---
title: 'Passport Audit'
description: A fleet-wide heatmap of device-passport completeness per corrector type and field — Qmax/Qmin, serials, firmware, battery, gas composition.
section: AI Analytics
weight: 9
related:
  - ai-analytics/fleet-reports/fleet-inventory
---

import Alert from '@/components/docs/Alert.astro';

A fleet-wide check: for each corrector type, the report measures how completely the device passport is filled in — maximum and minimum flow, serial numbers, firmware version, gas composition and other fields. It is for the operations manager and the data service, to see which device types have empty passports and where a crew needs to be sent for inventory.

This is a diagnostic of what data exists about the devices, not a check of the equipment itself on site. The report shows what is recorded in the database and does not confirm it physically.

## What it shows

A summary at the top (KPIs): how many points are in the fleet, how many device types were checked, how many points were actually polled, and the average passport completeness in percent.

The main element is a color-coded "device type x passport field" table. Each cell shows the share of checked devices for which the field is filled in (from 0 to 100 %). The redder it is, the worse. Fields are grouped by meaning:

| Field group | What it includes |
|---|---|
| Metrology | maximum and minimum flow (Qmax, Qmin) |
| Identification | serial numbers, firmware version |
| Safety | battery charge, pressure thresholds |
| Gas composition | density, fractions of nitrogen, CO₂, methane |
| Substitution protection | pressure substitution thresholds |

The rows are sorted so that **the worst types are at the top**: that is where to start putting things in order.

## How it works

1. Load the fleet's list of points (you can limit it to a single utility company).
2. Group the points by corrector type.
3. For each type, take a sample of several points (5 by default) and request their passport data (`archive_type=settings` — a request for the device's settings).
4. For each field, compute the share of devices where it is filled in.

Field names are written differently by different manufacturers, so the report maps them to 14 common names via a correspondence dictionary. Only points that actually responded to the request are counted; if a point did not respond, it does not spoil the statistics as "empty".

## Run parameters

| Parameter | Default | What it means |
|---|---|---|
| Sample nodes per type | 5 | how many points to poll per device type |
| Parallel fetch workers | 4 | how many points to poll simultaneously (faster, but loads the server more) |
| Utility company | All (whole fleet) | filter by a single utility company |

## How to read the result

Look at the red rows at the top of the table and at the average-completeness column (avg) for each type.

| Type's average completeness | What it means |
|---|---|
| 80 % and above | passport almost complete; calculations for this type are reliable |
| 50–79 % | partially filled; some operations are limited |
| less than 50 % | an inventory field visit and additional data entry are needed |

What to keep in mind:

- **A sample of 5 points** gives an approximate estimate for the type: if the group has 200 devices, only 5 are examined. This is a compromise between speed and accuracy.
- **0 % does not always mean "the field is absent".** The device may not report it, even though it is physically present.
- If a new corrector type appears with unfamiliar field names, the correspondence dictionary should be extended — otherwise its fields are not recognized.

<Alert type="note" title="Reads existing data only">
Passport Audit reflects what is recorded for each device, not the device's true on-site configuration. Use it to prioritize inventory visits, then confirm fields physically where completeness is low.
</Alert>