By clicking “Accept All Cookies”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts. View our Privacy Policy for more information.

Machine Learning

Use the Power of Machine Learning

Incorporating machine learning (ML) techniques into real-time digital twins takes their power and simplicity to the next level. It can be challenging to write analytics code that sifts out subtle issues emerging over time in a stream of telemetry. In many cases, the underlying processes which lead to device failures are not well understood or are complex to describe. However, application developers can train ML algorithms on thousands of known patterns to recognize abnormal telemetry. ML dramatically increases the power of digital twins to find issues and signal alerts.

For example, digital twins can use ML to track engine and cargo parameters, such as oil pressure and cargo temperature, for every vehicle in a large trucking fleet. To help spot issues, each digital twin can maintain unique training data associated with a specific type of truck.

A robot looking at mathematical equations representing machine learning

Flexible Alerting

When ML algorithms detect issues, digital twins can send alerts to many public alerting providers, including Splunk, Slack, and PagerDuty.

Many ML Algorithms

Developers can select from a wide variety of binary classification algorithms in ML.NET and evaluate them with training data before deployment.

Optional Business Rules

Digital twins can optionally make use of straightforward business rules to extend their functionality when incorporating ML algorithms.

A diagram showing machine learning for anomaly detection

Leverage Microsoft's ML.NET

ScaleOut’s digital twins incorporate Microsoft’s popular machine learning library called ML.NET to run ML algorithms. Using the ScaleOut Model Development Tool™, developers can select, train, evaluate, deploy, and test ML algorithms within their real-time digital twin models. Once deployed, the ML algorithm runs independently for each data source, examining incoming telemetry within milliseconds after it arrives and logging abnormal events.

ScaleOut’s ML algorithms can perform spike and trend detection for a single parameter or anomaly detection that looks for more complex patterns across multiple parameters. For example, after training with data acquired from live readings, anomaly detection can look at several parameters from a motor or pump to detect abnormal behaviors that might otherwise go unnoticed.

More Features