Shanghai University’s Novel Method Revolutionizes Wind Power Prediction for New Farms

In the rapidly evolving energy sector, the integration of clean energy sources like wind power is crucial for meeting growing energy demands sustainably. However, the intermittent nature of wind power poses significant challenges for grid management and energy distribution. A recent study published in *Power Construction*, led by YU Guangzheng from the School of Electrical Engineering at Shanghai University of Electric Power, introduces a novel method for short-term wind power prediction that could revolutionize how new wind farms are managed.

The research addresses a critical gap in the industry: the lack of accurate prediction methods for new wind power stations with limited data availability. Traditional prediction models often struggle with the dynamic and time-varying characteristics of wind power data, leading to inaccuracies that can disrupt grid stability and efficiency. “The construction of new wind power stations faces significant challenges in accurate wind power prediction owing to the limited data availability and dynamic nature of scene data,” explains YU Guangzheng, the lead author of the study.

To tackle this issue, the researchers developed a multi-faceted approach. First, they employed a similarity measurement method combining the Canberra distance and dynamic time warping algorithm to identify multiple source-domain wind power stations similar to the new target station. This step ensures that the prediction model is not overly reliant on a single source of data, enhancing its robustness and accuracy.

Next, the team established a pre-training model based on a multi-source transfer learning dilated convolutional neural network and bidirectional long short-term memory (DCNN-BiLSTM). This model leverages the collective experience of multiple similar source domain stations to predict the power output of the new target wind power station. “This approach transfers the experience knowledge of multiple similar source domain stations to the new target wind power station, thereby avoiding over-reliance on single-source domain data,” says YU.

The study also introduces an online adaptive module to account for the impact of time-varying scene data on prediction results. Two self-evolution methods were developed: data-matching update adaptation and weight-update adaptation. The output results of the basic prediction and online models in the online adaptive module are weighted to achieve short-term wind power prediction for the new wind power stations.

The proposed method was verified using data from a wind power station cluster in Northwest China, demonstrating its superior ability to screen source-domain wind power station data and achieve more accurate power predictions. This advancement is particularly significant for the energy sector, as it provides a feasible solution to the challenge of scarce power data in limited information conditions.

The implications of this research are far-reaching. Accurate wind power prediction is essential for grid stability, energy distribution, and the overall efficiency of the energy sector. By providing a reliable method for predicting wind power output, this study could broaden the application scope of such methods, making it easier to integrate new wind farms into the grid. “The method proposed in this article provides a feasible solution to the challenge of scarce power data in under limited information conditions and is expected to broaden the application scope of such methods,” YU notes.

As the energy sector continues to evolve, the need for innovative solutions to manage renewable energy sources will only grow. This research offers a promising step forward, demonstrating how advanced data analysis and machine learning can be harnessed to improve the predictability and reliability of wind power. With further development and implementation, this method could play a crucial role in shaping the future of clean energy integration and grid management.

Scroll to Top
×