K-Nearest Neighbors for Anomaly Detection and Predictive Maintenance in Water Pumping Systems.

Area of Science:

• Engineering
• Data Science
• Environmental Science

Background:

• Maintenance is crucial for water source quality and supply reliability.
• Water pipe corrosion leads to leaks and reduced water quality.
• Identifying hydraulic anomalies in pumping systems is essential.

Purpose of the Study:

• To identify hydraulic anomalies in water pumping systems.
• To develop a predictive model for pump status using sensor data.
• To determine the most important sensor parameters for accurate prediction.

Main Methods:

• Collected data from a water supply network with varied pipe characteristics.
• Installed sensor meters at diverse locations within the hydraulic system.
• Utilized the k-nearest neighbors (KNN) machine learning algorithm for prediction.

Main Results:

• Developed a model correlating sensor parameters with pump status.
• Identified key sensor parameters for predicting pump shutdowns.
• Achieved accurate prediction of pump shutdowns using the KNN algorithm.

Conclusions:

• The KNN algorithm effectively predicts pump status in water supply systems.
• Sensor data analysis can identify hydraulic anomalies and aid maintenance.
• Optimized models with minimal parameters provide accurate predictions.