FactoryPulse AI

The Problem

Industrial machines rarely fail without warning. Changes in temperature, vibration, pressure, current, and other sensor signals can indicate that a machine is moving toward an unhealthy state.

The challenge is turning these continuous sensor streams into actionable information before a small anomaly becomes an expensive failure.

Traditional monitoring systems often depend on fixed thresholds and can generate large numbers of alerts without explaining what is actually happening.

Our Solution

FactoryPulse AI is an AI-powered predictive maintenance platform designed to continuously monitor industrial machines, detect abnormal sensor behavior, estimate machine health, and provide actionable diagnostic recommendations.

Instead of simply displaying sensor values, FactoryPulse AI creates a live machine-health picture from multiple signals.

The platform can:

• Monitor multiple machines simultaneously
• Stream sensor data in real time
• Detect abnormal machine behavior
• Calculate machine health and risk levels
• Identify potentially abnormal operating patterns
• Simulate machine faults for testing
• Generate AI-assisted diagnostic recommendations
• Visualize machine conditions through an interactive dashboard

How It Works

FactoryPulse AI uses simulated industrial sensor streams representing real-world machine behavior.

The pipeline follows:

Sensor Simulation → Data Processing → Anomaly Detection → Risk Assessment → Diagnosis → Visualization

Each machine continuously produces signals such as:

• Temperature
• Vibration
• Current
• Pressure
• Rotational/operational measurements

The anomaly detection layer analyzes these signals and identifies behavior that differs from the machine's expected operating pattern.

The system then combines the detected anomalies with machine measurements to produce a health/risk assessment.

When abnormal behavior is detected, FactoryPulse AI provides a diagnostic explanation and recommended action instead of presenting the operator with raw sensor data alone.

Real-Time Fault Simulation

One of the key features of our demo is the ability to intentionally inject a machine fault.

When a fault is introduced, the simulated sensor behavior changes.

For example:

Normal State

Temperature → Normal
Vibration → Normal
Current → Normal
Risk → Low

Fault State

Temperature → Increasing
Vibration → Increasing
Current → Abnormal
Risk → High

This allows us to demonstrate the complete detection pipeline without requiring access to a physical industrial machine.

Technology Stack

Backend

• Python
• FastAPI
• WebSockets
• SQLite
• Machine-learning based anomaly detection

Frontend

• React
• Vite
• JavaScript
• Recharts
• Responsive dashboard UI

Deployment

• Docker
• Render
• Vercel
• GitHub

Architecture

Industrial Sensor Simulator
          |
          v
    Sensor Data Stream
          |
          v
    FastAPI Backend
          |
    +-----+------+
    |            |
    v            v
Anomaly      Machine
Detection    Health/Risk
    |            |
    +-----+------+
          |
          v
 AI Diagnostic Layer
          |
          v
   WebSocket Stream
          |
          v
 React Monitoring Dashboard

Built With

• anomaly
• api
• artificial
• data
• detection
• docker
• fastapi
• intelligence
• javascript
• learning
• machine
• maintenance
• predictive
• python
• react
• real-time
• rest
• sqlite
• vite
• websockets