← All workCase study No. 09Industrial IoT Web Platform09 / 17

Machine Monitoring.

Industry 4.0 monitoring for railway depots

Near real-time machine monitoring, alerts and maintenance for railway depots.

Client: CLF Projects
Industry: Railways / Industry 4.0
Engagement: Build — four audited phases
Relationship: A client partnership

Machine Monitoring

IThe ground

The brief.

Maintenance teams at railway depots in Pune and Nagpur needed one place to see the state of their machines — compressors, synchronised pit jacks, EOT cranes and retractable OHE — with alerts that tell a real maintenance need apart from a faulty sensor. The system had to run on the depot's own servers, work on everything from a phone to a desktop, keep a clear audit trail and be ready to receive signals from PLCs and edge gateways.

IIThe plan

Our response.

We built an on-premises Industry 4.0 platform: a NestJS API with BullMQ jobs, Socket.IO live updates, a built-in simulator and a correlation-based anomaly engine, and a responsive React 19 web app with light and dark themes. When related signals deviate together the platform raises "maintenance required"; when a single signal deviates on its own it flags a suspected sensor fault, and operators can confirm or correct every call. Machine types are data rather than code, and ingestion accepts HTTP push with site-scoped device keys or MQTT. Until PLC and gateway signals are connected, every machine runs on clearly labelled simulated data.

Product Design
Full-Stack Web Development
Real-time Data Pipeline
Anomaly Detection
On-premises Deployment

IIIThe buildStrategy → craft → engineering → care

Brick by brick.

Course 01
Machines as Data
Modelled machine types, signals, limits and correlation groups as configuration, seeded with four depot machine types, so adding a machine needs data, not code.
Course 02
A Live Pipeline
Built ingestion adapters for a physics-based simulator, HTTP push and MQTT, with tag mapping, near real-time updates over Socket.IO and verified downsampling for long-range history.
Course 03
Alerts Operators Can Trust
Designed the anomaly engine to catch spikes, drops, flatlines, missing data and interlock trips, with an acknowledge-to-resolve workflow and in-app, web push and email notifications.
Course 04
Secure, Audited, On-premises
Argon2id sign-in with rotating refresh tokens, role-based access for super admins, managers and operators, an audit log, backups and a Docker Compose production stack with TLS.

—A closer lookMachine Monitoring in detail

Machine detail page with safety interlock state, live values against limits and sparklines — Machine detail: interlock state and live values against their limits.

Daily analytics for the depot machines with charts and export — Daily analytics, ready to compare and export.

Machine Monitoring fleet overview in the graphite dark theme — The fleet overview in the dark theme.

IVThe outcomeWhat took shape

What it became.

Web App Routes: 16; Fleet, machine detail, alerts, analytics, maintenance and admin
Automated Tests: 349; 169 unit, 120 API end-to-end, 27 Playwright and 33 smoke tests
Load Test: 0 errors; 50 machines at 1 Hz, with every reading stored
Lighthouse: 98–100; Desktop performance, with 100 for accessibility

Project scope and implementation details from Edifio’s case study.

VSpecificationBuilt on the right foundations