Clean Solar Index
v0.3 — ML Engine
The Global Ranking Engine for Solar Panel Cleaning
Clean Solar Index is an open-source, ML-driven decision-support tool that scores and ranks 51+ photovoltaic module cleaning technologies from international literature using a mathematically rigorous Figure of Merit (FoM). Powered by Random Forest feature importance, it identifies the 5 most impactful variables from a 25-column dataset extracted from 90+ peer-reviewed publications.
Figure of Merit Engine
The FoM evaluates PV cleaning via 0-1 normalized variables dictated by our Top 5 ML Feature Importance. Numerator traits (benefits) maximize score; denominator traits (penalties) minimize it.
\[ FoM = \frac{(AI\_Level)^{\alpha} \times (Automation)^{\beta} \times (Lifetime)^{\gamma}}{(OPEX)^{\delta} \times (Scratch)^{\epsilon}} \]
Top 5 ML Feature Importance
A Random Forest regressor trained on 25 variables across 90+ peer-reviewed publications identified these five as the most impactful for predicting PV cleaning performance. They drive the FoM formula above.
1
AI Level 21.6%
↑ Benefit (Numerator)Measures the degree of artificial intelligence integration in the cleaning system. Ranges from 0 (no intelligence) — manual wiping or gravity-fed water — to 5 (full AI autonomy) — systems with computer vision, path planning, reinforcement learning, and real-time obstacle avoidance. Higher AI levels enable optimal scheduling, selective cleaning of soiled hotspots, and autonomous fault detection, directly improving cleaning efficiency per pass.
2
OPEX 17.0%
↓ Penalty (Denominator)The annual operating expenditure of the cleaning system, measured in USD/MW/year. Captures recurring costs including water consumption, electricity, labor wages, spare parts, and consumables (detergents, brushes). Systems with high OPEX are penalized because long-term operational cost is a primary barrier to adoption in large-scale solar farms, especially in arid regions where water is scarce and expensive.
3
Scratch Risk 15.7%
↓ Penalty (Denominator)Quantifies the likelihood that the cleaning mechanism causes micro-scratches, abrasion, or mechanical stress to the PV glass surface. Rated from 0 (no contact / zero risk) — electrostatic or airflow-based systems — to 5 (high abrasion) — stiff brush or manual scrubbing. Surface damage permanently reduces light transmittance and accelerates anti-reflective coating degradation, compounding energy losses over the panel's 25-year lifespan.
4
Automation Level 12.6%
↑ Benefit (Numerator)Classifies the system's operational autonomy from 0 (fully manual) — requiring human labor for every cleaning cycle — through semi-automatic — human-initiated but machine-executed — to 5 (fully autonomous) — the system schedules, navigates, cleans, and docks without human intervention. Higher automation reduces labor dependency, enables 24/7 cleaning schedules, and scales linearly across large solar installations.
5
Lifetime (Years) 7.8%
↑ Benefit (Numerator)The expected operational lifespan of the cleaning hardware before it requires full replacement or major mechanical overhaul. Ranges from 1–2 years for consumable-heavy systems (e.g., replaceable brush heads) to 15–25 years for contactless electrostatic or nanocoating-based solutions. Longer lifetimes amortize CAPEX over more cleaning cycles, improving the total cost of ownership and aligning with the PV module's own 25-year warranty period.