ZENERGY 众壹能源
DIGITAL TWIN · Plant lattice

Module-levelvisual cockpit

Computer vision and image processing precisely recognize and map plant modules, linking inverters and strings to module visuals — turning generation data into intuitive area-and-color insight.

Open Monitor demoRequest demo
Modules
384
Views
03
Anomalies
06
scroll
SETUP · 3 steps

Three steps to a visual cockpit

From orthophoto upload to a fully bound module layout — three modules connected end-to-end.

  1. 01
    STEP 01 / 03
    Upload orthophoto

    Aerial orthophoto + map-tile package, geo-aligned and overlaid as the recognition base.

  2. 02
    STEP 02 / 03
    Generate recognition

    AI auto-recognition over the orthophoto, or DXF engineering drawing snapped to map coords.

  3. 03
    STEP 03 / 03
    Create view & link devices

    Build a default visual layout, then bind every component into inverter → MPPT → string.

BINDING · Component link

Visual component binding configuration

Recognized visual components are bound to inverter-tier devices, building the bridge between modules and live generation data — the engine behind cockpit visualization.

Device hierarchy tree
Device hierarchy tree
Auto recognition grid
Auto recognition grid
01
01 · AUTO-RECOGNIZE

Auto-recognize sub-flow

  • Auto-recognize

    Click [Auto-recognize] — the system reads the orthophoto and produces a new recognition record with timestamp and module count.

  • Result list

    Recognition results render as a parent/child tree with name, time, and module count. Edit and delete supported.

  • Visual editing

    Click a result to reveal its modules on the right. Add, multi-select, move, and delete operations are supported.

Module binding view
02
02 · VISUAL BINDING

Visual component binding

  • Enter binding view

    Click [Visual component binding] to open the default view's device-binding page. White borders denote bound components, yellow denotes unbound.

  • String binding

    Pick a string in the left tree, click [Set] to enter binding mode. Multi-select target modules, click [Bind] — green borders indicate components bound to the active string.

  • Count validation

    Live indicator compares bound count vs. configured target. Match = green check, mismatch = yellow warning.

VIEWS · Three data layers

Three data views
Full-spectrum plant insight

From inverter efficiency to string-level yield to robot inspection anomalies — module-level data insight from every angle, mapped onto your visual layout.

Inverter heatmap
VIEW 01
Inverter heatmap

Heatmap data is mapped onto the module view. Color depth reflects efficiency at each timestamp — scrub the timeline, replay, or full-screen to pinpoint underperforming areas.

String-level yield
VIEW 02
String-level yield

Daily per-string generation visualized — color intensity reflects output, surfacing anomalies and root causes at a glance.

Robot anomaly inspection
VIEW 03
Robot anomaly inspection

Anomaly imagery captured during cleaning is analyzed and pushed to the platform. Issue type plus exact module location, with one-click [Clear anomaly] to update status.

FAQ · Digital twin

Frequently asked questions

Scale, granularity, and deployment cycle.

01

How does the digital twin reconstruct a 1.4 GWp-scale PV asset view?

ZenovaOS digital twin uses a 3-step module-level modeling flow that scales to large fleets: 1. Orthophoto upload: owner provides satellite imagery or UAV-captured orthophoto as base layer. 2. AI / DXF auto-recognition: computer vision detects each panel's boundary and orientation, or imports precise coordinates from a DXF design file; a single plant can be modeled in minutes. 3. View ↔ device binding: detected modules are automatically associated with inverters, strings, and combiner boxes. The twin supports three data views — inverter heatmap (colored by generation), string unit-generation view (detects mismatch / shading), and robot inspection anomaly view (overlays PCR-detected hot spots, micro-cracks, soiling, breakage). Across the 1.40 GWp we manage today (860+ plants, 360+ counties), every plant sits in one unified cockpit.
02

How significant is module-level vs string-level twin?

Granularity directly governs precision of loss attribution and insurance claims. String level looks at the aggregated electrical data of dozens of bundled modules — single-panel hot spots, micro-cracks, and per-cell mismatches are averaged out. A module-level twin models each panel as an individual entity; combined with PCR robot AI vision, you can pinpoint specific anomalies like "row 3, column 7, module 5#, hot spot detected on 2026-04-12." O&M cost typically drops 30-50%, and claim evidence is far stronger.
03

How long does modeling take? Is on-site support required?

Standard timeline: 1-3 days from orthophoto receipt to twin live; the pure AI auto-recognition step takes only tens of minutes, the rest is human spot-checks. No on-site support needed — owners provide: (1) orthophoto base layer or DXF design file, (2) device inventory (inverter / string / combiner IDs). We handle all data processing.
GET STARTED · Go digital

Start with a single scene
Let AI enter the workflow

Best first plays: alarm triage, trend analysis, weekly/monthly auto-reports — all three deliver results within 14 days.

Request Demo Open Platform↗