Improved Sparrow Search Algorithm with Support Vector Machine for Wind Turbine Fault Detection

Wind turbines are widely used across the globe as a primary means of renewable energy by using the power of wind, it converts kinetic energy into electrical energy, contributing to generating electricity. However, turbines are positioned far from control centers, operating in environments that continuously change and have uncontrollable factors and conditions lead to wind turbine failures. The proposed method using an Improved Sparrow Search Algorithm (ISSA) optimizer along with also used a Support Vector Machine (SVM) classifier helps to detect the faults in turbines. This method uses the Supervisory Control and Data Acquisition (SCADA) dataset for wind turbine fault detection, the optimization using ISSA which is improved by Uniformity Diversity Orientation Strategy (UDOS) to optimize model’s internal parameters. This helps to learn pattern during training period and helps to classify the normal and fault patterns. Then in the SVM classification process using the parameters this compares the obtained values with the current fitness value to classify patterns. The performance of the ISSA is analyzed by using accuracy, precision, f1-score, and recall. This shows that the proposed method outperforms existing methods such as K Nearest Neighbors (KNN), stacking classifier, and Feature Simplification Random Forest (FS_RF) in terms of accuracy 98.78%, precision 98.97%, recall 98.24%, and f1-score 98.00%.