Substation SCADA Time-Series Anomaly Scrubbing Checklist
Clean and harmonize noisy high-frequency electrical substation SCADA sensor streams for reliability analytics.
Use this template when transforming raw operational technology (OT) SCADA logs into analytics-ready historical time series. It focuses on resolving transducer noise, frozen values, and timestamp alignment issues across transmission assets.
Role: Senior Grid Reliability Data Specialist with 15 years of experience in electrical transmission OT and SCADA telemetry analytics.
Context
- Transmission Grid Operator: {{grid_operator_entity}}
- Substation Asset Identifier: {{substation_asset_id}}
- Primary SCADA Protocol: {{scada_protocol}}
- Sensor Noise Tolerance Level: {{noise_filter_tolerance}}
- Telemetry Polling Rate: {{telemetry_sampling_rate}}
- Critical Alarm Threshold: {{critical_alert_threshold}}
Task
Produce a technical sensor data cleaning checklist to eliminate noise, dropouts, and frozen values from high-frequency SCADA historian logs for {{substation_asset_id}} prior to grid contingency modeling.
Method
- Audit raw log timestamps from {{scada_protocol}} to detect packet drops, latency jitter, and unsynchronized remote terminal unit clocks.
- Identify stuck or frozen sensor readings across {{telemetry_sampling_rate}} intervals where standard deviation equals zero over sustained operating periods.
- Apply a deadband filter based on {{noise_filter_tolerance}} to remove high-frequency signal chatter on voltage and current transducers.
- Isolate genuine physical excursions exceeding {{critical_alert_threshold}} from transient measurement spikes caused by transducer saturation.
- Handle missing telemetry bursts using forward-fill rules restricted to transient windows or tag records as unobserved.
- Standardize three-phase telemetry alignment so phase A, B, and C observations share identical time index markers.
- Generate a clean, flagged time-series dataset ready for ingestion by dynamic line rating and power flow models.
Constraints
- Checklists MUST distinguish clearly between physical grid events and telemetry transmission artifacts.
- The data specialist MUST NOT smooth out genuine tripping events or breaker operations.
- Every verification checkpoint must specify the target telemetry tag type (e.g., MW, MVAR, kV, Bus Frequency).
- Avoid including commercial power market settlement rules.
Output format
1. Timestamp and Protocol Alignment (3-4 markdown checklist items)
2. Transducer Signal Noise & Spike Scrubbing (4-5 markdown checklist items with mathematical limits)
3. Missing Value & Deadband Handling (3 markdown checklist items)
4. Reliability Engineering Gate Checks (3 checklist items with pass/fail criteria)
Self-review
- Did I calibrate the checklist around the specified protocol {{scada_protocol}} and asset {{substation_asset_id}}?
- Are physical fault detections clearly protected from aggressive noise filtering?
- Is the sampling rate properly reflected in the temporal verification steps?
Explicit role, a named task, and discrete steps the model can follow.
Background, inputs and variables the model needs before it starts.
Hard boundaries — what the model must and must not do.
A named, field-level shape for the response.
Ordered work items that force analysis before an answer.
Length and structure that travel across frontier models.
Signal density — instruction weight without padding.
Documented variables so the scaffold adapts to new inputs.
Quality bar, assumptions and behaviour when inputs are thin.
How much real usage the template has behind it.