Solar Panel IoT Monitoring System — Performance Analytics & Fault Detection
A string-level IoT monitoring system for a 5MW solar farm — tracking performance ratio, detecting underperforming strings, and integrating with inverters via Modbus — increasing annual energy yield by 8%.
⚠The Problem
The operator of a 5MW solar farm was monitoring performance at inverter level only — they could see total AC output but had no visibility into which strings or panels were underperforming. Shading, soiling, and module degradation caused string-level losses that were invisible until they accumulated to inverter-level impact. Annual energy yield was 8–12% below modeled values (PVSyst). Soiling detection was manual — operators scheduled cleaning based on calendar, not actual panel soiling.
💡Our Solution
We installed ESP32 Modbus gateways at each combiner box (16 strings per box, 20 combiner boxes total). Hall-effect current sensors measure current for each individual string. A pyranometer and module temperature sensor provide irradiance and temperature data for performance ratio calculation. Inverter Modbus data (AC power, MPPT voltage, efficiency) is also ingested. Lambda calculates performance ratio per string every 15 minutes, flagging strings below 85% of expected value. A soiling index is derived from irradiance vs. output degradation over time. The React heatmap shows every string color-coded — green (good), yellow (degraded), red (fault).
🔗System Architecture
String Sensors (Modbus) → ESP32 Gateways → MQTT → AWS IoT Core → InfluxDB → Lambda (PR calculation) → React DashboardTech Stack
- ESP32 Modbus gateway per combiner box
- Hall-effect current sensor per string (20A)
- Voltage divider string voltage measurement
- Pyranometer (irradiance)
- Ambient + module temperature sensor
- SMA / Fronius / Huawei inverter Modbus interface
- Modbus RS485 (string to gateway)
- MQTT to cloud via site Wi-Fi / 4G
- Inverter Modbus integration
- AWS IoT Core
- InfluxDB (string-level time-series)
- Lambda (performance ratio + soiling calculation)
- S3 (daily performance reports)
- React solar farm dashboard
- String-level heatmap (color-coded performance)
- Inverter MPPT efficiency chart
- Automated O&M report generator
Key Features
Results Delivered
- 8% increase in annual energy yield from fault detection + soiling management
- 320 strings monitored across 20 combiner boxes
- 15 underperforming strings identified in first month (shading and soiling)
- Cleaning schedule optimized: 6 events/year → 3 targeted cleanings
- Fault MTTR (mean time to repair) reduced from 5 days to 18 hours
Technologies
Who This Is For
Solar farm operators, EPC contractors, O&M service providers, rooftop solar installers
Need a similar solution?
We've built production IoT systems like this across 15+ countries. Let's talk about yours.
Related Services
Frequently Asked Questions
What technology does this solar panel iot monitoring system use?+
Solar Panel IoT Monitoring System — Performance Analytics & Fault Detection is built on ESP32, Modbus, MQTT, React, InfluxDB. The stack was chosen for real-world reliability in a production deployment — full details are in the tech stack section.
Can CodeCaracal build a similar system for us?+
Yes. The same team that designed and deployed this renewable energy / solar system takes on new projects across the stack — firmware, connectivity, cloud and dashboards. Tell us your requirements and we will architect a solution for your environment.
Is this case study a real deployed system?+
Yes — every case study on this site describes a production system we designed and shipped. The problem, architecture, features and metrics are from the actual deployment.
Related Engineering Guides
Deep technical write-ups from the team behind this system.
More Projects
Smart Energy Metering IoT System — Sub-Metering for Commercial Buildings
LoRaWAN Livestock Tracking System — Cow Monitoring with GPS & Health Sensors
LoRa Fish Tracking System — Aquaculture Monitoring in Open Water
Got an IoT challenge?
We've shipped it.
Whether you need a fleet to track, a factory to monitor, or a farm to automate — our team has done it before and we'd love to build it with you. Typical response time: under 24 hours.