Time series power anomaly detection based on Light Gradient Boosting Machine

Time series anomaly data detection has always been a research hotspot in different fields and is also one of the prerequisite tasks for effective data analysis. At present, a large number of research models are based on machine learning methods, which fail to well extract the correlation between data, fail to obtain the development trend of data, and often lead to wrong abnormal judgment. However, due to the large number of parameters, the new deep learning method takes too long to compute and is difficult to deploy. Aiming at the above problems, Light Gradient Boosting Machine (LightGBM) is used in this paper to significantly improve the detection accuracy of the model. Through experimental comparison, this method can obtain high accuracy in a very short time.