HVAC Oct 27, 2026
Cross-Signal Consistency Checks for Chillers and Pumps
Comparing signals against each other catches what limits cannot. The relationships come from the engineer, and the output is a question.
HVAC Oct 27, 2026
An AHU Telemetry Data Quality Checklist Before You Analyse
Audit per unit, not per estate. Inventory, identity, units, completeness, flags and change history, read down the columns.
Building Automation Oct 27, 2026
Telling Communication Failures Apart From Equipment Anomalies
A data gap and a plant change look identical in a chart. Correlation across shared infrastructure is what separates them before dispatch.
Building Automation Oct 27, 2026
Monitoring Drift in a Live Building Anomaly Detection Service
A service gone quiet looks like a healthy estate. Watch inputs, output rate and usefulness separately, and version every retrain.
Building Automation Oct 27, 2026
Evaluating an Anomaly Detection Pilot on Building Data
Report counts, separate the data artefacts, state what you cannot know about misses, and compare against reviewed limits rather than nothing.
Building Automation Oct 27, 2026
Planning a BMS Anomaly Detection Pilot That Can Produce an Answer
A pilot that cannot fail has told you nothing. Set scope from review capacity, fix disposition categories, and write criteria first.
Building Automation Oct 27, 2026
Fixed Limits or Learned Baselines: Choosing a Detection Approach
The real difference is explainability and maintenance burden, not sensitivity. Review your existing limits before replacing them.
Building Automation Oct 27, 2026
Data Readiness for Anomaly Detection on BMS Telemetry
Most pilots fail on data history, not on method. Run the readiness checks per point first, and scope the pilot to what passes them.
Building Automation Oct 27, 2026
BMS Point Naming and Cross-System Joins: Why Names Fail as Keys
Names get corrected, restructured and reused. Join on a stable surrogate identifier and keep one owned, versioned, queryable crosswalk.
Reliability Oct 27, 2026
Linking BMS Faults to Asset History on One Timeline
Detected faults and work orders are two records of one asset. One chronological event record makes the pattern between them readable.