Self-organising systems with online monitoring and diagnostics
The evolution of manufacturing engineering systems takes place concurrently in many directions. Simultaneous progress can be observed in emergence of new technologies, as well as in the way the technologies are applied for manufacturing.
The economical requirements – e.g. customized products, high quality of products, small series of rapidly changing products, dynamic networks of subcontracting enterprises and suppliers – and the requirements presented by emerging manufacturing technologies to the production system – e.g. flexible readjustment, sensitivity to maintenance quality, partial rearrangements between processing of different products, high demands on the materials, processing instruments and environment – pose a necessity to develop and maintain persistent ability to monitor the status of the production system and rapid maintenance or replacing components, business partners, technologies in production system. Conventionally this ability has been provided by human managers and maintenance personnel.
This strategic development area (SDA3) – self-organising systems with on-line monitoring and diagnostics – studies the possibility to substitute humans in the status monitoring and maintenance of production systems in order to reach shorter lead time between different product series, faster detection of tools’ wear, detecting problems in manufacturing planning and initiating plan correction, etc. For building automated monitoring and maintenance systems the properties of softwareintensive components of the production system are slightly extended, combined with specially added motes of smart dust and integrated into ad-hoc networks for suggesting decisions and, if possible implementing those decisions.
Automated monitoring and maintenance systems for technical devices and technological processes have been studied for long time already. Progress has achieved with respect to status diagnosis and operation mode optimization. The sensing and measuring systems used have been hard-wired, and in many cases are not applicable on-line. Therefore IMECC focuses on the use of wireless and software-intensive monitoring technology that has gained international attention only ten years ago. Similar approach is applicable for monitoring status of PLM, manufacturing planning, and behavior in integrated enterprises (clusters), provided that super-Turing models of computation reach the sufficient maturity.
Objectives of SDA 3:
- To study and develop innovative methods and technologies applicable for monitoring and controlling the natural and artificial environments – e.g. smart dust, ad hoc sensor and actuator networks, agent-based technologies
- To extend and enhance the technological basis and processing methods used for monitoring and diagnostic of production systems to enable their on-line application.
- To develop methods for inserting self-organisation features into manufacturing technologies in concordance to the actual status and need of the manufacturing system
- To keep the price of innovative methods on the level that is feasible for SME-s
Description of methods:
Situation-aware models of interactive computing and rules of operation for ad-hoc networks are fine-tuned for this application.
The suitable sets of sensors are selected to match user and application requirements for on-line monitoring of manufacturing equipment – e.g. the profiles of devices, key parameters to be monitored, situations to be detected, balance between local and remote processing, data fusion and distribution requirements, required communication capabilities, etc. Intrinsic consistency of requirements is checked – e.g. situations to be detected are to cover all the involved levels of intelligent factory, sensory and fused information required to detect situations should meet the non-functional requirements, e.g. specified spatial and timing restrictions. The monitoring system has agent-based architecture.
The feasibility of sensors selected for the monitoring system is to be checked by physical industrial experiments. The same applies to the algorithms used for data fusion (i.e. for synthetic data). The available smart dust motes may need adaptation to pass the feasibility checks.