Author(s)

Yuan Ma^1*, JingYu Zhao²

Affiliation(s)

¹State Grid Energy Research Institute Co., Ltd., Beijing 102209, China.

²State Grid Talents Exchange And Service Center Company Limited, Beijing 100089, China.

Corresponding Author

Yuan Ma

ABSTRACT

As the core unit of the new energy grid, the operational reliability of the wind power system is directly related to the stability and power generation efficiency of the power grid, so it is urgent to improve its ability to accurately predict faults. Aiming at the complex timing fault characteristics of wind power generation systems, a multi-feature fusion operation fault prediction method based on LSTM is proposed. The wind farm SCADA system collects the operation data for preprocessing, screens the highly correlated features by the Pearson correlation coefficient method, and constructs a multi-feature input LSTM fault prediction model to improve the accuracy of wind power generation system operation fault prediction. Experimental results show that compared with the single feature model, the multi-feature fusion strategy can significantly improve the comprehensive performance of the prediction model, and the fault warning accuracy and F1 score are increased by 12.78% and 12.04% respectively.

KEYWORDS

Power system failure; Wind power systems; Fault prediction; LSTM

CITE THIS PAPER

Yuan Ma, JingYu Zhao. Fault diagnosis of wind power system based on LSTM prediction method. World Journal of Engineering Research. 2025, 3(3): 65-72. DOI: https://doi.org/10.61784/wjer3040.

REFERENCES

[1] Kozitsin V, Katser I, Lakontsev D. Online forecasting and anomaly detection based on the ARIMA model. Applied Sciences-BASEL, 2021, 11(7): 3194.

[2] Yang Bofan, Zhang Lin, Zhang Bo, et al. Online prediction method for dynamic multi-model exponential smoothing fusion. Systems Engineering and Electronic Technology, 2020, 42(09): 2013-2021.

[3] Zhang Z, Dong S, Li D, et al. Prediction and Diagnosis of Electric Vehicle Battery Fault Based on Abnormal Voltage: Using Decision Tree Algorithm Theories and Isolated Forest. Processes, 2024, 12(1): 136.

[4] Arumbu P V, Karthikeyan D. Reliability Assessment and Fault Prediction in a 13-Level Multilevel Inverter Through Machine Learning with SVM. Journal of Electrical Engineering & Technology, 2024, 20(1): 1-16.

[5] Sachin Kumar R, Sakthiya Ram S, Arun Jayakar S, et al. Failure prediction of turbines using machine learning algorithms. Materials Today: Proceedings, 2022, 66(3): 1175-1182.

[6] Gawali B M, Gawali S S, Patil M. Fault prediction model in wind turbines using deep learning structure with enhanced optimization algorithm. Journal of Control and Decision, 2025, 12(3): 471-488.

[7] Bharatheedasan K, Maity T, Kumaraswamidhas L, et al. Enhanced fault diagnosis and remaining useful life prediction of rolling bearings using a hybrid multilayer perceptron and LSTM network model. Alexandria Engineering Journal, 2025, 115, 355-369.

[8] Jin X, Chunyan L, Bo L, et al. Prediction of wind turbine blade icing fault based on selective deep ensemble model. Knowledge-Based Systems, 2022, 242.

[9] Daohua Z, Xinxin J, Piao S. Research on power system fault prediction based on GA-CNN-BiGRU. Frontiers in Energy Research, 2023, 11: 1245495. DOI: 10.3389/fenrg.2023.1245495.

[10] Zare A, Babalou M, Torkaman H. Open-circuit fault diagnosis of modular DC-DC converter based on multi-layer perception//2025 16th Power Electronics, Drive Systems, and Technologies Conference, PEDSTC. Tabriz, Iran, Islamic Republic of. IEEE, 2025, 1-5. DOI: 10.1109/PEDSTC65486.2025.10912043.

[11] Zhang W G, Li H R, Tang L B, et al. Displacement prediction of Jiuxianping landslide using gated recurrent unit (GRU) networks. Acta Geotechnical, 2022, 17(4): 1367-1382.

[12] Hochreiter S, Schmidhuber J. Long Short-Term Memory. Neural Computation, 1997, 9(8): 1735-1780.