Elevator IoT Monitoring System Architecture
The elevator IoT monitoring system is the infrastructure for achieving elevator intelligence. System design key points: (1) Sensor selection: vibration sensors (triaxial accelerometer, range ±16g, sampling frequency ≥1kHz), temperature sensors (PT100 or digital temperature sensors, accuracy ±0.5℃), current sensors (Hall effect type, accuracy ±1%), encoders (incremental type, resolution 1024P/R); (2) Edge computing gateway: one edge gateway per elevator (supporting 4G/5G/Ethernet communication, with local data processing and storage functions, capable of continuing operation for 2 hours after power failure); (3) Data transmission: wired Ethernet preferred (high reliability); when network conditions are unavailable, 4G/5G wireless transmission is used (encrypted data transmission, compliant with GB/T 22239-2019 information security classified protection requirements); (4) Cloud platform: private cloud or public cloud deployment (Alibaba Cloud/Tencent Cloud/Huawei Cloud recommended), server storage of no less than 3 years of historical data, database using time-series database (such as InfluxDB/TimescaleDB) to store high-frequency sensor data. The system can support up to 128 elevators per project, and the group-level platform can manage multiple projects (unlimited number of elevators).
AI Fault Prediction and Health Management
The elevator AI fault prediction system analyzes sensor data through machine learning algorithms to identify equipment anomalies in advance and predict faults. Technical approach: (1) Data preprocessing: filtering and denoising of raw sensor data (realtime denoising), feature extraction (time-domain features: RMS value, peak value, kurtosis; frequency-domain features: spectral envelope, FFT transform); (2) Health baseline establishment: accumulate normal elevator operation data (more than 3 months recommended) to establish health state baseline models for each component (bearings, guide rails, door operators, etc.); (3) Anomaly detection algorithm: adopt Isolation Forest or Autoencoder algorithms to detect anomalous data points; when the anomaly score exceeds the threshold (typically set at 0.75-0.85), an early warning is triggered; (4) Remaining useful life prediction: adopt LSTM (long short-term memory) or Transformer models to predict the remaining useful life (RUL) of key components (bearings, wire ropes, brakes), giving 30-60 days' advance warning for components requiring replacement. The system can automatically generate maintenance work orders based on fault prediction results, achieving predictive maintenance. AI models require continuous iterative optimization (retraining with the latest data quarterly is recommended).
AI Visual Recognition and Intelligent Security
The elevator car AI visual recognition system is a new technology for enhancing hotel safety management and service quality. Functional modules: (1) People counting: real-time counting of the number of people in the car based on deep learning (CNN or YOLO algorithm); when the rated passenger capacity is exceeded, an overload alarm is triggered and the monitoring center is notified. Counting accuracy ≥98%. (2) Passenger behavior recognition: recognition of abnormal behaviors such as falls (fall detection, response time <0.5 seconds), fighting (action recognition, accuracy >95%), and prolonged lingering (not closing the door for more than 10 minutes in the car), with alarms automatically pushed to the security center. (3) Dangerous object detection (optional): recognition of carrying large luggage, suspicious packages, etc. (enabled only in non-guest areas of the hotel). (4) Seamless access (linked with the hotel PMS system): automatic confirmation of VIP guest identity through facial recognition, with the elevator automatically lighting up the destination floor (no button press required), enhancing the high-end guest experience. The visual system must pay attention to privacy protection (image data processed locally, not uploaded to the cloud, retention time not exceeding 72 hours, no facial feature comparison to identify personal identity).
Smart Operation and Maintenance Management Platform Functions
The elevator intelligent operation and maintenance management platform is a comprehensive management tool integrating monitoring, early warning, maintenance, and analysis. Core functional modules of the platform: (1) Operation monitoring dashboard: displays real-time elevator status (running/standby/fault/maintenance) on the hotel BMS or in the engineering department, showing elevator location, running direction, speed, load, and other parameters; (2) Fault alarm management: classified by severity (Level 1: entrapment/emergency stop, immediate response; Level 2: operational abnormality, handled within 4 hours; Level 3: early warning information, handled per plan), automatically pushed to designated personnel and maintenance contractors; (3) Maintenance work order management: the system automatically generates maintenance work orders based on maintenance plans and equipment health status, and tracks work order execution progress (started-in progress-completed-accepted); (4) Data analysis reports: automatically generates daily/weekly/monthly/annual operation reports (elevator usage frequency, energy consumption statistics, fault statistics, maintenance completion rate, start/stop counts, etc.); (5) Mobile APP: operation status query, alarm push, maintenance work order processing, historical data query. The platform shall support unified management of multiple projects and multiple hotels, and be compatible with data access from different elevator brands (via standardized interfaces OPC UA or Modbus TCP).