src.dackar.RCA.equipment_similarity.equipment_spec_builder

equipment_spec_builder — EquipmentSpecBuilder.

Converts KG element_definition properties + failure mode data into a natural-language spec string suitable for embedding. No NLP, no LLM — pure deterministic string assembly. Missing properties are silently omitted.

Attributes

JsonDict

_DEFINITION_PROPS

Classes

EquipmentSpecBuilder

Builds a natural-language equipment spec string for embedding.

Module Contents

src.dackar.RCA.equipment_similarity.equipment_spec_builder.JsonDict[source]
src.dackar.RCA.equipment_similarity.equipment_spec_builder._DEFINITION_PROPS = [('domain_category', 'Type'), ('structural_kind', 'Structural kind'), ('nominal_size', 'Nominal...[source]
class src.dackar.RCA.equipment_similarity.equipment_spec_builder.EquipmentSpecBuilder[source]

Builds a natural-language equipment spec string for embedding.

The output is intentionally verbose — more context helps the embedding model place the component accurately in semantic space. Downstream similarity is purely cosine distance between these representations.

Example output:

Equipment: Main coolant pump (MCP-A)
Type: centrifugal pump
Structural kind: horizontal split-case, 4-stage
Nominal size: 500 GPM
Design pressure: 150 psig
Design temperature: 300°F
Material: 316 stainless steel
Manufacturer: Flowserve
Model: VPC-4-150
Failure modes: bearing wear, seal degradation, impeller erosion
Failure mechanisms: fatigue, abrasion, corrosion
build_spec_text(component_id, component_name, definition_props, failure_mode_names=None, failure_mechanisms=None)[source]

Build and return the spec text for one component.

Parameters:
  • component_id (str) – KG element_usage node ID.

  • component_name (Optional[str]) – Human-readable name of the component instance.

  • definition_props (JsonDict) – Dict of element_definition node properties. Keys expected: domain_category, structural_kind, nominal_size, design_pressure, design_temperature, material_spec, manufacturer, model_number. Missing keys are skipped.

  • failure_mode_names (Optional[List[str]]) – List of failure mode names linked to this component.

  • failure_mechanisms (Optional[List[str]]) – List of failure mechanism labels (e.g. “fatigue”, “corrosion”).

Return type:

str