As the most commonly used public transportation tool in urban buildings,elevators have brought great convenience to people's lives and production.But occasional elevator accidents also bring significant risks to personal safety.This paper studies the data design and processing methods of elevator multi-source data in response to the practical problems of multiple sources and parameters in the WEBSTAR elevator component fault prediction.By analyzing the typical faults and diagnostic methods of key components in elevators,a dataset for elevator faults is constructed,and a data design for elevator component fault prediction is carried out to preprocess elevator data,providing data support for improving the accuracy and flexibility of elevator fault prediction.