Search

Conventional industrial communication systems suffer from rigidness, inflexibility and lack of scalability. The environment is heterogeneous as the systems exchange data with a variety communication protocols, some of which are proprietary. This makes it laborious and expensive to reconfigure or upgrade the systems. As the solution, this article proposes a message-bus-based communication architecture to enable information exchange between systems regardless of their geographical location and position within the functional hierarchy of the plant. The architecture not only enables communication to cross the conventional physical borders but also provides scalability to growing data volumes and network sizes. As proofs of concept, the article presents a prototype in three environments: a copper smelter, a steel plant and a distillation column. The results suggest that the message-bus-based approach has potential to renew industrial communications, a core part of the fourth industrial revolution. publishedVersion

In mobile machines, as well as in manufacturing, the overall productivity is essential for business competitiveness. As the operation of a modern mobile machine is affected by various parameters, they need to be tuned to reach an optimal performance – however, due to machine complexity, parameter optimisation is difficult for a typical operator. To enable parameter optimisation locally in machines, this article presents a system architecture to generate information and knowledge from machine fleet data and to utilise them in machine operations in the field. Measurement data is collected and analysed to discover the associations between machine performance and parameter values. While some results are plain statistical distributions, any resulting more sophisticated domain knowledge is stored as rules. Rule-based reasoning enables a zone of interoperation between the information system and domain experts. Once information and knowledge have been generated, they are made available to machines that run the actual parameter assessment application. Results made with forestry data indicate that the system has a considerable potential to improve machine productivity. acceptedVersion

The modern mobile machinery has advanced on-board computer systems. They may execute various types of applications observing machine operation based on sensor data (such as feedback generators for more efficient operation). Measurement data utilisation requires preprocessing before use (e.g. outlier detection or dataset categorisation). As more and more data is collected from machine operation, better data preprocessing knowledge may be generated with data analyses. To enable the repeated deployment of that knowledge to machines in operation, information management must be considered; this is particularly challenging in geographically distributed fleets. This study considers both data refinement management and the refinement workflow required for data utilisation. The role of machine learning in data refinement knowledge generation is also considered. A functional cloud-managed data refinement component prototype has been implemented, and an experiment has been made with forestry data. The results indicate that the concept has considerable business potential. acceptedVersion

Performance improvement is a constantly important topic in manufacturing, and mobile work machines are no exception. In some mobile machines, machine behaviour is affected by various device parameters. Their effect can be examined by statistical analysis, but exploiting analysis results during machine operation requires a sophisticated distributed information system. This paper introduces an architecture for such a system suitable for geographically dispersed machine fleets. It covers aspects on data collection, storage, analysis and utilization. Once statistical analysis has been performed, its results are made available for applications that assess machine performance locally during operation. If there is room for improvement in parameter values, the operator of the machine is given suggestions to change them. A prototype implementation is presented. The results show that such an information system has a considerable potential of bringing competitive advantage to machine operators. Besides, added value is also expected to the manufacturer of the machine as performance-related knowledge is augmented. Thus, further service business development is also contributed.

The performance of mobile machine operators has a great impact on productivity that can be translated to, for example, wasted time or environmental concerns such as fuel consumption. In this paper, solutions for improving the assessment of mobile machine are studied. Usage data is gathered from machines and utilized to provide feedback for operators. The feedback is generated with rules that define in what way different measures indicate performance. The study contributes to developing an architecture to manage both data collection and inference rules. A prototype is created: rule knowledge is managed with decision tables from which machine-readable rules are generated. The rules are then distributed to application instances executed in various locations. The results of the prototype promote several benefits. Rules can be maintained independent of the actual assessment application, and they can also be distributed from a centrally managed source. In addition, no IT expertise is required for rule maintenance so the rule administrator can be a pure domain expert. The results bring the architecture towards a scalable cloud service that combines the benefits of both centralized knowledge and distributed data management.

In modern mobile machines, a lot of measurement data is available to generate information about machine performance. Exploiting it locally in machines would enable optimising their operation and, thus, yield competitive advantage and reduce environmental load due to reduced emissions. However, optimisation requires extensive knowledge about machine performance and characteristics in various conditions. As physical machines may be located geographically far from each other, the management of ever evolving knowledge is challenging. This study introduces a software concept to enable centralised management of data refinement performed locally in the machines of a geographically distributed fleet. It facilitates data utilisation in end user applications that provide useful information for operators in the field. Whatever the further data analysis requirements are, multiple preprocessing tasks are performed: it enables outlier limit configuration, the calculation of derived variables, data set categorisation and context recognition. A functional prototype has been implemented for the refinement of real operational data collected from forestry machines. The results show that the concept has considerable potential to bring added value for enterprises due to improved possibilities in managing data utilisation.