Power Theft Detection with Tamper Alerts and Utility Dashboard

Distribution utilities lose a measurable share of energy to theft — direct tapping, meter bypassing and tamper — but the theft is usually discovered by inspection, months late. The engineering principle behind detection is energy accounting: what the feeder transformer supplies must equal what the meters record plus known technical losses; a persistent gap is non-technical loss. This project demonstrates that principle on a scaled feeder model: a feeder-side CT measures total supply, reference meters measure each consumer, and reed-switch tampers watch the meter enclosures. ESP32 nodes publish all of it over Wi-Fi via MQTT to a utility dashboard with feeder-vs-metered graphs, loss-percentage analytics and a tamper event log. The project is careful about claims: it demonstrates the detection method and the analytics, not a utility-grade revenue-protection system, and the report discusses technical vs non-technical loss honestly.

1. The feeder CT measures total supply current while each consumer branch passes through its own reference meter on the demonstration model.
2. ESP32 nodes sample feeder and consumer readings, compute RMS values, and check tamper-switch states on every cycle.
3. Every 10 seconds (design target) the nodes publish feeder power, per-consumer power and tamper states as JSON over Wi-Fi via MQTT.
4. The dashboard aggregates metered total, subtracts it from feeder supply, and plots the unaccounted-loss percentage over 24 hours.
5. When loss crosses the configured threshold or a tamper switch opens, the dashboard raises an alert and writes to the event log.
6. A switchable 'theft tap' on the model lets the demonstrator introduce an unmetered load and watch the loss analytics respond.
7. The buyer calibrates feeder and meter channels against a reference meter with the documented procedure.

Tech stack:

• ESP32 development boards (feeder + meter nodes)
• SCT-013 feeder current transformer
• Reference energy meter modules x6
• Reed-switch tamper sensors
• MQTT broker + web dashboard
• Arduino IDE (C/C++ firmware)
• Scaled feeder demonstration model
• Switchable theft-tap load (demo)

• [Feeder-vs-metered accounting] A feeder CT and per-consumer reference meters let the system compute unaccounted loss continuously — the core theft-detection quantity.
• [Tamper sensing] Reed switches on meter enclosures and CT-bypass detection wiring raise immediate tamper flags with the meter ID and timestamp.
• [Utility vigilance dashboard] Feeder-vs-metered 24-hour graphs, loss-percentage trend, per-consumer comparison and the tamper event log in one view.
• [Loss analytics] The dashboard separates the measured gap into a persistent-loss estimate with configurable thresholds and alert levels.
• [MQTT telemetry] Feeder and meter ESP32 nodes publish readings as JSON over Wi-Fi via MQTT with a design-target 10-second interval.
• [Patrol-ready event log] Tamper and high-loss events log with values and timestamps, exportable for the report's case-study section.
• [Scaled feeder model] A demonstration feeder with transformer model, consumer branches and switchable 'theft' taps makes the accounting visible and testable.

• Working feeder demonstration model (feeder CT, 6 consumer meters, tamper switches, theft tap)
• Complete firmware source (RMS metering, tamper checks, MQTT)
• Live utility dashboard (graphs, loss analytics, tamper log) demonstrated with the model
• Circuit and wiring documentation
• Feeder/meter calibration procedure against a reference meter
• Component list with ratings
• Project report PDF (distribution loss background, energy accounting, tamper methods, methodology)
• PPT presentation for final review
• Viva Q&A preparation document (CTs, technical vs non-technical loss, tamper techniques, MQTT)
• Setup and demonstration guide

• This is a detection-method demonstration on a scaled model — not a utility-grade revenue-protection system and not certified for billing disputes.
• Loss analytics need stable calibration of all channels; drift in any meter channel appears as apparent loss, which the manual discusses.
• The model demonstrates direct-tap and bypass scenarios; sophisticated tamper (meter firmware attacks) is out of scope.
• Approximately +/-3% channel accuracy (expected) sets the floor for the smallest detectable persistent loss.
• Telemetry needs Wi-Fi; local tamper indication (LED/buzzer) works without network.