Data
Three inverters send 143. They mean three different things.
The same integer arrives from three manufacturers and means 14.3 amps, 14.3 amps and 1.43 amps. Nothing in the number tells you which — and a plausible wrong number is worse than an obviously wrong one.
Every manufacturer names measurements differently. Sungrow calls it MPPT 4 current. GoodWe calls it PV7 Current. Hopewind writes MPPT4 input current. All three roughly mean the same physical thing: how much electricity is flowing in through one input.
That much is manageable. A mapping table gets you most of the way, and an engineer can finish the job in an afternoon. The real difficulty is the second half of the problem, and it is the half that quietly ruins datasets.
The number itself gives you no clue
The three inverters can each send the integer 143. Vendor A intends it multiplied by 0.1. Vendor B intends it divided by 10. Vendor C intends it divided by 100. The result is 14.3 amps, 14.3 amps and 1.43 amps — from an identical payload, with nothing in the transmission to distinguish them.
A wrong number that looks obviously wrong is harmless. Someone spots it. A wrong number that looks plausible flows quietly into performance reports, contract calculations and carbon claims for years.
143 amps, 14.3 amps and 1.43 amps are all perfectly believable readings for a solar input. None of them will trip a range check. None of them will look wrong on a chart. The error survives every downstream sanity test precisely because it is the right order of magnitude for something.
Why this is not a mapping problem
It is tempting to treat this as configuration: write the scaling factor next to the register address and move on. That works until the fleet grows. Vendor labels, register addresses and scaling conventions multiply with every brand and every firmware revision, and a spreadsheet of them is a document nobody owns and everybody edits.
The alternative is to resolve the question once, at onboarding, into a shared vocabulary — and then never ask it again at runtime. Each vendor label is matched to one canonical measurement, with its unit and its scaling convention attached, and an engineer approves the mapping before it is saved.
Our dictionary currently holds 865 canonical measurements across 62 asset classes, built by parsing 8 manufacturer register documents rather than by guessing from field data. Those are counts of our own vocabulary, not claims about anyone's plant.
What this buys you
- A measurement means the same thing across every brand and vintage in the fleet, so a fleet-level comparison is actually a comparison.
- The translation decision is auditable: which label, which canonical measurement, which document it came from, and who approved it.
- Swapping an inverter or replacing a gateway does not orphan the history attached to that measurement.
None of this is glamorous, and none of it appears in a product demo. It is also the difference between a platform whose numbers an engineer will defend in a contract dispute and one whose numbers they will quietly re-derive in a spreadsheet before every meeting.
Start with how it works.
The architecture, the modeling approach, and where each part of the platform actually stands today.