When the Plant Goes Unattended: Who Still Looks Inside the Modules?

ⓘ This article is currently published in Chinese. An English version is being prepared — read the Chinese original.

Remote-operated, minimally-staffed solar plants are spreading fast across China's high-altitude and desert bases, where thin air, long winters and vast distances make permanent on-site crews expensive and slow to deploy. Remote operation works well for everything that leaves an electrical trace — inverter faults, string-level current deviation, tracker errors, communication dropouts — all of which travel over fibre to a central control room. What it quietly removes is the human eye that used to walk the array. That matters because the most consequential module-level damage, cracked cells and broken fingers, produces almost no electrical signature in its early life: the module still delivers, the string still matches its neighbours, and by the time the loss is large enough to surface in monitoring data the defect has usually matured into a hotspot or a measurable power deficit. This article maps that structural blind spot and proposes a practical way to close it without re-staffing the site: schedule electroluminescence imaging around four anchors — a commissioning baseline, post-extreme-weather verification, a periodic sampling cycle, and alarm-driven localisation — and choose an inspection format that matches a minimally-staffed operation. Drone-borne daytime EL fits that constraint better than per-module handheld work, because it needs no outage, no per-module wiring and few people on site. The piece closes with field notes specific to high-altitude plants — cold-weather power behaviour, wind and irradiance windows, payload and endurance derating with altitude — and argues that an EL result only completes the loop when it is written back into the asset ledger against a specific module ID.