The analog entrance finish with AI-based arc detection improves photo voltaic security by detecting arcs and lowering false alarms in photovoltaic methods.
TIDA-010955, a reference design from Texas Devices (TI) options an analog entrance finish (AFE) for DC arc detection in photo voltaic purposes, leveraging synthetic intelligence (AI) for enhanced efficiency. The string present is measured, filtered, and processed by an AI-based arc detection algorithm operating on the TMDSCNCD28P55X controlCARD, constructed across the C2000 F28P55x household of gadgets. The AI method presents larger accuracy and robustness in comparison with conventional detection strategies. An entire toolchain is offered for gathering arc information, coaching the embedded AI mannequin, and validating the system. This reference design could be ordered beneath half quantity TIEVM-ARC-AFE. The answer is appropriate for purposes comparable to photo voltaic arc safety, string inverters, and central inverters.
Because the variety of photo voltaic installations continues to rise, making certain system security turns into more and more essential. Particularly, arcing on high-voltage traces should be promptly detected, and affected photo voltaic strings should be de-energized to stop hazards comparable to electrical shock or fireplace. To handle this, requirements like UL 1699B require arc-fault safety circuits for all photo voltaic methods working at voltages beneath 1500V. This reference design presents an implementation of an analog entrance finish particularly for arc detection. Nonetheless, the design alone doesn’t meet the total necessities of the UL 1699B commonplace.
DC arcing generates an AC noise present inside the photovoltaic (PV) string cabling, spanning a broad frequency vary as much as a number of megahertz. The first problem of DC arc detection lies in reliably figuring out this noise improve with out triggering false alarms or pointless shutdowns. This calls for a low-noise, high-performance analog entrance finish, able to distinguishing small injected AC noise currents—sometimes just a few milliamps—superimposed on the a lot bigger DC string present, which might attain round 20A in residential methods and even larger in business purposes.
Furthermore, photo voltaic methods are uncovered to varied different noise sources that should not be mistaken for arcing occasions, comparable to inverter switching frequencies and energy line communication indicators alongside the PV cabling. Historically, detection algorithms have been employed to acknowledge arc signatures within the present measurements. These algorithms usually require system-specific tuning, as arcing signatures are inclined to range considerably between installations.
A more moderen and promising technique entails utilizing embedded Synthetic Intelligence (AI) fashions for arc detection. Correctly skilled, AI-based options can ship larger detection accuracy and improved resilience in opposition to false alarms.
To coach these AI fashions, high-quality labeled information is important. For that reason, this reference design incorporates an arc-labeling circuit. It leverages the measurable voltage drop throughout an arc to differentiate between arcing and regular working circumstances—a way made possible inside managed laboratory environments the place the arc voltage is instantly accessible.
TI has examined this reference design. It comes with a invoice of supplies (BOM), schematics, meeting drawing, printed circuit board (PCB) format, and extra. The corporate’s web site has further information in regards to the reference design. To learn extra about this reference design, click on right here.
