Abstract: Aiming at the problem of noise point and missing value in the operating status data of the coal mining machine drive system, a real-time data cleaning platform based on Storm is established. The platform uses ARIMA to establish a data cleaning model, uses the Spout component in Storm to read the point data in real time, encapsulates the data according to the set sample capacity and passes it to the Bolt component, and the Bolt component performs noise point determination, smoothing processing, model selection and other specific data cleaning tasks. The experiment proves that the platform can complete the real-time cleaning of the operating status data of the coal mining machine drive system.