SPICA NetworksSee PumpSpectra

SIIP / SPICA Industrial Intelligence Platform

From raw signals to asset intelligence.

SIIP is SPICA's physics-first industrial intelligence platform. It connects synchronized acquisition, deterministic C++ diagnostics, persistent asset state and bounded local AI in one traceable system.

Explore the architecture See the first SIIP product
SIIP / ACTIVE PATHPHYSICS → ACTION
  1. 01
    Physical AssetMotor · Pump · Process · Grid
  2. 02
    SPICA EdgeADE9000 · ESP32 · synchronized evidence
  3. 03
    Deterministic CoreC++ · FFT · MCSA · physics gates
  4. 04
    Evidence StoreMeasurements · spectra · alerts · context
  5. 05
    Temporal StateEMA · hysteresis · streaks · history
  6. 06
    DecisionBounded local AI · maintenance action
TRACEABLE EVIDENCEOPERATIONAL DECISION
30 sDefault asset-state refresh
2 h EMATime-aware trend horizon
90 daysOperational state history
Local AIConstrained explanation layer

One reusable architecture

Every conclusion stays connected to its source.

SIIP does not begin with an AI prompt. It begins with the physical asset, preserves the acquired evidence and advances each conclusion through explicit diagnostic, temporal and communication authorities.

  1. 01

    Physical Asset

    Motor · Pump · Process · Grid

  2. 02

    SPICA Edge

    ADE9000 · ESP32 · synchronized evidence

  3. 03

    Deterministic Core

    C++ · FFT · MCSA · physics gates

  4. 04

    Evidence Store

    Measurements · spectra · alerts · context

  5. 05

    Temporal State

    EMA · hysteresis · streaks · history

  6. 06

    Decision

    Bounded local AI · maintenance action

Platform capabilities

The reusable layers behind each SIIP product.

SIIP provides one multi-asset foundation for acquisition, analytics, state, history, integration and explanation. Each product adds the domain model and physical constraints for a specific asset class.

01

Synchronized Edge Evidence

Acquires calibrated electrical context and seven aligned waveform channels through explicit REST and TCP contracts.

02

Deterministic Diagnostics

C++ FFT and MCSA logic apply motor physics, bearing geometry, noise gates and corroboration before a fault can be asserted.

03

Temporal Asset State

Time-aware EMAs, persistence streaks and hysteresis turn individual measurements into stable SANO, VIGILAR or CRITICO states.

04

Structured Evidence Store

MongoDB preserves measurements, spectra, advanced analysis, alerts, bearing configuration and maintenance context without flattening the evidence.

05

Operational APIs

Versioned health, history and diagnosis endpoints expose current state, exact evidence and reports to applications and operators.

06

Bounded Local AI

A local model explains approved facts and recommended actions; response sanitization prevents it from overruling deterministic flags.

The SIIP asset-health record

Not just an alert. A traceable operational record.

  1. 01Acquired evidence
  2. 02Physics validity
  3. 03Operating context
  4. 04Fault flags
  5. 05Temporal trend
  6. 06Maintenance action

The engineering principle

Physics first.
AI second.

Deterministic C++ remains the diagnostic source of truth. A separate state engine then adds persistence, trend and maintenance context without rewriting the measured fault flags.

Local AI receives an allowed fact set and explains it for a human. It cannot confirm a fault that the deterministic layer rejected, and its response is sanitized before it becomes part of the asset record.

DETERMINISTIC COREEvidence → Context → DiagnosisEXPLANATION LAYERAI → Human decision
01DIAGNOSTIC AUTHORITY

Deterministic C++

Binary fault flags come from measured spectral evidence, physical validity checks and explicit thresholds—not from a language model.

02OPERATIONAL AUTHORITY

Temporal state engine

Time-aware EMA, hysteresis, persistence and maintenance urgency stabilize the asset state across captures.

03COMMUNICATION LAYER

Local AI explanation

The model summarizes allowed facts, trends and actions. Its output is sanitized and cannot create an unsupported diagnosis.

Engineering guardrails

Confidence includes knowing when not to decide.

SIIP preserves invalid, contaminated and ambiguous conditions as part of the result. That makes uncertainty inspectable instead of hiding it behind a single score.

01

Physics must be valid

Broken-bar and eccentricity decisions require valid frequency, poles, mechanical speed and slip before their flags can activate.

02

Close sidebands are rejected

A minimum separation guard suppresses broken-bar analysis when the expected line is too close to the electrical fundamental.

03

Grid contamination is explicit

Voltage/current coupling and proximity to grid harmonics are recorded so supply distortion is not silently presented as a mechanical fault.

04

Ambiguity remains visible

When drive-end and non-drive-end bearing frequencies are too close, SIIP marks localization as ambiguous and recommends physical inspection.

Operational interfaces

The evidence chain is available as an API.

The current SIIP service exposes immediate capture diagnosis, historical analysis and asset-state retrieval as separate contracts, so applications do not have to infer which evidence path produced a result.

GET/health/{motor_id}Current SANO / VIGILAR / CRITICO state, evidence and maintenance context.
GET/history/{motor_id}Retained state history and temporal deltas for an asset.
POST/diagnose_fftImmediate, stateless diagnosis of a supplied FFT result; no database read or write.
POST/diagnose_historyWindowed historical diagnosis using measurements, persistence and trend evidence.
GET/motorsDiscover the assets currently represented in the evidence store.

API paths describe the deployed SIIP service contract. They are not public web pages and therefore are intentionally shown as reference endpoints.

Platform → product

SIIP is the foundation. PumpSpectra is the first product.

SIIP supplies the reusable acquisition, analytics, event, history and integration layers. PumpSpectra applies those layers to motors, pumps and rotating equipment.

FOUNDATION / INDUSTRIAL PLATFORM

SIIP

Reusable signal-to-decision architecture
  • Synchronized edge evidence
  • Deterministic diagnostic authority
  • Temporal state, history and APIs
  • Bounded local AI explanation

FIRST PRODUCT / ROTATING EQUIPMENT

PumpSpectra™

Condition intelligence for motors and pumpsExplore PumpSpectra

Build on a traceable foundation

Discuss an industrial intelligence application.

Talk to Engineering