Research Background and Significance
With the rapid development of e-commerce and modern warehousing logistics, automated storage and retrieval systems and large-scale racking systems are being widely used. As the main load-bearing structure in a warehouse, racking safety directly affects operational safety, equipment stability, and warehouse efficiency.
Rack deformation, tilting, or collapse may cause cargo damage and even serious safety accidents. In many cases, these risks are related to excessive rack verticality or levelness deviation that has not been detected and handled in time.
Traditional rack inspection mainly relies on manual periodic checks. However, manual inspection has several limitations: long inspection cycles, high error rates, missed detection risks, high labor costs, and delayed reporting of abnormal conditions.
Therefore, it is important to develop an intelligent system that can automatically, continuously, and accurately monitor rack verticality and levelness, trigger alarms when abnormalities occur, and notify relevant personnel in real time through email and SMS.
Research Objectives
This study aims to develop a rack monitoring and alarm notification system with the following objectives:
Achieve high-precision rack tilt detection, with verticality and levelness detection accuracy better than ±0.1°.
Build a reliable data transmission link supporting both wired and wireless communication.
Establish a multi-level alarm mechanism and send alarm notifications within 3 seconds by email and SMS.
Provide a user-friendly configuration interface to make system deployment easier.
Overall System Design
The system adopts a three-layer architecture: perception layer, network layer, and application layer.
The perception layer consists of multiple sensor nodes installed on key rack columns and beams. Each node includes a MEMS IMU, a low-power MCU, and a wireless communication module. These nodes collect rack tilt data and perform local preprocessing.
The network layer includes an on-site gateway and cloud server connection. The gateway collects data from sensor nodes, performs protocol conversion, and uploads data to the cloud through 4G or Ethernet. It also supports local alarm logic, so alarms can still be triggered when the network is interrupted.
The application layer is deployed on the cloud server. It is responsible for data storage, threshold management, alarm decision-making, and notification sending. Users can monitor rack status, query historical data, and configure parameters through a web interface or mobile app.
Core System Modules
| Module | Function | Technical Implementation |
|---|---|---|
| Sensor node | Tilt angle acquisition and local filtering | MPU6050 + STM32L4 |
| On-site gateway | Data aggregation, protocol conversion, local alarm | Raspberry Pi 4B + LoRaWAN |
| Cloud server | Data storage, alarm decision-making, user management | Linux + MySQL + Python |
| Email notification | Alarm email sending and management | SMTP + TLS encryption |
| SMS notification | Alarm SMS sending | Third-party SMS API |
| User interface | Real-time monitoring and historical query | Web + Vue3 frontend |
Data Acquisition and Transmission
Sensor nodes collect raw IMU data at 10 Hz. After Kalman filtering, the processed tilt angle data is uploaded to the gateway at 1 Hz.
The data frame includes node ID, timestamp, verticality value, levelness value, battery level, and CRC16 checksum. LoRa communication uses a star topology, where each sensor node communicates directly with the gateway.
After data aggregation, the gateway sends the data to the cloud server through MQTT. The cloud server stores the data and triggers real-time alarm judgment.
Alarm Notification Mechanism
The system supports three alarm levels:
| Alarm Level | Color | Notification Method |
|---|---|---|
| Warning | Yellow | App push + email |
| Alarm | Orange | Email + SMS |
| Severe alarm | Red | Email + SMS + voice call |
When an alarm is triggered, the system automatically sends key information, including rack number, warehouse location, tilt angle, alarm level, and timestamp.
To improve reliability, the system also supports retry mechanisms, repeated alarm suppression, and alarm escalation. For example, if an orange alarm is not confirmed within 15 minutes, it can be upgraded to a red alarm.
Software Design
The sensor node firmware is developed based on FreeRTOS and includes several main tasks:
Data acquisition
Data processing
Communication
Local alarm
Power management
The cloud server adopts a microservice architecture, including:
Data ingestion service
Alarm engine service
Notification service
API service
Time-series data is stored in InfluxDB, while user information, equipment information, configuration parameters, and notification records are stored in MySQL.
The web management interface provides real-time monitoring, alarm management, device management, historical query, and notification configuration. The mobile app supports key monitoring and alarm functions on iOS and Android.
Research Conclusion
This study designs and implements an automatic detection and alarm notification system for rack verticality and levelness.
The system uses MEMS IMU sensor fusion and Kalman filtering to achieve high-precision rack tilt detection. It adopts a three-layer architecture and LoRa wireless communication to ensure reliable data acquisition and transmission. A single gateway can support up to 64 monitoring nodes.
The system also establishes a multi-level alarm mechanism, supports automatic email and SMS notifications, and provides real-time monitoring through web and mobile interfaces.
In practical warehouse operation tests, the system showed stable and reliable performance. It provides an effective technical solution for improving rack safety, reducing manual inspection risks, and supporting the long-term stable operation of automated warehouse systems.
