Overhead Crane Operator Behavior Monitoring DMS System with Camera
Solution Overview: This technical solution is purpose-built for overhead crane safety monitoring. By integrating sensors, AI algorithms, and Edge Computing, it delivers real-time monitoring, fault prediction, and safety warning for critical crane components—a key step in transitioning industrial cranes from scheduled maintenance to Predictive Maintenance.
System Architecture for Overhead Crane Safety Monitoring
The system follows an edge–cloud three-tier architecture. At the sensing layer, multiple sensor types—Vibration Sensors, Temperature Sensors, Current Transformers (CTs), industrial cameras, and microphone arrays—are mounted on critical crane components, continuously acquiring equipment status data at sampling rates from 1 to 10 kHz. At the edge layer, computing devices deployed on the crane (NVIDIA Jetson Orin NX or an industrial PC) perform real-time preprocessing and AI model inference on raw sensor data—including vibration spectrum analysis, acoustic diagnosis, and visual defect detection—so that over 90% of Fault Diagnosis is completed on site without cloud dependency. The cloud layer receives extracted feature data and alarm messages from the edge, running long-term trend analysis, cross-crane comparisons, and equipment health scoring to support management decisions.
Key Monitoring Technologies and AI Diagnostics
The system fuses multiple AI-based diagnostic techniques for comprehensive coverage. For vibration spectrum analysis, IEPE Acceleration Sensors (measuring range ±50g, frequency response 0.5–10 kHz) are mounted on the gearbox and motor bearing housing. FFT and envelope spectrum analysis identify gear mesh characteristic frequencies and bearing fault frequencies for the inner ring, outer ring, rollers, and Cage, enabling gear wear and bearing fault prediction 2–6 weeks in advance. For acoustic diagnosis, a 4–8 channel MEMS microphone array near the crane wheels captures wheel-to-rail friction signatures. A CNN-based convolutional neural network classifies 8 types of wheel rail gnawing/flange rubbing conditions (normal, mild, moderate, severe, single-side, double-side, periodic, intermittent) with over 95% detection accuracy. For AI visual inspection, a YOLOv8 deep learning model performs real-time video analysis of the Wire Rope surface, detecting wire breaks and wear at 0.1 mm-level Resolution—replacing manual visual checks.
Three-Level Warning and Interlock Protection
The system issues alerts at three severity levels. Yellow (caution)—parameters exceed the normal range but remain within safe limits (e.g., vibration velocity 2.8–4.5 mm/s); the system pushes an alert to the maintenance crew's mobile app and the remote operation and maintenance platform, recommending an inspection in the near term. Orange (warning)—parameters approach safety limits (vibration 4.5–7.1 mm/s); the system automatically decelerates the crane to 50% of rated speed and sends an urgent notification to the maintenance supervisor. Red (danger)—parameters exceed safety limits (vibration above 7.1 mm/s or bearing temperature above 85°C); the system cuts power to the Hoisting and Long Travel main circuits and locks them out. The crane cannot be re-energized until a maintenance technician confirms the fault is cleared on site and manually resets the system. The red alarm signal is wired via Hard Wiring (Relay contacts) directly into the crane's Safety Relay circuit—bypassing PLC software logic entirely—so the alarm protection remains fully functional even if the PLC fails.
FAQ
Q: How long does installation take? Will it disrupt production?
A: Physical sensor installation can be completed in batches during normal crane downtime (shift changes, lunch breaks, weekends). Vibration and Temperature Sensors are fixed with magnetic bases or adhesive—no Drilling or Welding on the crane is required. Total installation time for all sensors on a single crane is approximately 4–6 hours, which can be split across 2–3 sessions. Electrical wiring and control system commissioning require 1–2 days of crane downtime, typically scheduled over a weekend or during a slow production period. Overall, the impact on normal operations is manageable. Kelude coordinates the installation window with the customer in advance to minimize any disruption to production.
Q: Can the system be retrofitted to older cranes? Will sensors fit on aging equipment?
A: Yes. The suitability of retrofitting an older crane—even one 10–20+ years old—depends on three factors: ① Electrical system—an accessible 220V or 24V Power Supply must be available for the monitoring equipment; ② Safety Relay circuit—an available emergency stop expansion contact for red-alarm automatic shutdown interlock. If the existing Safety Relay lacks expansion contacts, an auxiliary contact Module can be added; ③ Mounting space—the Vibration Sensor measures just φ20×30mm, and the magnetic base takes up virtually no additional space, so suitable mounting locations can be found on nearly all cranes. Retrofitting an older crane with a monitoring system effectively gives aging equipment "digital senses," allowing its remaining life to be assessed and managed with far greater precision.
Q: How do you handle false alarms? Frequent nuisance alarms are worse than no system at all.
A: False alarms typically stem from two causes: ① Overly sensitive alarm thresholds—cranes vary widely in Work Duty and operating conditions, so a one-size-fits-all threshold inevitably triggers frequent nuisance alarms on some units. The solution is a "learning mode" during the first 1–3 months of operation: the system records normal vibration and temperature fluctuation ranges and automatically generates a personalized baseline for each crane. Alarm thresholds are then set at 1.5–2.0× the baseline rather than at fixed values. ② Transient sensor interference—such as electromagnetic interference from nearby Welding equipment. The solution is a "duration confirmation" setting in the software: a single instantaneous peak above the threshold does not trigger an alarm immediately. The value must remain above the threshold for 3–5 consecutive seconds (or 3 consecutive sampling cycles) before an alarm is raised, filtering out the vast majority of transient disturbances.
Q: Where is the monitoring data stored? Is it secure?
A: Data is stored using a dual edge + cloud architecture. At the edge, the industrial PC on the crane has a built-in 256GB SSD that stores full-resolution raw data at one record every 10 seconds (approximately 2KB per record), providing roughly 3 years of storage capacity. In the cloud, compressed and desensitized feature data—spectrum peak frequencies and amplitudes rather than full raw time-domain waveforms—is uploaded via 4G/5G to privately deployed servers on Alibaba Cloud or Huawei Cloud, with TLS 1.3 encrypted transmission. Edge data belongs to the customer and can be exported at any time via USB drive. Cloud data is used solely for backup and Remote Diagnostics, and the customer may request deletion at any time. Kelude does not own or share customer equipment data.
Kelude Heavy Industry: Overhead Crane & Gantry Crane Manufacturer
Kelude Heavy Industry is a professional manufacturer of overhead cranes, gantry cranes, and electric hoists, offering a full range of material handling equipment for industrial applications. With years of engineering experience, we deliver reliable, high-performance crane solutions tailored to the specific needs of workshops, warehouses, and production lines.
Quality Assurance and International Standards
All Kelude cranes are designed and manufactured in compliance with ISO 4301, ISO 12480, and IEC 60204-32 standards, ensuring safety, reliability, and performance. Our quality management system covers every stage of production, from raw material inspection to final load testing, so you can be confident in the long-term dependability of your equipment.
