Artificial intelligence-based predictive maintenance for various smart industry applications

Systems in the area of Industry 4.0, highly automated Cyber-Physical systems, or the Internet of Things (IoT) are significantly influenced by the use of artificial intelligence and the requirements of safety and security. The Austrian research centre AIT Austrian Institute of Technology offers a complete package for the effective use of predictive maintenance for smart industry applications: Predictive analytics for data-driven maintenance process planning on high-tech production machines , artificial intelligence based analysis of encrypted data streams in large communication networks (big data) as well as artificial intelligence based safety and security for Industry 4.0, IoT and highly automated Cyber-Physical Systems. By applying this technology, work can be brought forward or additional resources can be requested. An additional cost saving occurs if maintenance work is avoided when the machine is in good condition.

In the context of predictive maintenance, the creation of a forecast model requires a comprehensive analysis and preparation of the existing data. In this step, information from different data sources is unified, visualised and statistically examined in order to assess the data quality. The customer then receives an analysis of the data quality as well as methods for the systematic analysis of his data. Based on this, machine-learning models are created and evaluated. In particular, 'deep learning' approaches, which in many cases achieve excellent results, and classical models, which allow interpretation, have to be emphasised.

Interpretability is desirable in many applications, as decisions are thus comprehensible for the human user and hence confidence in the model is strengthened. The evaluation of the models takes place together with domain experts on the basis of real data of the customer and in reference to specifically created benchmarks. In order to be able to estimate the effort of an implementation for ongoing operations, a deployment concept is created that takes into account the requirements of the forecasting model and the local infrastructure of the customer.

AIT’s Center for Digital Safety & Security has unique know-how and experience in the development and operation of safe and secure systems. The centre's specialists have not only been actively involved in standardisation for many years, thus helping to determine standards to be adhered to in the future, but are also developing practical solutions. In addition, they are supporting companies with the introduction of standard-compliant processes and the implementation of engineering projects.