AI energy insights: anomalies, peak demand and savings explained

What the Insights page produces

Every insight combines a measured signal, a threshold, and a financial framing. Instead of a wall of readings, the console reports what changed, how confident the detection is, and what recovering it is worth per month at your tariff.

• Load anomaly drift — z-score scanning across 30 days of channel energy to catch spikes and step-changes.
• Off-hours waste — energy consumed outside the operating schedule, priced with time-of-use rates.
• Peak demand risk — rolling 15-minute demand versus the billing-period maximum, with pre-alarms.
• Power quality — power factor, phase imbalance and harmonic distortion trends.
• Carbon impact — CO₂e converted from measured kWh using grid emission factors.

Choosing an analysis window

Insights can be generated over one hour, one day, one week, one month, or the full retained buffer. Short windows expose process-level events; weekly and monthly windows separate a genuine efficiency change from normal weekday-to-weekend variation. Weather normalisation should be applied before comparing months.

Milesight sensors and smart parking sites

Parking structures and mixed-use sites see load driven by occupancy. Pairing Milesight LoRaWAN occupancy and environmental sensors with electrical channels makes it possible to correlate vehicle counts with lighting, ventilation and EV-charging demand, then schedule those loads against real utilisation instead of a fixed timer.

Frequently asked questions

What is an energy anomaly?

An energy anomaly is a reading that deviates significantly from the expected load pattern for that time of day. The console computes a rolling mean and standard deviation for each channel and flags samples whose z-score exceeds 2, which usually indicates equipment left running, a failing motor, or an unexpected process start.

How is off-hours waste calculated?

Off-hours waste is the energy consumed outside your configured operating schedule. The dashboard multiplies the average off-hours load by the number of closed hours in the period and prices it with your time-of-use tariff to produce a monthly cost estimate.

What is a peak demand pre-alarm?

Utilities bill demand on the highest rolling 15-minute average in the billing period. A pre-alarm fires when the current rolling window is trending toward a new maximum, giving operators time to shed load before the peak is recorded.

Why does power factor matter?

A low power factor means more current is drawn for the same useful work, increasing losses and often triggering utility penalties. The console tracks power factor per phase and flags sustained readings below 0.9 alongside the estimated correction benefit.

Can the dashboard monitor Milesight LoRaWAN and parking sensors?

Yes. Milesight LoRaWAN devices such as occupancy, parking and environmental sensors can be tracked next to electrical channels, so occupancy patterns can be correlated with lighting, ventilation and EV-charging load.

How accurate are the savings estimates?

Savings estimates use measured energy, your configured tariff rates and conservative recovery factors per insight type. They are directional planning figures, not guaranteed results; an IPMVP measurement-and-verification baseline is available for formal reporting.