Nobel Systems has developed a new Predictive Machine Learning algorithm that enables water utilities to forecast the likelihood of pipe leaks and system failures. The breakthrough algorithm allows services to create a preventative pipe replacement plan that will reduce costly ruptures.
Nobel IoT Engineer Tanmay Thakur created the new algorithm, which uses artificial intelligence (AI) or Machine Learning. The model gives a Prediction of Leak (POF) rating based on historical and present GIS/IoT data collected via GeoViewer. The data is exported into the model and analyzed. The results are pushed back to GeoViewer as a geospatial rating, indicating the POF for each analyzed pipe segment.
The POF enables water companies to create a preventative maintenance plan to avoid unforeseen water main breaks and costly disruption of water service.
The type of data collected includes pipe age, length, diameter material, install date, the pressure of pipe static, and elevation of installation. Other factors include soil type, weather, and road ratings, which differentiates Nobel’s method from other systems.
Nobel’s Predictive Machine Learning algorithm analyzes the correlation between historical leak data with infrastructure, environmental, and geographical parameters gathered by GeoVIewer. The model produces a highly accurate POF for a utility’s entire water distribution system.
A utility can use the POF to budget a pipe replacement plan where necessary.
Table of Contents
New Model Utilizes Neural Network Algorithm
Nobel’s Predictive Machine Learning model uses a neural network algorithm. Neural networks are loosely modeled on the human brain and designed to recognize patterns.
The neural network translates real-world data into numerical patterns, which helps cluster and classify data. Deep learning via classification enables the correlation of data. The more data available, the more likelihood of establishing relationships between past and future events.
Unique factors that Nobel’s Predictive Machine Learning algorithm uses is the use of detailed weather, soil, and road conditions.
Harsh or extreme weather conditions, such as it being sunny one day and cold the next, can cause pipes to expand and contract at higher rates, reducing the lifespan of pipes. High soil salinity can cause rapid pipe erosion. If pipes are located near a railroad, ground vibrations can cause higher probabilities of failures.
The predictive model takes into account all these factors, in addition to other data, to produce an accurate POF, or prediction of failure.
Utilities can develop a preventative maintenance program using the POF data to replace pipes that are more likely to leak. The predictive maintenance ensures uninterrupted water supply and distribution and reduces costly pipe system repairs.
To learn more about Nobel Systems Predictive Machine Learning module, contact 866-Nobel-SYS or send us a message.