Predictive Maintenance



In the area of Predictive Maintenance we develop an application called SysWatch. It is a Windows program that provides offline and online predictive insights. By offline analysis, we mean, on the one hand, the ability to visualize data, but even more important is the training mode, in which SysWatch reads a large, usually high-dimensional database with sensor time series from a technical or industrial source. That is usually a physical process, along with a database of historical events. The main goal in this mode is to determine correlations between the sensor values and the events in question. In online mode, SysWatch uses calculated and verified models for incoming data and calculates failure probabilies.

Application structure

The Explorer module is a professional visualization application for operational data. In addition, this makes it easier to feed technical data into the training step. A user-friendly PHP interface allows for hierarchical structuring of all parameters. The models module puts mathematical models into action. Once created models are stored for analysis purposes. In addition to the mathematical calculation, a structured visualization and representation of KPIs is the main functionality of this module. The dashboard visualizes the results of a Windows-based service on the generated and active models. New records are processed in the background and stored in an MS SQL database. The figure below shows the flow of data and processing steps up to visualizing the results. The SysWatch framework has four well-separate modules than executable files. From the user's point of view, there is mainly a separation into the SysWatch application, which includes all relevant data handling, visualization and services, as well as the application dashboard, a customized visualization implementation of the relevant Forecast models, managed.

Core competency

The core competencies of the framework lie in its algorithmic engine, including

• multivariate abnormality technology

• Multivariate regressions (Auto-associative kernel regressions, neurol networks and classical statistics)

• Spatial and temporal pattern recognition

• Empirical stochastic models

• Generating surfaces with high Dimensional Response surfaces

FCE focuses not only on the development of mathematical models on the technical feasibility of their mathematical ideas. The following points lead to a feasible implementation:

Flexibility in Data Source

Fault memory Analysis, Statistical Sensor Insights

Integration of System Know-How

Hierachical Structuring, Coverage of all dependencies

Application of predictive algorithms

State of the art algorithms, embedding of system-specific features


Automation of Data-Streams, Real time Monitoring, individual Dashboards conceivable


SysWatch has its origins in preventive maintenance and is primarily surveillance software. Over time, however, new areas emerged, such as response surface techniques related to adaptive process control, as well as multivariate time series forecasts for the energy industry. In 2018, SysWatch is living in a competitive environment that needs to bring many areas of expertise together. Here are some priorities that guarantee our competitiveness:

Development of supervised learning techniques

Development of high-performant and precise technique in the area of supervised learning

Microsoft Azure compatibility

SysWatch is a Microsoft-Azure compatible Software. Usage of application programming interfaces in order to integrate into existing industrial Solutions

Cooperations with Universities

Several cooperations with universities in the area of applied mathematics improves our competence in the predictive analytics field

Participation at international conferences

To ensure a high quality of our results, FCE regularly undergoes the peer review that emerges from the publication process in Journals and conferences