Intelligent Maintenance & Predictive AGL

Division 15 is the intelligence layer of the AGL maintenance system. It uses the automation, digital monitoring and advanced photometric intelligence established in Divisions 12–14 to answer the question that matters most: what is likely to fail, when, how serious is the risk, and what maintenance action should be taken before the failure affects AGL availability or operational performance. It covers 15 topics (PDM-01 to PDM-15).

What is likely to fail, when is it likely to fail, how serious is the risk, and what maintenance action should be taken before the failure affects AGL availability or operational performance — while the final maintenance decision remains under appropriate engineering and maintenance governance?

The 15 topics are Predictive AGL Maintenance, Failure Prediction, Remaining Useful Life Prediction, Asset Health Prediction, Condition-Based Maintenance, AI Maintenance Planning, Predictive Transformer Maintenance, Predictive CCR Maintenance, Predictive Fixture Maintenance, Predictive Cable Monitoring, Predictive Sign Maintenance, Smart Maintenance Alerts, Automated Maintenance Prioritization, Failure-Risk Prediction and Maintenance Optimization.

Division 15 is the natural convergence point of the technology divisions. It does not generate raw data itself — it converts the information produced upstream into maintenance intelligence and optimized intervention decisions.

Division 15 is recommended as the master architecture linking asset data collection through to the failure database and AI/reliability learning loop. AGL ASSETS │ ▼ DATA COLLECTION │ ┌───────────────┼────────────────┐ │ │ │ Inspections Testing Photometry │ │ │ └───────────────┼────────────────┘ │ IoT DATA │ ▼ ASSET DATA PLATFORM │ ▼ CONDITION ANALYSIS │ ┌──────────┼──────────┐ │ │ │ Health Trend Failure Score Analysis Prediction │ │ │ └──────────┼──────────┘ │ ▼ RUL / RISK │ ▼ MAINTENANCE OPTIMIZATION │ ▼ MAINTENANCE PRIORITY │ ▼ WORK ORDER │ ▼ MAINTENANCE │ ▼ TEST / VERIFY │ ▼ ASSET UPDATE │ ▼ FAILURE DATABASE │ └──────► AI / RELIABILITY LEARNING.

The ultimate objective is a continuous improvement loop rather than a one-time predictive-maintenance calculation. MONITOR; DETECT; DIAGNOSE; PREDICT; PRIORITIZE; PLAN; MAINTAIN; TEST; VERIFY; UPDATE; LEARN.

For each asset, the future platform should ideally maintain a continuously updated Asset Health Profile, grouped into six categories. Asset Identity; Asset ID; Asset type; Location; Manufacturer; Model; Serial number; Lifecycle; Installation date; Commissioning date; Operating age; Expected service life; Condition; Physical condition; Electrical condition; Functional condition; Photometric condition; Environmental condition; Reliability; Failure count; MTBF; MTTR; Failure frequency; Predictive; Health score; Degradation trend; Failure risk; RUL; Recommended action; Maintenance; Last maintenance; Next planned action; Maintenance priority; Work-order status; This connects directly with the Asset Lifecycle Master Database & Intelligence and Failure Records, Evidence & Intelligence frameworks.

Each AGL asset type is monitored against its own set of primary predictive indicators. Asset — Primary Predictive Indicators; Fixture — Photometry, LED condition, operating hours, temperature, failure history; CCR — Output current, alarms, trips, temperature, performance trend; Isolation Transformer — Test results, insulation condition, age, failure history; Cable — Insulation trend, test history, failures, environmental exposure; Sign — Illumination, electrical condition, physical condition, operating history; PAPI — Optical/functional performance, alignment-related data, maintenance history; Circuit — Current, voltage, insulation condition, faults, CCR performance; Control System — Alarms, communication status, event history, failures.

Division 15 positions AGL maintenance along a six-level maturity progression. Level — Stage — Description; 1 — Preventive — Maintenance based mainly on time or usage; 2 — Condition-Based — Maintenance based on measured asset condition; 3 — Predictive — Data is used to predict degradation or failure; 4 — Risk-Based — Failure probability and operational consequence are combined; 5 — Prescriptive — System recommends the most appropriate maintenance action; 6 — Intelligent Maintenance — AI, automation, asset intelligence and engineering judgment operate as one integrated maintenance system.

This is the long-term extension of Division 15. Predictive maintenance asks "What is likely to happen?" — prescriptive maintenance asks "What should we do about it?" This is more valuable than generating a simple "maintenance due" notification. Fixture Degradation Detected Failure Risk Calculated Operational Criticality Assessed Spare Availability Checked Maintenance Window Evaluated Repair vs Replacement Evaluated Recommended Action Engineer Approval.

Twenty registers are recommended to record and govern the outputs of the intelligent maintenance system. It includes Intelligent Maintenance Master Register, Predictive Maintenance Register, Failure Prediction Register, RUL Prediction Register, Asset Health Register, Condition-Based Maintenance Register, AI Maintenance Planning Register, Predictive Transformer Register, Predictive CCR Register, Predictive Fixture Register, Predictive Cable Register, Predictive Sign Register, Smart Maintenance Alert Register, Maintenance Priority Register, Failure-Risk Register and Maintenance Optimization Register.

Twenty-four KPIs are recommended across prediction accuracy, reliability, maintenance mix, asset health coverage and business impact. Prediction; Failure-prediction accuracy; False-positive rate; False-negative rate; Predictive-warning lead time; Prediction coverage; Reliability; MTBF improvement; MTTR reduction; Failure-frequency reduction; Repeat-failure reduction; Unplanned outage reduction; Maintenance; Predictive-maintenance percentage; Condition-based maintenance percentage; Planned vs reactive maintenance; Maintenance-priority accuracy; Maintenance backlog reduction; Asset Health; Assets with health scores; Assets with degradation trends; Assets with RUL estimates; Critical assets under predictive monitoring; Business / Operational; Maintenance cost optimization; Spare-parts optimization; Asset availability; Operational disruption reduction; Lifecycle-cost improvement.

Division 15 converts all upstream information into maintenance intelligence and optimized intervention decisions. The steps or layers are: Reactive Maintenance; Preventive Maintenance; Condition-Based Maintenance; Predictive Maintenance; Risk-Based Maintenance; Prescriptive Maintenance.

Move AGL maintenance from fixed schedules toward maintenance based on actual asset condition, performance trends, failure history and predicted risk. Potential Inputs: Asset condition, Inspection results, Test results, Photometric results, CCR performance, Transformer condition, Cable condition, Fixture performance, Sign performance and PAPI performance.

Determine the most appropriate maintenance action rather than simply generating more maintenance. Possible Decisions: Continue monitoring, Inspect, Test, Repair, Replace component, Replace complete asset, Change maintenance interval, Increase monitoring, Investigate root cause and Perform lifecycle replacement.