FIND ROOT CAUSE

Applies a KPI Library set against already-extracted Responses (from a simulation’s stored responses, a Response Template applied to a simulation, or an inline dataset) and emits Diagnoses keyed to each KPI Target that matched a Response. Source-agnostic: the worker doesn’t care whether responses came from a binout, a physical-test DAQ, a BMS scatter file, or upstream workflow output — extraction is upstream. Outputs the Diagnosis list, the Responses actually evaluated (for audit), and a model-level summary log.

When to use

Tagged: rca, root_cause, diagnoses, kpi, kpi_library, responses, template, dataset.

Inputs

Label ID Type Default Required Description
Simulation source_sim_id remote_lookup   Source: simulation id. Alone: loads all stored responses on the sim. With template_id: filters to that template’s response set on that sim.
Response Template template_id remote_lookup   Optional. Used together with source_sim_id to filter the sim’s responses to those produced by this Response Template.
Inline responses (JSON) dataset text   Inline { “responses”: [ { name, value, unit, type, channel_group?, meta? }, … ] } payload. Used for workflow composition. Pick exactly one of source_sim_id or dataset.
KPI Library set kpis_id remote_lookup   KPI Library set to evaluate. Pick from the existing platform library (Admin → KPI Library).
Rule filter rule_filter text   Restrict evaluated targets by glob (‘head_*’) or domain (‘domain=automotive’). Comma-separated filters AND-combine.
Enable internal derivation enable_internal string 1   When 1 (default), targets whose metric isn’t found in dataset/template responses are computed internally from the sim’s raw time-history via the operator packs.
Internal operator packs internal_packs text time_series,crash_structure   Comma-separated list of operator pack ids to enable for internal derivation. Available: time_series, crash_structure. (crash_occupant + bms ship in Phase D.3/D.4.)
Channel map (JSON) channel_map text   Optional JSON object mapping canonical roles to per-sim entity ids, e.g. {“driver.head”:”90001”,”front_rail”:”5”}. Used when KPI targets reference roles instead of concrete entities. Auto-discovered if absent (planned).

Outputs

Label ID Type Description
Diagnoses diagnoses dataset { items: [{ id, rule_id, rule_label, severity (‘info’/’warn’/’crit’), label, response_name, response_id, value, unit, threshold_op, threshold_value, breach, prescription, evidence }], severity_counts, hit_count }
Responses responses dataset { count, preview: first 5 responses evaluated }. Full set is in memory; surfaced for audit + downstream composition.
Provenance provenance dataset { kpi_set_id, kpi_set_name, kpi_set_regulator, kpi_set_revision, source_kind, source_ref, rule_filter, elapsed_ms, response_count, target_count, matched_count, unmatched_count }
Summary log summary_log dataset Model-level observations: severity counts, critical labels, unmatched targets, etc.

Disciplines

  • cae.postprocessing.diagnosis
  • cae.postprocessing.response
  • engineering.crash.injury

Auto-generated from platform schema. Worker id: find_root_cause. Schema hash: a1a309096022. Hand-curated docs in workerexamples/ override this page when present.