PRISM: Predictive Industrial Safety & Monitoring AI

PRISM is a comprehensive, AI-driven predictive maintenance and safety monitoring system for industrial environments. It simulates real-time sensor data from various industrial machines, detects anomalies, and triggers intelligent alerts, including detailed emails, emergency calls, and SMS notifications. The system also features an AI-powered chatbot for interactive diagnostics and operational support.

Key Features

• Real-time Machine Simulation: Generates continuous sensor data for four distinct types of industrial machines.
• Multiple Machine Modes: Supports different operational modes for each machine, including normal, maintenance, and sabotage, each with unique data generation patterns.
• Advanced Anomaly Detection: Implements a rule-based system to detect deviations from normal operating parameters and calculates an anomaly score.
• Intelligent Alerting System:
  • Email Alerts: Sends professionally formatted HTML emails for maintenance requests and critical sabotage incidents.
  • Voice Calls & SMS: Uses Twilio to make automated emergency calls and send SMS backup alerts to key personnel.
• AI-Powered Chatbot: An integrated chatbot (MaintenanceBot) that assists with:
  • Equipment diagnostics and troubleshooting.
  • Real-time status checks.
  • Safety protocols and maintenance advice.
• Web-Based Dashboard: A Flask-based web interface to visualize machine status, sensor readings, and recent alerts.
• Sensor Health Tracking: Monitors and simulates the degradation of sensor health over time, adding another layer of predictive maintenance.
• Comprehensive API: Offers a rich set of API endpoints for interacting with the system, fetching data, and triggering actions.

Monitored Machines

The simulation includes four pre-configured industrial machines, each with a unique set of sensors:

1. Chemical Reactor (R-001):
   • Sensors: Temperature, Pressure, Flow Rate, Level.
2. Biotech Fermenter (F-003):
   • Sensors: Temperature, pH, Dissolved Oxygen, Agitation.
3. Distillation Column (D-002):
   • Sensors: Top Temperature, Bottom Temperature, Pressure, Reflux Ratio.
4. Heat Exchanger (HX-005):
   • Sensors: Inlet Temperature, Outlet Temperature, Pressure Drop, Flow Rate.

Setup and Installation

Prerequisites

• Python 3.7+
• Flask and other Python packages (installable via pip).
• A Gmail account (for sending email alerts).
• A Twilio account (for making calls and sending SMS).

1. Clone the Repository

git clone https://github.com/joedanields/PRISM
cd PRISM

2. Install Dependencies

It is recommended to use a virtual environment:

python -m venv venv
source venv/bin/activate  # On Windows, use `venv\Scripts\activate`
pip install Flask Flask-SQLAlchemy twilio openai

3. Configure Credentials

You need to replace the placeholder credentials in app.py and/or ad.py with your actual service credentials.

Email Configuration (Gmail)

In the EmailConfig class, replace the $$ placeholders:

class EmailConfig:
    SMTP_SERVER = 'smtp.gmail.com'
    SMTP_PORT = 587
    EMAIL_USER = 'your-email@gmail.com'  # Your Gmail address
    EMAIL_PASSWORD = 'your-16-char-app-password'  # Gmail App Password
    MAINTENANCE_TEAM = 'maintenance-team-email@example.com'
    PLANT_MANAGER = 'plant-manager-email@example.com'
    EMERGENCY_RESPONSE = 'emergency-response-email@example.com'

Note: You must generate a 16-character "App Password" from your Google Account settings for EMAIL_PASSWORD. Your regular Google password will not work.

Twilio Configuration

In the TwilioConfig class, replace the $$, %%, and @@ placeholders:

class TwilioConfig:
    ACCOUNT_SID = 'ACxxxxxxxxxxxxxxxxxxxxxxxxxxxxx'  # Your Twilio Account SID
    AUTH_TOKEN = 'your_twilio_auth_token'          # Your Twilio Auth Token
    TWILIO_PHONE_NUMBER = '+1234567890'           # Your Twilio Phone Number
    EMERGENCY_CONTACTS = [
        '+19876543210',  # Plant Manager's number
        '+19876543211',  # Maintenance Chief's number
    ]

Running the Application

To start the system, run the app.py or ad.py file:

python app.py

Or, for the advanced version:

python ad.py

The application will start, create the necessary database (machines.db), and begin generating sensor data in the background.

You can access the web interface at:

• Landing Page: http://localhost:5000/
• Dashboard: http://localhost:5000/dashboard
• AI Chatbot: http://localhost:5000/chat

API and Debug Endpoints

The application exposes several API and debug endpoints to interact with the system programmatically.

Main API Endpoints

• GET /api/machines: Retrieve data for all active machines.
• GET /api/machine/<id>/latest: Get the latest sensor readings for a specific machine.
• POST /api/machine/<id>/mode: Set the operational mode (normal, maintenance, sabotage) for a machine.
• GET /api/alerts: Fetch the 20 most recent alerts.
• POST /api/chat: Interact with the AI chatbot.
• GET /api/machine/<id>/chart-data: Get historical data formatted for charts.
• GET /api/machine/<id>/sensor-health: Get the health status of all sensors on a machine.
• POST /api/emergency-call/<id>: Manually trigger an emergency call for a machine.