The invention discloses an elevator safety risk early warning method based on multi-modal data, and relates to the technical field of elevator safety operation detection. Comprising structured elevator sensor return data, unstructured elevator text data, elevator mechanical part operation audio data and elevator car passenger video data; performing feature extraction operation on the elevator text data, the elevator mechanical part operation audio data and the elevator car passenger video data; carrying out data noise reduction on the elevator sensor return data based on wavelet transform; and data obtained after noise reduction and the extracted text and audio and video feature data are input into a model-level feature fusion-multilayer long-short-term memory network, and an early warning result of the elevator safety risk in the next 24 hours is output and obtained. According to the method, multi-dimensional and intelligent comprehensive analysis, study and judgment of the elevator running state are achieved, and the accuracy prediction of the running state risk of the elevator in the future time is improved.
本发明公开了一种基于多模态数据的电梯安全风险预警方法,涉及电梯安全运行检测技术领域,包括以下步骤:获取电梯多模态数据,包括结构化的电梯传感器回传数据、非结构化的电梯文本数据、电梯机械部件运行音频数据和电梯轿厢乘客视频数据;对电梯文本数据、电梯机械部件运行音频数据和电梯轿厢乘客视频数据进行特征提取操作;基于小波变换对电梯传感器回传数据进行数据降噪;将降噪后得到的数据与提取的文本及音视频特征数据输入模型级特征融合‑多层长短期记忆网络,输出得到电梯安全风险未来24小时预警结果。本发明实现了对电梯运行状态的多维度、智能化综合分析与研判,提高了对电梯未来时间运行状态风险的准确性预测。
Elevator safety risk early warning method based on multi-modal data
一种基于多模态数据的电梯安全风险预警方法
2024-11-01
Patent
Electronic Resource
Chinese
IPC: | B66B Aufzüge , ELEVATORS |
Multi-modal feature fusion elevator early warning method in big data environment
European Patent Office | 2025
|Elevator risk early warning management method and system
European Patent Office | 2025
|