背景：车间里的”定时炸弹”
重机集团是一家做重型装备的制造企业，车间里有32台桥式起重机，大的起重量100吨，小的也有10吨。这些起重机每天吊着几十吨的工件在车间里来回跑，是生产的核心设备，但也是最大的安全隐患。
在我们进场之前，他们的安全管理是这样的：
- 靠人盯：每个起重机配一个司机，全靠司机的经验和责任心，有没有超载、有没有斜拉、有没有超速，全凭感觉；
- 靠巡检：安全员每天在车间里转，能看到的问题有限，而且很多隐患是看不见的；
- 靠事后查：真出了事故，调监控、问当事人，往往各说各的，很难还原真相。
去年就出过一次险：一台50吨的起重机吊工件的时候，因为斜拉导致工件摆动，撞到了旁边的设备，幸好没伤人，但直接损失就有几十万。
老板下了死命令：必须上一套安全监控系统，把起重机的安全管起来。
系统架构：四层设计，从感知到决策
我们给重机集团设计的起重机物联安全监控系统，整体分四层：
┌──────────────────────────────────────┐
│ 应用层：可视化 + 告警 │
│ 车间大屏 / 管理后台 / 小程序告警 │
├──────────────────────────────────────┤
│ 平台层：业务中台 │
│ 数据存储 / 告警引擎 / 统计分析 │
├──────────────────────────────────────┤
│ 网关层：边缘计算网关 │
│ 数据采集 / 本地告警 / 视频联动 │
├──────────────────────────────────────┤
│ 感知层：各类传感器 │
│ 重量 / 倾角 / 限位 / 风速 / 视频 │
└──────────────────────────────────────┘下面逐层说。
感知层：装什么传感器？
很多人一上来就问：“要装多少传感器？“其实不是越多越好，而是要针对最核心的安全风险来选。
我们给每台起重机装了这几类传感器：
1. 重量传感器
这是最核心的。超载是起重机事故的第一大原因。
怎么测重量？有两种方案：
- 销轴式传感器：替换起重机吊钩上的销轴，直接测受力，精度高，但安装麻烦，要拆吊钩；
- 旁压式传感器：夹在钢丝绳上，通过测钢丝绳的张力来换算重量，安装简单，不用拆设备，但精度稍差。
我们选的是旁压式，因为客户不想停产安装。精度大概在±3%，对于安全监控来说足够了——毕竟超载告警的阈值一般设在90%，3%的误差不影响安全判断。
2. 倾角传感器
装在小车上，测吊物的倾斜角度。斜拉歪吊是非常危险的，很容易导致起重机失稳。
倾角传感器可以实时监测吊物的偏摆角度，超过设定阈值就告警。
3. 限位传感器
包括起升高度限位、大车运行限位、小车运行限位。防止起重机冲到轨道尽头掉下去，或者吊钩冲顶。
这些限位很多老起重机上也有，但都是机械式的，到了位置直接断电，没有记录。我们加了电子限位，不仅能限位，还能记录每次限位触发的时间、位置，方便事后分析。
4. 风速传感器
装在起重机的顶部，测实时风速。室外的门式起重机对风速很敏感，风太大了要停止作业。室内的桥式起重机一般不需要，但如果是高大的厂房，也建议装。
5. 视频摄像头
在起重机的小车上装一个摄像头，对着吊钩位置。司机在驾驶室里就能看到吊钩的实时画面，不用仰着脖子看，也减少了盲区。
更重要的是，视频和告警联动——一旦触发告警，自动抓拍和录像，事后可以回放，还原事故现场。
网关层：为什么必须有边缘计算？
传感器的数据先传到装在起重机上的边缘网关，然后网关再传到中台。
为什么不直接传云端？有几个原因：
1. 实时性要求高
安全告警必须是实时的。如果超载了，要马上提醒司机，等数据传到云端再回来，可能已经晚了。
边缘网关本地就能做告警判断，一旦触发，马上通过驾驶室里的声光报警器提醒司机，同时在司机的操作屏上弹出提示。这个延迟在毫秒级，完全实时。
2. 网络不稳定
车间里的WiFi信号不一定好，起重机又是移动的，经常会有网络波动。如果所有数据都要实时上传，断网了就丢了。
网关本地存数据，断网了继续存，网络恢复了再补传。我们的网关本地能存3个月的数据，完全不用担心断网丢数据。
3. 视频数据量大
一路高清视频，每小时就是几个G。如果全传到云端，带宽和存储成本都很高。
边缘网关本地做视频处理：平时只传低码流的预览画面，触发告警的时候才传高清的抓拍和录像片段。这样带宽成本能降到原来的十分之一。
4. 本地联动
有些动作需要本地设备联动，比如超载了要限制起升动作，这个不能等云端指令，必须本地直接执行。
平台层：中台做什么？
数据到了中台之后，就不只是”看”了，还能做很多分析。
1. 实时监控
车间大屏上能看到所有起重机的实时状态：当前吊重、起升高度、大车位置、小车位置、有没有告警。
哪台起重机超载了、哪台在报警，一眼就能看到。
2. 告警管理
所有告警都有记录：什么时间、哪台起重机、什么类型的告警、持续了多久、当时的参数是多少。
告警分等级：
- 预警：接近阈值（比如90%额定载荷），提醒注意；
- 告警：超过阈值（比如105%额定载荷），必须处理；
- 严重告警：严重违规（比如130%额定载荷），自动上报安全主管。
3. 统计分析
这是很多客户没想到的价值——安全监控系统不仅能管安全，还能管效率。
中台积累了数据之后，可以分析：
- 每台起重机的利用率：每天工作多少小时、吊了多少吨、空载率多少；
- 每个司机的操作习惯：有没有经常超载、有没有经常急停、操作稳不稳；
- 故障趋势：哪些起重机的告警越来越多，是不是该检修了。
重机集团上线之后，通过数据分析发现有两台起重机的利用率特别低，一个月才用几次。后来调整了车间布局，把那两台起重机调去了别的车间，省了一大笔维护费。
4. 视频回放 + 数据同步
出了问题要查原因，光看视频不够，光看数据也不够。我们的系统可以把视频和数据同步回放——看视频的时候，旁边同步显示当时的重量、高度、位置等数据，一目了然。
上次有个司机说”我没超载”，调出来一看，视频里吊的工件不大，但重量数据显示确实超了。再一查，是工件里面藏了东西，司机自己都不知道。最后避免了一次可能的事故。
应用层：谁在用？怎么用？
系统有三个端，给不同的人用：
- 司机端：驾驶室里的操作屏，显示实时状态和告警信息，有语音提示；
- 管理端：Web管理后台，看所有起重机的状态、告警记录、统计报表；
- 移动端：微信小程序，安全主管和车间主任手机上就能看，有严重告警马上推送通知。
为什么要做小程序？因为管理人员不可能一直坐在电脑前。有了小程序，走到哪都能看，有告警马上收到，不用等别人汇报。
落地效果：安全和效率双提升
系统上线半年后，重机集团的变化：
- 安全事故：零起重伤害事故（之前平均每年2-3起）；
- 违规操作：超载、斜拉等违规操作下降了80%以上（司机知道被监控了，自然就规范了）；
- 维护成本：通过状态监测，从”定期维护”变成”按需维护”，维护成本降低了约25%；
- 生产效率：通过利用率分析优化了起重机调度，整体吊装效率提升了约15%。
写在最后
很多企业觉得”安全监控”是个成本中心——花钱但不产生效益。但实际上，好的安全监控系统，不仅能防事故，还能帮你优化生产、降低成本。
关键在于，你不能只把它当成一个”安监要求的摆设”，而是要真正把数据用起来。从数据里发现安全隐患，也从数据里发现效率提升的空间。
桥式起重机的物联监控，只是工业安全数字化的一个缩影。未来，工厂里的每一台设备、每一个环节，都会被数字化、被实时监控、被智能分析。安全不再是”靠人盯”，而是”靠数据管”。
Background: “Time Bombs” on the Factory Floor
Heavy Machinery Group manufactures heavy equipment, with 32 overhead cranes in their workshops — the largest lifting 100 tons, the smallest 10 tons. These cranes move workpieces weighing dozens of tons around the shop floor every day. They’re core production equipment, but also the biggest safety hazard.
Before we came in, their safety management looked like this:
- Relying on people: Each crane has an operator, and safety depends entirely on the operator’s experience and responsibility. Overload, side-pull, overspeed — all judged by feel.
- Relying on inspections: Safety officers walk around the shop floor every day. What they can see is limited, and many hazards are invisible.
- Relying on post-incident investigation: When an accident happens, you pull surveillance footage and interview people — but everyone has a different story, and it’s hard to get to the truth.
Last year there was a near-miss: a 50-ton crane was lifting a workpiece when a side-pull caused it to swing, hitting nearby equipment. Fortunately no one was hurt, but direct losses were several hundred thousand yuan.
The boss gave a strict order: we must implement a safety monitoring system to get crane safety under control.
System Architecture: Four Layers, From Sensing to Decision-Making
The crane IoT safety monitoring system we designed for Heavy Machinery Group has four layers:
┌──────────────────────────────────────┐
│ Application: Visualization + │
│ Alerts │
│ Shop Floor Dashboard / Admin / │
│ Mini Program Alerts │
├──────────────────────────────────────┤
│ Platform: Business Middle │
│ Data Storage / Alert Engine / │
│ Analytics │
├──────────────────────────────────────┤
│ Gateway: Edge Computing │
│ Data Collection / Local Alerts /│
│ Video Integration │
├──────────────────────────────────────┤
│ Sensing: Various Sensors │
│ Weight / Tilt / Limit / Wind / │
│ Video │
└──────────────────────────────────────┘Let’s go through each layer.
Sensing Layer: What Sensors to Install?
Many people immediately ask: “How many sensors do we need?” It’s not about more is better — it’s about selecting sensors for the most critical safety risks.
We installed these types of sensors on each crane:
1. Weight Sensors
This is the most critical. Overloading is the #1 cause of crane accidents.
How do you measure weight? Two approaches:
- Pin-type sensor: Replace the hook pin with a sensor pin, directly measuring force. High precision, but installation is complex — you have to disassemble the hook.
- Rope-clamp sensor: Clamps onto the wire rope, calculates weight by measuring rope tension. Easy installation, no disassembly needed, but slightly lower precision.
We chose rope-clamp sensors because the client didn’t want to halt production for installation. Precision is about ±3%, which is sufficient for safety monitoring — after all, overload alert thresholds are typically set at 90%, so a 3% error doesn’t affect safety judgments.
2. Tilt Sensors
Mounted on the trolley, measures the tilt angle of the load. Side-pull lifting is extremely dangerous and can easily cause crane instability.
Tilt sensors monitor load swing angle in real-time and alert when it exceeds set thresholds.
3. Limit Sensors
Include hoist height limit, bridge travel limit, and trolley travel limit. Prevent cranes from running off the end of tracks, or hooks from over-hoisting.
Many older cranes already have these, but they’re mechanical — they just cut power when triggered, with no record. We added electronic limits that not only stop the crane but also record each trigger’s time and position for post-incident analysis.
4. Wind Speed Sensors
Mounted on top of the crane, measures real-time wind speed. Outdoor gantry cranes are highly sensitive to wind — operations must stop when wind is too strong. Indoor overhead cranes generally don’t need them, but we recommend them for tall workshop buildings.
5. Video Cameras
Mount a camera on the trolley, pointed at the hook. Operators can see the real-time hook view from the cabin without craning their necks, reducing blind spots.
More importantly, video integrates with alerts — when an alert triggers, the system automatically captures snapshots and records video for later playback and incident reconstruction.
Gateway Layer: Why Edge Computing Is Necessary
Sensor data first goes to an edge gateway mounted on the crane, then the gateway transmits to the middle platform.
Why not send directly to the cloud? Several reasons:
1. Real-Time Requirements
Safety alerts must be real-time. If there’s an overload, you need to alert the operator immediately. Waiting for data to go to the cloud and come back might be too late.
The edge gateway handles alert judgment locally. When an alert triggers, it immediately notifies the operator through audio-visual alarms in the cabin and pops up a warning on the operator’s display. This happens in milliseconds — truly real-time.
2. Unstable Networks
WiFi signal in workshops isn’t always great, and cranes move around — network fluctuations are common. If all data requires real-time upload, you lose data when the network drops.
The gateway stores data locally. When the network goes down, it keeps storing; when it comes back, it resumes uploading. Our gateway can store 3 months of data locally — no worries about data loss from network outages.
3. High Video Data Volume
One HD video stream is several gigabytes per hour. If you upload everything to the cloud, bandwidth and storage costs are enormous.
The edge gateway handles video processing locally: normally it only transmits low-bitrate preview footage. When an alert triggers, it uploads HD snapshots and video clips. This reduces bandwidth costs to about one-tenth.
4. Local Control Integration
Some actions require local device coordination — for example, limiting hoisting motion during overload. You can’t wait for a cloud command — it has to execute locally.
Platform Layer: What Does the Middle Platform Do?
Once data reaches the middle platform, it’s not just about “viewing” — you can do a lot of analysis.
1. Real-Time Monitoring
The shop floor dashboard shows real-time status of all cranes: current load, hoist height, bridge position, trolley position, any active alerts.
You can instantly see which crane is overloaded, which one is alerting.
2. Alert Management
All alerts are recorded: time, crane, alert type, duration, parameters at the time.
Alerts have severity levels:
- Warning: Approaching threshold (e.g., 90% rated load), pay attention.
- Alert: Exceeding threshold (e.g., 105% rated load), must be addressed.
- Critical: Serious violation (e.g., 130% rated load), auto-escalate to safety manager.
3. Statistical Analysis
This is the value many clients don’t expect — a safety monitoring system not only manages safety but also improves efficiency.
After accumulating data, the platform can analyze:
- Utilization per crane: daily operating hours, tons lifted, idle rate.
- Operator habits: frequent overloads? frequent emergency stops? smooth operation?
- Failure trends: which cranes have increasing alerts — time for maintenance?
After launch, Heavy Machinery Group discovered through data analysis that two cranes had very low utilization — used only a few times a month. They rearranged the workshop layout and moved those cranes to other shops, saving a fortune in maintenance costs.
4. Video Playback + Data Synchronization
When investigating incidents, video alone isn’t enough, and data alone isn’t enough. Our system synchronizes video and data playback — while watching the video, you see the weight, height, position, and other data from that moment side by side. Crystal clear.
Last time an operator said “I wasn’t overloaded,” we pulled up the recording. The workpiece didn’t look big in the video, but the weight data clearly showed overload. Further investigation revealed something hidden inside the workpiece that even the operator didn’t know about. A potential accident was prevented.
Application Layer: Who Uses It? How?
The system has three interfaces for different users:
- Operator side: Display screen in the cabin, shows real-time status and alerts with voice prompts.
- Management side: Web admin dashboard, view all crane statuses, alert records, statistical reports.
- Mobile side: WeChat mini program — safety officers and workshop directors can view on their phones, and critical alerts push notifications immediately.
Why a mini program? Because managers don’t sit in front of computers all day. With a mini program, they can check from anywhere and get alerts instantly — no waiting for someone to report.
Results: Safety and Efficiency Both Improve
Six months after system launch, the changes at Heavy Machinery Group:
- Safety incidents: Zero crane-related injuries (previously 2-3 per year on average).
- Violations: Overload, side-pull, and other violations dropped over 80% (operators naturally follow the rules when they know they’re being monitored).
- Maintenance costs: Shifted from “scheduled maintenance” to “on-demand maintenance” through condition monitoring — maintenance costs reduced by ~25%.
- Productivity: Optimized crane scheduling through utilization analysis — overall lifting efficiency improved by ~15%.
Closing Thoughts
Many companies see “safety monitoring” as a cost center — you spend money but don’t generate revenue. But actually, a good safety monitoring system not only prevents accidents — it helps you optimize production and reduce costs.
The key is: you can’t treat it as just a “regulatory requirement checkbox.” You have to actually use the data. Discover safety hazards from data, and discover efficiency improvements from data too.
Crane IoT monitoring is just one example of industrial safety digitalization. In the future, every piece of equipment, every process in the factory will be digitized, monitored in real-time, and analyzed intelligently. Safety won’t rely on “people watching” — it will be “managed by data.”