Predictive maintenance is one of our cross-departmental core competencies on various levels. In particular, we work with the support of Artificial Intelligence (AI). Keywords here are Machine Learning (ML), neural networks and the Internet of Things (IoT) as a communication and data network.

Condition Monitoring and Predictive Maintenance

Data Acquisition, Analysis, Forecasting and Evaluation From a Single Source

Industry 4.0 and digitization offer great potential for operating production facilities efficiently and cost-effectively, for example, through condition monitoring and predictive maintenance. With customized solutions, we help companies to translate theory into industrial practice and save costs.


Predictive Maintenance Is Based Primarily on:

  • Avoiding downtime and keeping downtime to a minimum
  • Quickly repair or replace defective and faulty components
  • Plan maintenance at optimal costs and reduce energy costs
  • To reduce production costs with optimal quality
Predictive Maintenance
© Fraunhofer ITWM
Predictive Maintenance

Work Steps for Process Optimization and to Exploit the Potential of Predictive Maintenance

Various work steps with their production-specific requirements must be taken into account:

  1. Smart Sensor Data and Digital Twin
    The sensor data are collected, locally pre-processed and transferred to a computing unit for multivariate data analysis. For this purpose, the number of sensors has to be kept low on the one hand and on the other hand a high information content with best possible measurement quality has to be generated to create a digital twin of the system. A design-of-experiment helps to obtain the maximum information content with a minimum number of experiments.

  2. Condition Monitoring
    The quality analysis or monitoring of the product quality already allows first conclusions to be drawn about changes in the manufacturing process. The condition monitoring of plants allows the identification of critical events and conditions with high wear potential. Events and faults are classified and evaluated. Critical events can be eliminated immediately by a quick reaction in order to avert cost-intensive consequential damage. In addition, appropriate diagnoses of the causes are made.

  3. Anticipate Risks With Predictive Maintenance
    Predictive Maintenance predicts risks of undesired operating states and events based on the empirical values gained from condition monitoring. These forecasts enable demand-oriented planning of service and maintenance actions to optimize plant effectiveness.

  4. Predictive Maintenance Saves Costs and Time
    Predicitve Maintenance reduces downtime and saves costs, since service technicians, spare parts and logistics are provided in a targeted manner through appropriate diagnoses. Predictive maintenance helps to plan the availability of the systems and provides information at an early stage for targeted maintenance actions while taking manufacturing, service and sales prices into account.

Our Competences and Services

We support you in the development of modules to optimize the effectiveness of your systems and quality analysis of your products from a single source, step by step throughout the entire product lifecycle. Or we can supply you with the appropriate modules. Our services are modular in structure:
 

Module Data Acquisition, Data Transfer and Quality Analysis

  • We analyze existing and determine required information/measurement data for your applications, e.g. for quality analysis or creation of a digital twin. In doing so, we use existing sensor technology or, if required, we are happy to offer the special systems developed by us:
    • Innovative image-based complete solutions for automated quality analysis (e.g. surface inspection)
    • Terahertz technology for coating thickness measurements
    • Torsion detection by means of contactless torque sensors
  • We support you in data transfer from the sensor to the analysis unit.
     

Condition Monitoring Module

We identify, develop and integrate customized machine learning or deep learning algorithms for your data and information system for condition analysis and diagnosis. Examples are our online monitoring systems for torsional vibrations of rotating plants.

Predictive Maintenance Module

We prepare forecasts for the demand-oriented planning of service and maintenance actions, also based on information from plant parks. Together with you, we design solution-oriented predictive maintenance algorithms for your individual plant.

Module Cost Models

We create and consider market and cost models for the maintenance of your machines and plants for the total cost evaluation of the maintenance measure. In particular, we consider energy prices, which we calculate with innovative models based on financial mathematical methods. Special attention is paid to flexibilities that allow you to profit optimally from fluctuating energy prices.

Our solutions can be integrated into common IOT platforms. We also support you in the development of your individual solution.

Further Information on the Individual Modules

 

Contactless Torque Monitoring and Predictive Maintenance

We analyze information/measurement data for your applications e.g. for quality analysis or creation of a digital twin. In doing so, we use existing sensor technology or, if required, we are happy to offer the special systems developed by us, such as torsion detection using contactless torque sensors.

 

Market and Cost Models

We create and consider market and cost models for the maintenance of your machines and plants for the total cost evaluation of the maintenance measure. In doing so, we take energy prices, among other things, into account.

 

Virtual Image Processing

We also offer innovative image-based complete solutions for automated quality analysis (e.g. surface inspection) with virtual image processing.

Further Information From the Practice

Keynote on YouTube

The keynote »Implementation of a Condition Monitoring and Predictive Maintenance System« by Dr. Benjamin Andrian at the innovation forum PredictiveMaintenance@KMU is available for viewing on YouTube.

Dr. Benjamin Adrian presents the topic step by step and illustrates it with an example. He explains which questions can be answered with Predictive Maintenance.

Predictive Maintenance in the Context of Gas Turbines

The presentation by Dr. Benjamin Adrian and Dr. Andreas Wirsen to the ASUE Expert Group Gas Turbine Technology on »Condition-oriented Maintenance of Power Generation Plants - Condition Monitoring and Predictive Maintenance« from 30.09.2020 is now available for reading.