Designed for industrial machines and processes

Turbomechanica® is Mechademy’s big data platform specifically designed for industrial machines and processes. It integrates state-of-the-art deep learning and machine learning algorithms with proprietary physics-based performance models that empower plant personnel with domain rich predictive and prescriptive alerts to maximize equipment life and uptime.​

PHYSICS ASSET LIBRARY

Proprietary physics-based performance models​

These models compare the performance of equipment to the baseline performance established during testing or field operation. By continuously monitoring and comparing the actual performance to the expected performance, the platform provides early detection of impending faults with a high degree of accuracy.

Extensive asset library

OEM agnostic performance models

Drag and drop functionality allows fast, customizable equipment setup

Real gas equations of state for performance accuracy

State-of-the-art diagnostics engine provides prescriptive insights

MACHINE LEARNING

Domain informed machine learning & deep learning​

The platform utilizes deep learning and machine learning models that are trained on historical plant data or simulated data. These models analyze vast amounts of data to identify patterns, correlations, and potential failure modes.

Domain-informed use cases and feature engineering

Model building at scale using AutoEDA and AutoML

Model drift and data drift detection

Hybrid models using simulation-based transfer learning

Integrations with popular DL (PyTorch, TensorFlow, and Keras) and ML (scikit-learn, XGBoost, PySpark and more) frameworks

End to end lifecycle management of models (MLOps)

ORCHESTRATION & DIAGNOSTICS​

Sophisticated orchestration to enable state-of-the-art diagnostics​

The Turbomechanica platform uses a unique orchestration strategy that allows the seamless flow of data between physics and machine learning models. This allows the use of physics/ML generated synthetic sensors within models.

Significantly expanded scope of early fault detection

Richer insights into fault causality

Hybrid physics + ML digital twins

Use of physics generated dimensionless parameters allows better transferability of ML models

Prescriptive alerts

Predict machine failure, receive actionable insights and prescriptive alerts to minimize downtime

Info

Alert description and prescriptive insights

Plots

Automatically generated using diagnostic sensor groups

Timeline

Timeline of similar historical events

Service Requests

CMMS integration with service requests and work orders

Discussion Thread

Collaborate by sharing data, media files, and documents

Contact

Book a Demo

Our friendly team would love to hear from you.

+1-281-936-9525
675 Bering Drive, Suite 200 | Houston, TX 77057 | USA