lewy09-tm25/water-mains-prognosis

Water Main Prognosis

An AI-driven project to simplify the preventative management of water main segments. We focus on two main components: prediction of health scores for pipe segments and the optimized scheduling of repairs.

Project Organization

Use Python version >= 3.8 and run pip install -r requirements.txt.

├── data/    # Our datasets stored in csv format
├── graphs/  # SA and ML graphs
├── map/     # Demo map generation
├── models/  # ML models
├── prediction_and_survival/  # SA
├── research/   # SA and ML notebooks
├── scheduler/  # CSP and LP schedulers

Demo

• Run map/main.py
• This will generate the Plotly map that contains all prone-to-breakage pipes with assigned properties to them

Machine Learning and Survival Analysis

• Run prediction_and_survival/ML.py and prediction_and_survival/SA.py
• This will write the CSV output file in the data/ folder

Task schedulers

• /scheduler/workforce.py defines employees with work hours and tasks to be scheduled
• The output of the scheduling tasks are interactive gantt charts that open in a new browser window

Constraint Satisfaction Problem

• /scheduler/CSP/main.py solves the scheduling problem using backtracking
• /scheduler/CSP/csp.py contains the constraints

Linear Programming

• /scheduler/LP/main.py solves the scheduling problem using pulp's linear programming capabilities