Smart Maintenance & Incident Prediction System 🛠️🤖

Hackathon Problem Statement : Smart Maintenance, Incident Tracking & Prediction SystemA comprehensive, full-stack platform designed to streamline campus facility management. It features real-time incident reporting, intelligent technician assignment, interactive heatmaps, and a deterministic AI engine that predicts infrastructure failures before they happen.

🚀 Key Features🧠 AI & Predictive MaintenanceDeterioration Detection:

Uses Spearman Rank Correlation & Linear Regression to detect accelerating failure rates (e.g., an elevator breaking down more frequently).Wear-Out Analysis: Implements Weibull Survival Analysis to estimate the probability of equipment failure based on age.Seasonal Forecasting: Uses Chi-Square tests to identify recurring issues (e.g., AC filters clogging every 2 weeks or water leaks on Mondays).Spatial Clustering: Uses Poisson Distribution to detect infrastructure anomalies (e.g., 5 electrical failures in one block = System Failure).

👥 Role-Based Access Control (RBAC)Student/Staff Portal: Simple, multi-step "Wizard" for reporting issues with GPS location locking and photo uploads. Live status tracking timeline.Technician Dashboard: Dedicated job queue with "Accept", "Start", and "Resolve" workflows. Auto-refreshes and alerts on SLA breaches.Admin Command Center: Live heatmap of high-density zones, AI diagnostics trigger, and full user management. ⚙️ Intelligent AutomationSmart Priority Matrix: Auto-calculates priority (Low/Medium/High) based on keywords (e.g., "fire", "spark") and location context (e.g., "Server Room").Auto-Assignment: Algorithms assign the best-suited technician based on Specialty (Plumbing/Electrical) and Current Workload.Duplicate Shield: Prevents spam by checking for similar active reports within a 20m GPS radius.

🛠️ Tech StackFramework: Next.js 15 (App Router)Language: TypeScriptDatabase: MongoDB (via Mongoose)Styling: Tailwind CSSAuthentication: NextAuth.js (Credentials Provider)Maps: Leaflet (React-Leaflet)Icons: Lucide React

⚡ Getting Started1. PrerequisitesNode.js 18+ installed.A MongoDB Atlas connection string.2. Clone & Installgit clone https://github.com/your-username/smart-maintenance.git cd smart-maintenance npm install

3. Environment SetupCreate a .env.local file in the root directory:# MongoDB Connection String MONGODB_URI=mongodb+srv://:@cluster0.example.mongodb.net/smart_maintenance

NextAuth Secret (Any random string)

NEXTAUTH_SECRET=your_super_secret_key_here NEXTAUTH_URL=http://localhost:3000

4. Run the Development Servernpm run dev Open http://localhost:3000 in your browser.🧪 Seeding Data & Testing AITo demonstrate the AI capabilities, you don't need to wait for weeks of data. We have included a Smart Seed Script that plants specific mathematical patterns in the database.Seed the Database:Visit: http://localhost:3000/api/seedResult: Creates Users and injects 25+ incidents simulating deterioration, seasonality, and clusters.Login Credentials:Admin: [email protected] / password123Technician: [email protected] / password123Student: [email protected] / password123Run AI Diagnostics:Login as Admin.Go to the Dashboard.Click the "Run Diagnostics" button.Observe: The system will generate cards like "CRITICAL: System Deterioration" and "Seasonal Pattern: Mondays".

📂 Project Structuresrc/ ├── app/ │ ├── api/ # Backend Routes (Incidents, Auth, Predictions) │ ├── admin/ # Admin Dashboard Pages │ ├── student/ # Student Dashboard Pages │ ├── technician/ # Technician Dashboard Pages │ ├── report/ # Public Reporting Wizard │ └── login/ # Auth Pages ├── components/ # Reusable UI (Heatmap, Bell, Sidebar) ├── lib/ # DB Connection & Helpers ├── models/ # Mongoose Schemas (User, Incident, Prediction) └── middleware.ts # RBAC Security Logic

🏆 Hackathon Highlights: Python-Free AI: The prediction engine uses pure TypeScript implementations of statistical models, ensuring zero-latency analysis without external ML servers.Security: Middleware ensures strict separation of concerns. Technicians cannot access Admin routes, and Students cannot see internal tickets.UX: "Dark Mode" support and a mobile-first design for field technicians.

📄 License This project is open-source and available under the MIT License.

To test the platform deployed on vercel,

1. Create new user accounts by going to register
2. Use these existing credentials
   1. [email protected] - admin
   2. [email protected] - technician -Electrical
   3. [email protected] - technician - internet
   4. [email protected] - technician - civil
   5. [email protected] - technician - plumbing
   6. [email protected] - technician - Electrical
   7. [email protected] - Student
3. Password for all: password123

Developers

1. Javeria Taj