Detecting Abnormal of Coal mine informatization operation and maintenance system Based on Dynamic baseline anomaly

The coal enterprise information center is responsible for the operation and maintenance of the company’s basic network facilities, basic computing storage facilities, and important safety production and operation business systems. Currently, the operation and maintenance system have problems such as single monitoring methods, isolated operation and maintenance data, complex operation and maintenance processes, and excessive repetitive labor of operation and maintenance personnel. This paper proposes an improved dynamic baseline abnormal traffic detection method based on the LightGBM algorithm and autoregressive neural network interpolation method. the results show that this method can automatically calculate the collection frequency of each resource and indicator, with an accuracy of 100% through performance testing on a server with a 6-core CPU and 16GB of memory. It is still effective for indicators with many missing values and long-term missing values. It takes 58 seconds to calculate the collection frequency of 214 indicators. The research results can accurately detect abnormal traffic in coal mine information operation and maintenance systems, and improve the intelligence level of operation and maintenance systems.