Technical

Under the hood

Explore how Neraium blends multivariate statistics with deterministic governance to produce reliable infrastructure intelligence.

Signal geometry Interpretation rules

Sample algorithm

// Compute multivariate deviation
let diff = sensorVector - meanVector;
let invCov = invert(covarianceMatrix);
let distance = Math.sqrt(diffT * invCov * diff);

// Evaluate against drift threshold
if (distance > threshold) {
  if (passesPersistence &&
      passesCorrelation &&
      withinPhysicalLimits) {
    admit();
  } else {
    suppress();
  }
} else {
  voidResult();
}

Core methods

Signal geometry and system behaviour

Neraium models infrastructure behaviour through relationships between variables, helping distinguish structural change from ordinary operational motion.

Multivariate distance

Evaluate how far system behaviour has moved relative to its established operating geometry.

Covariance modelling

Capture how telemetry variables move together across assets, time, and operating conditions.

Adaptive baselines

Maintain relevance by updating expectations as valid infrastructure behaviour evolves.

Interpretation rules

Detection alone is not enough

Candidate deviations pass through deterministic filters before they become operational intelligence, helping reduce noise, ambiguity, and weak conclusions.

Physical validity

Signals must remain within plausible infrastructure interpretation boundaries.

Temporal persistence

Deviation must persist long enough to indicate meaningful system change.

Cross-signal agreement

Relevant changes require corroboration across related telemetry rather than one isolated channel.