Keep it clean+Keep it cool=increased profitability
Integrated Solar Systems is building patent-pending, condition-based cleaning for utility-scale solar farms. Our machine-learning approach is designed to decide what to clean, when to clean it, and how — recovering yield that fixed schedules leave on the panel.
Patent Pending in the United States.
A pending U.S. patent covers our machine-learning control of nozzle output and boom carriage movement. To our knowledge, the only such filing in condition-based panel cleaning.
Condition-based, not calendar-based.
The model is designed to decide per zone, per panel, per fluid, per pass — the boom acting only where the data says it should. Uniform cleaning gives way to intentional cleaning.
Verification by design.
Onboard cameras are built to re-scan each pass and confirm the clean before moving on, so every decision can be logged, every outcome audited, every cycle recorded.
How the system is designed to work.
Integrated Solar Systems is developing a condition-based cleaning system built around a predictive model. Here is the thinking behind it — and how we hope to prove it with early partners.
Condition-based soiling model
The model is designed to weigh losses against weather, latitude, and panel tilt, so cleaning is driven by actual soiling rather than a fixed calendar — with the goal of a projected yield delta in $/MW.
The cleaning system
The patent-pending boom + nozzle system is being engineered to work on existing arrays — multi-fluid, condition-based, and autonomously scanned, with every pass recorded.
Pilot partnerships
We're inviting a small group of operators to validate the system on a representative section of their array, with results measured against their current cleaning regime.
Built on a patent, engineered for the field.
Our approach is informed by direct observation of how panels actually behave when the dust falls — across the deserts, coasts, and snow lines the system is built to handle.
U.S. Patent Filed
Our control method for nozzle output and boom carriage movement is the subject of a U.S. patent under Attorney Docket 20859/160410.
Decisions per zone
The boom is designed to clean only where the data indicates a real yield gain — replacing calendar cleaning with intentional, model-driven action.
Right fluid, right pressure
Channels for water, deionized rinse, and a cooling-pulse regime, each designed to be governed by the model and triggered by panel state.
Every cycle on record
Onboard cameras are built to verify each pass, so every cleaning decision can be logged and reported.
It all reduces to one equation.
The innovation behind every cycle.
Our cleaning system is covered by a pending U.S. patent for the use of machine learning to control nozzle output and boom carriage movement in solar-panel cleaning. To our knowledge, it is the only such filing in the field of condition-based cleaning.
A machine-learning method for solar-panel cleaning.
Our patent-pending method is designed to control the cleaning system, panel by panel, using a predictive model rather than a fixed schedule.
- 01
Inputs the field actually gives you.
GPS, weather, image analysis, panel power output, cleaning regime, timers, manual input, and historical data.
- 02
A model that forecasts soiling losses.
The system predicts which zones lose the most yield next, and how much fluid and pressure will recover it.
- 03
Machine-learned activation of nozzles and boom.
Outputs are computed per zone, per pass: which nozzle fires, at what pressure, while the boom carriage moves at the chosen speed.
Three steps from soiling to spotless.
Every cleaning cycle the system is built to run follows a tight loop: sense the soiling, decide the right response, act on the panel, then verify it worked — with the model learning from every pass.
Sense.
Onboard cameras and live panel-output data are designed to map soiling and surface temperature across every zone of your array.
Decide.
The predictive model forecasts soiling loss and ranks zones by ROI — cleaning some today, queuing others for tomorrow, skipping many entirely.
Act & verify.
The boom is built to drive the carriage, open the right nozzle with the right fluid at the right pressure, then re-scan to confirm the clean.
Designed for high-soiling environments.
The system is built for the climates that lose operators the most yield: dust-laden air, infrequent rain, hard water, long daylight cycles. Every input the patent describes is designed to be sampled in those conditions.
One controller, three regimes.
From the patent itself: the model is designed to select between cooling water spray, light spray density, and heavy spray density — per zone, per panel, per cycle, using machine-learned activations of nozzle output and boom carriage movement.
- CoolingPulsed spray when panel temperature crosses a predetermined threshold.
- LightLow-density spray for routine condition-based passes.
- HeavyHigh-density spray for accumulated, high-loss soiling.
Let's talk about your farm.
We're looking for early partners to pilot the system. Tell us about your site and we'll start a conversation about working together.