Overhead Crane Remote Monitoring for 120 Workshops
Remote operation and maintenance platform for overhead cranes goes live, delivering real-time monitoring across 120 workshops in 18 provinces. As the most critical material handling equipment in heavy industrial workshops, overhead cranes directly impact production uptime and personnel safety.
Overhead cranes are the backbone of material handling in heavy industrial workshops, and their operating condition directly affects production rhythm and worker safety. Traditional crane maintenance relies on reactive repairs after equipment failure and periodic shutdown inspections, which means fault detection lags, spare parts response is slow, inspection records are kept on paper, and equipment health data remains largely opaque. Kelude Heavy Industry's independently developed remote operation and maintenance platform for overhead cranes has recently been officially launched. As of this writing, it has been connected to 120 industrial workshops across 18 provinces and municipalities, providing real-time monitoring of more than 300 overhead and gantry cranes. The platform collects over 2 million operational data points per day, has issued more than 500 early warnings for abnormal equipment conditions, and has reduced average fault response time from 48 hours to 2.5 hours.
This article examines the platform from four perspectives—system architecture design, data acquisition chain, remote diagnostic capability, and real-world operational results—to provide a systematic overview of the technical implementation and engineering value of the remote operation and maintenance platform.
Platform Architecture and Data Acquisition Chain
The remote operation and maintenance platform uses a hybrid architecture combining edge computing with cloud-based analytics. Each crane is equipped with a data acquisition terminal (DTU) that communicates with the crane PLC (Siemens S7-1200/1500 series) via PROFINET or Modbus TCP. The DTU reads key operating parameters including hoisting motor current, bridge/trolley travel frequency, brake status, overload limiter readings, and travel limit switch states, with a sampling frequency of 100 ms. In parallel, the DTU interfaces with vibration sensors (triaxial accelerometers, 2 kHz sampling rate), temperature sensors (PT100, ±0.5°C accuracy), and wire rope tension sensors (strain-gauge type, 0–500 kN measuring range), enabling synchronized acquisition of both electrical and mechanical parameters.
The DTU transmits encrypted data over 4G/5G cellular networks to industrial IoT gateways deployed as edge nodes at the crane's facility. After data cleaning, deduplication, and time alignment, the gateway forwards the data to the cloud platform via MQTT. The cloud platform runs on Alibaba Cloud ECS servers, using an InfluxDB time-series database to store operational data (retention period: 180 days) and a PostgreSQL relational database for equipment registers, maintenance records, and alarm events. The platform supports concurrent online monitoring of up to 5,000 devices, with an average end-to-end latency of 800 ms from PLC register read to cloud dashboard update. All communication links are encrypted with TLS 1.3, meeting the Level 2 protection requirements of GB/T 22239-2019, the Chinese standard for cybersecurity classified protection of information systems.
Real-Time Monitoring and Four-Tier Alarm System
The platform implements a four-tier alarm system covering the full equipment lifecycle, from early performance degradation to emergency shutdown:
Tier 1 — Parameter threshold alarms: For parameters with directly definable thresholds such as motor current, travel speed, and brake temperature, maintenance personnel can configure upper and lower alarm limits and duration conditions in the backend. For example, if hoisting motor current exceeds 85% of rated capacity for more than 10 seconds, the system automatically triggers an overload alarm. The false alarm rate for threshold alarms is kept below 3% through a configurable deadband mechanism.
Tier 2 — Trend analysis alarms: The platform performs linear regression analysis on long-term trends of key parameters, triggering early warnings when an indicator shows sustained deterioration within a configured time window. For instance, if brake actuation time gradually rises from 70% to 90% of the threshold range, the system flags it as a performance degradation warning even though the absolute threshold has not been exceeded. Trend analysis windows are adjustable across 7-day, 30-day, and 90-day periods, with alarm sensitivity configured via the coefficient of variation (CV value).
Tier 3 — Association rule alarms: Drawing on an expert knowledge base of equipment failure modes, the platform comes preloaded with over 80 association rules for identifying compound fault signatures. For example, when abnormal hoisting motor current fluctuation, rapid brake temperature rise, and wire rope tension oscillation all deviate from normal ranges simultaneously, the system diagnoses excessive brake shoe wear and recommends a scheduled shutdown for replacement.
Tier 4 — AI-based anomaly detection alarms: The platform incorporates an unsupervised anomaly detection model based on Isolation Forest, which scores anomalies across high-dimensional equipment operating parameters. The model builds its baseline using 60 days of historical operational data (learning rate 0.001, 100 trees), and each sample point is classified as either -1 (anomalous) or 1 (normal). The AI model is incrementally updated every 24 hours with newly acquired data, achieving an F1 score of 0.87 for anomaly detection on the test set. Since deployment, the AI anomaly detection module has identified 15 early-stage faults that traditional threshold alarms failed to catch, including gearbox gear pitting (11 days advance warning), motor bearing cage cracks (6 days advance warning), and VFD IGBT module characteristic drift (3 days advance warning).
Alarm events generated by the four-tier system are classified into three severity levels: Info (notification pushed to mobile app message feed), Warning (WeChat service account push plus SMS notification), and Critical (WeChat, SMS, and automated voice call simultaneously). Statistics from the past six months show a manual verification accuracy rate of 82% for Warning-level and above alarms, and daily alarm handling efficiency for maintenance personnel has improved 4.7-fold compared with traditional inspection methods.
Remote Diagnostics and Expert Knowledge Base
The remote operation and maintenance platform includes a remote diagnostic mechanism and an expert knowledge base system. When an alarm is triggered, on-site maintenance personnel can initiate a remote diagnostic request with a single tap on the mobile app. Platform technical experts then perform an initial remote diagnosis within 15 minutes and issue handling recommendations, supported by real-time data replay (reviewing parameter trend curves from the 30 minutes preceding the alarm), historical event correlation analysis (searching past resolution records for similar faults), and live video feed (via a 4G remote camera positioned by on-site staff toward the fault area).
The expert knowledge base currently contains more than 400 documented crane fault cases, organized into four major categories: electrical systems (VFDs, PLCs, brakes, limit switches, cable reels—124 cases), mechanical systems (gearboxes, couplings, drums, pulley blocks, hook blocks—156 cases), steel structures (main girder deformation, rail wear, end carriage cracks—58 cases), and wire rope (wire breaks, wear and corrosion, rope end termination, rope replacement—62 cases). Each case entry includes a description of the fault symptom, equipment model and operating conditions, diagnostic procedure, repair plan, required spare parts list, and estimated maintenance hours. On-site personnel can use the app's case search function to enter fault keywords (e.g., "hoisting abnormal noise," "VFD ERR12"), and the system returns the top-5 best-matching cases with corresponding standard handling procedures within 1 second.
The combination of remote diagnostics and the knowledge base enables approximately 65% of common faults (motor overload, brake gap adjustment, limit switch displacement, VFD parameter drift, etc.) to be resolved independently by on-site maintenance personnel under remote expert guidance, eliminating the need to wait for a manufacturer service engineer to arrive on site. Average repair time per fault has been reduced from 8 hours (including engineer travel time) to 1.5 hours.
Intelligent Maintenance Scheduling and Spare Parts Management
The platform automatically generates a differentiated maintenance plan for each overhead crane based on actual operating hours and the load spectrum of working conditions, replacing the traditional "one-size-fits-all" fixed-interval maintenance model. The maintenance scheduling engine considers the following inputs: cumulative equipment operating hours (read from the PLC runtime hour meter), heavy-load duty ratio (cumulative time percentage when hoisting motor current exceeds 70% of rated capacity), start/stop frequency (average starts/stops per hour for the crane bridge and trolley), ambient temperature/humidity (read from workshop environmental sensors), and wire rope/brake wear trend assessment (sourced from the trend alert module).
The scheduling engine outputs a four-tier maintenance plan: daily inspection (completed by the crane operator before each shift handover; the platform pushes a standardized 22-item daily checklist via the mobile app), weekly inspection (once per week, covering 18 items including guide rail cleaning, lubrication point servicing, and bolt tightening torque spot checks), monthly inspection (once per month, covering 15 items including brake gap measurement, wire rope diameter measurement, and gearbox lubricating oil sampling), and semi-annual inspection (every six months, covering 25 items including coupling shaft alignment re-verification, crane rail elevation measurement, main girder deflection measurement, and electrical cabinet cleaning with terminal tightening). Upon completion of each maintenance task, maintenance personnel confirm each item in the app and upload on-site photos, which the platform automatically archives into the equipment's digital maintenance record.
The spare parts management module is linked to the maintenance schedule: when a maintenance task is generated, the platform automatically identifies the required spare parts (such as brake shoes, wire rope, seals, filter elements, lubricating oil, and contactors), verifies whether the on-site spare parts inventory is sufficient, and automatically generates a purchase recommendation pushed to the procurement department if inventory falls below the safety threshold. The spare parts shortage warning is triggered 7 days before the scheduled maintenance date, allowing sufficient time for price comparison and delivery. Since the system went live, maintenance delays caused by spare parts shortages have dropped from an average of 4.5 incidents per month to 0.8.
Frequently Asked Questions (FAQ)
Q: Is the remote operation and maintenance platform compatible with older overhead cranes that lack a PLC interface?
A: Yes. For older cranes using relay-contactor control systems, Kelude offers an external sensor retrofit solution: current transformers (CTs) are installed on the motor incoming lines (measuring range 0–500A, accuracy class 0.5), proximity switches are added to the brakes (detecting engaged/released status), and signal conversion modules are fitted to the limit switches (converting normally open/normally closed contacts into dry contact signals fed to the DTU). The external sensor solution provides complete electrical isolation from the original equipment and does not affect any function of the crane's existing control system or safety circuit. In accordance with GB/T 28264-2017 Safety Monitoring and Management System for Lifting Appliances, the safety monitoring system must not interfere with the safety functions of the original control system.
Q: Is the 4G communication signal stable in the metal-rich workshop environment?
A: Steel structure workshops and dense metal equipment create significant signal shielding effects. Field measurements in a typical workshop with a 30m span and 20m height show that the 4G antenna signal strength at the crane end fluctuates between -85dBm and -105dBm, with a packet loss rate of approximately 1.5–3%. The platform addresses this with two redundancy measures: first, the DTU features dual-SIM, dual-module design (two carriers with automatic switching) — if one carrier's signal is lost, it automatically switches to the other within 15 seconds; second, the DTU has 16GB of onboard storage that can cache up to 72 hours of operational data, which is automatically uploaded once the network is restored, ensuring zero data loss. Statistics from 300 deployed units show a continuous data upload rate (including deferred uploads) of 99.7%.
Q: What is the typical deployment time and cost for the remote operation and maintenance platform?
A: The typical deployment cycle for retrofitting a single overhead crane is 3–5 working days (including DTU installation, sensor wiring, communication testing, and platform configuration). For cranes with a PLC interface, hardware costs (DTU + sensors + antenna + cabling) are approximately $450–$740 per crane. For older cranes without a PLC interface, the additional external sensors bring hardware costs to roughly $740–$1,190 per crane. The platform service fee is charged annually and includes cloud platform deployment, data storage (180-day rolling), alert push notifications (unlimited channels), and remote diagnostics — approximately $220–$445 per crane per year. Based on customer usage data, the remote operation and maintenance platform reduces unplanned downtime by about 60%. Given a production loss of $740–$2,970 per hour per crane, the typical payback period is 3–6 months.
Q: How is data security ensured on the platform? Will customer workshop operational data be exposed?
A: The platform has followed three data security principles since the design phase: minimal collection (only equipment operational parameters are collected — no video footage or personnel behavior data), on-demand sharing (customers can configure the data sharing scope themselves — for example, they may choose to share only alarm events with the service team without sharing full operational data), and customer data isolation (each customer's data is stored independently in the cloud using a multi-tenant architecture with physical-level database isolation rather than logical view isolation). The platform has passed Alibaba Cloud ISO 27001 information security management system certification and holds Level 2 data encryption filing. Customer data is used exclusively for remote diagnostics within the authorized scope and is never used for commercial analysis or AI model training.
