TimesNet Elevator Operation Accident Prediction Fusing DLinear and Deformable Convolution

Elevator is a convenient building transportation for people to travel, and more and more elevators are being registered and put into use, and the ACCOMPANYING problems of elevator failure and maintenance are becoming more and more prominent. In this study, the Kalman filter algorithm is used to optimize the feature extraction performance and prediction accuracy of the deformable convolutional TimesNet model for elevator operation time series data, and the improved TimesNet model is fused with the DLinear model to construct the TimesNet DLinear model for predicting elevator operation accidents. Finally, the TimesNet DLinear model is used as the main analysis modu le to design the elevator operation accident prediction system. After testing, it is found that the TimesNet DLinear model can maintain a low error in the prediction of elevator operation data, with an average absolute error of 0 167 , and the prediction ac curacy is better than other prediction models. It is also found that the elevator operation accident prediction system is able to predict the accidents in the elevator operation in a certain district and make a warning according to the current error thresh old situation. The elevator operation accident prediction system proposed in this study is able to realize real time monitoring and early warning of elevator failures, providing an effective solution for real time decision making and scheduling of elevator maintenance.