Elevator Fault Prediction and Adaptive Maintenance Using Nonlinear IoT with CNN-LSTM and BP Neural Networks

Abstract

With the acceleration of urbanization, elevators, as an indispensable vertical transportation tool for high- rise buildings, have attracted increasing attention for their safety and reliability. However, the traditional elevator maintenance method often relies on regular maintenance and post-maintenance, and it is difficult to effectively prevent the occurrence of failures. Therefore, this study proposes an intelligent elevator fault prediction and adaptive maintenance algorithm based on nonlinear IoT (Internet of Things), aiming to improve the level of elevator operation and maintenance through advanced technology. In this study, a nonlinear IoT platform was constructed, and sensors were used to collect elevator operation data in real time, covering multi-dimensional information such as speed, acceleration, and door opening and closing status, and the dataset was derived from 500 commercial elevators in multiple cities in China, and a standardized dataset containing 1 million records was formed after preprocessing. At the level of algorithm construction, CNN-LSTM and BP neural network fusion models are adopted, the former uses convolutional neural network (CNN) to extract local features of data, combined with long short-term memory network (LSTM) to process time series information, and the latter is used to optimize the parameters of the error backpropagation optimization model, revealing the potential characteristics of faults through in-depth mining of elevator operation data. The designed intelligent fault prediction algorithm can accurately identify abnormal patterns in elevator operation and predict potential faults. In order to verify the effectiveness of the algorithm, the traditional fault prediction method was used as the baseline for comparative experiments, and the prediction accuracy of the proposed algorithm was 95% after the double-tailed t-test, which was highly statistically significant compared with the 80% accuracy of the baseline method (P <0.01 ） . At the same time, the adaptive maintenance algorithm can automatically adjust the maintenance strategy based on the prediction results, and the actual application verification shows that after the use of adaptive maintenance, the elevator maintenance cost is reduced by 30% and the downtime is reduced by 50%. The intelligent elevator fault prediction and adaptive maintenance algorithm based on nonlinear Internet of Things proposed in this study not only significantly improves the accuracy of elevator fault prediction, but also realizes the intelligence and automation of maintenance strategies, which provides a strong guarantee for the safe and efficient operation of elevators.