Module alert threshold: false alerts and fault detection

Compares each module with the median of its peers once a day. All noise figures are your assumptions, not vendor figures. Use a point or a comma for decimals; write whole numbers without a decimal part.

Modules in the array

Noise, module-day vs peers (% sd)

Alert if below peers by (%)

Days in a row required

Days checked per month

Missing readings (% of module-days)

Fault: module shortfall vs peers (%)

The default 1.803% (about 1.8%) combines 1.0% reading error and 1.5% day-to-day module difference (assumptions). Days are treated as independent, so real false-alert rates will be higher when weather, snow or shade hit several modules together.