About  ·  Engineering Story

We were tired
of cleaning panels
the wrong way.

Integrated Solar Systems is a solar technology company. We're developing a patent-pending cleaning system for utility-scale solar arrays. Our thinking started in the field — where every kilowatt-hour earned during the day depends on the dust that fell the night before.

01  ·  The Mission

Solve the high cost
of manual cleaning.

A solar farm is a financial instrument. Dust is the spread that eats the yield. Manual cleaning is the maintenance bill that erases the spread.

Most utility-scale farms treat cleaning as scheduled maintenance, a fixed monthly cost regardless of weather, season, or local soiling rate. The crews arrive. They scrub the whole array. The water tank empties. Nobody knows whether half those panels needed the visit.

We spent time with cleaning crews in the field while the patent first took shape. The simplest insight in this whole company is that soiling is non-uniform. Cleaning, until now, has always been uniform.

Our system is designed to invert that. The model decides per zone, per panel, per fluid, per pass. The boom does only what the model asks. The result is meant to be cleaner glass, smaller water bills, fewer cycles, and a software trail that proves it.

What the patent covers
  • Machine-learned control
    A predictive model decides what to clean, when, and with which regime.
  • Per-zone activation
    Nozzle output and boom carriage movement are controlled independently for each section of the array.
  • Three spray regimes
    Cooling pulse, light density, and heavy density, selected by the model per panel and per cycle.
  • Field-trained model
    The system is built to collect field data on every input and improve its decisions over time.
03  ·  Patent Transparency

The work we filed,
in plain language.

Trust comes from being specific. Below are the patent claims that underpin our system: what we filed, when, and what each one covers.

Docket 20859/160410-US

Method to control a solar power cleaning system

Inputting GPS, weather, image analysis, panel power output, cleaning regime, timer, manual input, historical data, and predictive model value; activating scans and cleaning cycles; outputting machine-learned activations of nozzle output and boom carriage wheel movement.

Method claim  ·  Cooling

Cooling cycle on temperature threshold

Determining whether the solar panels have reached a temperature above a predetermined threshold; activating a cooling water spray pulse in response to the affirmative determination.

System claim  ·  ML Loop

Predictive model from collected field data

Entering variable data to a machine-learning database; collecting and storing field data for each system input; defining data set characteristics; creating a predictive model based on datasets for selected desired outputs.

04  ·  Schematic

The three
cleaning regimes.

From the patent itself: three regimes of spray density. The model picks the right one for each zone, each cycle.

Regime A

Single-point dispense. Light spot cleaning for low-soiling zones.

Regime B

Multi-point staggered dispense. Standard cleaning regime for normal soiling.

Regime C

High-density saturation. Deep cleaning + cooling-pulse regime for heavily soiled or overheated arrays.

Source  ·  Attorney Docket No. 20859-160410-US  ·  Customer No. 42798  ·  "Cleaning Devices and Systems for Solar Panel Arrays"

PARTNERSHIPS · ENGINEERING · PILOTS

Want to see the system on your farm?

We're inviting select operators to pilot the system on a representative section of their array. Tell us where your dust comes from and we'll start a conversation.

Talk to engineering