Research on Evaluation Method of Elevator Safety Index Based on Bayesian Network

Abstract

In the elevator industry, inadequate maintenance is the main cause of elevator failures. A kind of new maintenance model for elevators, “on-demand maintenance”, is an inevitable trend in the development of the elevator maintenance industry. Based on the characteristics of many elevator brands, complex user environment, and different safety management conditions, this paper fully considers the complex situation of multi-parameter joint influence and multi-level coupling influence, and introduces Bayesian network to solve the uncertainty in the complex environment of electromechanical equipment. With many advantages of correlation analysis, a Bayesian network-based elevator safety index evaluation method is proposed. Then we establish four types of bottom-level impact factors that affect elevator safety conditions: elevator quality factor, maintenance quality factor, inspection factor, and use management factor as the child nodes for building a Bayesian network-based elevator safety operation evaluation model, and then form a Bayesian-based the evaluation model of the actual operation of the elevator on the network. Based on the actual elevator operation data and elevator management status in the model, the Bayesian network based on probabilistic reasoning is used to solve the uncertainty and incompleteness problems to construct the elevator safety index evaluation method, and also to make maintenance early warning and intelligent response scientifically and reasonably. At the same time, the maintenance plan is tailor-made for the elevators that the maintenance units participate in the pilot. Compared with the current maintenance model with a fixed maintenance cycle of 15 days, it greatly reduces labor costs and improves maintenance efficiency. It is estimated that the annual maintenance costs will be reduced by 20% once “on-demand maintenance” takes action.