PODA: Positive Active Power Outlier Detection Based on Auto Machine Learning

As China continues to strengthen the implementation of the "double carbon" development strategy, the power industry is in urgent need of upgrading its industrial structure. The identification and processing of inevitable outliers in power data is a prerequisite for all optimization and adjustment work. However, traditional identification methods are often ineffective, and machine learning methods are often associated with high learning and labor costs. To address this problem, this paper proposes a positive active power outlier detection method based on automated machine learning techniques, which can realize automated feature extraction and outlier identification of power data. Firstly, the power outlier detection and automated machine learning are introduced. Then the related working process is designed and experiments and analysis are conducted. The experimental results show that the method can significantly improve the efficiency of power data analysis and ultimately improve the quality of power business services.