---
title: 'Fleet Battery Forecast'
description: Battery charge snapshot across the whole fleet, with a composite 0-100 score per node so the weakest batteries surface first.
section: AI Analytics
weight: 10
related:
  - ai-analytics/node-reports/battery-forecast
---

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

This report gives an instant snapshot of battery state across the entire fleet of metering nodes at once. For each node it assesses the current charge and builds an overall score from 0 to 100, then sorts the table so the weakest batteries are at the top. It answers an operations question: where in the fleet batteries are running out and who to replace first.

<Alert type="info">
This is a snapshot of the current state, without a discharge-date forecast. To get a service-life forecast for a specific node, use the [Battery Forecast](/en/platform/v3/ai-analytics/node-reports/battery-forecast) report.
</Alert>

## What it shows

A table of all nodes with batteries, sorted by ascending score (worst at the top). For each node, two charge sources and the overall score are visible.

| Column | What it means |
|---|---|
| Node | metering node number |
| Type | metering device (corrector) type |
| Metrological charge, % | the device's main battery charge in percent |
| Telemetry voltage, mV | the link modem's battery voltage in millivolts |
| Score | the node's overall battery score 0-100, the lower the worse |
| Last session | how long ago the device came online |

The charge is taken from two sources: the device's main (metrological) battery and the telemetry (link) battery. The overall score is the worse of the two sources: one battery running out is enough to make the node problematic.

## How it's calculated

First the report gets the list of all fleet nodes and keeps only those that have a battery (mains-powered devices are excluded). Then for each node, in parallel across several workers, two indicators are loaded:

- the main battery charge in percent — from the device's settings slice (a request for the passport settings, where there is a charge field);
- the link battery voltage in millivolts — from the last link session (records of the device coming online).

Each source is converted to a 0-100 scale: 0 is the critical level, 100 is a fresh battery. The thresholds depend on the unit of measurement: for percent the critical level is 20%, for link voltage — 3300 mV. The node's overall score is the smaller of the two: the weakest battery determines the node's state. If one source is unavailable, only the other is used.

## Run parameters

| Parameter | Default | Meaning |
|---|---|---|
| Period from | 30 days ago | start of the window (needed to compute the charge drop) |
| Period to | yesterday | end of the window |
| Parallel fetch workers | 8 | how many nodes to poll at once; lower to 2-4 on HTTP 429 |
| Stations cap | 0 (whole fleet) | a limit for a trial run, e.g. 50 |
| Utility company | from context | fleet filter by company |

## How to read it

- Look at the **top of the table**: nodes with the lowest score are closest to discharge — plan them for replacement first.
- Compare the **two charge sources**: if the score is low because of the link battery, the node may still count correctly but will soon drop off the network.
- **Devices without a battery** (mains-powered) do not appear in the report — that is normal, they should not be in the table.
- On a large fleet (800+ nodes) with many workers the platform may throttle requests (HTTP 429 error). The report tracks data availability and notes it at the bottom of the page; on failures, lower the number of workers.
- This is a **snapshot of the current state**, not a forecast. For a specific node's discharge date, use the [Battery Forecast](/en/platform/v3/ai-analytics/node-reports/battery-forecast) report.

## Related reports

- [Battery Forecast](/en/platform/v3/ai-analytics/node-reports/battery-forecast) — a detailed single-node breakdown with a service-life forecast.
- [Battery Comparison](/en/platform/v3/ai-analytics/node-reports/battery-comparison) — a side-by-side comparison of a small group of nodes.