Wattminder is developing a web platform that uses algorithms to monitor, detect, and diagnose performance issues in photovoltaic plants, improving system availability and efficiency. The system sends actionable notifications to quickly restore plant performance after degradation is identified. The project aims to enable performance contracts for PV investments through enhanced oversight and real-time data analysis.

1. The Wattminder Vision Mission Statement, Pitch & Teaser Demo Internal confidential Apr 13 2009 wattminderVision.gdoc We create value by building a collection of intellectual properties & prototypes that will drive optimal performance of photovoltaic plants to minimize under-performance and downtime, resulting in enhanced system availability, efficiency, and assured return on investment.

2. The Wattminder group has built a web platform to assimilate knowledge and is developing a framework to deploy heuristics, expert system, statistical, and data-mining algorithms that can detect, analyze, and diagnose faults that impact performance of a photovoltaic plant. Notification with actionable paging message will then enable remedy to be applied to bring a plant back to full production promptly. This web service will be the enabler for PPA, or performance contract type of PV investments. Elevator Pitch

3. The state of the art PV performance monitoring services gather the same set of measurement –AC Power, DC Voltage & String Currents, Irradiance, Ambient & Cell-temperature, etc. Leading operators mostly rely on SOC staff to watch over wall-ful of display monitors to keep an eye on performances. Given any sudden or gradual de-gradation in output, there is no indication of under-performance, nor the faulty condition that is causing this de-gradation in performance. Problem Statement

4. Working with live measurement data stream, backed by a data-warehouse of historic measurement records; a set of algorithms is being developed to: 1. Detect the occurrence of de-gradation. 2. Diagnose the cause of the de-gradation. 3. Deliver an actionable message to enable prompt restoration of full performance. Wattminder Solution

5. A few illustrative, perhaps conceptual demonstra-tions, or statements of solution will be articulated to present a convincing case to a prospective funding, or collaboration-joint venture party. Example: Our output mathematical model is 'corroborated' with actual real-time and archived data from a known reference PV site--involving shading by some objects (Light-tower, or layer of dust) Teaser Demonstrations

6. Example: A leaf has dropped onto a PV module, covering much of a cell. Within seconds, a message 'Warning: Unexpected shading upon String-XYZ on Array ABC, resulting in 888 Watts of loss--Jul 25, 2009 at 13:45:43. Advisory to field technician: 'String current indicates 33% below baseline. Bring pole grabber'. Note: Benchmark was calculated to be 2650 Watts, actual output was 2666 Watts observed a minute ago, while -3% change in irradiance occurred. Example: An array has been gathering dust for the past 3 months. A message-- 'Alert: Possible layer of dust on Array Gplx resulting in 8888 Watts of loss --Jul. 25, 2009 at 11:50:45' is received by maintenance staff. Note: Maintenance log indicated it has been 3 months since last power wash, and no rain was recorded since then, while output is 3.4% below benchmark . Enter in batch advisory to schedule power wash for array DEF. Teaser Demonstrations –SMS to iPhone

7. Hardware : ♦ Real-Time DSP Data-Acq. Module ♦ Inverter fault detector ♦ Vandalism Detector ♦ Thin-film String Reconfigurator ♦ Array Fire Safety-Shutdown Circuit Software/Services : ♦ KnowledgeBase with heuristic acquisition ♦ Mathematical model for prediction & fault-detection ♦ Distributed DC instrumentation Network ♦ Fault Dictionary -observables, diagnostic scheme ♦ PV performance benchmark & metrics ♦ Intelligent Diagnostic Algorithms ♦ Live IV-Chart of PV Array ♦ Maintenance tracking & tie-in with KnowledgeBase Intellectual Properties, Patents

8. Foundation : ♦ full set of design parameters of PV plant ♦ fault dictionary with observables, probability functions ♦ mathematical models with contigent terms ♦ heuristics and refinement mechanism ♦ data warehouse of history of logged data ♦ real-time stream of instrument data Algorithmic Output (IM, SMS): ♦ detect and notify of under-performance ♦ diagnose and notify of nature of fault ♦ provide possible remedial instructions Operation with constant update Track loss of output, mean-time-to-repair, update system availability metrics, algorithmic verification & refinement A living frame-work

9. Innovations in PV Monitoring & Diagnostics True Real-time data display -live IV plot Mathematical model-based PV monitoring –irradiance, cell temperature, solar & incident angle calculated every minute Artificial Intelligence & data-mining detection algorithms on live and historic data, alert sent by paging when anomaly occurs, with diagnosis & advisory Closely coupled tracking mechanism to minimize down-time, assure Service Level Availability – & return-on-investment on PPA contracts

10. Hardware : ♦ More intelligent inverters, combiner boxes; merged functions ♦ Micro-Inverter (Distributed DC-AC conversion) ♦ String-Sub-Ar ray coupling to inverters ♦ Smart Module / String ♦ Array/String Shutdown Circuit for Fire Safety Value-Added Services Bundling : ♦ Energy Management with shared savings ♦ Efficiency improvement, new and retrofit buildings ♦ Guaranteed Service Level (Performance) contracts Financing : ♦ Feed-in-Tariff replacing Performance based incentive ♦ Lease, or PPA contracts become dominant ♦ Third-party, R.E.C. Cap & trade Emerging Macro Trend in PV Electronic content >> $5B in 2012

11. A Multi-disciplinenary Team Hardware Bob Getsla, RF and Analog wizard Dave Kolden, PCB, animation designer Software Chin Li, PhD Computer Sci. Datamining Dennis Feick Electronic, Java C coder Achim Steinhagen, MSEE data manager ZhongSheng Luo, PhD Mat.Physics, KLA Adm-Marketing Jeff Blake, MSCS, Web 2.0 evangelist Steve Yang, MSE, Intelligent System

12. MileStones Funding Abstract -18 months 100 Printed Circuits & BlackBoxes $600K Development of AI & DM algorithm $850K Crystal Report UI, QB, eCommerce $100K Gen & Adm. Staffing $300K U/L Value & Certificate Engin'ring $400K Patents Applications, Legal fees $150K total: $2,400K '======= =======