction of numerical weather prediction

Wind power forecasting based on a machine learning model: considering a coastal wind farm in Zhejiang as an example