Thermal drone inspection · panel-level digital twin

Fully autonomous solar thermal inspection and asset management platform.

Recover your production loss.

Every panel is measured by thermal drone, and every anomaly enters your plant's panel-level digital twin with its place and priority. Maintenance, production loss and panel history are managed on one record.

How the platform works ↓

  • 100% panel coverage
  • 30 m flight altitude
  • ≥ 5×5 px resolution per cell
  • RGB + thermal two channels
  • 17 anomaly classes
  • 24 hours to a quote
  • One flight for the whole site and all panels
  • 100% developed in Türkiye

Catch anomalies while they are small.

In a single flight, every panel in your plant is imaged with a thermal camera from dozens of angles and measured on an AI-assisted processing pipeline. The type, priority and estimated annual production loss are determined for each finding. Monitoring systems cannot provide this, because they measure at inverter or string level. Panel-level losses can go unnoticed inside those totals for years.

Anomalies detected early can be resolved before they grow, which prevents the loss from extending over years.

FLIGHT

How the digital twin is built

The route covers the whole site

The route is planned over the site boundary with lines crossing the panel rows, and it covers the whole plant in a single flight. A flight altitude of 30 m gives the resolution IEC TS 62446-3 requires.

RGB orthomosaic of an entire solar plant. Rows and tables in visible light.
FLIGHT

01

The frames from the flight are stitched into one map

The frames from the flight are stitched into a single map covering the whole plant. The location and identity of each panel on the site are determined from this map.

RGB orthomosaic of an entire solar plant. Rows and tables in visible light.
RGB ORTHOMOSAIC · 30 m · Visible-light map

02

The whole plant is shown in a single thermal map

The thermal map is aligned on the same pixel grid as the visible-light map, so temperature differences can be read directly on the map.

Thermal orthomosaic of the same plant. The whole site in one map, with temperature differences distinguishable at module level.
THERMAL ORTHOMOSAIC · Thermal map on the same pixel grid

03

AI detects the panels

Image segmentation detects the modules in the plant one by one. Each module's outline is recorded together with its row, table, tier and column position.

An anonymised region from a real measurement of the 8 MW demo site. 1,236 modules detected, 95 flagged.

The thermal image of a few rows of the same plant. Detected module outlines and the flagged outliers over the region.
ROWS · DETECTION · Detected module outlines

04

Measurement flags the anomalies

Each module's temperature difference, normalised to irradiance, is compared with the neighbouring panels, and panels that show a clear deviation are flagged.

An anonymised region from a real measurement of the 8 MW demo site. 1,236 modules detected, 95 flagged.

The thermal image of a few rows of the same plant. Detected module outlines and the flagged outliers over the region.

ΔT · normalised to irradiance · K

−2.1+68.8

Flagged

ROWS · ΔT · Temperature difference normalised to irradiance, with flagged deviations

05

Model names the anomaly type

A classification model assigns each finding to one of the 17 anomaly classes and gives it a priority.

The thermal image of a few rows of the same plant. Detected module outlines and the flagged outliers over the region.

ΔT · normalised to irradiance · K

−2.1+68.8

Flagged

  1. 1+51.0 K · DIODE ×3
  2. 2+48.7 K · DIODE ×3
  3. 3+68.8 K · PHYSICAL DAMAGE
  4. 4+8.7 K · STRING FAULT ×12
  5. 5+16.1 K · MULTIPLE DIODES ×2
  6. 6+15.6 K · DIODE ×2
ROWS · CLASS · Anomaly types and ΔT labels
Thermal cut of the same table. Module-level temperature differences on the same pixel grid.Visible-light cut of one mounting table. 2×12 modules, with the neighbouring tables.
ORTHOMOSAIC · LOCATION · The finding's location on the orthomosaic

The finding's location on the orthomosaic

Physical damage. Thermal image from a real inspection, peak temperature difference against its neighbours +68.8 K.
RAW THERMAL FRAME · FULL RESOLUTION
Physical damage. Visible-light image of the same module.
RAW RGB FRAME · SEPARATE SHOT

A finding is verified when it measures consistently across several frames that see the same module from different angles.

RAW FRAMES · THERMAL + RGB · The same module, examined in the drone's raw frames at full resolution

The same module, examined in the drone's raw frames at full resolution

06

A separate page is prepared for every finding

The page holds the finding's position, its RGB and thermal close-ups, its class, the number of modules affected and its estimated annual loss. Every finding is verified across several images.

A finding page from the report. The panel's position, RGB and thermal close-ups, the measured temperature difference, its class and its estimated loss.
FINDING PAGE · A finding page from the same inspection

07

The report is prepared after the flight

After the flight, an inspection report is prepared and the same findings are added to your panel record on the platform. Because every inspection is recorded against the same panel identities, a temperature history is built for each panel. We recommend one inspection a year.

Sample report · PDF · 50 pages50 pages of a 315-page reportThe sample report contains the executive summary, inspection conditions, method, loss and revenue analysis, site observations, findings maps, the complete anomaly index and the finding cards. The finding cards cover all findings at the highest priority and examples at high, medium and low priority.

Enter your email. The download link opens at once and stays valid for 7 days. Your address is used only for this request and our reply.

The first two pages of a delivered inspection report, cover and executive summary.
INSPECTION REPORT · The opening pages of a delivered report

Platform modules

What you manage on the platform

Every step of the maintenance process, from detection to resolution, is managed on one platform. The SolarDetech platform gives every panel in your solar power plant a persistent identity and updates the panel record with every flight. All of the modules below work on that same record.

  • Portfolio management

    The fleet map, last inspection dates, open findings and priority breakdown of all your plants are shown on one screen.

  • Panel-level digital twin

    Each panel's persistent identity, geographic position (block, row, table), electrical position (transformer, inverter, combiner, string) and measurement history are kept. The record is updated with every flight.

  • Autonomous thermal inspection

    The inspection provides 100% panel coverage. Every panel in the plant is imaged, measured with a thermal camera and recorded under its persistent identity. Orthomosaic, panel segmentation and anomaly detection run without human intervention. The model assigns each anomaly to one of 17 classes, and an expert confirms the class.

  • Work and maintenance management

    Every anomaly is listed with its probable root cause. A work order opened from a finding is assigned, tracked and closed. The field team is guided on the map to the relevant panel. The panel's status is tracked from anomaly to repair, and a resolved anomaly is closed on the same record at the next inspection.

  • Serial number records

    Each panel's serial number is entered into its record by scanning the label with a camera (OCR) and can be corrected manually if needed. For warranty claims and panel replacements, the serial number is available together with the panel's position and measurement history.

  • Inventory management

    Panels, inverters and combiner boxes are kept in one inventory with their serial numbers and positions. Spare parts and stock levels are tracked. The history of replaced parts is kept on the record.

  • Warranty claims

    The serial number, measurement history and raw frames are gathered into one claim file. The verification process with the manufacturer is managed on the platform.

  • Production loss and efficiency

    The estimated production loss of every anomaly is calculated. Plant efficiency is tracked by period and by anomaly group.

  • Performance and trends

    The yield and findings of the plants in a portfolio are monitored comparatively. The degradation trend and the estimated loss for the coming period are calculated from the panel history.

  • Fire risk

    Junction box heating, string faults with arc traces and advanced hot spots can carry fire risk. These anomalies go to the top of the list with very high priority.

  • History and comparison

    Measurements of the same panel from different inspections are displayed side by side. This allows comparison after a storm, before and after a warranty claim, and between consecutive flights.

  • Measurement conditions

    The irradiance and weather of every flight are recorded and documented in the report.

  • Teams and roles

    Teams are defined per plant, with separate roles for the field team, the operator and the investor.

  • Integration

    An API is available for partners. You can add your own users to plant teams and create one-click sign-in links for them.

More capabilities

Planning and field
Periodic inspection planning, Flight plan generation, Mobile use for field teams, Notifications and alerts
Maintenance
Spare parts and stock tracking, Cleaning planning, Vegetation and shade management
Integration and reporting
SCADA and inverter data integration, Data API, Export, Multilingual reports, Insurance report, Finance and investor report
Inspection types
Commissioning and acceptance inspection, Pre-warranty-expiry inspection, Rapid damage inspection after storms and hail
Solution partner services (optional)
I-V curve and insulation measurement, Maintenance and panel replacement
All capabilities

Explore the platform Request a demo account

Who it is for

The same panel record for every stakeholder

The platform is used by investors, EPC contractors, O&M teams and panel manufacturers through the same panel-level digital twin. Each user group finds the information relevant to its own responsibilities in that record.

  • Investor and plant owner

    • You can monitor the last inspection date, open findings and priority breakdown of every plant in your portfolio on one screen.
    • You can audit the work of your EPC and O&M providers with independent, panel-level measurements.
    • You can identify anomalies that cause production loss or may carry fire risk at an early stage. Anomalies that grow between consecutive inspections are highlighted.
    • You can support warranty claims with the serial number, measurement history and raw frames.
  • EPC contractor

    • Every panel is measured before commissioning (acceptance) and at the end of the warranty period.
    • Installation and workmanship anomalies, such as physical damage, reversed polarity, junction box heating and missing panels, are documented with the panel address.
    • Panel serial numbers are entered into the record and kept on the same panel record after handover.
    • Acceptance and end-of-warranty measurements are displayed side by side on the same panel identity.
  • O&M provider

    • Findings are presented as a work list sorted by priority and estimated loss. The probable root cause is stated for every anomaly.
    • A work order opened from a finding is assigned to a team member and tracked until close-out.
    • The field team is guided on the map to the relevant panel by block, row, table and panel.
    • The repair is closed on the same panel record at the next inspection, and the panel's history is kept on the record.
  • Panel manufacturer

    • Anomaly types that recur under real operating conditions in the field can be monitored.
    • Warranty claims are submitted with the panel's serial number, position, dated measurements and raw frames.
    • The claim verification process is managed on the platform.

Start on one of three paths

The platform is free. The inspection is priced by quote for your site.

  1. With your own drone

    The platform is hardware-agnostic. Your own team carries out the flight with your RTK-equipped drone and radiometric thermal camera, under the conditions we set in line with IEC TS 62446-3. Once approved, you receive an upload link.

    Start with your own drone
  2. Our professional field team flies

    The flight is carried out by our professional field team. The details are set out in a quote prepared for your site.

    Request a quote
  3. With your existing data

    If you have previously captured thermal and RGB data, write to us. We will assess whether the data is suitable.

    Write to us

The autonomous line that feeds the panel record

The data from every flight is processed in this pipeline and transferred to your panel record. Once the data is uploaded to the platform, the orthomosaic, segmentation and anomaly detection run without human intervention. An expert confirms the class of every anomaly.

  1. Data upload

    automatic

    The thermal and RGB flight frames are uploaded to the platform.

    Time about 2 hours

  2. Orthomosaic

    automatic

    The flight frames are merged into a single map.

    Time about 6 hours

  3. Segmentation

    automatic

    Panel outlines are extracted and every panel gets a permanent identity.

    Time about 1 hour

  4. Classification

    automatic

    The model sorts anomalies into 17 classes and prioritises them. An expert confirms every class.

    Time about 2 hours

  5. Report

    automatic

    The report and the panel-level platform record are prepared from the same data.

    Time about 30 minutes

The times shown are approximate values for a 10 MW plant. The data passes an acceptance check before processing. The report is delivered within at most 5 business days of the data entering the system.

Explore the platform

8 MW demo site · anonymised data · read-only

Explore the measured region yourself

An anonymised region from a real measurement of the 8 MW demo site. 1,236 modules detected, 95 flagged.

Detected module outlines
1,236
Flagged
95

Open the map

95 flagged

  • Very high
  • High
  • Medium
  • No priority assigned
  • Healthy
The thermal image of a few rows of the same plant. Detected module outlines and the flagged outliers over the region.

What we detect

Each of the seventeen anomaly classes is shown with a real cut from a delivered inspection. Priority is set on the basis of safety risk as well as production loss.

Detection channel

17 classes shown

  1. Single hot cell. Thermal image from a real inspection, peak temperature difference against its neighbours +4.9 K.
    THERMAL
    ΔT +4.9 K

    Single hot cell

    Cell

    Channel
    Thermal
    Priority
    Medium → Very high
  2. Hot spot. Thermal image from a real inspection, peak temperature difference against its neighbours +18.4 K.
    THERMAL
    ΔT +18.4 K

    Hot spot

    Hot-Spot

    Channel
    Thermal
    Priority
    Medium → Very high
  3. Multiple hot cells. Thermal image from a real inspection, peak temperature difference against its neighbours +31.3 K.
    THERMAL
    ΔT +31.3 K

    Multiple hot cells

    Cell-Multi

    Channel
    Thermal
    Priority
    Medium → Very high
  4. Single diode / substring. Thermal image from a real inspection, peak temperature difference against its neighbours +64.2 K.
    THERMAL
    ΔT +64.2 K

    Single diode / substring

    Diode

    Channel
    Thermal
    Priority
    Medium → High
  5. Multiple diodes. Thermal image from a real inspection, peak temperature difference against its neighbours +17.4 K.
    THERMAL
    ΔT +17.4 K

    Multiple diodes

    Diode-Multi

    Channel
    Thermal
    Priority
    High
  6. Physical damage. Real images from delivered inspections.Physical damage. Visible-light image of the same module.

    Physical damage

    Physical-Destruction

    Channel
    RGB + thermal
    Priority
    Very high
  7. Offline module. Thermal image from a real inspection, peak temperature difference against its neighbours +13.1 K.
    THERMAL
    ΔT +13.1 K

    Offline module

    Offline-Module

    Channel
    Thermal
    Priority
    High
  8. String fault (3 or more consecutive modules). Thermal image from a real inspection, peak temperature difference against its neighbours +18.3 K.
    THERMAL
    ΔT +18.3 K

    String fault (3 or more consecutive modules)

    String-Fault

    Channel
    Thermal
    Priority
    High · very high if arcing
  9. Junction box / terminal. Thermal image from a real inspection, peak temperature difference against its neighbours +17.9 K.
    THERMAL
    ΔT +17.9 K

    Junction box / terminal

    Junction-Box

    Channel
    Thermal
    Priority
    Very high
  10. Reversed polarity. Thermal image from a real inspection, peak temperature difference against its neighbours +14.6 K.
    THERMAL
    ΔT +14.6 K

    Reversed polarity

    Reversed-Polarity

    Channel
    Thermal
    Priority
    Very high
  11. Glass breakage. Thermal image from a real inspection, peak temperature difference against its neighbours +64.0 K.Glass breakage. Visible-light image of the same module.
    ΔT +64.0 K

    Glass breakage

    Glass-Breakage

    Channel
    RGB
    Priority
    Very high
  12. Missing / detached module. Thermal image from a real inspection, peak temperature difference against its neighbours +2.6 K.Missing / detached module. Visible-light image of the same module.
    ΔT +2.6 K

    Missing / detached module

    Missing-Module

    Channel
    RGB
    Priority
    High · very high if hanging
  13. Soiling. Thermal image from a real inspection, a whole table with its neighbours.Soiling. Visible-light image of the same module.

    Soiling

    Soiling

    Channel
    RGB
    Priority
    Low
  14. Bird droppings. Thermal image from a real inspection, a whole table with its neighbours.Bird droppings. Visible-light image of the same module.

    Bird droppings

    Bird-Droppings

    Channel
    RGB ≤1 cm/px
    Priority
    Low
  15. Biological growth. Thermal image from a real inspection, peak temperature difference against its neighbours +3.8 K.Biological growth. Visible-light image of the same module.
    ΔT +3.8 K

    Biological growth

    Biological-Growth

    Channel
    RGB ≤1 cm/px
    Priority
    Low
  16. Vegetation shading. Thermal image from a real inspection, peak temperature difference against its neighbours +4.2 K.Vegetation shading. Visible-light image of the same module.
    ΔT +4.2 K

    Vegetation shading

    Vegetation

    Channel
    RGB + thermal
    Priority
    Low
  17. Structural shading. Thermal image from a real inspection, a whole table with its neighbours.Structural shading. Visible-light image of the same module.

    Structural shading

    Shadowing

    Channel
    RGB + geometry
    Priority
    Low

Priority is set on the basis of safety risk as well as production loss. A burnt junction box and reversed polarity are rated very high even where the loss is small. Classes that cannot be told apart with confidence from the image are flagged with a low confidence value.

Real images from delivered inspections.

All classes and detection channels

What is delivered after a flight

For every anomaly the report gives the row, table and panel, its type and its priority. Your plant's panel-level platform record is delivered together with the report.

  1. The Anomalies screen of the anonymised 8 MW demo site on the SolarDetech platform: the selected finding's card and the prioritised anomaly list. Panel IDs, totals and the panel model are blurred.

    17 anomaly classes

    Prioritised anomaly list

    Findings are listed with their type and priority, together with their probable root causes. Every finding is verified across several images.

  2. The Efficiency screen of the anonymised 8 MW demo site on the SolarDetech platform: the monthly production loss chart. All figures are blurred.

    Per finding

    Production loss analysis

    The estimated annual production loss is calculated for every finding.

  3. One report per flight

    Inspection report

    The report consists of a summary section for investors and operators and the technical finding list.

  4. The map screen of the anonymised 8 MW demo site on the SolarDetech platform: panels coloured by priority over the thermal orthomosaic. Totals are blurred.

    Portfolio · map · panel record

    Panel-level digital twin

    Every panel is recorded on the platform under its own identity. You can view findings on the map, convert them into work orders, assign them to your team, track their status and record their close-out. Every inspection is added to the same panel record, and a resolved anomaly is closed at the next inspection.

  5. The whole site

    Orthomosaic mapping

    The whole plant is stitched into a single map.

  6. The Panel Details drawer of the anonymised 8 MW demo site on the SolarDetech platform: measured temperatures, anomaly type and priority. Panel ID, position and model are blurred.

    Platform

    Panel-level record

    The position, measured temperature difference and deviation from neighbouring panels are recorded for every panel.

Panel record on the platform

Your request is forwarded to us by email, and an invitation to the sample site is sent within one business day.

Log in to the platform

The full scope and the flight standard

SCADA monitors the plant's total output. The SolarDetech platform measures each panel individually. The two systems complement each other.

Comparison of methods

Every finding in the report has a known basis

The same three rules apply to every inspection. No figure appears in the report without a stated source, and findings based on a single frame are not included.

  1. The thermal image of a few rows of the same plant. Detected module outlines and the flagged outliers over the region.
    ROWS · ΔT

    Based on measurement

    A panel is not counted as anomalous unless it shows a measurable temperature difference from its neighbours. Measurements are normalised to irradiance.

  2. Physical damage. Thermal image from a real inspection, peak temperature difference against its neighbours +68.8 K.Physical damage. Visible-light image of the same module.
    SAME MODULE, SEPARATE FRAME

    Verified across many images

    Every panel is assessed from many images taken from different angles. A finding is included in the report only when it is measured consistently in more than one image.

  3. A finding page from the report. The panel's position, RGB and thermal close-ups, the measured temperature difference, its class and its estimated loss.
    REPORT · FINDING PAGE

    Every finding has a recorded source

    For every finding in the report, the image and measurement it is based on are recorded. Where needed, the panel's temperature data can be examined down to pixel level.

IEC TS 62446-3

Flown, measured and reported to IEC TS 62446-3.

This statement does not imply certification. Measurement conditions are documented on every flight.

Standards

Common questions about solar thermal inspection

All eighteen questions and their answers
What is SolarDetech?

The SolarDetech platform is a panel-level digital twin and asset management platform for solar plants. Its thermal inspection is fully autonomous.

The platform's modules all work on the same record: portfolio management, a panel-level digital twin, autonomous thermal inspection, work and maintenance management, serial number records, inventory management, warranty claims, production loss and efficiency, fire risk, history and comparison, measurement conditions, teams and roles, integration. In the thermal inspection, the orthomosaic, the panel segmentation and the anomaly detection are produced from thermal and RGB drone data without human intervention. The model assigns each anomaly one of 17 classes. An expert confirms the class of every anomaly in the report. Findings are shown on the map, ordered by priority and converted into work orders. The platform is free. The inspection is priced by quote for your site. The flight is carried out by our professional field team or by your own team with a drone that has RTK and a radiometric thermal camera. Measurement and classification are carried out in line with the IEC TS 62446-3 methodology.

What is a digital twin, and what does it cover on the SolarDetech platform?

On the SolarDetech platform the digital twin is panel-level. Every panel in your plant keeps its persistent identity, its position and its measurement history in one record.

Each panel is assigned a persistent identity that does not change from flight to flight. Its geographic position is recorded as block, row and table, and its electrical position as transformer, inverter, combiner and string. The measurement of every flight is added to the same panel identity, which builds the panel's measurement history. The digital twin is this record, updated with every flight. Findings, work orders and close-outs are linked to it. When a finding is opened, the panel's location on the map and the inverter and string it is connected to are shown on the same screen. Whether an anomaly has grown, and whether it has closed after repair, is also tracked on this record.

What is solar thermal inspection and how is it done?

Every panel in your plant is measured with a thermal camera, and every anomaly detected is listed with its type and location.

A drone with thermal and RGB cameras images the whole site from an altitude of 30 m in a single flight. The frames are stitched into a single map, and AI detects the panels one by one. Each panel's temperature difference is normalised to irradiance and compared with the neighbouring panels, and panels that show a clear deviation are flagged. The anomaly type of each flagged panel is determined by a classification model, and every finding is verified across several images. At the end of the inspection a report is delivered and the findings are added to your panel record on the platform. For every anomaly the report gives the row, table and panel, its type, its priority and its estimated annual loss. Flight, measurement, classification and reporting are carried out in line with the IEC TS 62446-3 methodology.

Is the platform free?

Yes, the platform is free. Only the inspection is charged.

There is no charge for platform access, and only the inspection is charged. The inspection is priced by quote for your site. The quote arrives within 24 hours at most. The report and the platform record are included in the inspection. If you have a drone with RTK and a thermal camera, your own team can carry out the flight. Your panel record, your map and your earlier inspections are kept on the platform.

How is the price calculated?

Every plant gets its own quote. The quote arrives within 24 hours at most.

The platform is free. The inspection is priced by quote for your site. The quote is prepared from the plant's panel count and location. Classified and prioritised findings, the inspection report and the platform record are included in the inspection. If our field team carries out the flight, the number of flight days is also stated in the quote. If you have a drone with RTK and a thermal camera, your own team can carry out the flight. To receive a quote, you only need to provide the panel count and the plant's province. The quote arrives within 24 hours at most. By the industry average, a plant loses 5.77% of its production a year, and most of this loss is concentrated in a small number of high-priority findings. This is why a prioritised report matters.

Which provinces do you cover?

Our field team carries out flights across Türkiye on ground-mounted and rooftop plants. For plants abroad, the platform can be used by uploading the data.

We work at sites in all provinces, particularly Konya, Karaman, Şanlıurfa, Ankara, Kayseri and Niğde, where installed capacity is concentrated. The number of flight days depends on the size of the site and is stated in the quote. If you have an RTK-equipped drone with a thermal camera, your own team can also carry out the flight. Data from plants abroad is processed as well. In that case the flight is carried out by your own team, and the data is uploaded through the link we send after approval. In every case the data is collected to the same standard and passes the same acceptance check.

Every panel measured, every anomaly verified.

  1. Request

    The panel count and province are sufficient. We reply to your request within 24 hours.

  2. Quote

    The platform is free. The inspection is priced by quote for your site. If our field team flies, the number of flight days is also written in the quote.

  3. Flight

    The flight is carried out from 30 m with RGB and thermal cameras by our field team or your own team, and the data passes the acceptance check.

  4. Report

    A prioritised finding list, a loss analysis and a panel-level platform record are delivered within 5 business days of the data entering the system.

Request a quote