From remote signal to field inspection: a solar O&M triage workflow

Effective Solar O&M starts before a truck is dispatched. The operational goal is to turn a detected symptom into the smallest defensible next step: validate the data, determine what can be checked remotely, estimate the value of further investigation, and give field personnel a concise evidence package when an inspection is justified.

1. Validate the alert before assigning work

Confirm that the affected interval is during meaningful expected production and that actual telemetry, weather inputs, and timestamps are usable. Review persistence and recurrence rather than escalating one unusually poor interval. A communications gap belongs in a data-recovery workflow until independent evidence supports a physical production problem.

• Verify the site, equipment boundary, timezone, and measurement cadence.
• Compare expected and actual output only across valid aligned intervals.
• Review inverter status, plant-controller limits, and recent maintenance context when available.
• Record supporting and contradicting evidence instead of preserving only the alert label.

2. Exhaust focused remote checks

Remote investigation should test the most plausible explanations without treating them as confirmed. Compare peer equipment, inspect alarms and status changes, verify communications freshness, and check whether curtailment or an export ceiling explains the observed shape. Document what was checked and which uncertainty remains.

3. Make the dispatch decision explicit

Compare the modeled avoidable impact with site-configured inspection cost, likely repair cost, persistence, and diagnostic confidence. A high-impact persistent outage may warrant urgent inspection, while a low-value derate with weak evidence may justify continued monitoring or additional remote work. The recommendation remains decision support, not an automatic work order.

4. Prepare a field-ready evidence package

1. State the observed symptom, affected period, severity, and confidence.
2. Include expected-versus-actual evidence and relevant data-quality limitations.
3. List the probable explanations to investigate without presenting them as confirmed.
4. Identify equipment boundaries, applicable drawings, known alarms, and safe-access requirements.
5. Define what evidence should be captured before and after any repair.

5. Close the loop with the field outcome

Record the confirmed cause, resolution action, repair cost when known, whether production was restored, resolver, and resolution timestamp. Preserve the original detector prediction. Comparing historical predictions with confirmed outcomes exposes where rules are useful, where data quality is limiting, and where future diagnostic logic needs review without automatically training a model.