TOLKKU - From measurements data to information; data mining from sensor and control data

Focus

From various machines different variables relating to their state, operational environment, production process etc. are continuously measured. Very large amounts of measurement data are thus generated and they should be processed into information that serves efficiently the parties involved in the product process of the system (e.g. product development, operation, maintenance).

The main research questions are

- how to manage the huge amount of data which are generated in the control and sensor systems of the mobile work machines and

- how to extract the essential information without increasing the number of sensors.

The project aims at developing methods and tools that enable processing the measurement data masses into information on the machines’ operation, state, loads, causal connections etc.

The improved knowledge on the operation and behaviour of the product/facility in the different circumstances throughout its entire life-cycle enables advanced optimization and risk management and supports e.g. implementation of the maintenance service.

Main research targets

Project creates tools (software and methods) for data mining from sensor and control systems, which

- are generic and product independent

- extract the essential information out of the data

- automate the measurement data processing, which enables the utilization of the obtained information more extensively

- fuse data coming from multiple, heterogeneous, independent sources

- taking into account the uncertainty in the data

- produce information of the machine status and use, which is easy to understand and use

- give information about causal connections and answers to why and what-if questions for the decision support

Another target is to study feasibility of different data mining methods and data sources in various applications.