Skip to content
⁨NDDev RnD⁩, home page
Community
Language: English
Display
Theme
Motion

Equipment and sensor data processing

NDDev processes data from equipment and sensors: we take in telemetry and measurements, check them, clean out faults and turn them into figures that engineers and managers work from. Our RnD practice writes these pipelines in C++, Rust and Python. After our work for Traktorodetal, measurements of the required product range run five times faster.

Who it is for

Manufacturers, laboratories and equipment owners whose machines and sensors already produce data that is hard to use because of gaps, noise and mixed formats.

  • Every measurement checked
  • One format from many sources
  • Heavy computation in C++ and Rust
  • Export to your systems

A measurement's path from sensor to report

Raw equipment data is almost always messy: gaps, outliers, clock drift, different units on different machines. Together with your process engineer we describe what a valid measurement looks like, then write checks that filter out faults and flag doubtful values. An unknown value is recorded as a gap with a flag, so that a person can review it.

  • intake of telemetry and measurements from equipment protocols and files
  • validation rules: allowed ranges, outliers, gaps
  • data from different machines brought to the same units and clock
  • derived indicators calculated for process engineers
  • a data store and an API for reports
  • a dashboard or an export to your business system

Heavy processing is written in C++ and Rust so it keeps pace with the stream, and analysis and reports in Python. The output goes to a PostgreSQL database, to files in the format you need, or to an API for your dashboards and business systems.

Working with your data, stage by stage

We start from a real slice of your data, so the first findings come early.

  1. Sample

    We take a real slice of data and go through it with your process engineer: what each field means and where faults occur.

  2. Quality rules

    We write down the validation rules and show what share of the data they filter out, and why.

  3. Processing pipeline

    We write intake, checks and calculations and run them over the accumulated history.

  4. Live stream

    We connect the pipeline to live equipment data and deliver the output to your systems.

Work in this area

  • HarvesterLib

    A C++20 library for Traktorodetal: it processes log diameter and length measurements from forestry harvesters and works out trunk width from them. With it, measuring the required product range runs five times faster.

Questions about equipment data

Which formats do you work with?

Whatever your equipment produces: export files, communication protocols, controller and manufacturer APIs. If a format is closed, we start by studying the documentation and a trial export.

What result can we expect?

We name the target figures after reviewing a sample of your data and choose them with you: the share of faults filtered out, processing time, measurement speed. The same figures are checked at handover.

Do we need our own server?

Processing can run on your server or on a computer next to the equipment when the data has to stay on site. We choose where the pipeline runs during the review stage.

Can we build forecasts on this data later?

Yes. Clean data with a history becomes the basis for predictive analytics: failure forecasting and anomaly detection. That is the next stage, and we estimate it separately.

How much does data processing cost?

The price depends on the number of sources and formats, the volume of the stream and the calculations your engineers need. The starting price is shown in the price block on this page; we fix the total after reviewing the sample.

from ₽420,000

Free warranty
60 days

Prices are starting prices for information only and are not a public offer. The final price is fixed in the specification.

Software for equipment and industry

Send us an export from your equipment

Even a small file shows what we will be working with. We will go through it and propose a processing pipeline.

We reply within 24 hours

Assistant

Loading the assistant…