Prescriptive maintenance for machinery and industrial processes


Event Generation

Explainable AI

First Principle Models



Event Management

Mobile App

Our product features

Anomaly Detection using First Principle Models

Detection of changes without witnessing faults

The general monitoring strategy is to detect changes of the equipment relative to a state that is defined healthy. With this strategy it is possible to monitor equipment without witnessing faults.

Accessible useful data

In many cases the accessible, useful data does not have enough faults to use classic predictive maintenance strategies that rely on a degradation model.

Thermodynamic Process Models

Industrial Analytics Thermodynamic Process Models are designed to be fitted without geometric data from the equipment, instead it uses measured data of the process. A key necessity are the gas properties that are taken from an efficient implementation of the NIST REFPROP model. They provide additional insights of the monitored equipment and feature reasonable extrapolation into regions where the model is not trained.

IoT enabled Vibration Monitoring

Vibration signatures as sensors for machinery health

Vibration signatures are the most sensitive indicator for machinery health. Any defects are typically first indicated by a change of the vibration spectrum.

Cost-effective edge device

Industrial Analytics has engineered a cost effective edge device which is built from industrial approved components. It is possible to use accelerometers or typical displacement sensors for shafts vibration. Other additional sensors can be added to the highly modular system.

Signal Analysis software

The signal analysis software is developed by Industrial Analytics. It is possible to extract features relative to the rotational speed of the machine and a broad representation of the spectrum. 

Transmitting the data

The data rates are reduced using a highly effective data compression algorithm. The data is transmitted via secure standard protocols like OPC UA or can be fed directly into a PI System by OSIsoft Message Format. It is possible to use wired or wi-fi connections.


AI-assisted Event Management

Event management

The event management assists with information aggregation, classification and learns to support operators. With the UI user annotations are gathered in a way that enables training of AI models. The trained AI is able to give advice to the operator on what to do next in the actual event by classification of the events.

Usage of a graph data model

This is possible by using a graph data model that represents the functional relationships of the plant and can distinguish equipment types semantically. With the models and the event management it is possible to retain knowledge of machinery and process.

Explainable AI & Optimization

Maximizing evidence results while minimizing uncertainties

Explainable AI models give not only the best answer, but obtain statistical information on the accuracy. The anomaly detection uses ensemble learning with training of multiple models. The results are a better estimation of expected value. This expected value is compared with the actual measured value. 

Maximizing extrapolation capabilities while minimizing uncertainties

Industrial Analytics uses a reliable way to estimate the uncertainty of the model and the data. Maximizing Bayesian model evidence results in maximizing extrapolation capabilities while minimizing uncertainties. Generation of indicative events is based on hypothesis testing, which uses a quantifying effect size and probability of the errors and an automatic adjustment of the evaluation period.

Collaboration with the Hasso-Plattner- Institute

Optimization is performed by Evolutionary Algorithms with whom it is possible to tackle very complex multi objective problems with many variables. These types of algorithms are known for effectively finding a global optimum. Industrial Analytics is collaborating with Prof. Friedrich of the Hasso Plattner Institute, who specializes in these algorithm types.


How the product is integrated in the existing environment

Step 1

Analysis of sensors and P&ID

We analyse the provided process and instrument diagrams (P&ID) and equipment data sheets by the client. Additional instrumentation is suggested depending on the specific use case. For a maintenance use case often additional vibration measurements are needed that can be integrated using an edge device for vibration monitoring.

Step 2

Data modelling & AI training

We are building a graph type data model of your asset. This is a very flexible method to mapping the sensor tags on a data model that represents the functional relationships and controls the preprocessing, visualisation and model training. The initial training and consistency checks of the first principle model are performed typically on historic data of the asset.

Step 3


We support and perform the integration of our solution into your existing IT infrastructure. We are experts on edge computing solutions that are integrated in the plant environment (OSIsoft PI, ABB, Siemens or other). The deployment is performed by using the platform solution Docker and needs a virtual machine that is typically supplied by the operator.


Book a demo with our experts

Within a 30 min appointment, we will give you a quick intro depending on your area of interest and covering your questions. We will demonstrate our solution and will develop a solution for your business.

Get a demo