The present invention relates to the field of monitoring escalators, specifically a cloud-based platform for online monitoring and warning system and method for escalators, the system comprising a line network cloud platform, a data integrated service system located at the central level of the line network, a communication system, and a station-level monitoring and warning system; the station-level monitoring and warning system is set up in each station; each escalator is set up with an escalator code, and the station-level monitoring and warning system transmits the risk warning information and escalator code to the line network cloud platform through a communication system; said data integrated service system retrieves said parts risk warning information from said line network cloud platform and classifies the escalator based on a cluster analysis method. The station-level monitoring and warning system transmits the parts risk warning information and the escalator code to the line network cloud platform through the communication system; said data integrated service system retrieves said parts risk warning information from said line network cloud platform and categorizes the escalators based on a cluster analysis method. The present invention overcomes the technical problems of the prior art of being unable to provide timely warning before the occurrence of faults and unable to predict the development trend of escalator faults in advance by detecting the operation status of the key components of all escalators in the station and analyzing the characteristics and patterns of the abnormal information of the escalators of the same type by means of the clustering analysis method.