The invention belongs to the technical field of elevator monitoring and computer vision, and discloses a non-intrusive multi-type elevator operation centralized monitoring system and a detection method. According to the method, aiming at real-time collection of multiple types of elevator operation data sequences, a dynamic time warping method is adopted to calculate and process elevator operation and car vibration data sequences, and then the data sequences are input into an elevator operation model to detect the elevator operation state in real time; according to the abnormal elevator taking behavior detection system, a DeepSort neural network model is optimized and trained to track and detect targets of passengers and articles, detection and recognition of abnormal elevator taking behaviors of the passengers are achieved on the basis, and transmission of elevator mechanism operation information and personnel abnormal elevator taking alarm information is achieved. The non-intrusive multi-type elevator scene centralized monitoring system is acquisition equipment oriented to multi-type elevator application scene operation monitoring, and a foundation is laid for large-area operation monitoring and maintenance of multi-type elevators in different years.
本发明属于电梯监控技术和计算机视觉技术领域,公开了一种非侵入式多类型电梯运行集中监控系统及检测方法。该方法针对多类型电梯运行数据序列实时采集,采用动态时间规整方法对电梯运行及轿厢振动数据序列进行计算处理,再输入至电梯运行模型实时检测电梯运行状态;异常乘梯行为检测系统通过优化并训练DeepSort神经网络模型对乘客及物品的目标追踪与检测,在此基础上实现对乘客异常乘梯行为的检测与识别,实现对电梯机构运行信息和人员异常乘梯报警信息的传输。该非侵入式多类型电梯场景集中监控系统是面向多类型电梯应用场景运行监控的采集设备,为多类型不同年代电梯大面积运行监控维保奠定基础。
Non-intrusive multi-type elevator operation centralized monitoring system and detection method
一种非侵入式多类型电梯运行集中监控系统及检测方法
2024-11-01
Patent
Electronic Resource
Chinese
Centralized management and evaluation method for elevator operation data
European Patent Office | 2021
|