Adopt Physics Informed Machine Learning, with Sensor Data, to find more Robust Models for the Prediction of Structural Failures

The increased connectivity and availability of sensor data is driving new opportunities for both the design engineering and asset operations communities. Historical data stored in a big data system can be coupled with analytics to build a mathematical model, which can then be used to predict upcoming failures based on day-to-day usage data. Predicting the remaining useful life of components can improve readiness and enable optimized maintenance activities.

However, the adoption of such predictive analytics can be jeopardized by too many false positives. So, the model must be trained and made robust by incorporating as much historical data as possible. This requires to run high performance data fusion, cleaning, transformation and engineering analytics, as workflows on datasets.

nCodeDS is a highly scalable data processing tool that applies dedicated engineering analyses at high speed to huge quantities of time series data, whether evenly sampled or not. These analyses include handling NaN (Not a Number) and outliers, producing histograms, statistics, fatigue analysis, etc.

This webinar will present a few use cases to illustrate how one can efficiently use engineering algorithms to reduce sensor data for machine learning applications, and how a model can then be deployed as a digital twin to make predictions.

Date: 2023 December 13th

Instructors: Fred Kihm

Duration: 30 minutes

Language: English

No more result to load

Modern Slavery Statement Imprint Cookie Policy Privacy Policy Sitemap Ethical Business