AI assisted
Image and video recognition where it is reliable: lane and egress obstruction monitoring, reading combustion-performance grades of linings and insulation, with human review of results.
Industry framework briefing · Updated
In one sentence: When every vendor scores a building by its own rubric, no two results can be compared - the shared data layer that lets buildings, platforms and assessors speak one language about fire safety.
At a glance
Assessment quality is decided before anyone scores anything - by the completeness and honesty of the fields beneath the score.
The reference framework fixes the collection scope before any scoring is discussed. Each row below is a family of fields with a declared primary source; profiles feed every later judgement, including whether a protection system is even required.
| Group | Size | Covers | Primary source |
|---|---|---|---|
| Building fire prevention | 13 items | External environment, fire lanes, separation distances, fire resistance, height and storey type, building age, compartments, egress, layout, occupant density, interior linings, insulation, electrical equipment | Manual / AI |
| Fire protection systems | 7 systems | Outdoor and indoor hydrants, automatic alarm, automatic sprinklers, smoke control and exhaust, extinguishers and other means, maintenance regime | Manual / IoT |
| Fire safety management | 5 duties | Safety policy and archives, emergency and evacuation plans, training and drills, daily and periodic inspection, on-site response capability | Manual |
| Common - building profile | 16 fields | Type, nature, height, storeys above and below grade, total and footprint area, volume, largest single floor, hazard class, seat and bed counts, central air conditioning, fire-resistance level, construction date, use type | Manual |
Every field carries a declared source type, and the mix matters: data completeness, value density and the manual / IoT / AI ratio are themselves evaluation dimensions in the downstream program.
Image and video recognition where it is reliable: lane and egress obstruction monitoring, reading combustion-performance grades of linings and insulation, with human review of results.
Field-verified facts entered after site checks and document review: profiles, external environment, separation distances, policies, plans, training, inspections and response staffing.
Live facility status reported by connected sensors: electrical load and detection results, hydrant pressure, alarm and sprinkler health, online status and long-unresolved faults.
The boundary is explicit: people submit raw fields; judgement is computed. This is what keeps two platforms from reaching two different truths from the same building.
Answers draw on one shared dictionary so values compare across platforms and years. Samples:
| Enumeration | Values |
|---|---|
| Fire-resistance level | Level 1 / Level 2 / Level 3 / Level 4 / below Level 4 |
| Combustion performance | A / B1 / B2 / B3 |
| Fire hazard class | Class A / B / C / D / E (Jia-Yi-Bing-Ding-Wu) |
| Storey type | Single-storey / multi-storey / high-rise / underground |
| Structure type | Reinforced concrete / frame / steel structure / temporary |
The IoT-sourced fields are only as real as the devices behind them. The mappings below show published Wanlin device families against the framework fields they feed; model selection always follows the site survey.
| Framework field | Device family | What it proves |
|---|---|---|
| Protection-system status (hydrant, sprinkler, tank) | WANLIN-770 4G pressure gauges, WANLIN-780 4G level gauges | Mains pressure and tank levels reported live instead of graded from an annual reading |
| Electrical load and detection (indicator 1.13) | Smart electricity set: host 705 with breakers 710 / 711 | Design versus measured load and electrical fault detection at distribution boxes |
| Alarm system networking (indicator 2.3) | WANLIN-921 user information transmission | Networks an existing fire alarm panel; online and linkage status supervised (GB 26875.3) |
| Detection coverage in corridors, sheds and homes | WANLIN-307 / 309 / 310 smoke, WANLIN-340 heat | Battery-powered points whose installed-and-online state is itself verifiable data |
| Maintenance evidence (indicator 2.7) | Cloud maintenance tickets and patrol records | Faults, repairs and monthly upkeep leave timestamped records rather than claims |
Yes. Fire-resistance levels, combustion-performance classes, hazard classes, building classifications and structure types all draw on one shared enumeration dictionary, so values are comparable across platforms.
Conditional fields such as seat count or bed count are left empty when not applicable; whether a protection system is required is judged from the building profile rather than assumed.
Inventory what already exists - drawings, records, live devices - map it to the field list, then close the gaps: manual items first, IoT coverage where continuous monitoring pays back.
Three source types are declared per field: manual entry after site checks, IoT devices reporting live facility status, and AI-assisted recognition such as lane monitoring and material-grade reading, with human review of AI results.
GB/T 41020-2021, the national guideline for fire risk assessment of building property insurance, is the declared minimum granularity for the collection items.
The cheapest moment to fix data quality is before the first assessment. Map existing records to the framework, mark the IoT-able fields, and close the manual gaps in one pass.
Scope disclosure: this page explains a published industry data framework in plain terms for fire-safety practitioners; it is educational and does not assert any participation status, designation or certification for Wanlin products. Program participation, assessments and certificate scope must be confirmed through the organizing bodies and, for devices, against the exact model and configuration. To discuss how your building data could meet the framework, contact wanlinfirecontrol@163.com or 16710807777.
Yes. Fire-resistance levels, combustion-performance classes, hazard classes, building classifications and structure types all draw on one shared enumeration dictionary, so values are comparable across platforms.
Conditional fields such as seat count or bed count are left empty when not applicable; whether a protection system is required is judged from the building profile rather than assumed.
Inventory what already exists - drawings, records, live devices - map it to the field list, then close the gaps: manual items first, IoT coverage where continuous monitoring pays back.
Also on this site: SDOS / SPP Data Asset Management · Fire Prevention Capability Index · Fire Industry Internet Demonstration Site, the aging-community renovation program, the product center for datasheets, and the site FAQ section., and the topic guides index