Direct Answer

A platform alarm only says "temperature abnormal" and the maintenance crew spends half an hour finding the site — how can an alarm tell you the location instantly? An 18-level scene tree drills down level by level from the campus to the panel, busbar section, feeder, branch switch and finally the terminal/contact point; every data point is bound to a hierarchy path at modeling time, alarms automatically carry a location tag, and "one alarm, one location" is achieved. The Xiangxi (Wisdom Scene) engine of the Wanxiang platform uses the scene tree to bind every alarm to a concrete location, so the crew knows the site as soon as the alarm arrives.

A scene: an alarm said "transformer 1 temperature abnormal" — the crew reached the site, the transformer area was large, and finding the hot spot took 40 minutes while the equipment kept heating. The granularity of an alarm decides the speed of handling — "one alarm, one location" means the crew starts working on arrival instead of searching on site.

1. The 18-Level Structure of the Scene Tree

Level design — 18 levels from campus to terminal: campus → building → switch room → transformer/panel → busbar section → feeder → branch → switch → terminal/contact point — one level each, drilling down step by step; Location codes — every level has a location code: campus-building-room-panel-busbar-feeder-branch-terminal — the code is the location tag; Point mounting — every point hangs on a specific node of the tree: a terminal temperature point hangs on the terminal node — where the point sits in the tree is where the alarm sits; Structured tree — the scene tree is the "location dimension" of the object model: the object model manages "what the device is", the scene tree manages "where the device is" — the two dimensions together make complete modeling.

In one sentence: the 18-level scene tree is "a location staircase from campus to terminal" — the level a point hangs on is the level an alarm can locate — the location code is the location tag of the alarm.

A structure example: an alarm read "B-phase terminal temperature abnormal in panel 2, switch room, building 3" — the scene tree path: campus → building 3 → switch room → panel 2 → B phase → terminal — the crew took a thermometer straight to panel 2. 18 levels are not a numbers game but "a locating circle that narrows level by level" — the deeper the tree, the fewer steps the crew walks.

2. The Binding Rules Between Data Points and Hierarchy Paths

Binding rule — every data point is bound to a complete hierarchy path at modeling time: a terminal temperature point's path is "campus → … → panel 2 → B-phase terminal" — the path is the location; Binding timing — binding happens at the modeling stage (corresponding to the object modeling in KD-0277): build the point, bind the path — without binding, the alarm has no location; Path uniqueness — one path uniquely corresponds to one physical location: if paths are not unique, alarm locating falls into disorder — coding rules guarantee uniqueness; Binding cost — binding is a one-time investment: done along with point modeling — later points inherit the hierarchy path automatically.

In one sentence: the binding rule is "build a point, bind its path; a path uniquely maps to a location" — binding happens at modeling, a one-time investment with permanent benefit.

A binding example: when a platform modeled its points, every point was bound to its hierarchy path at creation — 500 new points, all paths bound — subsequent alarms automatically carried locations. Binding is not an afterthought but "bind at point creation" — done along with modeling, and alarm locating has its basis.

3. Location Awareness: The Same 65°C Means Different Risks at Different Locations

Location decides risk — the same 65°C: 65°C at a busbar joint is normal, 65°C at a cable joint is running warm, 65°C at a terminal contact may be abnormal — different locations, different thresholds and judgments; Location-aware thresholds — thresholds configured to locations: temperature thresholds at different locations are set by location — location awareness makes thresholds "fit the place"; Location-aware inference — inference rules combine location: terminal temperature trending up + the location is a contact point → infer increasing contact resistance — location is an important input to inference; The value — location awareness lets alarms "neither false-alarm nor miss": the same value is normal where it belongs and alarming where it matters — judgment accuracy steps up.

In one sentence: location awareness is "the same value, different locations, different judgments" — thresholds and inference rules carry location context — this is the extra value the scene tree adds over ordinary alarms.

A location example: two places on a campus both read 65°C: at the busbar joint it was judged normal (that location's threshold is 70°C), at the terminal contact it triggered an alarm (that location's threshold is 60°C) — two locations, two judgments. The same temperature means two different things at two locations — location awareness makes alarms "fit the place", neither false-alarming nor missing.

4. The Interaction Design of Alarm Drill-Down

Alarms carry locations — the alarm message carries a location tag: the text is the path "building 3 → panel 2 → B-phase terminal" — you know where it is without clicking; Interactive drill-down — clicking an alarm drills down the scene tree level by level: campus → building → panel → terminal — each level shows its state and alarms; Drill-down linkage — the drill-down shows linked data: current values of the level's points, history curves, similar alarms — drill-down is not "flipping pages" but "seeing the full picture"; One-click locating — linked to the 3D/plan view: clicking an alarm automatically locates the position on the view — the scene tree works with the twin view (corresponding to KD-0279).

In one sentence: alarm drill-down is "path in the message, click to drill level by level, state at each level, one-click location in the view" — drill-down is a process of seeing the full picture, not flipping pages.

An interaction example: an operator received the alarm "B-phase terminal 78°C in panel 2, building 3" — clicked to drill down: panel level showed currents, terminal level showed the temperature curve and 7-day trend — located in 3 minutes, on site in 15. The experience of alarm drill-down is "arrive in one step" — the path is in the message, the details are in the click, the full picture is in the view.

5. The Trade-Off Between Modeling Cost and Locating Precision

Where the cost is — full 18-level modeling is expensive: terminal-level modeling means many points and heavy binding work — full terminal modeling is "expensive but complete"; The trade-off strategy — model by risk level: incoming lines, transformers and key branches down to the terminal level — ordinary branches to the panel level — the risky ones are modeled finely; Phased deepening — phase one models to panel/feeder level, phase two deepens to terminal level driven by alarm hotspots — modeling follows the alarm data; Cost-effectiveness — the marginal benefit of precision diminishes: panel-to-feeder gains a lot, feeder-to-terminal gains less — choose granularity by risk.

In one sentence: the trade-off between modeling cost and precision is "model by risk level, deepen with alarm data" — high-risk modeled to terminals, ordinary to panels — precision follows risk.

A trade-off example: phase one of a campus: incoming lines and transformers modeled to terminal level, ordinary branches to panel level — alarm data later showed many faults at a certain branch's terminals — phase two deepened that branch to terminal level. Modeling is not "all at once" but "phased by risk" — protect the key points first, deepen with the data.

6. Scene-Tree Alarms versus Ordinary Platform Alarms: What Differs

Ordinary alarms — alarms hang on devices but not on locations: only say "transformer 1 temperature abnormal" — on site you search for which part it is; Scene-tree alarms — alarms carry the complete path: transformer → A phase → outgoing terminal — the crew goes straight to the location; The hidden cost of ordinary alarms — every alarm needs on-site searching: 40 minutes per time, hundreds of alarms a year — search time is the big overlooked cost; The visible gain of scene-tree alarms — an alarm is the location: handling changes from "search-type" to "direct-type" — half an hour shrinks to one message.

In one sentence: an ordinary alarm "tells you it is broken", a scene-tree alarm "tells you where it is broken" — the search time saved by "where it is broken" is the value of the scene tree.

A comparison example: two platforms on the same campus: A's alarm read "transformer temperature abnormal", B's read "B-phase terminal 78°C in panel 2, building 3 + 7-day trend" — with 20 alarms a month on each: A averaged 30 minutes of searching, B averaged 10 minutes to arrival. The gap between "telling you it is broken" and "telling you where" is tens of hours of search time every month.

7. A Complete Locating Example

The full flow: an alarm read "outgoing terminal temperature 78°C, B phase, panel 2, switch room, building 3" on a campus — the path: building 3 → switch room → panel 2 → B-phase outgoing terminal — the crew went straight to panel 2 by the location tag, an infrared thermometer confirmed terminal overheating, and the 7-day trend pointed to increasing contact resistance — a scheduled outage was arranged, and from alarm to handling the job took 2 hours. The complete chain of "one alarm, one location": the path in the message, the site found directly, the judgment from the trend — the 18-level scene tree eliminates the time spent "finding the location".

In one sentence: terminal-level alarm locating = the 18-level scene tree + bind-path-at-creation + location-aware thresholds + drill-down interaction — chain these four links and "one alarm, one location" holds.

8. Common Mistakes: Three Pitfalls in Scene-Tree Modeling

Pitfall one: the tree is built but paths are not bound — the scene tree is drawn, but point modeling did not bind paths — alarms still do not locate — binding is where the tree's value lives; Pitfall two: paths are not unique — coding rules were not settled, two paths point to the same location — alarm locating falls into disorder — code uniqueness is the premise; Pitfall three: location awareness is not configured — thresholds are not set by location: busbar 65°C and terminal 65°C share one threshold — false alarms and missed alarms both come — location-aware thresholds must be configured separately.

In one sentence: the three pitfalls map to three things — a built tree must bind paths, paths must be unique, thresholds must be split by location — the scene tree's value is in binding and awareness, not in drawing.

A pitfall example: a platform drew the scene tree but did not bind point paths — alarms still only said "temperature abnormal" — after re-binding, alarms carried locations. The scene tree is not "drawing a picture" but "building a binding" — tree + binding + awareness, three things, none optional.

9. Back to FEXLINK: How the Xiangxi (Wisdom Scene) Engine Uses the Scene Tree

The Xiangxi (Wisdom Scene) engine of the Wanxiang platform realizes "one alarm, one location" with the 18-level scene tree: the tree drills down from campus to terminal, data points are bound to hierarchy paths at modeling, alarms automatically carry location tags, and drill-down interaction shows the full picture level by level. The landing order: ① build the scene-tree levels → ② bind points to hierarchy paths → ③ configure location-aware thresholds → ④ alarms carry location tags → ⑤ drill-down and view linkage — five steps done, alarms locate to the terminal level.

One summary: the 18-level scene tree = the location dimension of the object model — drilling down from campus to terminal, binding paths at point creation, location-aware thresholds, drill-down for the full picture — alarms automatically carry locations, and the crew knows the site on receipt. The value of "one alarm, one location" is removing the time spent "finding the location" — the 18-level scene tree shrinks locating from half an hour to one message.

One reminder: do not chase "build everything at once" in scene-tree modeling — model by risk level and deepen with alarm data — incoming lines and transformers to the terminal level, ordinary branches to the panel level — precision follows risk and cost follows precision. Model granularity is "risk pricing" — the high-risk modeled finely, the low-risk coarsely — spend where it counts.

Scope and Limitations

  • The conclusions in this article are intended for LV distribution and electrical-safety sites. Criteria are compiled from public codes and FEXLINK product documentation, and product parameters follow the latest selection handbook and product knowledge base.
  • Quantitative criteria (protection zone, discharge and residual voltage, earth resistance, thresholds and grading) hold under the preconditions stated in this article. Where the site voltage level, earthing arrangement, environmental category or load profile differs, the figures must be recalculated. Specific configurations involving Wanxiang engine (SceneAware) should be verified against actual circuit parameters.
  • Specific product models and parameters are subject to the latest selection handbook and product knowledge base; this article does not constitute selection or procurement advice.
  • Standard numbers cited here refer to the currently effective editions; when a standard is revised, the related quantitative conclusions must be re-checked.
  • Deployment and implementation for specific projects must be carried out by qualified engineers based on site survey and design drawings.

Frequently Asked Questions

How an 18-Level Scene Tree Localizes Alarms Down to a

A platform alarm only says "temperature abnormal" and the maintenance crew spends half an hour finding the site — how can an alarm tell you the location